THE EVOLUTION OF SHELTER: ECOLOGY AND ETHOLOGY OF CHIMPANZEE NEST BUILDING. A dissertation submitted to the University of Cambridge in partial fulfilment of the conditions of application for the Degree of Doctor of Philosophy By Fiona Anne Stewart July 2011 Department of Biological Anthropology University of Cambridge Corpus Christi College Cambridge ii To my friends, family, and Alex, for all their support © Fiona A. Stewart, July 2011 iii Contents Preface ............................................................................................................................ xviii Summary ........................................................................................................................... xix Acknowledgements ........................................................................................................... xx Abbreviations and notations used .................................................................................... xxv Chapter 1 Introduction ......................................................................................................... 1 Thesis ......................................................................................................................................................... 1 Significance ............................................................................................................................................... 1 Animal building ........................................................................................................................................ 4 Great ape nest building ........................................................................................................................... 6 Thesis outline ............................................................................................................................................ 8 Chapter 2 General methods ................................................................................................ 11 INTRODUCTION ................................................................................................................................... 12 STUDY SITE ............................................................................................................................................. 12 Fongoli ..................................................................................................................................................... 12 History and geography .................................................................................................................................... 12 Vegetation ......................................................................................................................................................... 13 Fauna ................................................................................................................................................................. 14 Climate .............................................................................................................................................................. 15 Subjects ............................................................................................................................................................. 18 Issa ............................................................................................................................................................ 19 History and geography .................................................................................................................................... 19 Vegetation ......................................................................................................................................................... 21 Fauna ................................................................................................................................................................. 22 Climate .............................................................................................................................................................. 23 Subjects ............................................................................................................................................................. 29 DATA COLLECTION ............................................................................................................................ 30 Nest data ................................................................................................................................................. 30 Nest and nest tree characteristics .................................................................................................................. 31 Nest characteristics .......................................................................................................................................... 31 Nesting tree characteristics ............................................................................................................................... 33 Nest shape and architecture ........................................................................................................................... 37 iv Nest shape ...................................................................................................................................................... 37 Nest architecture .............................................................................................................................................. 40 Nest data summary .......................................................................................................................................... 42 Vegetation plots ..................................................................................................................................... 44 Phenology ................................................................................................................................................ 45 Behaviour ................................................................................................................................................ 45 Group night nesting ........................................................................................................................................ 46 Individual nest construction .......................................................................................................................... 46 Nocturnal observations .................................................................................................................................. 47 Genetic sampling ................................................................................................................................... 47 Insect trapping experiment ................................................................................................................... 49 ANALYSES ................................................................................................................................................ 49 Data included in analyses ...................................................................................................................... 49 Statistical analyses .................................................................................................................................. 49 Chapter 3 Characteristics of nests and nesting trees, nest shape and architecture, in two chimpanzee populations, compared .................................................................................. 51 INTRODUCTION ................................................................................................................................... 52 METHODS ................................................................................................................................................. 54 RESULTS .................................................................................................................................................... 54 Nest groups ............................................................................................................................................. 54 Nest characteristics ................................................................................................................................ 54 Nest type ........................................................................................................................................................... 54 Support type ..................................................................................................................................................... 56 Nest integration ............................................................................................................................................... 56 Number of trees used for building ............................................................................................................... 56 Nest cover ......................................................................................................................................................... 58 Nest position .................................................................................................................................................... 58 Nesting tree characteristics ................................................................................................................... 60 Nesting species ................................................................................................................................................. 60 Phenological stage ........................................................................................................................................... 63 Tree size and morphology .............................................................................................................................. 63 Nest shape ............................................................................................................................................... 64 Nest Architecture ................................................................................................................................... 66 DISCUSSION ............................................................................................................................................ 70 Overall differences and similarities ..................................................................................................... 70 v Nest characteristics .......................................................................................................................................... 70 Nest type ......................................................................................................................................................... 70 Support type .................................................................................................................................................... 71 Nest integration and number of trees used ........................................................................................................ 72 Cover .............................................................................................................................................................. 72 Nest position ................................................................................................................................................... 73 Nesting tree characteristics ............................................................................................................................ 73 Nest shape ........................................................................................................................................................ 75 Nest architecture .............................................................................................................................................. 76 Summary ........................................................................................................................................................... 78 Seasonal differences in Issa .................................................................................................................. 79 Nest characteristics .......................................................................................................................................... 79 Nesting trees ..................................................................................................................................................... 80 Nest shape and architecture ........................................................................................................................... 80 Summary ........................................................................................................................................................... 81 Conclusion .............................................................................................................................................. 81 Chapter 4 Why sleep in a nest? Empirical testing of the function of simple shelters made by wild chimpanzees .......................................................................................................... 82 INTRODUCTION ................................................................................................................................... 83 Sleep quality ............................................................................................................................................ 83 Anti-predation ........................................................................................................................................ 84 Anti-pathogen ......................................................................................................................................... 84 Thermoregulation .................................................................................................................................. 85 METHODS ................................................................................................................................................. 85 Study site ................................................................................................................................................. 85 Data collection ....................................................................................................................................... 85 Sleep quality ...................................................................................................................................................... 86 Anti-pathogen .................................................................................................................................................. 86 Thermoregulation ............................................................................................................................................ 86 RESULTS .................................................................................................................................................... 86 Sleep quality ............................................................................................................................................ 86 Anti-pathogen ......................................................................................................................................... 88 Thermoregulation .................................................................................................................................. 88 DISCUSSION ............................................................................................................................................ 89 Sleep quality ............................................................................................................................................ 89 vi Anti-pathogen ......................................................................................................................................... 90 Thermoregulation .................................................................................................................................. 90 Conclusions ............................................................................................................................................. 91 Chapter 5 Do chimpanzee nests have an anti-predator function? ..................................... 92 INTRODUCTION ................................................................................................................................... 93 Predator avoidance in other primates ................................................................................................. 93 Are chimpanzees and other large apes at risk of predation? ........................................................... 94 Shelter construction and predator defence ........................................................................................ 96 METHODS ................................................................................................................................................. 98 Predator presence .................................................................................................................................. 98 Nest data ................................................................................................................................................. 99 Analyses ................................................................................................................................................ 101 RESULTS ................................................................................................................................................. 102 DISCUSSION ......................................................................................................................................... 106 Chapter 6 Thermoregulatory nest function: variation in nest characteristics, shape, and architecture in response to weather ................................................................................. 110 INTRODUCTION ................................................................................................................................ 111 METHODS .............................................................................................................................................. 113 Study sites and climate ....................................................................................................................... 113 Data collected ...................................................................................................................................... 114 Nest characteristics ........................................................................................................................................ 114 Nest shape and architecture ......................................................................................................................... 115 Analyses ................................................................................................................................................ 117 RESULTS ................................................................................................................................................. 117 Weather conditions ............................................................................................................................. 117 Variation in nest position and type with weather conditions ....................................................... 120 Fongoli ............................................................................................................................................................ 120 Issa ................................................................................................................................................................... 121 Principal components of shape and architecture ........................................................................... 122 Cross-site comparison of principal components of nest shape and architecture ................................ 125 Variation in shape and architecture with weather conditions ...................................................... 125 Fongoli ............................................................................................................................................................ 125 Issa ................................................................................................................................................................... 129 DISCUSSION ......................................................................................................................................... 131 vii Chapter 7 Sex-bias and social influences on nest-building techniques and behaviour .. 136 INTRODUCTION ................................................................................................................................ 137 METHODS .............................................................................................................................................. 139 Nesting behaviour ............................................................................................................................... 139 Group behaviour ........................................................................................................................................... 139 Individual behaviour ..................................................................................................................................... 140 Nest and tree characteristics, shape and architecture .................................................................... 141 Age sex class comparison .................................................................................................................. 142 Inter-individual comparison .............................................................................................................. 143 RESULTS ................................................................................................................................................. 144 Behaviour ............................................................................................................................................. 144 Nest group ...................................................................................................................................................... 144 Group composition ......................................................................................................................................... 144 Time of construction ....................................................................................................................................... 144 Duration of construction ................................................................................................................................ 145 Initiator and final builder of group nest-building ............................................................................................. 146 Individual nest-building behaviour summary ............................................................................................ 146 Differences between age and sex classes ......................................................................................... 148 Nesting behaviour ......................................................................................................................................... 148 Nest and tree characteristics ........................................................................................................................ 148 Nest shape ...................................................................................................................................................... 150 Nest architecture ............................................................................................................................................ 150 Individual differences ......................................................................................................................... 153 Nest and tree characteristics ........................................................................................................................ 153 Nest shape and architecture ......................................................................................................................... 153 DISCUSSION ......................................................................................................................................... 156 Nesting behaviour of Fongoli chimpanzees ................................................................................... 156 Time and duration of group nest building ................................................................................................. 156 Initiator of nest building ............................................................................................................................... 158 Duration of construction .............................................................................................................................. 158 Activities .......................................................................................................................................................... 158 Sex differences ..................................................................................................................................... 159 Age differences .................................................................................................................................... 161 Individual differences ......................................................................................................................... 162 Conclusions .......................................................................................................................................... 163 viii Chapter 8 Living archaeology: Artefacts of specific nest-site fidelity in wild chimpanzees .......................................................................................................................................... 164 INTRODUCTION ................................................................................................................................ 165 METHODS .............................................................................................................................................. 168 Study site .............................................................................................................................................. 168 Data collection .................................................................................................................................... 168 RESULTS ................................................................................................................................................. 170 Scars ...................................................................................................................................................... 171 Branch re-growth ................................................................................................................................ 173 Nest decay ............................................................................................................................................ 174 Re-use ................................................................................................................................................... 175 DISCUSSION ......................................................................................................................................... 176 Chapter 9 Conclusion ....................................................................................................... 181 Introduction ......................................................................................................................................... 181 Proximate functions of nests ............................................................................................................ 181 Great ape nests and the evolution of shelter .................................................................................. 185 A culture of nest-building .................................................................................................................. 187 Niche construction ............................................................................................................................. 189 Implications for hominin evolution ................................................................................................. 190 References ........................................................................................................................ 192 ix List of Tables Table 2-1. Description of vegetation types in Fongoli [from Baldwin et al. 1982; Pruetz et al. 2008; Pruetz et al. 2002], and terms used for comparison across sites in this study. ......................................................................................................................................... 13 Table 2-2. Sex and age class composition of the Fongoli community during the study period. ....................................................................................................................................... 19 Table 2-3. Vegetation types in Issa [from Hernandez-Aguilar 2006; 2009], and terms used for comparison across sites in this study. ............................................................................ 22 Table 2-4. Large- and medium-sized mammal species recorded in Issa during the study period [edited from Hernandez-Aguilar 2006; 2009] (* observed once, - no evidence, bolded newly observed). ....................................................................................... 24 Table 2-5. Nests found and measured at Fongoli and Issa. Night nests had fresh faeces or urine beneath or were seen built and slept in. Abandoned nests were seen built but left the night before, or discovered to have been abandoned during the night (e.g. if the group moved or the builder was seen to arise from another nest). Day nests were seen built for temporary use during the day. Unknown nests had no faeces or urine beneath, but due to un-withered leaves were likely either fresh or abandoned. ..................................................................................................... 43 Table 2-6. Nest status and number of nests accessed for shape and architectural measurements. ......................................................................................................................... 44 Table 2-7. Measurements taken of nests included in analyses. ............................................................ 44 Table 2-8. Summary of genetic data already generated (HV1, HV2, and Amelogenin sexing tested) from reference faecal samples collected from 26 weaned individuals in Fongoli. In the future, in order to identify unknown samples collected from nests, the following hierarchical analyses, from step 1-4, will be conducted in order to rapidly identify nest-builders. Letters indicate hyper-variable region one and two (HV1 & HV2) haplotypes and sexes (M: male; F: female). Grey cells indicate analyses to be conducted, whilst unfilled cells indicate analyses unnecessary to identify to individual level. ......................................................................... 48 Table 3-1. Nest position compared between Fongoli and Issa dry season, and between seasons in Issa. ......................................................................................................................... 59 Table 3-2. Nesting tree size compared between Fongoli and Issa dry season, and between seasons in Issa. ......................................................................................................................... 64 x Table 3-3. Nesting tree morphology compared between Fongoli and Issa dry season, and between seasons in Issa. ......................................................................................................... 64 Table 3-4. Nest shape, including all nests measured, in Fongoli and Issa. ........................................ 65 Table 3-5. Nest shape of tree nests built in Fongoli in dry and in Issa dry and wet season. .......... 66 Table 3-6. Variation in nest architecture between all nests deconstructed in Fongoli and Issa. ............................................................................................................................................ 67 Table 3-7. Nest architecture between tree nests built in Fongoli and Issa dry and wet season. ....................................................................................................................................... 68 Table 3-8. Nest support between tree nests built in Fongoli and Issa dry and wet season. ........... 68 Table 3-9. Nest architecture between tree nests built in Fongoli in dry and Issa dry and wet season. ....................................................................................................................................... 69 Table 3-10. Mean nest and tree heights across chimpanzee study sites. ............................................ 74 Table 3-11. Bonobo and chimpanzee tree nest architecture, compared (P. paniscus nest measurements are taken from [Fruth 1995]). ..................................................................... 78 Table 4-1. External and internal sources of disturbance during sleep experiments under the two conditions of nest or ground sleep. .............................................................................. 87 Table 4-2. Comparison of median sleep amount (measured in total hrs of sleep and sleep quality) and sleep disturbance (measured by waking events and mean sleep bout length) between nest and ground conditions. ..................................................................... 87 Table 5-1. Evidence of carnivore species in Fongoli and Issa. ........................................................... 98 Table 6-1. Variables in the Principal Components Analysis of nest shape and architecture. ...... 116 Table 6-2. Mean overnight weather conditions compared at Fongoli and Issa across vegetation types. ................................................................................................................... 118 Table 6-3. Varimax rotated component matrix of nest shape and architecture. Loadings greater than 0.5 of each variable onto components of shape and architecture are highlighted in bold. Component labels and percentage of variance are highlighted in italics. ............................................................................................................ 124 Table 6-4. Fongoli, all nests: Model statistics of stepwise backward multiple regression of each principal component of nest architecture against possible predictors of mean overnight temperature, mean overnight relative humidity, and mean overnight gust speed. R2 value represents the amount of variation explained by the model (e.g. in model with R2 value of 0.09, the significant variables explain 9% of variance). Constant and significant variables of the best model are presented through b values, standard error of b and standardised β. The value xi and the sign (+/-) of Beta indicate the strength and direction of the proportional relationship between the variables. ............................................................ 127 Table 6-5. Fongoli, ground nests excluded: Model statistics of stepwise backward multiple regression of each principal component of nest architecture, in addition to further models including selected and constructed support diameter, against possible predictors of mean overnight temperature, mean overnight relative humidity, and mean overnight gust speed. R2 value represents the amount of variation explained by the model (e.g. in model with R2 value of 0.09, the significant variables explain 9% of variance). Constant and significant variables of the best model are presented through b values, standard error of b and standardised β. The value and the sign (+/-) of Beta indicate the strength and direction of the proportional relationship between the variables. ................................ 128 Table 6-6. Varimax rotated component matrix of mean overnight weather conditions in Issa. Loading scores >0.5 are highlighted in bold. Component labels and variance explained are highlighted in italics. .................................................................... 129 Table 6-7. Issa: Model statistics of stepwise backward multiple regression of each principal component of nest architecture, in addition to further models including selected and constructed support diameter, against possible predictors of principal components of weather conditions; wetness and temperature/wind. R2 value represents the amount of variation explained by the model (e.g. in model with R2 value of 0.09, the significant variables explain 9% of variance). Constant and significant variables of the best model are presented through b values, standard error of b and standardised β. The value and the sign (+/-) of Beta indicate the strength and direction of the proportional relationship between the variables. Stepwise backward multiple regression of each principal component of nest architecture against principal components of weather conditions; wetness and temperature/wind. ............................................................................................................... 130 Table 7-1. Frequency of nest building recorded and nests measured of known individuals of each sex and age class. .................................................................................................... 143 Table 7-2. Nest and nesting tree characteristics of each age/sex class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). ................................................ 149 Table 7-3. Nest shape for each sex and age class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). ......................................................................................... 150 xii Table 7-4. Summary of nest architecture and support of each sex and age class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). .................................... 151 Table 7-5. Detailed measures of nest architecture of each age and sex class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). .................................... 152 Table 7-6. Comparison of individual nest and nesting tree characteristics between males and females. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). ....................................................................................................................................... 154 Table 7-7. Comparison of nest shape and architecture between two adult males. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). .................................... 155 xiii List of Figures Figure 2-1. Woodland vegetation in Fongoli, with small strip of forest running from upper left to centre. ............................................................................................................................ 14 Figure 2-2. Monthly mean temperature and mean daily minimum and mean daily maximum temperature in three vegetation types of grassland, woodland, and forest, from October 2007 to March 2008, Fongoli. ......................................................... 16 Figure 2-3. Monthly mean relative humidity and mean daily minimum and mean daily maximum relative humidity in three vegetation types of grassland, woodland, and forest, from October 2007 to March 2008, Fongoli. ................................................. 17 Figure 2-4. Mean monthly mean and mean daily maximum wind speed in the vegetation types of grassland, woodland, and forest, from October 2007 to March 2008, Fongoli. ..................................................................................................................................... 18 Figure 2-5. Map of western Tanzania. Issa Valley study area is highlighted in yellow. The two national parks of Gombe and Mahale Mountain are outlined in green on the lake shore. The colours of the image also indicate vegetation cover: brown woodland, green forest, and orange suitable nesting habitat calculated from GIS modelling [Pintea and Plumptre 2006]. Rivers in blue indicate possible barriers to chimpanzee movement, and the north and east boundaries of the Greater Mahale Ecosystem plus the town of Mpanda in the southeast. (Image contributed by Lilian Pintea, Jane Goodall Institute). ....................................................... 20 Figure 2-6. Miombo woodland and thin strip of forest in Ugalla [from Moyer et al. 2006]. ......... 21 Figure 2-7. Monthly rainfall in Issa during the study period. .............................................................. 25 Figure 2-8. Woodland, Issa: monthly mean temperature and mean daily minimum and mean daily maximum temperature on topographic levels of valley, slope, and plateau. ...................................................................................................................................... 26 Figure 2-9. Forest, Issa: monthly mean temperature and mean daily minimum and mean daily maximum temperature in forest on topographic levels of valley, slope, and plateau. ...................................................................................................................................... 26 Figure 2-10. Woodland, Issa: monthly mean relative humidity and mean daily minimum and mean daily maximum relative humidity on topographic levels of valley, slope, and plateau. ................................................................................................................... 27 xiv Figure 2-11. Forest, Issa: monthly mean relative humidity and mean daily minimum and mean daily maximum relative humidity on topographic levels of valley, slope, and plateau. .............................................................................................................................. 28 Figure 2-12. Woodland, Issa: monthly mean and mean daily maximum wind speed in valley, slope, and plateau. ....................................................................................................... 28 Figure 2-13. Forest, Issa: monthly mean and mean daily maximum wind speed in forest valley, slope, and plateau. ....................................................................................................... 29 Figure 2-14. Nest types (schematic): 0 – Ground nest, 1 – 1st fork, 2 – Outer fork, 3 – Tree top, 4 – Tree to tree (one or more) side branches, 5 – Tree top to (one or more) tree tops, 6 – Tree top to (one or more) tree side branches, 7 – Tree crux, 8 – Lianas. ....................................................................................................................................... 31 Figure 2-15. Types of branch structure supporting nest. ..................................................................... 32 Figure 2-16. Measures of nest and tree characteristics; numbers correspond to descriptions of each measure - 7) nest height, 8) nest height above trunk base, 9) distance to trunk base, 10) distance to main stem, 11) distance to tree crown edge, 17) tree height, 18) tree crown base height, 19) tree crown height. ............................................... 34 Figure 2-17. Tree branch morphology .................................................................................................... 35 Figure 2-18. Author using (a) DRT (Double Rope Technique): friction knots and attachment over branches allows greater horizontal movement (photo by Alex Piel), and (b) SRT (Single Rope Technique): mechanical ascenders and descenders allow rapid ascent (photo by Jim Moore). ...................................................... 38 Figure 2-19. Nest shape measurements. a) Length: longest diameter of the nest; b) Width: perpendicular to length; c) Radii 1-4: length/width intersect to nest edge; d) Depth: vertical distance from central surface of nest bowl to the height of the nest edges, and Depth unsprung: many nests spring up when empty so I measured depth again after depressing the centre of the nest; e) Central thickness: vertical distance through the nest centre from the ventral to the dorsal surface of the nest bowl; f) Thickness 1-4: vertical distance through the nest 5cm in from the edge of the nest along each radius. ....................................................................................... 39 Figure 2-20. Nest deconstruction and measures of a) stem diameter, b) distance to stem, c) selected support diameters and type (“horizontal Y-shaped branch”), d) bend count and diameter, e) break count and diameter. ............................................................ 41 Figure 3-1. Nest types built in Fongoli and Issa (wet and dry seasons). ............................................ 55 Figure 3-2. Nests built on different types of support (branch configurations). ............................... 55 xv Figure 3-3. Nests integrated out of multiple sources of material in Fongoli in dry and in Issa in wet and dry season. .................................................................................................... 57 Figure 3-4. Nests built using different numbers of trees in Fongoli in dry and in Issa in dry and wet season. ........................................................................................................................ 57 Figure 3-5. Nests with full, partial, or no cover overhead in Fongoli in dry or Issa in dry and wet season. ........................................................................................................................ 58 Figure 3-6. Nest heights in Fongoli in dry and Issa in dry and wet season. ...................................... 59 Figure 3-7. Nests built in the upper, middle, and bottom third of the tree crown in Fongoli in dry and Issa in dry and wet season. ................................................................................. 60 Figure 3-8. Nesting tree species in Issa in vegetation types of woodland and forest (uid = unidentified). ............................................................................................................................ 61 Figure 3-9. Nesting tree species in Fongoli in vegetation types of grassland, woodland and forest (uid = unidentified, more remain uid because only feeding tree species have been identified in Fongoli). .......................................................................................... 62 Figure 3-10. Nests built in fruiting trees in Fongoli in dry and in Issa in dry and wet season. ...... 63 Figure 4-1. Mean minimum overnight ambient temperature recorded during bi-weekly periods from October to March with permanently deployed loggers (x symbol) and mean differential temperature recorded in a nest (closed circles) and on the ground (open circles) below. Error bars indicate max and min. ..................................... 88 Figure 5-1. Nest position as a proportion of tree crown height and nest position as a proportion of tree crown radius were calculated using the depicted measurements. 1) Nest position = c/a (c = d-b). 2) Nest position = e/g (g = e + f). ........................................................................................................................................ 101 Figure 5-2. Frequency of fresh nest groups recorded in closed (Issa, 1.5%; Fongoli, 2%) or open (Issa, 98.5%; Fongoli, 98%) vegetation in Fongoli and Issa. .............................. 103 Figure 5-3. Nests with alternative escape routes in closed (Issa, 1.5%; Fongoli, 2%) or open (Issa, 98.5%; Fongoli, 98%) vegetation in Fongoli and Issa. .............................. 103 Figure 5-4. Nest position in tree crown as a proportion of tree crown height in open and closed vegetation types in Fongoli versus Issa. (┌─*─┐indicates significant difference). Bars outside boxes indicate range. ............................................................... 104 Figure 5-5. Nest position in the tree crown as a proportion of tree crown radius in different vegetation types in Fongoli and Issa. (┌─*─┐indicates significant difference). Bars outside boxes indicate range, excluding outliers, which are indicated by circles. .............................................................................................................. 105 xvi Figure 6-1. Mean overnight temperature compared between Fongoli dry and Issa dry and wet season (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by circles. ................................ 119 Figure 6-2. Mean overnight relative humidity compared between Fongoli dry and Issa dry and wet season (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range. ........................................................................................................... 119 Figure 6-3. Mean overnight gust speed compared between Fongoli dry and Issa dry and wet season (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by circles and stars. ............... 120 Figure 6-4. Proportions of nest types built per month with mean monthly overnight relative humidity. .................................................................................................................. 121 Figure 6-5. Proportions of nest types built per month across months of varying rainfall. .......... 122 Figure 7-1. Onset of nesting for nest groups, at intervals of 10 min (n = 55). ............................. 145 Figure 7-2. Duration of nest building in small-, medium- and large-sized nest groups (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by stars or circles. .................. 146 Figure 7-3. Duration of building in seconds of all overnight, abandoned, and re-used nests (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by stars. ................................... 147 Figure 7-4. Building duration of males (10 individuals), females (4 individuals), and immature males (2 individuals) (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers. .................................... 148 Figure 8-1. Four types of scar, or macro-use-wear, described as follows: A) old break – note the new shoots sprouting around the break. Above, shoot continues growing vertically, and to left, shoot freshly broken into new nest; B) old bend – normal growth continues (horizontally) in direction of main branch. Two side branches continue growing aberrantly perpendicular to main branch; C) dead xvii end – note characteristic tail at old detachment point and two new shoots have sprouted. Above, one continues to grow vertically, and to left, other freshly detached and used in nest building; D) old frame branch - several dead branches remain, some detached and others broken, forming an old frame. ............ 171 Figure 8-2. Number of scars around fresh nests and around random control suitable nest spots (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, indicated by stars and circles. ................................. 172 Figure 8-3. Percentage of nests with re-growth and amount of re-growth at each nest increases over time. Re-growth measured in categories of few (0-5), some (5- 10), or many (>10) branches or twigs. .............................................................................. 173 Figure 8-4. Nest survival over time. Nests built in forest in dry season (n = 26) decay faster than nests built in wet season (n = 25), whilst nests built in woodland in dry season (n = 49) take longer to decay than nests built in rainy season (n = 91). Nests built in woodland overall take longer to decay than those in forest. ................ 174 Figure 8-5. Percentage of specific nest locations not re-used over time in woodland and forest trees. Specific nest sites in forest were re-used at faster rate than in woodland. Polynomial trend lines indicate possible continued re-use in forest (y = -0.08x3 - 1.51x + 100, r2= 0.998) and woodland (y = 0.011x3 - 0.29x + 100, r²= 0.989). ............................................................................................................................. 176 xviii Preface The work described in this thesis was conducted from the Department of Biological Anthropology, University of Cambridge, under the supervision of Prof. William C. McGrew. This thesis is the result of my own research and includes nothing that is the outcome of work done in collaboration, except where specifically indicated in the text. No part of this thesis has been submitted to this or any other university for any degree or diploma. The text does not exceed 80,000 words. Fiona A. Stewart Cambridge, July 2011 xix Summary Human beings of all cultures build some form of shelter, and the global distribution of Homo sapiens depends on this basic trait. All great apes (chimpanzee, bonobo, gorilla, and orangutan) build analogous structures (called nests or beds) at least once a day throughout their adult lives, which suggests that this elementary technology was present before the hominid lines separated. This thesis investigates the variability and function of specifically wild chimpanzee shelters. I compared characteristics of chimpanzee nests, nesting trees, nest shape, and architecture in two savanna-dwelling populations on opposite sides of Africa: Fongoli, Senegal, and Issa, Tanzania. Savanna habitats are the most extreme habitats in which chimpanzees survive today, and may represent a similar environment to that in which early hominins evolved in the Plio-Pleistocene (Chapter 2). Investigating variation in nest-building within and between these two extreme habitats made it possible to tackle hypotheses of the shelter function of nests (Chapter 3). The influence of environment, specifically the role of protection from disease vectors and fluctuating temperatures, was assessed through a novel experiment in which I slept overnight in arboreal chimpanzee nests and on the bare earth (Chapter 4). To assess whether or not nests serve as an anti-predation function, I compared nesting in Issa, where predators are abundant, to Fongoli, where they are absent (Chapter 5). I provided further support for the thermoregulatory function of nests by showing that chimpanzees build more insulating nests in adverse weather conditions (Chapter 6). Nest-building is a learned behaviour, but its ontogeny is little known. I investigated social sources of variation in nest building in Fongoli to examine whether sex and age differences exist in nest building duration, nest position, shape and architecture (Chapter 7). Finally, ecosystem engineering is a consequence of animal construction, from ants to humans. I investigated use-wear traces around nests to assess niche construction of nest- building. I showed that chimpanzees repeatedly re-used these specific nest-spots within trees, which are pre-fabricated for future building through repeated pruning and shaping of these structures (Chapter 8). Nest building in great apes may be the foundation of constructivity in hominids. This thesis describes proximate functions and influences on nest-building variation in wild chimpanzees that help to model the evolution of shelter in hominids. xx Acknowledgements This research was only possible through the help and support of a great many people and organisations. To everyone below and anyone I have accidentally omitted... Thank you! This work was done under the supervision of Prof. William McGrew. To him, I will be forever grateful for his taking a chance on a young scot, after meeting in Edinburgh Airport, and sending her off to Senegal to chase wild chimpanzees for her first African experience. I am grateful for his continued support and friendship since that day. Throughout all the emotional, logistical, and financial oscillations of the last five years, I am deeply grateful for all of his advice, support, feedback, endless patience... and generous round-buying in the weekly pub! Thanks also to Bill for careful reading and editing of every word of this thesis. I thank him, in particular, for the brainstorming sessions, and persuading me that to understand nest sleep, I must first experience chimpanzee sleep! This work would not have been possible without permission from several organisations in Senegal and Tanzania. I thank the Republique du Senegal, Departement du Eaux et Forets du Senegal for granting research permission to the Fongoli Chimpanzee Research Project. I thank the Tanzania Commission for Science and Technology and Tanzania Wildlife Research Institute for permission to work in Issa, Ugalla. Many thanks also to the Convention in Trade of Endangered Species (CITES) management authority in the UK, Senegal, and Tanzania, and to Department of Environment, Food, and Rural Affairs (DEFRA) UK, for assistance in application for permission to export/import chimpanzee hairs and faecal samples. I am extremely grateful to all of the funding organisations who supported this project: Carnegie Trust for Universities of Scotland (Carnegie Scholarship, Carnegie Research Grant), Corpus Christi College for a Taylor Bursary, Harold Hyam Wingate Foundation, International Primatological Society, L.S.B. Leakey Foundation, and the Wenner Gren Foundation. Many thanks also go to NewTribe, Tree-Climbing Northwest, who provided me with tree-climbing equipment. Huge thanks go to Tim Kovar, who taught me all the basics of tree- climbing and patiently answered all my questions from near and far! I am very grateful to Dr. Jill Pruetz for first offering me the opportunity to study wild chimpanzees, teaching me methods in primatology field research and for inviting me back to Fongoli for the PhD. To her I am grateful for logistical support and a nest of my own in Fongoli village. I am grateful to her, Maja Gaspersic, Stacey Lindshield, and Dondo Kante for providing data on who nested where, and when, and for help with day to day project decisions and xxi methods. Many thanks go to Wali Camara and Eladjh Saho, for their assistance with data collection and patience with requests, which are ‘vraiment fous’, of staying out after dark and sometimes sleeping in the bush! I thank Sita and Nene Camara for all the work they did for us in the village, and for sharing their food with us. I thank everyone above, Mboule Camara and the entire Fongoli village, plus Julie Lesnick, Nene Kante, and all her family, especially Nadege and Alex, for their wonderful company and keeping me sane during the research. To Nene and Dondo Kante, I thank them for being maid of honour and best man, respectively, and for hosting an unforgettable party at their home, for our wedding. I thank Dondo for welcoming me into his home in Kedougou and into his life, for his special and enduring friendship, and our many shared memories. I am extremely grateful to the many friends and colleagues who helped us during our study period and ongoing research at Issa. In Tanzania, I thank Gabriel S. Laizier (Tanzania Botanical Research and Conservation Programme, National Herbarium of Tanzania, Arusha) and Henry J. Ndangalsi (Department of Botany, University of Dar es Salaam) for identification of plant species. Many thanks to Roy Gereau for putting us in touch with, and facilitating, our relationship with Gabriel. I thank all the people at the Jane Goodall Institute in Kigoma (JGI) who helped often with logistical challenges, exporting faecal samples to collaborators and occasional rides to our camp! Specifically, thanks to Dr. Anthony Collins, Mzee Emile Kayega, Soodi Ndimuligo, Dr. Shadrack Kamenya, Mzee Emmanual Mtiti, Joyce, and Ambrose Hermegast, among others. Thanks to Oskar at High Tech Lodge for providing our home away from home in Kigoma and for frequently storing so much of our gear! In Uvinza, thanks go to Moshi Rajabu and Mama Shabani for looking after our house there, and for always taking care of us and our friends when we or they passed through. Many thanks go to Adriana Hernandez- Aguilar and Jim Moore, for our ongoing collaboration to keep long-term research going at Issa. Many people helped on the project at Issa. I thank Shedrack Lucas and Busoti Juma for always keeping me safe when tree-climbing, for their indefatigable enthusiasm and hard work with meticulous measurements, unsafe night time flooded river crossings, and most importantly for their loyalty and integrity to the work. I thank Ndai Sammwely for taking care of camp and cooking. Many thanks go to Abdalla Kanimba, Msigwa Rajabu, Abdalla Said, and Moshi Rajabu for all their help on the project. I thank everyone at Issa too, for their stories and friendship. Thanks to Deborah Moore, who made periodic visits to the Issa camp, and helped with locating nests, and faecal sample collection. Similarly, many thanks to Marina Llorente (a fellow nest- deconstructor), who embraced life at Issa, becoming an integral member of our team for several months, translated parts of her thesis, and spent hours discussing nests with me in the forest. I xxii will be forever grateful to Alex Piel (who is the ‘we’ in Tanzania), for finding numerous nests, collecting faecal samples, help measuring nest groups, for his company, support, fantastic (nest- finding) acoustic surveillance system, for being there for everything, and all our shared memories. I am grateful to all the strangers who helped a pair of stranded researchers to hitch- hike to Kigoma! Many people visited us at the Issa camp and I am grateful for their company during the year; my parents, Jane Piel, Ken Piel, Mary-Lou Bova, Jim Moore, Margaret Schoeninger, Klara Petrzelkova, Charlie and Nikki Jackson, Julia Rist, Mario Rucke, and Rob and Pol Summerhayes. Many primatologist colleagues and friends, in the Department of Biological Anthropology, provided invaluable advice throughout the dissertation process. I am grateful to everyone in the Primate Journal Club, for hours of primatey debate and talk in the office and in the pub. Many thanks go to Dr. Sonja Koski, for patiently answering all my interruptions and questions about statistics whilst we shared an office. I will be forever grateful to Dr. Adriana Hernandez-Aguilar for her detailed comments and advice on funding applications in my first year, for reading and improving Chapter 8 of this thesis, and for her ongoing friendship. I am extremely grateful to our cohort of chimpers for all of their advice, support, discussions, and friendship; Paco Bertolani, Susanna Carvalho, Kat Koops, and Caroline Phillips. I thank Paco for his frankness, and humour, and for all his support since our arrival together in Cambridge years ago. I am very grateful to Susanna for reading Chapter 8, her archaeological expertise and novel ideas greatly improved the writing and discussion. To Paco, Susanna, and Caroline, I am grateful for the camaraderie of our office, despite all the stress of ‘finishing’! I will be forever appreciative to Kat for our shared development of some of the ideas in this thesis during our first year, in particular brainstorming about the ‘shelter’ functions of nests, and our joint development of the nest and tree characteristics measurement methods in the hopes of future comparisons across sites. I thank her for sharing her Nimba chimpanzee faecal and hair samples with me for our learning of genetic analysis methods. I am extremely grateful to Caroline Phillips, in particular, for her enduring friendship and patience, for always having an ear to listen to and offer advice on the thesis (and life!) and for being my lifeline during the write up. I am indebted to many other friends and colleagues for all their support and ideas: to Chris Woolley, who spent many hours over a bottle of wine discussing statistics and helping me think about how to deal with this huge database! To Andrea Gibson, of the University of Zurich, who visited Cambridge and helped develop some of the methods together with Kat Koops; to Mike Haslam for his thoughts and advice on use of archaeological terms and fossil tree scars! I thank everyone in the Large Animal Research Group in the Department of Zoology (and xxiii partners), who were a network of support during the most trying time of this write up and specifically with statistical and organisational guidance: Sinead English, Raff Mares, Dieter Lukas, Andrew Bateman, and Kirsty Macleod. In the Department of Biological Anthropology, many other people helped with parts of this project, including parts I was not able to include! I thank Rie Goto for her class on SPSS and statistics, and meeting with me to talk through some of the methods. I thank Noreen von Crammon, Steve Lycett, and Colin Shaw, whose various ideas of how to measure (whether a skull, a stone tool, a bone, or a tree!), in my first year helped develop the methods used in the project. I also thank Vilnis Termanis and Fabio Lahr for computing assistance and for rescuing my computer and this very thesis from the brink of oblivion! I am grateful for the help of everyone in Dr. Leslie Knapp’s Primate and Immunogenetics (PrIME) laboratory with whom I have worked and enjoyed many discussions. In particular, huge thanks go to Jo Osborn, who patiently taught me laboratory methods and genetic analyses, especially during my initial stages of lab work, and to Pascaline le Gouar, and Catia Traca, who helped a great deal with development and use of some methods. Huge thanks go to Maggie Bellatti and Jo Osborn, for so much help with last minute orders, sending numerous supply packages to me in the field, and all of their help back in Cambridge. Finally, I am extremely indebted to Dr. Leslie Knapp, for all of her enthusiasm and help in the genetic component originally planned for this project, which although it could not be included in this thesis, has taught me many skills and shaped the direction of future research questions into savanna chimpanzees. I will be forever grateful to Leslie for always being available no matter how big or small the problem, for welcoming me into her lab group, and for the many, many hours spent discussing genetics, savanna chimpanzees, and future research. I will be forever indebted to all my early mentors in the Department of Zoology, at the University of Glasgow, including Dr. Stewart White, Dr. Isobel Coombes, Prof. Graeme Ruxton, Mr. Geoffe Hancock and Prof. Mike Hansell, for all of their support, experiences offered in the field, and their contagious appreciation for the natural world. In particular, I am grateful to Prof. Mike Hansell, who encouraged me throughout my undergraduate degree, who introduced me to the diversity of animal building behaviour, and who suggested chimpanzee nest architecture as a first research project. I am deeply grateful to Prof. Jim Moore and his family, Sue, Ben and Maeve Moore, for taking me into their family during my application to begin my PhD at Cambridge and for making this research possible through supporting my first year. I am indebted to Jim for the opportunity to first begin research in Ugalla, continued feedback on ideas, chapters and manuscripts, his xxiv insightful and creative ideas and shared field experience, and overall for his past and ongoing emotional and academic support. And who can forget his unique, but messy, method of finding fresh chimpanzee nests! Finally, I thank my family for all their enduring support. My parents, I thank for always encouraging me to follow my dreams and ambitions, for letting me run wild in the Malaysian jungle as a child, and for providing me immeasurable emotional and financial support throughout my life. One person has been with me for every stage of this process and experienced every part of this PhD. I cannot describe all of the things I have to be grateful to Alex Piel. I am grateful for the feedback he provided on chapters and help overseeing data collection for Chapter 8 from October 2009 to June 2010. I thank him for taking over my hut in Senegal when we met and for sharing every moment since. I thank him for his laughter, for always picking me up when I stumble, for being Lahalahato and le médicament pour ma vie. I thank Alex deeply, for all of his love and support, since we met and forever. Finally, I thank the chimpanzees of Fongoli and Issa, for tolerating theft of their nests and allowing me to peer into their world. xxv Abbreviations and notations used BN Bend BR Break CBH Crown Base Height cm centimetres DBH Diameter at Breast Height DET Detached DRT Double Rope Technique HB Horizontal Branch hr Hour hrs Hours LCA Last Common Ancestor LSB Lone Side Branch M Median m metres m/s metres per second Min Minutes Mya Million years ago n number NH Nest Height PASW Predictive Analytic Software for Windows PCR Polymerase Chain Reaction REM Rapid Eye Movement SB Side Branch sec Seconds SRT Single Rope Technique TH Tree Height TW Twig UCSD University of California, San Diego x ̄ Mean 1 Chapter 1 Introduction Thesis Shelter construction is a human universal. Great apes build analogous constructions termed nests. Shelter often indicates an overhead barrier, but in its broadest sense the noun ‘shelter’ simply refers to a structure, or feature, that provides protection from environmental challenges, safety, or refuge from danger, and as such great ape nests are shelters. They build at least one a day for their entire lives (post-weaning), as these sleeping shelters are built for both day-time rest and over-night sleep. Thus in a lifetime an individual may build more than 19,000 nests [Fruth and Hohmann 1994b], making this behaviour the most pervasive form of technology in the great apes. These structures have been termed ‘nests’ since the very first field studies [e.g. Nissen 1931], but later researchers have offered alternative terminology to clarify this distinction, referring to them as “beds” [Hiraiwa-Hasegawa 1989], or “sleeping platforms” [McGrew 1992]. I will continue to use the term nest here, despite its misleading sense (a nest is usually a structure for rearing offspring [Hediger 1977]), in regard to the historic use of the term. Through detailed study of chimpanzee nest-building this study aimed to understand, not only how they are made, but also why this behaviour has evolved and been conserved in all extant great apes. Through specific tests and examination of how the environment influences structural variation, this study aims to determine proximate functions of nest-building, influences on variation, and how this may have influenced the evolution of shelter. Significance Great apes build a new nest (or bed) at least once a day throughout their lives, from weaning onwards. The four species of great apes make broadly similar nests, despite great differences among them in habitat and social life [Fruth and Hohmann 1996]. All great apes build these structures nightly as sleeping-platforms and such construction is absent in all other anthropoid primates [Kappeler 1998]. Nest-building is thus the most pervasive form of material culture in living apes, and it was likely present long before the hominid and hominin lines separated, making it an important deep ancestral trait when seeking to model the behaviour of our earliest ancestors and the evolution of shelter-construction in humans. Humans in all cultures on every continent make shelters [Brown 1991], and the global distribution of Homo sapiens depends ultimately on this basic trait for protection from the environment. Human shelters can range from a simple depression in the ground to the complex 2 high-rise architecture seen in cities today. Similarly, in other animal species, some homes or shelters are enclosed structures, others may be as simple as a scraping in the ground (e.g. in some ground nesting birds) [Hansell 2005]. Despite nightly nest-making in the great apes, no hypothesis-driven test yet has been done to determine the functions of these primitive shelters. The primary function of shelter construction in great apes is arguably to create a secure platform for sleep. Fruth and Hohmann [1996] proposed a “cognitive leap” to have occurred in the great apes with the evolution of nest-building, as the ability to sleep safely and comfortably in a relaxed, recumbent, supine posture likely increased REM sleep, which may have aided memory consolidation and enabled cognitive evolution in hominids. McGrew [2004] framed three hypotheses of nest function in addition to that of recumbent sleep: anti-predation, anti- pathogen, and thermoregulation. This thesis investigates these hypothesized functions of nests through experimental tests, and using a comparative and a correlational approach. Chimpanzees have long served as effective models for reconstructing behaviour of our last common ancestor (LCA) [Moore 1996; Wrangham 1987]. Wrangham [1987] highlighted the importance of the phylogenetic method for modelling behaviour of the LCA, and by its presence in four extant apes, we can thus attribute nest-building to the LCA. Fossil evidence shows many early hominins to have apelike anatomical adaptations to arboreality (Ardipithecus ramidus: [White et al. 2009b]; Australopithecus afarensis: [Alemseged et al. 2006]; A. africanus: [Berger and Tobias 1996]; Homo habilis: [Richmond et al. 2002; Ruff 2009]). Although, the lack of arboreal morphological traits found in the newly discovered Australopithecine, A. sediba [Berger et al. 2010], suggest that the transition from the more arboreal-adapted, less-habitual bipeds (Ardipithicus, A. afarensis, and A. africanus cited above) to more committed terrestrial and large bipeds, like H. erectus [Ruff 2009], may have occurred in a ‘mosaic fashion’ [Berger et al. 2010]. It is reasonable to infer that traits for arboreality are found in early hominins who continued to sleep in trees, long after becoming terrestrial, perhaps until the controlled use of fire. Homo erectus may have been the first hominin to sleep terrestrially, as post cranial skeletal elements show no arboreal adaptations [Brown et al. 1985; Ruff 2009]. The timing and emergence of the controlled use of fire remains controversial, however archaeological and ecological analyses support the hypothesis that H. erectus was the earliest hominin to use fire [Clark and Harris 1985; Goren-Inbar 2004; Karkanas et al. 2007]. H. erectus was also the first to leave Africa [Anton and Swisher 2004]. Given the lack of unequivocal evidence of use of fire [James 1989] in the early Pleistocene, dispersals out of Africa seem impossible without sheltered protection from temperate conditions. Prior to this, nests may have served as shelters. Nest- building by early hominins has thus been hypothesized by several researchers [McGrew 1992; 3 Sabater Pi et al. 1997; Sept 1992; 1998], so that these primitive shelters of apes may be analogous to the earliest manifestations of material culture in hominids. Evidence for shelter construction by early hominins is necessarily scarce as structures comprised of organic material are not easily preserved in the fossil record. Present evidence thus remains ambiguous and debated, e.g. foundation of wooden hut construction dated to approximately 380kyrs ago in Terra Amata [Gamble 1999]. Although hominin shelter construction cannot be verified, it can neither be assumed absent. Sabater Pi et al. [1997] outline further factors that support hypothesised nest-building in early hominins, several of which are important in this thesis: transition from forested to wooded savanna environment, poor nocturnal and crepuscular vision common to diurnal primates, greater density of and predator pressure in an open environment, lack of evidence of fire use by early hominins preceding late Homo erectus and archaic H. sapiens, and a likely requirement for recumbent relaxed sleeping postures. Here I address whether or not nests of chimpanzees function as shelters in a dry, open, wooded savanna habitat, an environment similar to those characterizing many reconstructions of early hominin habitats [Alemseged 2003; Reed 1997; 1998; Schoeninger et al. 2003; White et al. 2009a], where chimpanzees face ecological pressures similar to those of nest-building extinct hominins. The ubiquitous presence of shelter construction across the globe in human societies is marked by variation in building techniques and architecture, indisputably both functional and cultural. The ability for skilled object manipulation has been the defining characteristic of hominins, yet most studies on apes have focused on tool use and tool production, leaving nest- building lagging behind [McGrew 1992]. Comparisons of shelter-making across populations and individuals may reveal further insights into behavioural transmission processes within [e.g. ant- dipping or termite-fishing skill acquisition: Humle et al. 2009; Lonsdorf 2005; 2006; Lonsdorf et al. 2004] and across sites [e.g. ant-dipping, Humle and Matsuzawa 2002; Schöning et al. 2008]. The results of this study inform on motivation and technique of rudimentary constructivity by chimpanzees and likely early hominins. Groves and Sabater Pi [1985] preliminarily compared ape nesting to hunter-gatherer sleeping site and home-base use, yet they did not include comparison of constructivity. Sept [1992; 1998] later reported re-use of nesting sites by savanna chimpanzees [see also Hernandez-Aguilar 2009], thus refuting preconceived differences between chimpanzee, hunter-gatherer, and hominin ranging. Hernandez-Aguilar’s [2006] work highlighted the role of tree morphology in nest site selection by chimpanzees, but did not address microclimatic influences, nest function, or structure. Future comparisons can be made between shelter function and structure in hunter-gatherer groups, including sex differences 4 and patterns of variation. This thesis provides critical information on how and why these analogous constructions vary in chimpanzees. Animal building Construction behaviour is widely distributed throughout the animal kingdom, in both invertebrates and vertebrates [Hansell 2005]. Use of tools is relatively rare, and some authors argue that building should be included within the realm of tool use [Fruth and Hohmann 1996], or the converse that tool use would be more effectively studied within a framework of all animal construction behaviour [Hansell and Ruxton 2008]. Animal-built structures fulfil three main functions: home or shelter, food trap, and communication [Hansell 2005]. Both in number and diversity, most animal structures function as a home or shelter and may be built for the long- or short-term, for individual use, or for rearing of young [Hansell 2005; Hediger 1977]. The most common functions of animal built shelters include protection against temperature extremes and predation [reviewed in Hansell 2005]. Hansell [1984; 2000] defined several different construction techniques that are applicable across all species of builders. A species tends to use predominantly one method, but may use more than one in the complete repertoire of construction. He hypothesised that simple and repetitive actions are selected for in building behaviour across species [Hansell 2005]. The simplest method of building is “fetch-and-drop”, which involves collection and placement of materials and is seen in species from ants to fish and some birds [Hansell 2005], although birds may employ additional methods to secure materials in place, e.g. shuddering of twigs (innate shaking movement) to secure their placement in the nest by Corvidae [Goodwin 1976]. “Inter-locking and weaving” is often required to make sure a structure remains intact and secure. Hansell [2000] characterised several techniques from comparisons of bird nest building. The first is “velcro”, which relies on the properties of two materials naturally fastening to secure the structure; e.g. lichen and spider silk. The other two techniques, “stitches and pop-rivets” and “entangle” result from the behaviour of the builder to create a fastening effect. Stitching is rare, but found in some birds and invertebrates, whilst entangle is the most widespread building behaviour in birds and involves folding of twig ends back into the nest after placement. The most complex form of building is “weaving”, which involves a number of attachments to begin and threading material in and out of the structure. Weaving is found in just two taxa of birds and 5 there is some evidence that the behaviour is learned, e.g. in Ploceus cucullatus [Collias and Collias 1964].1 The behavioural elements of construction are generally few in number, simple, and repetitive across species [Hansell 2005]. Analysis of repetition or sequence of behaviours can quantify the behavioural complexity. For example, Byrne [2003] put forward the “behaviour parsing” model of imitative learning, in which statistical regularities present in a behaviour may aid observational learning. It is theoretically possible that Byrne’s [2003] model reflects learning also of nest-building behaviour. A pre-requisite for this possibility is a hierarchical organisation of behaviour which has been shown in preparation of difficult to process plants in both mountain gorillas Gorilla g. beringei [Byrne and Byrne 1993; Byrne et al. 2001] and chimpanzees Pan troglodytes schweinfurthii [Corp and Byrne 2002a]. They identified over 200 separate elements of behaviour in thistle processing by mountain gorillas [Byrne et al. 2001]. These elements varied idiosyncratically, each individual having a preferred set of elements. However techniques, which were defined as structured sequences of functionally distinct elements, numbered only about 250, which is much less than the number of techniques possible from combinations of the repertoire of behavioural elements and so suggests a high degree of selectivity and complex organisation of behaviour [Byrne et al. 2001]. Through these studies of manual skill in great apes it has been determined that great apes (including ourselves) show: hierarchically organised behaviour composed of regular sequences of elementary actions, bimanual role differentiation, with a modular organisation in which some stages may be omitted or repeated [Byrne 2007]. Byrne [2007] calls for the application of such studies of technologically complex behaviours to understand cultural transmission processes in wild apes. This approach may be broadly applicable to the study of animal architecture, and an initial goal of this study was to differentiate, through deconstruction of nests, the building steps involved. The building behaviour of animals, whether complex or simple, can have significant effects on the physical state of the environment, available resources, and other species within the ecosystem. As a result, many species are termed “ecosystem engineers” [Jones et al. 1994]. A similar concept was also framed by Dawkins [1982] through the term “extended phenotype” whereby the impacts of behaviours of organisms extend beyond their bodies. Extended phenotypic effects may influence the environment in ways that alter the fitness of other organisms, but also feed back to the individual. Hansell [1993] proposed provocative hypotheses 1 A further five methods are highly specialised in certain taxa and so are less relevant here, e.g. secreted materials to either combine materials (“Sticking together”), shape (“moulding”), or make webs and cocoons (“spinning”), and “Folding or rolling”, which uses secreted glue to create an enclosed shelter from a sheet of material, like a leaf. Finally, “sculpting” or burrowing is common across a number of taxa and these species have specialised anatomy Hansell M. 2005. Animal Architecture. Oxford: Oxford University Press. 314 p.. 6 that through the added protection provided by built structures, animal architects are able to extend their range and reduce possibilities of extinction. Laland et al. [2000] expanded these concepts to include not just influences of builders on ecosystems and individual builders, but also influence on the individuals’ descendents through ecological inheritance. They framed these concepts into a branch of evolutionary theory, termed “niche construction” [Odling-Smee et al. 2003]. Great ape nest building The nests of all four species of great ape are said to be constructed in a similar manner despite differences in habitat and social organisation [Fruth and Hohmann 1996]. Great apes spend over half their lives in nests, yet there are few detailed descriptions of the behaviour patterns employed. Nest-builders create platforms by bending and breaking branches over a foundation of stronger tree limbs, interweaving branches to varying extents, and then folding smaller twigs over the edge to form a rim. The finished structure is a circular platform or bowl, to which detached leafy twigs are often added for lining (Pongo pygmaeus: [MacKinnon 1974; Schaller 1961]; Pan paniscus: [Fruth 1995; Horn 1980]; Pan troglodytes: [Goodall 1962]; Gorilla gorilla: [Bolwig 1959; Schaller 1963]). Bolwig [1959] analysed the method of construction for 45 gorilla nests, finding that although the principle behind construction is uniform, the structure varied depending on vegetation and slope gradient. Schaller [1961] analysed 13 orangutan nests, finding variation in the number of broken supporting branches, whilst Mackinnon [1974] cut down several intact nests and found only bent branches, which he classified according to their structural function, e.g. ‘rimming’, ‘hanging’, ‘pillaring’, or ‘loose’. Mackinnon [1974] also reported orangutans to transport detached material for nest-building from parts of the tree far away from the nest, or other trees; this has been recently observed again in sanctuary dwelling orangutans [Russon et al. 2007], but has not been reported for chimpanzees. Goodall [1962] noted that some chimpanzees work methodically in a circular motion, however others alternate bending of branches from opposite sides of the structure. Research into nest building since these early studies mainly has used nests as indicators of ape presence and distribution, leading to development of methods to assess population densities [e.g. Ancrenaz et al. 2004; Kouakou et al. 2009; Plumptre and Reynolds 1996; Plumptre and Reynolds 1997; Tutin and Fernandez 1984]. Some studies have compared variables such as nest height, age, and tree species used, and have yielded more comprehensive knowledge of factors influencing nest material and location selection [Baldwin et al. 1981; Fruth and Hohmann 1994a; Fruth and Hohmann 1996; Hernandez-Aguilar 2006; Wrogemann 1992]. There is some 7 evidence also that great apes adjust nest-building to improve thermoregulation; e.g. gorillas build fuller and arboreal structures [Mehlman and Doran 2002; Tutin et al. 1995] and chimpanzees build higher, more open, nests in the wet season [Baldwin et al. 1981]. Despite the important link between shelter construction by humans and nest-building by apes, no in-depth study of the function of these shelters in apes has previously been done. Several authors have hypothesized that the ability to sleep more comfortably and safely may have influenced the type and quality of sleep, and increased energy available the following day, which in turn enabled cognitive evolution in hominids [Baldwin et al. 1981; Fruth and Hohmann 1996]. Videan [2006b] revealed that sleep/waking patterns in captive chimpanzees mimic closely those of humans, and that the apes adjusted their sleeping sites in relation to temperature and humidity. Predation on apes has been reported from several long-term field sites [Boesch 1991; D'Amour et al. 2006; Fay et al. 1995; Tsukahara 1993]. Previous studies found across site differences in nest height within trees that may reflect differential predation pressure [Baldwin et al. 1981; Pruetz et al. 2008]. Hausfater and Meade [1982] showed baboons to stagger visits to sleeping-sites to reduce endo-parasitic transmission. The same may be true in chimpanzees, which have also been hypothesized to build fresh nests nightly to reduce parasite transmission [McGrew 2004]. Chimpanzees at Fongoli continue to nest arboreally despite a lack of predator pressure [Pruetz et al. 2008], perhaps due to additional benefits of arboreal nesting such as reduced numbers of biting insects or disease vectors. Sleeping separately in a nest might also decrease transmission of contact, or air-borne, transmitted infections, in contrast to the huddling behaviour of many monkey species. If nests serve the hypothesized functions, specific nesting-sites may also be selected for optimal conditions relating to thermoregulation, anti-predation, or anti-vector roles. For example, previous researchers have noted chimpanzee preference for nesting on slopes [Groves and Sabater Pi 1985; Suzuki 1969]. In savanna habitats, in particular in Issa, chimpanzees are highly selective of the areas of the landscape and tree species and tree morphology used for nesting [Hernandez-Aguilar 2006]. However, neither morphological characteristics of trees nor species present explains differential selection of topographic features or specific areas for nesting-sites [Hernandez-Aguilar 2006]. There is evidence for intra-specific level variation in nest characteristics [Baldwin et al. 1981] and intra-community variation of nest architecture [Stewart et al. 2007]. No detailed comparison of nest architecture across populations has previously been done, despite demonstration of such variation in other behavioural patterns [Whiten et al. 1999]. Universal categories such as nest-building were discarded from Whiten et al.’s [1999] analysis on the 8 assumption of no cross-population variation, although behavioural variation exists in nest- building by orangutans [van Schaik et al. 2003]. Nest-building in captivity requires learning and experience [Bernstein 1967; Videan 2006a], and in a learned behaviour, mimetic drift may be expected in separated populations, or variation may occur due to habitat differences. Within populations, the nature and nurture of nest-building in the wild remains unstudied. Sex-differences in acquisition of termite-fishing skills are known in chimpanzees [Lonsdorf 2005; Lonsdorf et al. 2004], so presumably there are differential fitness benefits to males and females. Females therefore may be expected to invest more in acquisition of nest-building skills, as their nests must frequently provide a safe and secure haven for offspring. If individuals learn nest-building through trial and error, then heterogeneity is expected in nest-building techniques and nest structure. However, if social learning occurs vertically from maternal models, or obliquely from other individuals, then more homogeneity in techniques is expected within lineages or communities. If learning is horizontal, then similarity may be expected between frequently-associating individuals. Alternatively, nest- building techniques may vary in response to the environment, and similarities could reflect convergent problem-solving, regardless of kinship or association patterns. Thesis outline To address the above thesis, I compare characteristics of chimpanzee nests, nest trees, nest shape, and architecture in two savanna-dwelling populations on opposite sides of Africa: Fongoli, Senegal, and Issa, Ugalla, Tanzania. Savanna habitats are the most extreme habitats in which chimpanzees survive today, and may represent a similar environment to that in which early hominins evolved in the Plio-Pleistocene [McGrew et al. 1981; Reed 1998; Schoeninger et al. 2003]. I describe the extent of these conditions in Fongoli and Issa in the methods of Chapter 2, in addition to describing all methods employed to study nests, trees, nest shape and architecture. In comparing variation in nest-building within and between these two extreme habitats in Chapter 3, I have two goals: First, to investigate whether or not differences exist in nest characteristics or building techniques across two sites. This is a useful first step to determine whether or not there is cultural variation in a specific behaviour across populations [sensu Whiten et al. 1999]. Second, by describing variation between the two sites and between seasons in Issa, I pose hypotheses of possible functions of nest-building in chimpanzees to be tested further in this thesis. 9 In Chapter 4, I describe an experiment in which I test two of the hypothesized functions of nests described by McGrew [1992]: anti-pathogen and thermoregulation. The anti-predation function of nests is discussed in Chapter 5. Issa has many predators [Hernandez-Aguilar 2006; pers. obs.], whereas Fongoli has none [Pruetz et al. 2008]. Issa chimpanzees may therefore be expected to nest in more inaccessible locations within trees than Fongoli chimpanzees if they are vulnerable to predation. The vulnerability of great apes to predation is highly debated, due to their large body size and arboreal habits. In Chapter 5, I also review the evidence that great apes have evolved anti-predator strategies and are currently vulnerable to predation. There are seasonal influences on nest-building, such as height or type of nest built. Through detailed architectural analyses, I test the hypothesis that nests vary structurally with more adverse weather conditions in accordance with hypothesized function in Chapter 5. Specifically, I test whether or not chimpanzees in both sites build nests that are more insulating through greater thickness, depth, amount of material, or lining twigs in colder or wetter conditions and more supported in windier conditions. Nest-building is a learned behaviour, but its ontogeny is little known. Thorough study of ontogeny of behaviours requires longitudinal study of individuals as they acquire a skill [sensu Lonsdorf 2005] or a cross-sectional analysis of a number of individuals of different ages and stages of development [McGrew 1977]. In Chapter 7, I test whether or not nest building varies according to individuals and by sex and age class, to present preliminary data on development of building skills, but also to test the hypothesis that female chimpanzees invest more in nest building. Finally, ecosystem engineering is a consequence of animal construction, from ants to humans, and through its feedback influence on builders and descendants of builders is termed ecological inheritance or niche construction [Laland et al. 2000]. Human niche construction is arguably the most pervasive and influential on the planet. The potential for apes’ nest building to influence future building behaviour of individuals and their descendants was first proposed by Fruth and Hohmann [1994b]. They hypothesised that nests are ‘prefabricated for future use’. After noting such artefacts of past nest building events around fresh nests, I tested in Chapter 8 whether or not nest sites regenerate through repeated pruning and shaping of nest spots and whether or not these specific nest spots are preferentially re-used over time. Nest-building in the great apes is hypothesised to have evolved initially through the construction of temporary feeding nests as branches were broken in towards the ape, whilst feeding, and subsequent benefits of monopolisation of food sources through nearby nest building [Fruth and Hohmann 1996]. Fruth and Hohmann [1996] also propose that nests 10 became ‘social information centres’ and in turn influenced the cognitive evolution of hominids through ‘information exchange’ and the benefits of a good night’s sleep. Hansell [1993] hypothesised that animal built structures influence the range, diversification, and survivability of building species. The evolution of nest-building in hominids may also have permitted them to expand their range and buffer them against likelihood of extinction through the benefits of shelter, which reduces the impact of the environmental conditions and changes. 11 Chapter 2 General methods Author measuring a low nest in Issa, western Tanzania; photo by Busoti Juma. 12 INTRODUCTION In this chapter I describe the study sites and the general methods of data collection that form the basis of the thesis. Two chimpanzee study sites in similar environments, but with differing levels of habituation were selected for comparative study; Fongoli, in Senegal, and Issa, in Tanzania. Here I present published data on the study sites, but I also describe in detail the climate data that I collected during the study periods at each site. Climatic measurements will be used further in Chapter 6 to analyse variation in nests with respect to weather. Several types of data were collected at both sites. I also include data that were collected but have not yet been analysed, and so are not included in the following chapters. Here I outline all the data collected on nest groups, characteristics, shape and architecture, plus other data collected on vegetation structure, phenology, behaviour, genetics, and insect density. Specific methods are described in each chapter. STUDY SITE Fongoli History and geography The Fongoli study site (12° 13.90 N 12° 11.30 W) is at the north western edge of the chimpanzee species’ distribution and is in south-eastern Senegal, about 35km north of the border with Guinea and 85km west of the border with Mali [Pruetz 2006]. Fongoli is 10km northwest of the town of Kedougou and 45km southeast of the Assirik study site in the Parc National du Niokolo-Koba [Pruetz 2006]. Research at Assirik began in 1976 and provided the only previous long-term behavioural and ecological study of savanna-dwelling chimpanzees, but lapsed in 1979 [McGrew et al. 1981]. Pruetz et al. [2002] returned to conduct a nest count survey to assess the distribution and density of chimpanzees within the park and the surrounding areas, and to identify areas most suitable for long-term research. Chimpanzees in Assirik were found to occur at a density of 0.13 individuals / km2, which is higher than that found outside the park [Pruetz et al. 2002]. However, small areas were found to have similarly high concentrations of chimpanzees, and the area of Fongoli, within the Tomboronkoto population was identified as a suitable area for long-term research. Later, the Fongoli research site was established to study this population. Research has been ongoing since 2001, and the chimpanzees were habituated in 2005 [Pruetz 2006]. 13 Vegetat ion The vegetation of south-eastern Senegal is a mosaic woodland-savanna habitat (see Figure 2-12) dominated by woodland and wooded grassland, interspersed with areas of bamboo, plateau, and thicket, plus small patches of gallery forest (including ‘ecotone’) that make up <1% of the landscape [Pruetz 2006; Pruetz et al. 2002]. Vegetation types were classified similarly for the purpose of this study and are described in Table 2-1. Pruetz and Bertolani [2009] report vegetation composition at Fongoli to be 46% woodland, 36% grassland, 12% bamboo, 4% field, and 2% forest. Previous work has shown that Fongoli chimpanzees nest most frequently in woodland (65% of nests), followed by grassland (23%) and gallery forest (8%). However, gallery forest is the preferred nesting habitat when the availability of this vegetation is considered [Pruetz et al. 2008]. Table 2-1. Description of vegetation types in Fongoli [from Baldwin et al. 1982; Pruetz et al. 2008; Pruetz et al. 2002], and terms used for comparison across sites in this study. This study Fongoli vegetation types Forest Ecotone: Evergreen woody vegetation at locations of water runoff from plateau edges Gallery forest: Tropical semi-deciduous lowland forest, usually along seasonal water courses. Multi- layered continuous canopy cover, and trees >10cm in DBH average 12.5m in height. Woodland Woodland: Drought-deciduous lowland woodland with mostly grass understory. Canopy cover discontinuous and trees over 10cm DBH average 7.6m in height. Wooded grassland Wooded grassland: Narrow-leafed savanna with isolated palms and deciduous trees Bamboo Bamboo: Flat-leafed savanna with isolated deciduous trees. Grassland Plateau grassland: Narrow-leafed savanna with isolated deciduous shrubs Swamp not present 14 Figure 2-1. Woodland vegetation in Fongoli, with small strip of forest running from upper left to centre. Fauna The Fongoli chimpanzees’ home range extends to within 6 km of the town of Kedougou [Howells et al. 2010]. In 2001, when the project began, only 4-5% of the chimpanzees’ home range was disturbed by anthropogenic activity, which was primarily due to horticulture and settlements of Malinke, Bassari, Diahanke, and Puhlar people [Howells et al. 2010]. However, since this time harvesting of wild fruit, gold mining, and grazing of livestock have intensified, along with increased immigration of humans. The fauna has been heavily affected and those species that remain occur only at very low densities [Pruetz et al. 2002]. Most frequently- observed fauna include other primates, such as patas monkey (Erythrocebus patas), vervet monkey (Cercopithecus aethiops), Guinea baboon (Papio papio) and Senegal bush baby (Galago senegalensis). Other species of mammal sometimes seen include bushbuck (Tragelaphus scriptus), warthog (Phacochoerus aethiopicus), duiker (Cephalophus spp.), and banded mongoose (Mungos mungo). Other large mammals are no longer seen outside the national park, although traces of hippopotamus (Hippopotamus amphibius) were found as recently as 2001 (Kante, pers. comm.). Fongoli has few large mammals and no evidence of large predatory species (lion, Panthera leo; leopard, P. pardus; 15 spotted hyena, Crocutta crocutta; wild dog, Lycaon pictus) were found in systematic surveys [Pruetz et al. 2008], although hyena and side-striped jackal (Canis audustus) vocalisations are occasionally heard. Climate This region has been described as the hottest, driest, and most open habitat for chimpanzees [McGrew et al. 1981]. In Assirik an absolute temperature range of 16-44°C was recorded over a four year period, with mean daily maximum of 35°C and mean daily minimum of 23°C [McGrew et al. 1981]. Mean monthly maxima were highest in April (>40°C), and mean monthly minima were lowest in November/December (<20°C) [McGrew et al. 1981]. Thus, the most extreme range of temperatures in the region occurs during the dry season. Rainfall averaged 954 mm per annum over four years in Assirik [McGrew et al. 1981]. Long-term climatic data have not yet been reported for Fongoli beyond annual daytime temperatures in woodland (mean, 30°C; maximum 38°C), grassland (mean, 25°C; maximum, 42°C), and gallery forest (26°C, 37°C) [Pruetz 2007]. Mean annual rainfall over a three year period was 786mm in Fongoli [Pruetz and Bertolani 2009]. Measurements from Kedougou show that rainfall averages 900-1100 mm annually, with October through May constituting the dry season, and the highest monthly average temperature of 33°C in May and the lowest minimum of 25°C in December [Pruetz 2006]. I deployed assemblages of data loggers at six locations in representative vegetation types within the Fongoli home range at typical nesting heights, which vary across each vegetation type used for nesting [from Pruetz et al. 2008]; Ecotone (7m) , Gallery (7m), Woodland (8m), Bamboo (10m), Grassland (8m), Plateau (8m). One aim for these data was to investigate how nests vary with weather conditions; I therefore deployed loggers at mean nesting height to be representative because climatic measurements may vary with deployment height. Each assemblage included a Hobo 4-channel micro weather station logger (H21-002) with attached Hobo wind speed smart sensor (S-WSA-M003), and a Hobo temperature and relative humidity logger (H8 Pro series). Each logger recorded data every 30 min. Two measures of wind speed were recorded; the mean wind speed for each 30 min sampling interval and the maximum 2-second wind gust speed in each 30 min sampling interval. The six vegetation types can be classified into three main vegetation types in the region, which are typically described in south eastern Senegal as ‘forest’ (ecotone and gallery), ‘woodland’ (woodland and wooded grassland), and ‘grassland’ (bamboo and plateau) [see Baldwin 1979]. I therefore took mean climatic measures for these three vegetation types. 16 This study was done during the dry season from October 2007 – March 2008, when there was almost no rainfall (12.2mm in October and 4.7mm in November), and absolute temperatures ranged from 7–45°C. Using the mean temperatures across loggers, mean daily maximum temperature was lowest in October (32°C) and highest in March (41°C), whilst mean daily minima were lowest in December and January (14°C) and highest in October and March (22°C, 23°C). Mean, minimum, and maximum temperatures were lower in forest than woodland, and in woodland than grassland (Figure 2-2). Relative humidity ranged from 3-100% and followed an inverse pattern through the dry season; mean relative humidity was lowest in February (22%), and highest in October (82%). Humidity was also highest in forest, and lowest in grassland (Figure 2-3). Mean daily maximum relative humidity remained close to 100% for the first two months during the dry season, whilst temperatures also remained low, and dew formed on the leaves and grass each morning. Figure 2-2. Monthly mean temperature and mean daily minimum and mean daily maximum temperature in three vegetation types of grassland, woodland, and forest, from October 2007 to March 2008, Fongoli. 11 16 21 26 31 36 41 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 Mar-08 T em pe ra tu re ° C Grassland Monthly Mean Woodland Monthly Mean Forest Monthly Mean Mean daily minimum Mean daily maximum 17 Figure 2-3. Monthly mean relative humidity and mean daily minimum and mean daily maximum relative humidity in three vegetation types of grassland, woodland, and forest, from October 2007 to March 2008, Fongoli. An absolute range of wind speeds from 0 - 4.6 m/s, and gust speeds from 0 – 14.5 m/s were recorded. Mean wind and gust speeds increased from 0.07 m/s and 0.88 m/s in October to 0.29 m/s and 1.9 m/s in March (averaged across all loggers). Rodents several times chewed through the wind speed sensor wires, which lead to lost data in some vegetation types and months. At least one sensor recorded continuously in forest and in grassland for the duration of the study; however, both woodland sensors were severed and so lacked data for February to March. Mean wind speed is higher in grassland than woodland and forest from October to December (Figure 2-4). Mean wind speeds year round remained close to zero in forest, likely due to the closed vegetation, but increased in January in woodland and grassland (Figure 2-4). Mean daily maximum wind speeds increased suddenly in January in woodland and grassland. This dry season wind is called ‘the harmattan’ or ‘West African trade winds’ and blows south from the Sahara to Guinea. 0 10 20 30 40 50 60 70 80 90 100 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 Mar-08 R el at iv e hu m it iy ( % ) Grassland Monthly Mean Woodland Monthly Mean Forest Monthly Mean Mean daily minimum 18 Figure 2-4. Mean monthly mean and mean daily maximum wind speed in the vegetation types of grassland, woodland, and forest, from October 2007 to March 2008, Fongoli. Subjec ts The Fongoli chimpanzees range over a minimum area of 65 km2 [Pruetz and Bertolani 2009]. Pruetz and Bertolani [2009] reported differences in the behaviour of these dry habitat chimpanzees in comparison to forested communities; specifically, daytime party sizes are much higher, the community uses their home range cyclically across seasons and are more active during periods or times of day of low temperatures. The chimpanzees were fully habituated in 2005, but only males are followed as focal individuals, in order to protect females from possible stress during pregnancy and lactation. During the study period, the community numbered 33-34 individuals, following the birth of an infant female, and the death of an old male. Table 2-2 gives the age and sex composition of the community during the study period. For most of the study period, 27 individuals were of nest-building age. 0 1 2 3 4 5 6 7 8 Oct-07 Nov-07 Dec-07 Jan-08 Feb-08 Mar-08 M ax im um g us t s pe ed ( m /s ) Grassland Monthly Mean Woodland Monthly Mean Forest Monthly Mean Mean daily maximum 19 Table 2-2. Sex and age class composition of the Fongoli community during the study period. Adult Sub-adult Adolescent Juvenile Infant Total ♂ 10 1 2 5 2 20 ♀ 7 2 0 1 4 14 Total 17 3 2 6 6 34 Issa History and geography The Issa study area (05° 23.34 S 30° 35.04 E) was established by Hernandez-Aguilar [Hernandez- Aguilar 2006] in 2001 within the Ugalla region in western Tanzania. Research presence was absent from 2003-2005, but has been permanent since 2008. Research is focused within an 85km2 study area (Figure 2-5). Issa lies in the west of the Ugalla region, 81 km inland and East of Lake Tanganyika. Kano [1972] conducted several surveys across western Tanzania to determine the extent of chimpanzee distribution. He also defined several regions in the area, including Ugalla and Masito. These regions now form part of the Greater Mahale Ecosystem, and there is potentially chimpanzee habitat continuity from Mahale National Park to Ugalla (Figure 2-5). Although Ugalla is a term coined by Kano [1972] in the primatological literature, most of the Ugalla region falls within Tongwe East Forest Reserve. However, Issa is situated on ‘general land’, which has no official status and is currently unprotected. Ugalla is likely the easternmost distribution of chimpanzees in Africa and may represent the eastern most ecological edge [Massawe 1992]. Ugalla is a 3352 km2 region of broad valleys broken up by steep mountains and flat hilltop plateaus of 900-1800 m in elevation. Most streams are seasonal. Two permanent rivers, the Malagarasi to the North and the Ugalla to the west, form the boundaries of the region, and chimpanzees are not known to exist on the east side of the Ugalla [Massawe 1992]. 20 Figure 2-5. Map of western Tanzania. Issa Valley study area is highlighted in yellow. The two national parks of Gombe and Mahale Mountain are outlined in green on the lake shore. The colours of the image also indicate vegetation cover: brown woodland, green forest, and orange suitable nesting habitat calculated from GIS modelling [Pintea and Plumptre 2006]. Rivers in blue indicate possible barriers to chimpanzee movement, and the north and east boundaries of the Greater Mahale Ecosystem plus the town of Mpanda in the southeast. (Image contributed by Lilian Pintea, Jane Goodall Institute). 21 Vegetat ion The vegetation of the Ugalla region is miombo woodland, named for the dominant tree genera of Brachystegia and Julbernardia (Fabaceae). Hernandez-Aguilar [2006; 2009] described the vegetation of the Issa study area as: swamp, dry grassland, wooded grassland, woodland, gallery forest, thicket forest, and hill forest (Table 2-3). The first four vegetation types are “open” vegetation, which covers 98.5% of the study area, whilst the other three are broadly classified as “forest” and comprise only 1.5% [Hernandez-Aguilar 2009, Figure 2-6]. In this study and in previous research, the apes used only forest and woodland vegetation types for nesting. Hernandez-Aguilar [2009] measured >5000 nests within the Issa study area, 93% of which were built in woodland and 7% in forest. In both vegetation-types, nests were associated with steep slopes, although some also were built in valleys or on plateaus [Hernandez-Aguilar 2009]. All vegetation types occur in each topographic level, except for ‘swamp’, which is found only in broad open valleys. Figure 2-6. Miombo woodland and thin strip of forest in Ugalla [from Moyer et al. 2006]. 22 Table 2-3. Vegetation types in Issa [from Hernandez-Aguilar 2006; 2009], and terms used for comparison across sites in this study. This study Issa vegetation types description Forest Hill forest: Evergreen and semi-deciduous species growing on the edges of escarpments or following points of seasonal runoff on hillsides. Thicket ‘msitu’ forest: Evergreen and semi-deciduous vegetation, dominated by lianas and climbers. Gallery ‘kabamba’ forest: Evergreen forest with open understory, usually along seasonal water courses. Woodland Woodland: Deciduous trees and shrubs with grass understory and discontinuous canopy. Wooded grassland Wooded grassland: Dominated by grasses with isolated shrubs and trees. Bamboo Bamboo: Present in Ugalla, but not Issa. Dense patches of woodland or wooded grassland with bamboo understory. Grassland Dry grassland: Short grasses with isolated shrubs in broad valley lowlands or high plateaus, which are seasonally inundated. Swamp Swamp ‘mbuga’ grassland: Tall grasses up to 3m. Permanently inundated with few scattered trees or shrubs. Fauna The Ugalla region remains an intact ecosystem, but in recent years, the area has been under greater pressure from human disturbance. The greatest threats outlined in a conservation action plan for the region include agricultural expansion, cattle herding, fire, logging, and poaching. Snare poaching is most common, but large game such as buffalo (Syncerus caffer) or elephant (Loxodonta africana) are hunted with firearms. Elephant numbers have declined sharply over Tanzania in just the last five years, along with increased illegal ivory trade [Wasser et al. 2010]. Table 2-4 includes large mammal species that were seen or indirectly evidenced by Hernandez- Aguilar [2006] in 2001-2003 within the Issa study area. During the present study, some of the largest species were absent, including elephant, zebra (Equus burchelli), and eland (Taurotragus oryx). Further data were collected on mammal density (Piel and Stewart, unpublished data), in order to provide a baseline from which to measure future changes in mammal density. 23 Unlike Fongoli, Issa has a full complement of potential chimpanzee predators, including lion, leopard, hyena, and wild dog (Table 2-4). We twice heard a lion’s roar during the study period, and twice found faeces. We found evidence of leopards more often; vocalisations were heard on six occasions, and scat or footprints observed at least once a month. Dr. Jim Moore saw a melanic leopard just outside the study area during a visit to the site in May 2009 (Moore, pers. com.). We found hyena traces on several occasions, however vocalisations were not heard. Wild dogs are extremely rare, and were observed only once during the study period by a field assistant. Climate The Ugalla region is also described as one of the driest habitats where chimpanzees live [Kano 1972; Moore 1992]. The only long-term field study in Issa by Hernandez-Aguilar [2006] reported that mean daily maximum temperature was highest in August (34°C) and lowest in November (28°C). Mean daily minimum temperature was highest in January (17.2°C) and lowest in August (14.4°C; Hernandez-Aguilar, 2006; 2009). There is a rainy (October-April) and a dry season (May-September), with dry months defined as having <100 mm of rainfall. Rainfall averages <1000 mm per annum [Hernandez-Aguilar 2009]. Thus, although conditions are more extreme and dry than similar data from forested sites, Ugalla has milder conditions than south eastern Senegal [McGrew et al. 1981; Pruetz 2007]. I deployed a Hobo data-logging rain gauge (RG-3), which uses a tipping-mechanism to record rainfall continuously allowing the measurement of rain rates, times, duration and volume. I also deployed the same six assemblages of data loggers that were used in Fongoli to record temperature, relative humidity, and wind speed in six locations in representative vegetation types and topographic levels used for nesting within the Issa study area: Woodland plateau, slope, and valley, and forest plateau, slope, and valley. A second sensor (Hobo leaf wetness, S-LWA-M003) was also added to the weather station, which is a flat sensor that uses a capacitive grid to measure surface moisture; the logger records this as the % surface area (leaf) wetness. Although research began in October 2008, these data loggers were not deployed until the 8th January 2009 due to delayed arrival of their shipment to Tanzania, so data presented here are from February 2009 to January 2010. 24 Table 2-4. Large- and medium-sized mammal species recorded in Issa during the study period [edited from Hernandez-Aguilar 2006; 2009] (* observed once, - no evidence, bolded newly observed). Scientific name Common name Evidence Artiodactyla Alcelaphus lichtensteinii Lichtenstein's Hartebeest Observed Cephalophus mont i co la Blue duiker Observed Damal i s cus lunatus top i Topi Observed Hippotragus equinius Roan antelope Observed Hippotragus niger Sable antelope Observed* Oreotragus oreotragus Klipspringer Observed Ourebia ourebi Oribi - Phacochoerus aethiopicus Warthog Observed Potamochoerus porcus Bushpig Observed Redunca r edunca Bohor reedbuck Observed Rhynchotragus kirki Kirk’s dikdik Observed Sylvicapra grimmia Grey duiker Observed Syncerus caffer African buffalo Observed Taurotragus oryx Eland - Tragelaphus scriptus Bushbuck Observed Carnivora Canis mesomeles East African black-backed jackal Observed Civettictis civetta African civet Observed Crocuta crocuta Spotted hyena Faeces & foot prints Felis sylvestris African wild cat Observed* Genetta genetta Common genet Observed, carcass identified Herpestes ichneumon Lesser mongoose Observed Lycaon pictus East African wild dog Observed* Mellivora capensis East African honey badger Observed Panthera leo Lion Vocalisations & faeces Panthera pardus Leopard Vocalisations, faeces, & footprints ? Mongoose Observed Hyracoidia Heterohyrax brucei Bush hyrax Observed Dendrohyrax sp . Tree hyrax Vocalisations Perissodactyla Equus burchelli Zebra - Pholidota Manis temminckii Ground pangolin Observed* Primates Cercopithecus aethiops Vervet monkey Observed Cercopithecus ascanis Red-tailed monkey Observed Cercopithecus mitis Blue monkey Observed Galago senegalensis Senegal galago Vocalisations Otolemur c rass i caudatus Greater galago Observed Pan troglodytes Chimpanzee Observed Papio cynocephalus Yellow baboon Observed Proboscidea Loxodonta Africana African elephant - Rodentia Hystrix africae-australis Porcupine Observed Orycteropus afer Ardvark - 25 Data for the rain gauge were lost from October to December 2010 due to a technical fault, so I could not record a full annual cycle of rain fall. I made a temporary rain gauge, using a funnel and measuring cup, in order to record rainfall volume in the absence of the data logging rain gauge from October 2008 to January 2009. Rain volume recorded with this method may be less accurate than the rain gauge, but I combined these data to present monthly rainfall in Figure 2-7. The study period was an unusually rainy year with (possibly over-estimated) 1500mm of rainfall, which is much higher than the average rainfall/year [Hernandez-Aguilar 2009]. Figure 2-7. Monthly rainfall in Issa during the study period. An absolute range of 11–35°C temperature was recorded across an annual cycle in Issa, which is less extreme than conditions in the dry season in Fongoli. Based on mean temperatures across loggers, mean daily maximum temperature was lowest towards the end of the rainy season from February to April (24°C) and highest in October (29°C) at the end of the dry season, whilst mean daily minimum temperature was lowest mid-dry season in July (14°C) and highest in October (18°C). Mean daily mean and maximum temperatures were lower in forest than woodland, higher in valleys, and lowest on plateaus (Figure 2-8 & 2-9). However, mean daily minimum temperatures were similar in forest and woodland vegetation types, and while minimum temperatures were coldest on plateaus and valleys, mean daily minimum temperature seems to remain higher on slopes. Chimpanzees in Issa prefer to nest on slopes [Hernandez- Aguilar 2009] and higher minimum temperatures on slopes in Issa may influence this preference for slope-nesting. 0   50   100   150   200   250   300   350   400   450   500   To ta l  r ai nf al l  ( m m )   26 Figure 2-8. Woodland, Issa: monthly mean temperature and mean daily minimum and mean daily maximum temperature on topographic levels of valley, slope, and plateau. Figure 2-9. Forest, Issa: monthly mean temperature and mean daily minimum and mean daily maximum temperature in forest on topographic levels of valley, slope, and plateau. 13 15 17 19 21 23 25 27 29 31 T em pe ra tu re ° C Woodland Valley Monthly Mean Woodland Slope Monthly Mean Woodland Plateau Monthly Mean Mean Daily Maximum Mean Daily Minimum 13 15 17 19 21 23 25 27 29 31 T em pe ra tu re ° C Forest Valley Monthly Mean Forest Slope Monthly Mean Forest Plateau Monthly Mean Mean Daily Maximum Mean Daily Minimum 27 Relative humidity ranged from 0-100% and followed an inverse pattern through the dry season; mean relative humidity was lowest in September (50%), and highest in December (91%). Humidity was similar across vegetation types and topographic levels, but was higher in forest during the dry season and dropped lowest in woodland valley and forest slopes (Figure 2-10 & 2- 11). Mean daily maximum relative humidity was close to 100% during the wet season whilst temperatures were lower (Figure 2-10 & 2-11). Leaf wetness followed a similar pattern to relative humidity and rain fall, ranging from 0-100%. Mean leaf wetness was highest in the wet season from November to April (range: 21-44%) and lowest from May to October (range: 3-13%). An absolute range of wind speeds from 0 – 3.4 m/s, and gust speeds from 0 – 11.4 m/s were recorded. Mean wind and gust speeds were lowest during the rainy season and increased through the dry season from 0.03 m/s and 0.81 m/s in March/April to 0.17 m/s and 1.5 m/s in September (averaged across all loggers). In Issa, wind speed sensor wires were protected with metal conduit and no data were lost from rodents chewing the wires. However, in October the woodland slope logger recorded negligible wind speeds, and these data were excluded, as it is likely some error occurred, such as the turbine being prevented from turning by a branch. Mean wind speed was higher in woodland than forest, likely due to denser vegetation (Figure 2-12 & 2- 13). Mean daily mean and maximum wind speeds increased during the dry season and were highest in valley in woodland and in valley and slope in forest (Figure 2-12 & 2-13). Figure 2-10. Woodland, Issa: monthly mean relative humidity and mean daily minimum and mean daily maximum relative humidity on topographic levels of valley, slope, and plateau. 0 20 40 60 80 100 R el at iv e H um id it y % Woodland Valley Monthly Mean Woodland Slope Monthly Mean Woodland Plateau Monthly Mean Mean Daily Maximum Mean Daily Minimum 28 Figure 2-11. Forest, Issa: monthly mean relative humidity and mean daily minimum and mean daily maximum relative humidity on topographic levels of valley, slope, and plateau. Figure 2-12. Woodland, Issa: monthly mean and mean daily maximum wind speed in valley, slope, and plateau. 0 20 40 60 80 100 R el at iv e H um id it y % Forest Valley Monthly Mean Forest Slope Monthly Mean Forest Plateau Monthly Mean Mean Daily Maximum Mean Daily Minimum 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 W in d Sp ee d m /s Woodland Valley Monthly Mean Woodland Slope Monthly Mean Woodland Plateau Monthly Mean Mean Daily Maximum 29 Figure 2-13. Forest, Issa: monthly mean and mean daily maximum wind speed in forest valley, slope, and plateau. Subjec ts The Issa community of chimpanzees was unhabituated, so little information was garnered about community structure or ranging patterns. However, from genetic analyses conducted on faecal samples collected from beneath night nests, I have some information about the community. Preliminary genetic analyses were done as part of an independent study on the prevalence of the Simian Immuno-deficiency Virus (SIVcpz), which was 31% in this chimpanzee community [Rudicell et al. in press]. SIVcpz is known to cause AIDs like symptoms, decreased reproductive success, and increased mortality in the chimpanzees of Gombe [Keele et al. 2009]. Rudicell et al. [in press] conservatively identified 67 different individuals, 31 females and 27 males (while 9 individuals could not be sexed definitively), from 332 samples collected throughout the study area; 1-12 samples were collected per individual. I constructed a matrix of genetically identified individuals against temporally and geographically distinct associations, i.e. day follows and night nest groups. This matrix links 64 individuals, whilst the other three are geographically situated within the 64, suggesting that all 67 individuals are part of one community that ranges over a minimum home range of ~40km2 [Rudicell et al. in press]. However, the maximum home range estimate is much larger at ~478km2; calculated by dividing the minimum number of individuals in the community by the density of chimpanzees in the area 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 W in d Sp ee d m /s Forest Valley Monthly Mean Forest Slope Monthly Mean Forest Plateau Monthly Mean Mean Daily Maximum 30 (0.14/km2) [sensu Baldwin et al. 1982]. This analysis of community membership presumes that savanna chimpanzee community structure resembles that of forest chimpanzees, whereby communities are defined behaviourally and spatially, with members of a single community ranging in fluid sub-groups over geographically circumscribed areas [Herbinger et al. 2001; Williams et al. 2002]. DATA COLLECTION Data were collected in two periods: At Fongoli data were collected only during the dry season from October 2007 to March 2008, but at Issa data for a whole annual cycle from October 2008 to 2009. Nest data In both sites, only fresh night nest groups were selected for measurements. Night nests were focussed on, because these were built for long duration, overnight, in colder conditions, compared to day nests which were used for just short periods of daytime rest. Therefore night nests are expected to vary more with respect to function with climatic conditions. Nest groups were defined as all nests built on the same night within a 100m search radius. Previous studies of forest-dwelling apes have defined post hoc a nest group to span a 30m radius [e.g. Fruth 1995]; I expanded this due to the more open habitat which allows greater visibility. In Fongoli, focal individuals were followed to the nesting site, and nests were often seen or heard built before I returned to measure the following morning. In Issa, fresh nests groups were found opportunistically or after hearing vocalisations at night or in the early morning. Several nest groups were located through the aid of a concurrent study by A. Piel. He deployed an acoustic monitoring system across the area, which relayed all sounds to base-camp via radio. We were thus able to monitor these units from up to 8km distance and hear chimpanzees vocalising from their nests in the vicinity of the unit [Piel and Moore 2010]. Nests were judged to be freshly- built, from the night before, if the nest was seen built the previous night, or if fresh faeces or urine were found beneath the nest. In some cases nests, without fresh faeces or urine beneath were recorded as fresh if individuals were seen arising in the morning. Nests without faeces or urine were classed as ‘abandoned’ if nest-building was observed, but either the group or the individual then moved during the night. Other nests within the group without fresh faeces or urine were also measured (labelled ‘unknown’ in Table 2-5) and thought to be fresh, because of the state of the broken branches that cannot survive 24 hr in dry season conditions in Fongoli or Issa without the leaves withering. 31 Nest and nest tree character i s t i c s I recorded data on nest and nesting tree characteristics. These variables are listed below and were modified from Fruth, [1995], Hernandez-Aguilar [2006], Baldwin et al. [1979], Gibson [2005] and were developed in collaboration with Kat Koops (University of Cambridge) and Andrea Gibson (University of Zurich) for future comparative research. Variables were recorded for all nesting trees, but data presented here are from primary trees only (defined as trees contributing most branches to the nest structure). Data on nests and nesting trees were taken from a single place where possible, and I used a compass, clinometer, and measuring tape to map the relative positions of nests (these data will be entered into commercial software and resultant measures of inter-nest distance and vertical or horizontal group dispersion will be analysed at a later date). All height measures (numbers 7, 8 and 17-19 below) were taken with a clinometer (Suunto PM-5/360PC, range 0º +/-90º and 0% +/-150, precision 1º and 1%) to the nearest metre. All horizontal distances (numbers 9-11 and 20 below) were measured to the nearest metre using a tape measure. All diameters (numbers 15-16 below) were measured to the nearest cm using a metric diameter tape. Nest characteristics 1) Nest type Nest types were modified from Fruth [1995] with the addition of two other distinct types: 7. nests in the fork of main tree stems (branch crux); 8. nests made on a tangle of lianas with little or no tree support (Figure 2-14). Figure 2-14. Nest types (schematic): 0 – Ground nest, 1 – 1st fork, 2 – Outer fork, 3 – Tree top, 4 – Tree to tree (one or more) side branches, 5 – Tree top to (one or more) tree tops, 6 – Tree top to (one or more) tree side branches, 7 – Tree crux, 8 – Lianas. 32 2) Support type The type of branch structure supporting the nest was classified into six types described in Figure 2-15. Figure 2-15. Types of branch structure supporting nest. 3) Nest integration Nests were sometimes integrated using multiple trees as described above (Figure 2-14; 4-6), but in some cases lianas, shrubs, saplings, or grasses also were actively integrated, so nests were also categorised as Integrated or Non-integrated (with any plant materials). 4) Nest cover The percentage of canopy cover above the nest was estimated from the ground and later categorised as absent (0%), partial (<50%), or present (>50%). 5) Number of trees used for building Number of trees used for building was noted, and later classified into categories of Single or Multiple. 6) Number of decisions to nest location Number of decisions made to nesting location was counted. A decision was defined as a possible choice made if approaching the tree from the trunk base, e.g. a ground nest would be no decisions, or the nest in Figure 2-16 had seven decisions; one decision to climb the trunk and subsequent decisions made to bypass each branch to climb higher, or horizontally further out from the trunk. 33 7) Nest height. Nest height was measured to the nearest metre from the ground beneath the nest to the centre of the nest using a clinometer and a tape measure (Figure 2-16). 8) Nest height above the trunk base Nest height above the trunk base was recalculated post hoc by combining clinometer measurements to the primary tree trunk base and to the centre of the nest. This measure may differ from 7) Nest height, when the ground is sloping, and allows calculations to be made of the height of the nest within the tree crown (Figure 2-16). 9) Distance to trunk base The horizontal distance from the nest centre to the trunk base was measured to the nearest metre using measuring tape (Figure 2-16). Where the ground sloped, measurements were made parallel to the ground and the slope angle used to calculate the horizontal distance. 10) Distance to main stem The horizontal distance from the nest centre to the ground beneath the Main Stem (as defined above) was measured to the nearest metre using a tape measure (see number 9 above; Figure 2- 16). 11) Distance to tree crown edge The horizontal distance from the nest centre to the tree crown edge was measured to the nearest metre using a tape measure (following number 9 and 10 above; Figure 2-16). Nesting tree characteristics 12) Nesting tree species Local names of trees were recorded and identified from known tree species lists [Fongoli: Pruetz, unpublished data; Issa, Hernandez-Aguilar 2006]. In Issa unknown species were identified by Gabriel S. Laizier (Tanzania Botanical Research and Conservation Programme, National Herbarium of Tanzania, Arusha) and Henry J. Ndangalsi (Department of Botany, University of Dar es Salaam). Several species were differentiated in the field but remain unidentified (‘uid 1’ through ‘uid 16’). For each species I also measured mean leaf length and width, and relative branch strength, and bark/sap characteristics, however, these data are not yet analysed. 34 Figure 2-16. Measures of nest and tree characteristics; numbers correspond to descriptions of each measure - 7) nest height, 8) nest height above trunk base, 9) distance to trunk base, 10) distance to main stem, 11) distance to tree crown edge, 17) tree height, 18) tree crown base height, 19) tree crown height. 13) Tree branch morphology The orientation of the majority of tree branches was recorded as either Vertical (including inclined) or Horizontal, whilst branching pattern was recorded as Opposite or Alternate (see Figure 2-17). The canopy type was recorded as single or multiple, where several layers of canopy occur in one tree. 35 Figure 2-17. Tree branch morphology 14) Phenological stage Phenological stage of nesting trees was recorded as sterile, flowering, or fruiting. I recorded the percentage and stage of leafing (young, mature, or old leaves), percentage of flowers, and number of fruits (ripe and unripe). These data have not yet been analysed, but will in the future be used to address nesting tree selection. Here only the proportion of nesting trees with fruit (including both ripe or unripe fruit) were compared between sites. 15) Diameter at Breast Height (DBH) Diameter at Breast Height (1.5m) was calculated by summing the DBHs of all stems of a tree, which were measured to the nearest cm using a metric diameter tape. 16) Diameter of largest stem at breast height From the measures taken of each stem in number 15 above, I also recorded separately the DBH of the largest stem. This is not independent from number 15, but may allow comparison across sites where this measure alone has been recorded. 36 17) Tree height Tree height was measured from the base of the tree to the highest leaf (Figure 2-16). 18) Tree crown base height The tree crown base height was measured from the base of the tree to the lowest branch at the point where it forks from the trunk (Figure 2-16). 19) Tree crown height The height of the crown was calculated post hoc by subtracting the tree crown base height from the tree height (Figure 2-16). 20) Tree crown diameter The horizontal length of the crown was measured parallel to the ground along the longest length of the tree crown to the nearest metre using a tape measure. Where the ground sloped away, the distance was measured parallel to the ground and the angle of the tape recorded to later calculate the horizontal distance. 21) Tree crown area Crown area was derived from the crown diameter (length) multiplied by the crown width (measured horizontally at the midpoint of and perpendicular to crown diameter). 22) Number of main stems at breast height The number of stems at 1.5m was counted. 23) Number of main stems The number of main stems was counted; these were branches leading to the tree crown which fork below 1/3rd of the height of the tree. 24) Canopy cover Percentage of canopy cover was estimated to the nearest 5% and was defined as the proportion of the tree crown closed to the sky if the tree was in full leaf. 25) Leaf amount Leaf amount was estimated to the nearest 5% and was defined as the proportion of tree crown in leaf. 37 26) Canopy connectivity Percentage of canopy connectivity was calculated from the number of tree crown branches overlapping with another tree crown divided by the number of main branches in the tree crown. 27) Number of nests in tree The number of fresh nests (of the same nest group) in a tree was counted. Nest shape and archi tec ture In order to record in detail the shape and architecture of nests in Fongoli and Issa, I climbed up to a sub-sample of fresh nests using arborist tree-climbing equipment. In Fongoli, some low tree nests and ground nests were measured without climbing-equipment. Nests were selected according to their safe accessibility, by either free-climbing or using a combination of single-rope technique (SRT) and double-rope technique (DRT). Some inaccessible nests introduced a bias into the nest sample, but this bias should be similar across the study sites and climatic conditions. Houle et al. [2004] described SRT, which requires use of mechanical ascenders to climb a single line installed in the canopy, thus permitting rapid ascent and descent without damage to the tree. DRT, although slower, relies on the use of friction knots rather than mechanical ascenders and permits a greater degree of horizontal movement in the canopy. Figure 2-18 shows the use of these two methods. I used both methods together in order to efficiently access nests, depending on their position within the tree. During nest architectural measurements, I tried to deconstruct the nest in reverse order from that in which it was built, in order to parse the behaviour [sensu Byrne 1993; Byrne 2003; Byrne et al. 2001; Corp and Byrne 2002a; Corp and Byrne 2002b], using flow-charts describing the construction sequence. For each building-step the type of material, type of manipulation, and branch diameter (to the nearest cm using a metric diameter tape) was recorded as outlined below. Measurements were summarised into the variables described below for the purposes of this study; detailed data on nest building steps and sequence will be analysed at a future date. Nest shape After accessing nests I recorded detailed measurements of shape as described in Figure 2-19. Two hand-made metre sticks were used to take all measurements to the nearest cm. 38 Figure 2-18. Author using (a) DRT (Double Rope Technique): friction knots and attachment over branches allows greater horizontal movement (photo by Alex Piel), and (b) SRT (Single Rope Technique): mechanical ascenders and descenders allow rapid ascent (photo by Jim Moore). (a) (b) 39 Figure 19. The following variables were calculated post hoc from nest shape measurements: 1) Mean thickness Mean thickness of the nest was calculated by taking the average of central thickness, and thickness 1-4, (see (e) and (f) in Figure 2-19. 2) Nest misshapenness The range between the minimum and maximum radius (from (c) Figure 2-19) is taken as a measurement of nest misshapenness, as a perfectly circular nest would have a score of 0 if all radii are equal. 3) Nest circularity Nest circularity was calculated as a ratio, by dividing nest width by the nest length. A ratio closer to 1 indicates a circular nest, whilst a ratio closer to 0 indicates a more oval shaped nest. a b c1 c2 c3 c4 d e f1 f3 f2 f4 2-19. Nest shape measurements. a) Length: longest diameter of the nest; b) Width: perpendicular to length; c) Radii 1-4: length/width intersect to nest edge; d) Depth: vertical distance from central surface of nest bowl to the height of the nest edges, and Depth unsprung: many nests spring up when empty so I measured depth again after depressing the centre of the nest; e) Central thickness: vertical distance through the nest centre from the ventral to the dorsal surface of the nest bowl; f) Thickness 1-4: vertical distance through the nest 5cm in from the edge of the nest along each radius. 40 Nest architecture Detailed measurements of nest architecture were recorded by taking the nest to pieces and recording each step in reverse construction sequence (Figure 2-20). Exact measurements of all material involved in nest building were taken to the nearest 10th of a cm with a metric diameter tape at point indicated by the arrows in figure 2-20. Measurements are summarised below. The following variables were recorded: 1) Number of construction steps a. Total steps: total number of steps counted during deconstruction. b. Lining steps: number of steps of lining (including manipulation of lining material) c. Support steps: number of steps contributing to the nest structure and shape d. Mattress steps: number of non-lining steps contributing leafy material to the nest (total steps minus lining and support steps). 2) Lining pieces Lining was defined as any detached material placed on the surface of the nest and was always represented by the first building-steps deconstructed. a. Number of TW, LSB, SB, MB (see definitions of material type below) and diameters recorded. 3) Complexity: two possible indicators for nest complexity were recorded. a. Ratio of branches placed to branches interwoven. b. Time taken to deconstruct the nest. 4) Stem diameter (a, Figure 2-20) and distance to stem (b, Figure 2-20) The stem was defined as the source branch from which the nest support branch diverged. Measurements are depicted in Figure 2-20. Diameter was measured using a metric diameter tape, and the distance to stem with a tape measure. 5) Selected support (c, Figure 2-20) Selected support branches were defined as weight-bearing un-modified branches or branch formations on which the nest was built. a. Support type: recorded following definitions in nest characteristics 2) support type (Figure 2-15). 41 b. Sum support diameter: calculated from diameters of all selected support branches beneath the nest (depicted in Figure 2-20). c. Mean support diameter: calculated by dividing b, above, by d, below. d. Number of supporting branches: counted selected support branches beneath the nest. 6) Constructed support Constructed support branches were defined as the initial building-steps that provided structure and support to the nest, but contributed little leafy material. a. Main support diameter: recorded beneath the nest where most weight is likely borne. b. Sum support diameter: calculated from diameters of all constructed support branches beneath the nest. c. Mean support diameter: calculated by dividing b, above, by d, below. d. Number of supporting branches: counted all constructed support branches beneath the nest. Figure 2-20. Nest deconstruction and measures of a) stem diameter, b) distance to stem, c) selected support diameters and type (“horizontal Y-shaped branch”), d) bend count and diameter, e) break count and diameter. 42 7) Manipulation (d & e, Figure 2-20) For each step of the deconstruction (or reverse-building), sequence of the type of manipulation and material was recorded. a. Bend (BN): material remained attached and pliable and <50% broken. b. Break (BR): material was >50% severed. c. Detach (DET): material was completely severed. 8) Type of material manipulated Material was classified into the following branch categories and diameters were measured at the proximal end or point of return after the manipulation (see Figure 2-20). a. Main branches (MB): material providing additional side-branches that were also incorporated into the nest. - Number of MBs - Mean diameter of MBs b. Side branches (SB): all non-twig material incorporated into the nest from MBs. - Number of SBs - Mean diameter of SBs c. Lone side branches (LSB): all non-twig material incorporated into the nest but independently from any MBs. - Number of LSBs - Mean diameter of LSBs d. Twigs (TW): material terminating in leaves or leaflets and not any subsequent branching (as SBs or MBs) - Number of TWs - Mean diameter of TWs Nest data summary I found and measured 997 nests; 476 nests in 60 nest groups in Fongoli and 521 nests in 90 nest groups in Issa (Table 2-5). As this study focussed on night-nest building behaviour only two day- nests were recorded in Fongoli and all of the nests in Issa are assumed to be night-nests, as these were found usually in the morning after hearing chimpanzee vocalisations that morning or the evening before (the two day nests in Fongoli were excluded from analysis). Some cases of nest re-use were determined from the ground (6% of nests in Ugalla and 11% of nests in Fongoli), if fresh leaves had been added to a recent or old nest and fresh faeces or urine were found beneath 43 the nest. However, because the variables of nest characteristics reflect nest site and type selection within the tree, rather than structural variation, all fresh nests were included for analysis of nest characteristics. Abandoned and unknown nests from Fongoli and Issa were excluded from analyses, in order that nests from Fongoli and Issa were sampled in the same way, i.e. nests in Issa were more likely to be determined fresh only through faeces or urine, because nest-building could not be observed. In addition, ‘abandoned’ and ‘unknown’ nests from Fongoli were significantly lower in height (t = 5.9, n1 = 409, n2 = 63, p < 0.001), and more likely to be ground nests (χ2 = 100.4, n = 470, df = 16, p < 0.001), than those classed as definite ‘night nests’. Certain individuals, or age/sex classes, may be more likely to build lower night nests and not defecate or urinate before leaving. However, to be conservative, I excluded these abandoned nests from analyses of nest shape and architecture and assumed them not to have been built for overnight sleep, but rather to have been temporary resting platforms similar to day nests. Table 2-5. Nests found and measured at Fongoli and Issa. Night nests had fresh faeces or urine beneath or were seen built and slept in. Abandoned nests were seen built but left the night before, or discovered to have been abandoned during the night (e.g. if the group moved or the builder was seen to arise from another nest). Day nests were seen built for temporary use during the day. Unknown nests had no faeces or urine beneath, but due to un-withered leaves were likely either fresh or abandoned. Study site Nest status Total Abandoned Night Day Unknown Fongoli dry season 30 411 2 33 476 Issa wet season 0 204 0 4 208 Issa dry season 0 307 0 6 313 Total 30 922 2 43 997 I accessed 286 nests for shape and architectural measurements (Table 2-6). Re-used nests were excluded from analyses of nest shape and architecture, as these do not represent fresh or whole nest construction on the night before the nests were found. Five abandoned and eleven nests of unknown status were also excluded from the Fongoli nest sample, and one nest of unknown status was excluded from the Issa sample, in order to ensure that all nests were whole and freshly built for overnight sleep. However, due to time constraints, or if nests sprang apart rapidly, it was not always possible to take each measurement for each nest, thus samples sizes vary across 44 analyses depending on the measures recorded. Table 2-7 describes measurements taken for different sample sizes of nests following exclusion of the above nests. Table 2-6. Nest status and number of nests accessed for shape and architectural measurements. Study site Nest status Total Abandoned Fresh night Day Unknown Re-used night Fongoli dry season 5 146 2 11 9 173 Issa wet season 0 36 0 0 0 36 Issa dry season 0 70 0 1 6 77 Total 5 252 2 12 15 286 Table 2-7. Measurements taken of nests included in analyses. Study site Measurements of nests analysed Total Shape & all architecture Shape only Shape & main branches only Main branches only Fongoli dry season 113 27 3 3 146 Issa wet season 29 1 5 1 36 Issa dry season 70 0 0 0 70 Total 212 27 9 3 252 Vegetation plots At each site a randomly-selected 7km-long transect was walked and vegetation type and topographic level was recorded each 100m. Ten by ten metre plots were measured for different vegetation types in Fongoli and Issa (see below) and vegetation described, including understory species and count, canopy cover, and tree measurements. In Issa, an additional transect of 2km was walked through open valley to include vegetation types on this topographic level. A total of 150 plots were measured and described. In Fongoli, ten plots were done in each vegetation type described under study site Table 2-1 (Fongoli; forest, woodland, wooded grassland, bamboo, and grassland). In Issa, ten plots were done in each vegetation type, but also topographic level (where vegetation types existed): forest plateau, forest slope, forest valley, woodland plateau, woodland slope, woodland valley, wooded grassland plateau, wooded grassland slope, wooded grassland valley, and grassland plateau. There was no swamp, or valley grassland, or plateau grassland found on the two transects. Tree measurements were taken following the above section on nesting tree characteristics for all trees over 2m in height and 5cm DBH [which follows minimum 45 measurements of suitable nesting trees, from Fruth 1995]. These data provide a measure of available tree species, density, and characteristics, in addition to comparable measurements of vegetation types. Data on vegetation plots have not yet been entered or analysed, so could not be included in this thesis and will be analysed in the future. Phenology In Fongoli, all trees in the vegetation plots and an established transect of 2km length and 10m width including only feeding-tree species of >10cm DBH [Pruetz, unpublished data] were visited monthly to record phenology. Vegetation plot trees were included for phenology, because nesting trees were sometimes smaller than 10cm DBH and also included non-feeding tree species. In Issa, two stratified randomly placed 1-2km-long transects of 10m width including all trees over 2m height and 5cm DBH were established, one through open habitats and one in forest, and phenology recorded monthly. Four hundred trees were monitored at each site. Phenology was recorded using the following method [following Pruetz 2006]: % leaf amount (to nearest 5%) and age (young leaves, mature leaves, old leaves), number of fruits and age (unripe or ripe), % of flowers (to nearest 5% flower buds, mature flowers). Data on phenology have not yet been analysed, as it was not necessary for the topics included in this thesis and will be analysed in the future. Behaviour Behavioural data were collected only in Fongoli, as observations of chimpanzees in Issa were few and difficult. Only once was a chimpanzee seen in a nest at Issa, when a female and her juvenile were seen huddling in their nests in the rain into the morning. In Fongoli, behavioural data was collected only around night nesting and un-nesting times, evening and morning, because data were collected on fresh nest characteristics, shape, and architecture later in the mornings. With the use of GPS radios (Garmin Rino 530HCx) I located researchers with chimpanzees in the afternoon. As only males are followed as focal subjects I selected a focal male to follow to nesting from those present in the party on a switching focal basis, seeking to conduct a similar number of follows for each male. Data on females and juveniles were collected opportunistically. 46 Group night nest ing Group composition and number were recorded and counted every 15 min during daytime follows; the final scan providing a measure of nest group composition and number of individuals. Data were collected with the assistance of researchers or research assistants present. If night fell so that writing notes was difficult, data were recorded onto a voice recorder (Memory Stick ICD- MS1 Recorder) and later transcribed onto data-sheets. I remained at the nest site until after all sounds of nest building had ceased, and when observations could not be made I recorded whatever data were possible from sounds. The duration from the first branch broken to the cessation of all nest construction sounds was taken as the group nest building duration. I recorded the time from stop travelling to begin nest building, and the association and activities (feed/rest/groom) ad libitum in that period. I recorded the time and identity of the first and last builder in the group, whenever possible. At the nest site, all visible nest-builders were recorded opportunistically to note the identity or sex or age-class of nest-builders and the location of their nests (to be measured the next morning). In the morning I re-recorded party composition and number of individuals, identity and un-nesting time of first and last individuals, and the time that the group left the nest site. Some of these data are described further in Chapter 7. Indiv idual nest construct ion Initially this project aimed to follow focal individuals to nesting time. However, Fongoli chimpanzees so often nested after darkness had already fallen that I followed focal males only to ensure nest groups were not biased towards certain individuals or large group sizes. However, I switched focal data collection to the first individual to begin building in order to ensure some observations of how nests are constructed while enough light remained. Whenever possible, I recorded nest-construction on video. These data have not yet been analysed, but provide complementary information to that collected etho-archaeologically (on nest architecture). Methods will follow Byrne and colleagues [Byrne 1993; Byrne 2003; Byrne et al. 2001; Corp and Byrne 2002a; Corp and Byrne 2002b], who used video analysis techniques to separate complex food processing skills into detailed flow-charts of behaviour. Such flow-charts are analysed for statistical regularities and patterning in the behaviour. I aim to use a similar procedure to compare nest-building techniques across individuals with respect to sex, age, or kinship. I videotaped nest-building on 32 occasions, and these data will be analysed at a later date. For several nest-building events within the nest group I recorded whenever possible: builder identity, sex, or age class, construction start and end time, duration of construction, duration of use (if 47 abandoned), activity (rest, groom, feed, play) before building, activity in nest, resting position (supine, right-side, left-side, prone). These data are combined with data on nest characteristics, shape, and architecture to investigate nest variation with respect to builder sex, age, and identity in Chapter 7. Nocturnal observat ions In Chapter 4, I describe an experiment in which I slept out in the Fongoli chimpanzee range to investigate the functions of nest-building. During these sleep-outs, I opportunistically collected data on chimpanzee nocturnal behaviour, if chimpanzees were present. On three other occasions I stayed out with the chimpanzees late into the night and recorded their behaviour. These data were collected ad libitum and were limited as even in moonlight individual chimpanzees cannot be identified, but included vocalisations, movement, activity (feed, rest, travel), and nest-building. These data are not presented here, but indicate that Fongoli chimpanzees sometimes build multiple nests per night, nest for only few hours a night, after 22:00 hr, arise in the middle of the night to forage and sometimes to travel, and may forgo nest-building altogether. Twice entire nest-groups were abandoned during the night. I hypothesise that nights of greater activity are related to high daytime temperatures that prevent Fongoli chimpanzees from foraging enough during the day, or to full moonlight that affords greater night time visibility, or presence of oestrous females that males pursue at night when temperatures are cooler. Future study to investigate these behaviours more systematically could be with night-vision technology. Genetic sampling At the start of this project, genetic analyses were proposed to investigate variation in nest characteristics, shape, and architecture, with sex, individual identity, and relatedness. Shed hairs were collected from each nest accessed, and fresh faecal samples were collected from beneath fresh night nests. In Fongoli, 86 reference samples were collected from 26 known individuals of nest-building age, and in Issa samples also were collected from daytime encounters. In all, 484 samples were collected from Fongoli and 701 samples from Issa. Three collection techniques were used for sample storage in the field in Fongoli: Silica gel only, 100% ethanol, and a combination 2-step method [Nsubuga et al. 2004]. In Issa only 100% ethanol and Nsubuga et al.’s [2004] method were used as these were found in preliminary analyses of Fongoli samples to amplify products most successfully. Two previously published DNA extraction protocols were assessed for efficacy in extraction of amplifiable mitochondrial DNA: Qiagen stool extraction kit [Bradley et al. 2000] and a new 2CTAB/PCI method developed by Vallet et al. [2008]. The latter extraction 48 technique was found to be the most effective in producing DNA extracts useful for amplification of the HV1 region of mitochondrial DNA. Preliminary genetic analyses have focused on using reference samples, in order to begin building a bank of information of all the nest-building individuals in the community and formulating a method to rapidly screen and identify samples collected from nests. I have amplified the hyper variable region one (HV1) and two (HV2) of mitochondrial DNA, which reveals haplotypes that indicate maternal lineages [Gagneux et al. 1999; Goldberg and Ruvolo 1997]. Two of these haplotypes have not previously been found in wild populations of Pan troglodytes verus. Further pilot work has shown the efficacy of PCR (Polymerase Chain Reaction) amplification of amelogenin to identify sex [Bradley et al. 2001] and a multiplex PCR for genotyping individuals at variable microsatellite loci [Roeder et al. 2006], allowing genetic identification of the individual identity of unknown samples. Faecal and hair samples collected from below, or within, nests will then be assigned individual identities by conducting these analyses hierarchically to generate haplotypes, sex, and individual genotypes respectively, as necessary (Table 2-8). These analyses remain to be done, which I hope to complete in the future in order to address questions about nest construction, in addition to savanna chimpanzee population genetics. Table 2-8. Summary of genetic data already generated (HV1, HV2, and Amelogenin sexing tested) from reference faecal samples collected from 26 weaned individuals in Fongoli. In the future, in order to identify unknown samples collected from nests, the following hierarchical analyses, from step 1-4, will be conducted in order to rapidly identify nest-builders. Letters indicate hyper-variable region one and two (HV1 & HV2) haplotypes and sexes (M: male; F: female). Grey cells indicate analyses to be conducted, whilst unfilled cells indicate analyses unnecessary to identify to individual level. Step 1: HV1 type A B C D F G H I Step 2: HV2 type B F A E D F G H C E Step 3: Amelogenin sex M F F F M F M F F F M F M M Step 4: Microsatellite genotype ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ ↓ Individual BI KL TI NN NI LI BN JM SI FO LU LT BO YO TM NA DA DW DF LP KM FA FR DV MM MI 49 Insect trapping experiment This project proposed to measure biting insect density using two vector-specific traps deployed over five day periods, in rotating pairwise comparisons of different combinations of vegetation type and topographic level to investigate the influence of vectors on nest site selection. Moore (pers. comm.) hypothesised that lower densities of vectors may be found on slopes due to orographic winds, which may influence Issa chimpanzee’s preference to nest on slopes. In another study I sought to test whether freshly-made nests reduce biting insect density by running experiments comparing two traps deployed in the same tree, one surrounded by a ‘nest’ of broken branches, the other with no ‘nest’. Traps were deployed using attractants that mimic mammalian signals: granular media CO2, octenol, and a human skin lure. However, very few trials garnered valid results due to technological difficulties with the insect traps. For example, if one trap in the pairwise experiment stopped working during the experiment, the results had to be discarded. The experiment was discontinued due to these problems, but the influence of possible disease vectors was investigated through self-experimentation described in Chapter 4. ANALYSES Data included in analyses In the following chapters only some of the data described above is presented where relevant to the investigation of the function of nest construction. Data on nest characteristics, shape and architecture are described and presented in the following Chapter 3, but these data are also used further in specific hypothesis focussed Chapters 5-7. Some behavioural data is presented in Chapter 7 where relevant to variation in nest characteristics, shape, and architecture with sex, age, or identity. Statistical analyses All statistical analyses were conducted using Predictive Analytic Software for Windows (PASW). I tested normality of continuous data using Kolmogorov-Smirnov tests. Parametric tests included Multiple regression, Pearson’s correlation, t-test, and one way ANOVA. Where data were non-normally distributed (strong kurtosis or right skewed), square-root or log- transformations were used to normalise data. Where data could not be normalised, I used the non-parametric equivalent tests, such as Spearmans rank correlation, Mann-Whitney, Wilcoxon’s matched pairs, and Kruskal-Wallis tests. If multiple-comparisons were necessary to test two or more hypotheses with the same data, I used Bonferroni corrections to control for Type 1 error. 50 Results of tests that were significant prior to, but not after, Bonferroni correction was applied, were interpreted as non-significant ‘trends’. Categorical data were analysed using Chi-square tests. All correlations were two-tailed, and alpha was set at 0.05 throughout. 51 Chapter 3 Characteristics of nests and nesting trees, nest shape and architecture, in two chimpanzee populations, compared Above, tree nest from Issa. Below, ground nest from Fongoli. 52 INTRODUCTION Over 50 years of research into wild and captive chimpanzee behaviour, a vast data-set has amassed regarding the cultural nature of chimpanzees, which is surpassed only by the scope and variation of culture in humans [McGrew 2004; Whiten 2011]. For comparative study across species, culture is often defined as “a distinctive behaviour pattern shared by two or more individuals in a social unit, which persists over time and that new practitioners acquire in part through socially aided learning” [Fragaszy and Perry 2003, p. xiii]. This definition readily encompasses a single cultural behaviour in a range of species, but in chimpanzees, a multitude of behaviours differ across populations [Whiten et al. 1999]. However, determining whether or not social learning is operating in wild primates is difficult, and study of the patterns of behavioural variation in wild primates commonly determines behaviours as cultural through the ‘method of exclusion’, whereby genetic or environmental factors can be eliminated [sensu Whiten et al. 1999; 2001]. The debate over the extent to which patterns of behavioural variation in chimpanzees across their range can be attributed to culture over genetic, geographic, or local ecological influences continues [Kamilar and Marshack 2011; Langergraber et al. 2011; Lycett et al. 2007; Lycett et al. 2011]. This debate (and the use of a ‘method of exclusion’ to detect culture) has limitations, as culturally transmitted behaviours also are likely to be linked to their function, to the specific limitations and pressures of different environments, and to parallel genetic transmission. Thus, environmental influences on behaviour patterns do not preclude transmission of the behaviour through social learning [Laland and Janik 2006], and a correlation between genetic and behavioural similarities does not eliminate the likelihood that behaviours are culturally transmitted [Lycett et al. 2011]. In their seminal paper describing patterning of variation across long-term chimpanzee field sites, Whiten et al. [1999] disregarded universal behaviours like nest-building as the analysis focused on gross differences in presence or absence of behaviours. More recently, studies have begun to investigate nuanced variation within a behavioural pattern. For example, Humle and Matsuzawa [2002] found that the use of different lengths of ant-dipping tool and harvesting technique was related to the ferocity of the species of ant dipped. This correlation might suggest that ant-dipping is environmentally determined rather than culturally transmitted. However, even if a specific behaviour is socially learned, its use and technique may be linked to function and environmental constraints. A wider geographic analysis of the patterning of such nuanced variation in ant-dipping tools and technique suggests the behaviour is not determined by genetic or environmental variation, but most likely is socio-cultural [Schöning et al. 2008]. 53 Nest-building is ubiquitous across all the great apes, in all studied populations, and preliminary comparisons suggest much similarity in construction between species and populations [Fruth and Hohmann 1994a; 1996]. The original goal of this Chapter was to determine possible cultural variants through a detailed analysis of nuanced variation in nest- building, influenced and uninfluenced by the environmental conditions of two disparate chimpanzee study communities: Fongoli, Senegal, and Issa, Tanzania. However, controlling for all environmental influences is nearly impossible and beyond the scope of the study to date. Future comparisons incorporating more data on the environmental differences may be able to address this goal more thoroughly. I now aim in this Chapter to describe and compare the nest and nesting tree characteristics, nest shape and architecture, of chimpanzees in Fongoli and Issa, and between the wet and dry season in Issa. Analysis of seasonal variation in nests may suggest possible functions of nests. These data form the basis for further specific tests of nest function in this thesis, and provide a base-line for more detailed and controlled comparisons in the future. Most studies of nest building are done in order to improve great ape census techniques and most frequently entail counting nests and calculating numbers of individuals, by dividing nest density by the rate of nest decay [Ancrenaz et al. 2004; Furuichi et al. 2001; Hashimoto 1995; Kouakou et al. 2009; Mathewson et al. 2008; Plumptre and Reynolds 1996; Plumptre and Reynolds 1997; Tutin et al. 1995]. Comparisons of basic variables, such as nest height, age, and tree species used have yielded wider knowledge of factors influencing nest material and location selection. Chimpanzees are now known to be highly selective of tree species, tree morphology, nesting heights within trees, and locations on the landscape [Baldwin et al. 1981; Fruth and Hohmann 1996; Furuichi and Hashimoto 2004; Hernandez-Aguilar 2006; Hernandez-Aguilar 2009; Stanford and O'Malley 2008; Wrogemann 1992]. Site and tree selection is not addressed in this study, which Hernandez-Aguilar [2006] has studied in detail within the same community in Issa. Instead, I compared the characteristics of the trees and nesting locations that are presumed to be selected, between the two populations. Detailed descriptions of the behavioural patterns, techniques, and skills employed in nest-building are few [cf. Bolwig 1959; Fruth 1995; Goodall 1962; MacKinnon 1974; Schaller 1963]. Thus, I also aim to elucidate the skills entailed in making a nest by providing the first detailed description and comparison of chimpanzee nest shape and architecture across two communities. Further comparison of nest shape and architecture between the wet and dry season in Issa is hypothesized to reflect the thermoregulatory function of nests. Nests may therefore be larger, thicker, and built with more material in the wet season, in order to provide more insulation from more chilling conditions in the wet season in Issa. 54 METHODS Data collection methods in Fongoli and Issa are described in full in Chapter 2. Here I describe all nest data: Nest characteristics, nesting tree characteristics, nest shape, and nest architecture. Normal data were analysed using one-way analysis of variance in order to compare simultaneously across Fongoli dry season, Issa dry and Issa wet seasons, seeking to control for Type 1 error. Non-normal data were analysed using Kruskal-Wallis tests and Bonferroni corrected multiple Mann-Whitney comparisons. All significant differences (p < 0.05) reported in tables are highlighted in bold. Categorical data were analysed using Chi-squared tests. RESULTS Nest groups I measured 140 fresh night nest groups; 58 in Fongoli, 40 in Issa in the dry season, and 42 in Issa in the wet season. Group sizes were not larger in Fongoli (M = 6, range = 1-18), than Issa (M = 4.5, range = 1-26) in the dry season (Mann-Whitney, z = 1.59, p = 0.22), nor during the dry compared to the wet season (M = 2.5, range = 1-16) within Issa (Mann-Whitney, z = 1.07, p = 0.57). Nest characteristics Nest type Proportions of different nest types built or selected differed across Fongoli dry, Issa dry and Issa wet seasons (χ2 = 124.41, df = 16, n = 913, p < 0.001; Figure 3-1). There was no difference between seasons in the proportions of different nest types built in Issa (χ2 = 7.84, df = 6, n = 503, p = 0.25). The difference between Fongoli and Issa was due in part to the construction of overnight ground nests in Fongoli (12% of nests), which did not occur in Issa. The relationship remains significant if ground nests are excluded (χ2 = 47.12, df = 6, n = 856, p < 0.001). In Fongoli and Issa night nests were made most often on outer forks of side branches (34% and 60% respectively). However, these proportions were the greatest difference between nest types in Fongoli and Issa (Figure 3-1); in Fongoli most nests were of other types, whilst in Issa the most nests were built on outer forks of side branches. In Fongoli nests were built slightly more frequently in tree cruxes (7%), 1st forks (15%), and tree tops (13%) than in Issa (4%, 8%, 10% respectively). Similar proportions of integrated nests of types tree to tree side branches, tree top to tree top, and tree top to side branch were similar in Fongoli (5%, 4% and 8% respectively) and Issa (7%, 3%, and 8% respectively). 55 Figure 3-1. Nest types built in Fongoli and Issa (wet and dry seasons). Figure 3-2. Nests built on different types of support (branch configurations). 56 Support type Type of selected support beneath the nest differed between Fongoli and Issa dry and wet seasons (χ2 = 135.81, df = 10, n = 897, p < 0.001; Figure 3-2). In Fongoli the greatest proportion of nests were built on single horizontal branches (24%), or Y-shaped horizontal branches (21%); whereas in Issa a greater proportion of nests were built upon a branch-cup shaped support (wet: 38%; dry: 31%), fewer on Y-shaped vertical branches (wet: 8%; dry: 5%) and none on the ground. Proportions of different support types used did not differ between seasons in Issa (χ2 = 7.41, df = 4, n = 496, p = 0.12; Figure 3-2). Nest integrat ion In Fongoli, 34% of nests were integrated from multiple plants (including grass, saplings, shrubs, and trees) compared to 27% of nests in Issa in the dry season and 20% in Issa in the wet seasons (Figure 3-3). The proportion of nests integrated between Issa and Fongoli (dry seasons; χ2 = 3.47, n = 715, df = 1, p = 0.06) and between the wet and dry season in Issa did not differ (χ2 = 3.43, n = 508, df = 1, p = 0.06). In Fongoli, ground nests (53%) were more frequently integrated than tree nests (31%; χ2 = 9.20, n = 411, df = 1, p = 0.002). Ground nests (n = 49) were made of grass (24%), saplings (35%), shrubs (14%), lianas (33%), and trees (33%). In Fongoli, 96% of integrated arboreal nests included trees, 45% included lianas, 3% included shrubs, but none included saplings or grass (n = 139); in Issa, no nests were built with saplings, shrubs, or grass, but 57% of integrated nests included lianas and 71% included trees (n = 124). Number o f t rees used for bui lding The above measure of nest integration of different plant types may have included inadvertent integration of lianas or other material, so the proportion of nests built by integrating multiple trees was also analysed. In Fongoli, the number of trees used for building (excluding ground nests) ranged from 0-4 (M = 1) and in Issa ranged from 0-5 (M = 1; Figure 3-4). In Fongoli, 19% of nests integrated multiple trees, 18% in Issa in the dry season, and 20% in Issa in the wet season. Nests built with zero trees were built upon only a tangle of lianas, with no other support beneath. The proportion of tree nests integrated did not differ between sites or seasons (χ2 = 14.54, df = 10, p = 0.15). Most integrated nests in Fongoli and Issa were built with two trees (15% of nests), followed by three trees (2%), four trees (Fongoli: 0.2%; Issa: 0.6%), and five trees (Issa: 0.2%). 57 Figure 3-3. Nests integrated out of multiple sources of material in Fongoli in dry and in Issa in wet and dry season. Figure 3-4. Nests built using different numbers of trees in Fongoli in dry and in Issa in dry and wet season. 58 Nest cover A similar proportion of nests had cover overhead in Fongoli (full, 32%; partial, 32%) and Issa dry (full, 32%; partial, 26%) and wet seasons (full, 34%; partial, 30%; Figure 3-5). There was no difference in cover across sites or seasons (χ2 = 3.97, df = 4, n = 910, p = 0.41). Figure 3-5. Nests with full, partial, or no cover overhead in Fongoli in dry or Issa in dry and wet season. Nest pos i t ion The median number of potential decisions differed across sites and seasons (Kruskal-Wallis, H = 158, n = 910, df = 2, p < 0.001), with a greater number of decisions in Fongoli than Issa (Table 3-1). Median nest height was greater in Issa than Fongoli, and no different between seasons (Table 3-1; Figure 3-6 demonstrates the extent of this difference between the two sites). Most nests in Issa were built on slopes, so that nest height above the ground was on average 2m taller than the nest height above the trunk base (Wilcoxon’s matched pairs z = 9.25, n = 493, p < 0.001). Nests in Issa were built at a greater absolute distance from the tree trunk, main stem, and tree crown edge than those in Fongoli (Table 3-1), but this is due to the difference in tree size (see nesting tree characteristics below and comparisons controlling for tree size in Chapter 5); there was no difference between dry and wet seasons in Issa. Most nests in Fongoli were also built within the middle and lower third of the tree crown height, whilst in Issa most nests were built in the upper and middle third of the tree crown 59 (Figure 3-7; χ2 = 40.33, df = 2, p < 0.001). There is no difference in the proportion of nests built in upper, middle, or lower portions of the tree crown between seasons in Issa (χ2 = 1.04, df = 2, p = 0.60). Table 3-1. Nest position compared between Fongoli and Issa dry season, and between seasons in Issa. Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p Nest position M Range n M Range n M Range n Number of decisions (n) 4 0-16 408 6 2-14 202 <0.001 6 2-15 300 >0.05 Nest height [NH] (m) 5 0-16 409 13 2-43 304 <0.001 12 2-26 204 >0.05 NH above trunk base (m) 5 0-17 343 11 0-43 290 <0.001 11 2-24 203 >0.05 Nest - trunk base (m) 1 0-9 358 2 0-13 302 <0.001 2 0-14 203 >0.05 Nest - main stem (m) 1 0-10 339 2 0-12 299 <0.001 2 0-9 203 >0.05 Nest - tree crown edge (m) 1 0-6 327 2 0-8 297 <0.001 1 0-9 203 >0.05 p-values are Bonferroni corrected Mann-Whitney multiple comparisons Figure 3-6. Nest heights in Fongoli in dry and Issa in dry and wet season. 60 Figure 3-7. Nests built in the upper, middle, and bottom third of the tree crown in Fongoli in dry and Issa in dry and wet season. Nesting tree characteristics Nesting spec i es Trees are the primary plant type used for construction in both Fongoli and Issa. However, in Fongoli the primary plant material used to build 10% of nests is non-tree (i.e. shrub, liana, sapling, or grass). In Issa, although lianas are often included in construction, non-trees are the primary source in building only 0.2% of nests. In Issa 44 species of tree were used for nesting (Figure 3-8) and in Fongoli 37 species were used (Figure 3-9). In Issa, 10 were unidentified and 10 could be identified only to genus. The most frequently used species in Issa could not be differentiated: Brachystegia puberula, B. stipulate, and B. utilis. Julbernardia unijugata, an evergreen forest-growing tree, is therefore likely the most frequently used species in Issa. In Fongoli, 17 species were not identified, including 3 of the most frequently used species. (Unfortunately, I was unable to hire a botanist for this work and only feeding tree species have been identified by previous research [Pruetz 2006]). In Fongoli, the most frequently used tree species, Pterocarpus erinaceus and Hexalobus monopetalus, grow primarily in woodland and grassland. In both Fongoli and Issa, the top 10 most frequently used tree species represent a large proportion of all nesting trees (78% in Issa, 67% in Fongoli) and represent species growing in different vegetation types used for nesting (Figures 3-8 and 3-9). n = 98 n = 37 n = 46 n = 137 n = 67 n = 108 n = 51 n = 67 n = 106 0 10 20 30 40 50 60 70 80 90 100 Fongoli dry season Issa wet season Issa dry season Pe rc en ta ge o f n es ts (% ) upper crown middle crown bottom crown 61 Figure 3-8. Nesting tree species in Issa in vegetation types of woodland and forest (uid = unidentified). n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 2 n = 2 n = 2 n = 2 n = 2 n = 3 n = 3 n = 3 n = 3 n = 4 n = 4 n = 4 n = 4 n = 4 n = 5 n = 5 n = 6 n = 8 n = 8 n = 8 n = 11 n = 12 n = 15 n = 15 n = 21 n = 26 n = 46 n = 62 n = 69 n = 70 0 2 4 6 8 10 12 14 16 Albizia antunesiana Annona senegalensis Craibia grandiflora Croton sp. Diplorhynchus condylocarpon Drypetes sp. Ficus sp. Tarenna sp. uid1 uid17 uid4 uid6 uid7 Uvaria sp. Anisophyllea boehmii Englerophytum magalismontanum Trichilia sp. uid8 uid9 Brachystegia longifolia Strychnos sp. Thespesia garckeana uid3 Drypetes gerrardi Garcinia huillensis Pericopsis angolensis Pleurostylia africana Thecacoris lucida Cordia sp. uid5 Parinari curatellifolia Brachystegia microphylla Diospyros sp. Lannea schimperi Newtonia buchananii Pterocarpus tinctorius Isoberlinia tomentosa uid2 Salacia erecta Combretum molle Brachystegia spiciformis Brachystegia bussei Julbernardia unijugata Brachystegia puberula/stipulata/utilis % of nests Forest Woodland 62 Figure 3-9. Nesting tree species in Fongoli in vegetation types of grassland, woodland and forest (uid = unidentified, more remain uid because only feeding tree species have been identified in Fongoli). n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 1 n = 2 n = 2 n = 2 n = 2 n = 3 n = 3 n = 4 n = 4 n = 5 n = 5 n = 5 n = 5 n = 6 n = 6 n = 7 n = 8 n = 9 n = 13 n = 16 n = 16 n = 17 n = 19 n = 19 n = 19 n = 20 n = 20 n = 25 n = 26 n = 43 n = 50 0 2 4 6 8 10 12 14 Baissea multiflora Cordyla pinnata Tamarindus indica uid13 uid14 uid4 uid9 Adansonia digitata Spondias mombin uid7 Vitex madiensis Lannea microcarpa Terminalis sp. Ficus sp. uid5 Nauclea latifolia Parkia biglobosa uid3 uid5 uid1 uid10 uid11 Grewia lasiodiscus uid6 Allophylus africanus Cola cordifolia Diospyros mespiliformis Khaya senegalensis Afzelia africana Daniella oliveri Piliostigma thonningii Combretum molle uid8 uid15 uid12 Hexalobus monopetalus Pterocarpus erinaceus % of nests Forest Woodland Grassland 63 Phenolog i cal s tage A similar proportion of nesting trees were in fruit in Fongoli (12%) and Issa in the dry season (10%; χ2 = 1.4, df = 1, p = 0.24; Figure 3-10). Fewer nesting trees were in fruit in Issa in the wet (0.02%) than the dry season (χ2 = 19.4, df = 2, p < 0.001). Figure 3-10. Nests built in fruiting trees in Fongoli in dry and in Issa in dry and wet season. Tree s ize and morphology Nesting trees were larger in Issa than in Fongoli across all measures of size measured and described in Table 3-2. Nesting tree size did not differ between seasons in Issa (Table 3-2). Nesting trees in Fongoli had on average less crown cover, fewer leaves, but greater connectivity in the canopy than nesting trees in Issa (Table 3-3). Nesting trees in Fongoli were also often shrub-like, with many more stems than nesting trees in Issa. In Issa in the dry season nesting trees had more stems, less canopy connectivity, and slightly more leaves and crown cover than nesting trees in the wet season (Table 3-3). 64 Table 3-2. Nesting tree size compared between Fongoli and Issa dry season, and between seasons in Issa. Tree measurements Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p x ̄ M Range n x ̄ M Range n x ̄ M Range n DBH (cm) 21.6 21.8 2-354.8 296 28.8 28.9 3-134.9 218 <0.01 27.5 27.8 6.8-199.5 171 >0.05 DBH largest stem (cm) 15.4 15.2 2-354.8 297 26.2 26.4 3-134.9 218 <0.01 26.6 26.8 6.8-199.5 171 >0.05 Tree height (m) 9.2 8 1-27 302 16.8 15 4-53 217 <0.01 15.8 15 2-41 170 >0.05 Crown base height (m) 2.4 2 0-10 296 5.7 5 0-20 215 <0.01 6.0 6 0-20 170 >0.05 Tree crown height (m) 6.8 6 0-22 296 11 11 1-37 215 <0.01 9.8 9 1-36 170 >0.05 Tree crown diameter (m) 7.5 7 2-22 243 10.6 10 2-30 208 <0.01 10.0 10 2-26 165 >0.05 Bonferroni corrected p-values: Mann-Whitney multiple comparisons, cf. DBH analysed using ANOVA and post hoc comparisons Table 3-3. Nesting tree morphology compared between Fongoli and Issa dry season, and between seasons in Issa. Tree measurements Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p x ̄ M Range n x ̄ M Range n x ̄ M Range n Stems at 1.5m (n) 2.0 1 1-17 296 1.24 1 1-6 218 <0.01 1.1 1 1-2 171 <0.05 Main stems (n) 2.1 1 1-19 266 1.82 1 1-10 217 >0.05 1.4 1 1-5 168 <0.01 Canopy cover (%) 47.8 50 0-90 253 72.3 75 35-100 211 <0.01 64.0 70 5-90 168 <0.001 Leaf amount (%) 55.3 60 0-90 249 76.5 80 40-95 215 <0.01 73.1 75 20-90 155 <0.01 Canopy connectivity (%) 58.3 60 0-100 296 41.5 25 0-100 210 <0.01 45.4 40 0-100 168 >0.05 Bonferroni corrected p-values: Mann-Whitney multiple comparisons Nest shape Size and shape of nests did not differ overall between the two study sites (Table 3-4). Overall nests were oval shaped, with mean length about one fifth longer than nest width. However, in Fongoli, nests tended to be wider, despite no difference in circularity index. This may be accounted for by greater misshapenness of nests in Fongoli, where a circular or oval shape was therefore often less defined. The relationships described in Table 3-4 are presented as an overall comparison of nest shape between the two sites, however, 35 (25%) of the nests analysed in Fongoli were built on the ground. As this proportion is greater, and so not representative of the overall proportion of ground nests (12%) built in Fongoli during the study period, ground and tree nests were further 65 compared within-site and analysed separately across-sites. However, ground nesting itself represents a major difference between the two sites. Overall in Fongoli, ground nests are smaller, thinner and more oval in shape (width: n1 = 34, n2 = 109, t = 2.49, p = 0.014; central thickness: n1 = 34, n2 = 108, t = 6.99, p < 0.001; mean thickness: n1 = 34, n2 = 108, t = 5.59, p < 0.001; circularity: n1 = 34, n2 = 108, t = 2.27, p = 0.025). Table 3-4. Nest shape, including all nests measured, in Fongoli and Issa. Shape measurement Fongoli, dry season Issa overall test value p x ̄ M Range n x ̄ M Range n Length cm 88.9 88 38-138 143 85.6 87 36-135 105 t=1.59 >0.05 Width cm 71.1 70 32-112 143 68.0 68 34-102 105 t=1.72 0.087 Misshapenness cm 10.1 10 0-36 142 7.5 7 0-28 97 t=3.26 0.001 Circularity ratio 0.8 0.8 0.5-1.0 143 0.8 0.8 0.5-1.0 105 t=0.18 >0.05 Depth cm 7.9 8 0-30 138 8.7 7 0-31 96 z=0.52 >0.05 Depth un-sprung cm 15.2 16 0-40 127 14.9 15 0-32 94 t=0.27 >0.05 Central thickness cm 14.5 15 0-40 142 14.8 15 0-33 105 t=0.29 >0.05 Mean thickness cm 13.0 13 1-40 142 12.6 12.2 1.6-30.2 105 t=0.47 >0.05 In addition, 35 nests measured in Issa were built during the wet season, which are hypothesised to differ from nests built in the dry season. One-way ANOVA tests reveal differences between tree nests built in Fongoli in dry and Issa in dry and wet seasons in width (df = 2, 211; F = 4.8; p = 0.009), misshapenness (df = 2, 202; F = 7.3; p = 0.001), central thickness (df = 2, 210; F = 11.9; p < 0.001), and mean thickness (df = 2, 210; F = 10.8; p < 0.001). Circularity also differed across the two sites and seasons (Kruskal-Wallis test, H = 8.1, df = 2, n = 214, p = 0.018). Nest length and depth did not differ across sites or seasons (length, df = 2, 211; F = 1.6; p = 0.20; depth unsprung, df = 2, 184; F = 0.20; p = 0.81; depth, Kruskal-Wallis test, H = 4.67, df = 2, n = 200, p = 0.097). Post hoc comparisons between tree nests built in Fongoli in the dry and in Issa in the wet and dry season are reported in Table 3-5. Results show that tree nests in Fongoli tended to be wider, were more misshapen, and were thicker than nests built in Issa in the dry season. Tree nests built in Issa in the wet season were also rounder, and thicker than nests built in the dry season. Differences in nest thickness between Fongoli and Issa in the dry season disappear if ground nests are included in the analysis. 66 Table 3-5. Nest shape of tree nests built in Fongoli in dry and in Issa dry and wet season. Shape measurements Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p x ̄ M Range n x ̄ M Range n x ̄ M Range n Length cm 89.6 88 38-138 109 86.1 87.5 53-135 70 >0.05 84.8 83 36-115 35 >0.05 Width cm 72.8 73 32-112 109 66.2 65 45-98 70 0.008 71.7 70 34-102 35 >0.05 Misshapenness cm 11.4 10 0-36 108 7.4 6 0-20 68 0.001 10.5 9 0-28 29 0.062 Circularity ratio 0.8 0.8 0.5-1.0 109 0.8 0.8 0.5-1.0 70 >0.05 0.9 0.9 0.6-1.0 35 0.014 Depth cm 7.4 8 0-29 104 9.6 10 0-31 67 >0.05 6.6 0 0-27 29 >0.05 Depth un-sprung cm 15.6 16 0-40 93 15.0 15 0-32 65 >0.05 14.7 14 2-28 29 >0.05 Central thickness cm 16.9 17 1-40 108 12.7 13 0-29 70 0.001 19.0 18 5-33 35 0.001 Mean thickness cm 14.5 14 1-40 108 11.5 10.6 1.6-29.0 70 0.005 17.2 16 4.2-30.2 35 0.001 Nest Architecture Fongoli and Issa nests differed architecturally when all nests were compared (Table 3-6.). Fongoli nests tended to be built with fewer steps than Issa nests. This may reflect a greater number of lining and support steps in Issa, in addition to more lining pieces in Issa nests, as there was no difference in the number of mattress steps. However, Fongoli nests tended to be more complex, with a greater proportion of branches or twigs interwoven rather than placed during nest building. Architecturally, differences may be expected between the two sites, as Fongoli chimpanzees nest on the ground, which requires little structural support and uses different materials. Ground nests were built using trees as the primary source of material in just 35% of nests, whilst shrubs, saplings, lianas or grass were the primary source of material in 65% of nests. Ground nests in Fongoli were built with fewer steps overall, reflected in the number of lining and support steps, whilst the number of mattress building steps was similar (n1 = 28, n2 = 85; total steps: x ̄1 = 20.9, x ̄2 = 34.5, t = 3.46, p < 0.001; mattress steps: x ̄1 = 21.9, x ̄2 = 26.9, t = 1.63, p > 0.05; lining steps: M1 = 0, M2 = 2, z = 2.23, p = 0.026; support steps: M1 = 0, M2 = 4, z = 6.81, p < 0.001). Ground nests were also less complex, as they took less time to deconstruct and had a lower proportion of steps interwoven than placed (complexity time: M1 = 0.75 min , M2 = 1.25 min, n1 = 16, n2 = 52, t = 4.11, p < 0.001; complexity interwoven: M1 = 0.03, M2 = 0.11, n1 = 86, n2 = 26, z = 2.99, p = 0.003). If grass clumps, lianas, or saplings were the equivalent to main branches used in tree nest construction, then there was no difference in the amount of main building material used in ground and tree nest building (M1 = 2.5, M2 = 3, n1 = 28, n2 = 86, z = 1.02, p > 0.05). 67 Table 3-6. Variation in nest architecture between all nests deconstructed in Fongoli and Issa. Architectural measurement Fongoli, dry season Issa overall test value p x ̄ M Range n x ̄ M Range n Number of steps Total n 30.83 32 1-86 114 35.27 38 5-104 101 t=1.79 0.075 Lining n 4.52 2 0-52 114 6.71 5 0-29 103 z=3.56 0.001 Mattress n 25.65 26 0-68 113 26.41 28 0-74 100 t=0.39 >0.05 Support n 3.02 3 0-21 113 4.74 4.5 0-14 100 z=3.48 0.001 Lining pieces n 4.2 2 51 114 5.92 4 0-27 101 z=3.53 0.001 Complexity time hr 1.22 1.25 2.25 68 1.37 1.5 0.25-3 66 t=1.57 >0.05 Complexity weave ratio 0.10 0.10 0.46 112 0.09 0.06 0-0.35 99 z=1.92 0.055 Main material n 2.78 2 0-14 119 3.79 4 0-11 99 z=3.68 0.001 One-way ANOVA tests revealed differences across tree nests built in Fongoli and Issa in dry and wet seasons in complexity time (df = 2, 115; F = 5.0; p = 0.008), main branch diameter (df = 2, 163; F = 2.0; p = 0.025), side branch diameter (df = 2, 181; F = 28.9; p < 0.001), lone side branch diameter (df = 2, 171; F = 10.2; p < 0.001), number of lone side branches (df = 2, 184; F = 4.0; p = 0.020), number of twigs (df = 2, 179; F = 12.8; p < 0.001), sum of constructed support diameter (df = 2, 159; F = 3.15; p = 0.045), and mean constructed support diameter (df = 2, 159; F = 11.49; p < 0.001). Post hoc comparisons and Bonferroni corrected multiple Mann- Whitney tests of non-normal variables revealed some similarities and differences between Fongoli and Issa dry seasons and between seasons within Issa (Tables 3-7, 3-8, and 3-9). Tree nests built in the dry season in both Fongoli and Issa were architecturally similar (Tables 3-7, 3-8, and 3-9). The median number of building steps ranged from 35-36, of which a median of 2-4 were lining, 4-6 were support steps and were similarly complex as nests took on average 1 hr and 15 min to deconstruct (Table 3-7). However, nests in Fongoli were more complex in terms of the proportion of steps interwoven. Diameter and number of branches forming the selected support location, and the number of branches manipulated for support were also similar across sites (Table 3-8). There was also no difference in the number of breaks, bends or detachments per nest (Table 3-9). Across sites tree nests were built with 18-19 bent, 6-9 broken, and 6-7 detached pieces of material. Branches of different diameter were also bent, broken, or detached with similar frequency in nest construction. On average the mean diameter of all manipulated material in a nest was ~1 cm across both sites. The frequency of use and diameter of different types of material did, however, differ between sites (Table 3-9). Nests in Issa were built with more main branches, side-branches, and lone side-branches of smaller 68 diameter, whereas nests in Fongoli were built with a greater number of twigs. Mean diameter of branches used in support was greater in Fongoli, and although a one-way ANOVA of the sum of constructed support branch diameters was significant (see above), post hoc comparisons were not significant (Table 3-8). Nests built in the wet season in Issa differed from those built in the dry season. Wet season nests were more complex in terms of time taken to deconstruct, were built with larger diameter support branches, and had a greater number of support steps, breaks, bends >3cm diameter, breaks between 1 and 3cm diameter, and twigs (Table 3-7, 3-8, and 3-9). Table 3-7. Nest architecture between tree nests built in Fongoli and Issa dry and wet season. Architectural measurements Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p x ̄ M Range n x ̄ M Range n x ̄ M Range n Number of steps Total n 34.5 35 7-86 86 33.42 36 5-104 70 >0.05 39.64 46 10-90 31 >0.05 Lining n 5.44 2 0-52 86 5.90 4 0-26 70 >0.05 8.42 7 0-29 33 >0.05 Mattress n 26.89 27 0-68 85 25.70 26.50 0-74 70 >0.05 28.07 31.5 5-55 30 >0.05 Support n 3.76 4 0-9 85 4.16 4 0-14 70 >0.05 6.10 6 1-13 30 0.01 Complexity Time hr 1.35 1.25 0.25-2.25 52 1.29 1.25 0.25-2.25 57 >0.05 1.88 2 1.40-3 9 0.006 Weave ratio 0.12 0.11 0-0.46 86 0.08 0.06 0-0.35 70 0.01 0.10 0.06 0-0.35 29 >0.05 Table 3-8. Nest support between tree nests built in Fongoli and Issa dry and wet season. Architectural measurements Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p x ̄ M Range n x ̄ M Range n x ̄ M Range n Constructed support Sum diameter cm 7.0 6.9 2.1-20.8 70 6.1 6.7 1.9-10.0 70 >0.05 7.2 7.7 1.4-10.9 29 >0.05 Mean diameter cm 1.9 1.8 0.8-5.8 70 1.6 1.6 0.6-3.3 70 0.001 2.0 2.2 1.1-2.7 29 0.001 Branches n 3.3 3 1-4 70 3.2 4 0-4 70 >0.05 3.2 3.5 1-4 29 >0.05 Selected support Sum diameter cm 11.5 11.6 1.7-59.5 81 10.5 1.0 2.5-50.8 70 >0.05 10.4 1.0 3.5-38.6 29 >0.05 Mean diameter cm 4.6 4.6 1.1-29.8 81 4.6 4.1 1.1-23.2 70 >0.05 5.4 5.8 0.9-13.0 29 >0.05 Branches n 2.7 3 1-4 81 2.6 2 1-4 70 >0.05 2.2 2 1-4 29 >0.05 69 Table 3-9. Nest architecture between tree nests built in Fongoli in dry and Issa dry and wet season. Architectural measurements Fongoli, dry Issa, dry Across s i t e s p Issa, wet Across s easons p x ̄ M Range n x ̄ M Range n x ̄ M Range n Bends n 17.6 18 2-51 86 18.6 20.5 2-51 70 >0.05 21.4 23 3-50 29 >0.05 Breaks n 10.5 9 0-35 86 8.1 6 0-29 70 >0.05 11.3 12 2-26 29 0.05 Detachments n 11.0 6 0-70 86 8.7 7 0-31 70 >0.05 11.7 10 1-31 29 >0.05 Bends <1cm n 10.1 8 0-34 86 12.7 11.5 0-37 70 >0.05 12.9 12 0-34 29 >0.05 1-2cm n 6.5 6 0-17 86 5.6 5 0-16 70 >0.05 6.9 6 1-18 29 >0.05 2-3cm n 1.6 1 0-7 86 1.2 1 0-4 70 >0.05 1.7 1 0-6 29 >0.05 >3cm n 0.4 0 0-3 86 0.3 0 0-3 70 >0.05 0.8 0 0-4 29 0.05 Breaks <1cm n 4.2 3 0-17 86 3.5 3 0-23 70 >0.05 4.1 3 0-13 29 >0.05 1-2cm n 4.0 3 0-17 86 3.2 2 0-14 70 >0.05 4.7 4 0-13 29 0.05 2-3cm n 1.4 1 0-10 86 1.0 1 0-6 70 >0.05 1.7 2 0-5 29 0.05 >3cm n 0.6 0 0-5 86 0.4 0 0-4 70 >0.05 0.8 0 0-4 29 >0.05 Detachments <1cm n 8.3 4 0-61 86 7.0 5.5 0-24 70 >0.05 9.4 8 1-31 29 >0.05 1-2cm n 2.1 1 0-19 86 1.6 1 0-19 70 >0.05 1.8 1 0-8 29 >0.05 2-3cm n 0.2 0 0-5 86 0.2 0 0-2 70 >0.05 0.2 0 0-2 29 >0.05 >3cm n 0.1 0 0-2 86 0.1 0 0-3 70 >0.05 0.2 0 0-4 29 >0.05 Mean diameter All material cm 1.0 1.0 0.5-2.3 86 1.0 0.9 0.6-1.6 67 >0.05 1.1 1.1 0.4-1.5 29 >0.05 Main branches cm 2.6 2.5 1.1-4.7 76 2.3 2.2 1.3-3.7 61 0.049 2.6 2.7 1.2-3.8 29 >0.05 Lone side branches cm 1.5 1.5 0.5-7.5 81 1.2 1.1 0.7-2.4 69 <0.001 1.3 1.3 0.7-2.5 25 >0.05 Side-branches cm 1.4 1.4 0.5-4.1 86 1.00 1.0 0.5-1.9 70 <0.001 1.1 1.1 0.6-1.6 29 >0.05 Number of Main branches n 2.7 3 0-7 89 3.6 3 0-11 70 0.05 4.3 4 1-9 35 >0.05 Lone side branches n 6.0 6 0-25 86 8.4 7 1-27 70 0.021 7.9 6 0.99 31 >0.05 Side-branches n 6.1 5 0-23 86 11.5 10.5 0-53 70 <0.001 12.3 12 0-32 32 >0.05 Twigs n 15.7 15.5 2-44 86 9.7 9 0-33 67 <0.001 13.7 14 1-55 29 0.042 Leaves or Leaflets n 5.0 0.5 0-59 86 2.0 0 0-24 67 >0.05 1.7 0 0-10 31 >0.05 70 DISCUSSION Overall differences and similarities Nest character i s t i c s Group sizes were similar between the two sites compared here, Fongoli and Issa, and also fall within the range of published night nest party sizes for other sites: Median night nest party sizes range from 2-5 across several chimpanzee populations studied [Fruth and Hohmann 1996]. Fruth and Hohmann [1996] found no pattern of variation in group size across chimpanzee sub- species, but in an earlier cross-species comparison, they reported that bonobos seem to aggregate in larger night nest than day parties, whilst chimpanzees split up at night [Fruth and Hohmann 1994a]. The two populations studied here live in habitats characterised as dry, open, hot savanna woodlands. Chimpanzees living in these habitats are hypothesised to differ in important ways from forest-dwelling chimpanzees [Moore 1992; 1996]. One hypothesis is that savanna chimpanzees may aggregate into larger night nest groups in order to maintain social relationships, whilst fissioning to forage during the day [Moore 1996]. Pruetz and Bertolani [2009] reported diurnal party sizes in Fongoli of 18 in the wet season and 12 in the dry season, which is more than twice as large as the night nest party size found here. Thus, these savanna chimpanzees appear to split up at night, at least into nest clusters more than 100m apart, despite ranging more cohesively and using their large home range cyclically together [Pruetz and Bertolani 2009]. It would be necessary to compare day to night nest party sizes collected on the same days to address this question thoroughly, but these data are not yet available for Fongoli or Issa. Presence of predators may also be expected to influence grouping patterns, as reported for Tai chimpanzees [Anderson et al. 2002], and this is discussed further in Chapter 5. Nest type The first step in detailed study of tools or architecture often involves categorisation of subjects into type. Few studies have reported nest type for great apes, and those that do differ in the typology used to characterise nests. Kano [1979] classified nests into five categories, which were also used by Fruth [1995]: side-branch, tree-top, two or more tree to tree-tops combined, two or more side-branches and tree-tops combined, and two or more side-branches combined. The categories used in this study were similar, although side-branch was split into two categories of 1st fork, outer fork and tree crux, and other types of ground and nests built wholly from lianas were included. Despite differences in typology used, I am able to make some comparisons between chimpanzee nests in this study, and bonobo nests reported by Fruth [1995]. 71 Fruth [1995] found that bonobos in Lomako built night nests on side branches 37% of the time, and 19% in the top of a tree. Proportions of nest types built in Fongoli were similar to those of bonobos in Lomako with 34% built on side branches and 13% in the tops of trees. Fruth [1995] did not differentiate between nests built on the 1st fork compared to outer forks of side-branches, but if 1st fork nests are included, then 49% of nests in Fongoli were built on side- branches, compared to 68% in Issa. In Issa overall, more nests (60%) were built on side- branches and only 10% were built in the tops of trees. Wrogemann [1992] found that chimpanzees in Lope (Gabon), most frequently built nests on side-branches (78.4%). However, she did not differentiate between nests of side-branch type alone, or those incorporating material from other trees. If the same categories are used here, then the proportions of nest types in Issa are similar to those in Lope, as a total of 83% of nests are classed as side-branch type. Use of these nest types may reflect selection of a certain nesting height within the tree. For example, in Lomako most nests (45%) were built within the middle third of the tree crown [Fruth 1995], which is similar to the proportions built in this position within the crown in Fongoli (48%), whereas in Issa, a greater proportion of nests were built in the top of the tree (40% overall). Given that Issa chimpanzees sleep higher within the tree crown than Fongoli chimpanzees, it is surprising that few nests are of the tree-top type. Thus, although nests are most frequently built on outer forks of side-branches, nests of this type may be at any level of the tree crown. Support type I found no previous studies that described the type of foundational support used to construct nests. However, the substrate used for building often exerts the most influence over variability in the architecture of other animals as the repertoire of building behaviours or actions used to ‘get started’ on a structure necessarily must be more flexible than the actions used to build upon an existing structure [Hansell 2005]. All branch configurations of support used in Fongoli are also used in Issa (cf. ground), which suggests that places where nests can be built within a tree must fit some criteria of weight- bearing support. Data were not collected on the availability of each of these types of supporting branch configurations within trees, so I cannot say if some types are selected over others. However, branch cup supports were more frequently used for nest building in Issa than in Fongoli. This also may be influenced by selection of locations of nests higher within the tree crown, or more peripheral along the horizontal axis of the tree crown, if branch cups are more frequent in these positions within the tree. 72 As Fruth [1995] noted in her detailed study of nesting in bonobos, social factors also may influence the selection of specific nest sites within trees. The support used for nesting may influence the architecture of a nest and reveal flexibility and variation. This hypothesis is not tested here, but data provide a baseline for investigations of flexibility in chimpanzee nest building in the future. Nest integration and number of trees used Fruth [1995] also hypothesised that if specific nest spots are a limited resource bonobos may integrate more trees in their nest building, in order to select specific sleeping spots in close proximity to other individuals. She found that 40% of bonobo night nests were integrated in Lomako, which is similar to the proportion reported by Kano [Kano 1992] for Wamba (37%). This is much higher than has been previously reported for chimpanzees, and Fruth [1995] suggested that this difference could be due to the greater sociality of bonobos. Wrogemann [1992] found that fewer than 10% of chimpanzee nests in Lope incorporated more than one tree, and no nest integrated more than two. Fruth and Hohmann [1996] reported questionnaire results from a number of chimpanzee study sites and most nests (90-95%) were reported to be built within single trees, and the maximum number of integrated trees was most often two. In this study, the proportion of integrated nests was higher and the maximum number of trees used was five in Issa and four in Fongoli. Hernandez-Aguilar [2006] reported that only 5% of nests were integrated, with up to a maximum of three trees, in Issa. However, she included nests of all stages of decay in her sample, and nests integrating multiple trees may spring apart and decompose faster than non-integrated nests [Fruth 1995]. McGrew reported in Fruth and Hohmann [1996] that up to four trees were used for nest-building in Assirik. An alternative hypothesis to explain integrating multiple trees could be as a solution to low availability of building materials: twice the proportion of nests are integrated in the dry habitats of Fongoli and Issa compared to the Lope rainforest habitat. In this study, the total proportion of integrated nests was high in both Fongoli and Issa, but I included all raw materials used in nest integration. Ground nests likely account for the tendency for more integrated nests in Fongoli than Issa, which is not significantly different when only the number of trees used is compared. Cover Most nests (68%) in Fongoli and Issa had no or only partial cover overhead. This proportion of open nests mirrors that reported from other dry sites, e.g. 70% in Assirik [Baldwin et al. 1981]. 73 However, if only completely open nests are counted, then the proportion is lower (ca. 33%). In Fongoli, a lower proportion of nests may be open compared to Assirik, as most nests are built in the middle or lower crown in Fongoli. However, in Issa a greater proportion of nests are built in the upper crown, but yet have partial or complete cover overhead. The amount of cover is likely influenced by more open vegetation in these drier sites; in the forests of Equatorial Guinea only 17% [Baldwin et al. 1981], and in Lope only 28% [Wrogemann 1992], of nests were open. A roughly equal proportion of nests built uncovered, under partial cover, or under complete cover in Fongoli and Issa may indicate that chimpanzees did not select nesting positions based on amount of cover overhead. Nest position Nest heights have been recorded for a number of chimpanzee populations (Table 3-10). It is at first unsurprising that nests are built higher in Issa than in Fongoli, as the trees are significantly taller in Issa. However, when the relative position of the nest within the tree crown is taken into account, chimpanzees in Issa are nesting higher even within the crowns of taller trees in Issa. Baldwin et al. [1981] found that chimpanzees in Assirik nested higher than chimpanzees in Equatorial Guinea, where trees were larger and proposed that nesting higher within the tree crown may be an anti-predator strategy (see Chapter 5). However, a previous study of the same community of chimpanzees in Issa found that most nests were built in the middle and lower crowns, which conflicts with the results of this study [Hernandez-Aguilar 2006]. However, as Hernandez-Aguilar [2006] measured nests of all stages of decay, it is possible that differential decay rates of nests built at different heights within trees influenced this result. A detailed study of orangutan nest decay found that several factors including nest height and tree species influenced the rate of nest decay [Mathewson et al. 2008]. Similarly, the number of decisions was greater in Issa than in Fongoli and likely reflects the difference in tree size. This contrast suggests that chimpanzees in Issa are not nesting in the first available locations within trees, but instead higher and further out in the tree crown. This is congruent with the difference in the distance of the nest to the trunk and to the main stem between Fongoli and Issa. In order to compare nest building across populations of chimpanzees, one must control for differences in tree size and morphology. Nest ing tree character i s t i c s Chimpanzees, and other great apes, are selective in the tree species used for nesting [Baldwin et al. 1981; Fruth and Hohmann 1996; Furuichi and Hashimoto 2004; Hernandez-Aguilar 2006; 74 Hernandez-Aguilar 2009; Stanford and O'Malley 2008; Wrogemann 1992]. The same is likely in both Fongoli and Issa, as although 37-44 tree species were used for nesting, most nests are built in only a few species. Data on available tree species are not yet analysed in this study, but in Assirik (nearby to Fongoli) and in Issa researchers reported selectivity in the use of nesting species [Baldwin 1979; Hernandez-Aguilar 2006]. Previous studies of chimpanzees and bonobos noted that night nests are rarely built within ripe fruit trees [Brownlow et al. 2001; Furuichi and Hashimoto 2004]. Fongoli and Issa chimpanzees were similarly found to construct nests in fruiting trees only ~10% of the time. Table 3-10. Mean nest and tree heights across chimpanzee study sites. Study site Tree height (m) Nest height (m) Proportion of tree height (ratio) Source Assirik, Senegal 14.1 11.3 0.80 Baldwin [1979] Assirik 14.5 13.6 0.94 Pruetz et al. [2008] Fongoli, Senegal 11.1 8.3 0.75 Pruetz et al. [2008] Semliki, Uganda 14 11 0.79 Hunt and McGrew [2002] Issa, Tanzania 17.7 13.4 0.76 Hernandez-Aguilar [2006] Fongoli, Senegal 9.2 4.9 0.53 This study Issa 16.7 13.1 0.78 This study Many of the differences in nest characteristics may be attributed to the difference in selected nesting tree size between the two sites. Trees in Issa are almost twice as tall as those in Fongoli, whilst measurements of nest position are also over twice as large in Issa compared to Fongoli. (DBH of nesting trees in Fongoli is only a third smaller than Issa nesting tree DBH, which may be due to the use of baobab, Adansonia digitata, for nesting in Fongoli; a tree that has an unusually large circumference). Few previous studies of nest characteristics in chimpanzees have also reported measurements of tree size. Table 3-10 compares nesting and nesting tree height across several savanna chimpanzee research sites and studies. My results in Fongoli differ from those of Pruetz et al. [2008], perhaps because I collected data only in the dry season, or because we used different data collection methods; they estimated nest heights rather than measuring them directly and included nests of different stages of decay. Pruetz et al. [2008] also found differences between the findings of Baldwin [1979] in the same region, which they attributed to a possible difference in predator prevalence. The nest heights of chimpanzees in these dry sites fall within 75 the range of 10-20m reported in a previous cross-community comparison [Fruth and Hohmann 1996] with the exception of Fongoli (5-8 m). However, trees used for nesting in Fongoli are also shorter than reported elsewhere (Table 3-10), despite having taller available trees than Assirik [Pruetz et al. 2008]. Comparisons of nest characteristics across sites must also therefore control for differences in tree size (see Chapter 5) and available tree size. Nest shape Overall nests were very similar in shape between Issa and Fongoli; however, this comparison includes a large proportion of ground nests in Fongoli and tree nests from both the wet and dry season in Issa. That nests, overall, were similar suggests that other factors may influence differences found when a smaller subset of only tree-nests built in the dry seasons is compared. For example, dry season conditions in Fongoli and Issa may not be comparable, given the differences in climate; e.g. in Fongoli temperatures were more variable, reaching higher maxima and minima than Issa dry season temperatures. Few measurements of body size are available for wild chimpanzees; however, the available data suggest that eastern chimpanzees (P. t. schweinfurthii) are the smallest subspecies of chimpanzee [Morbeck and Zihlman 1989]. Nests did not differ in length between seasons in Issa, across the two communities, or between ground and tree nests in Fongoli. Nests varied in width, with larger tree nests built by Fongoli chimpanzees, which may be consequence of larger body size. Ground nests need not support the whole weight of the builder and so do not need to be as large, which may have contributed to the lack of difference overall across all nests. Alternatively, width may increase as a result of other sources of variation, for example inclusion of more material to make a thicker nest could also increase the width of the nest as a by-product. Few studies have measured the shape of great ape nests. Baldwin [1979] estimated the size of some nests in Assirik and reported estimates similar to the measurements reported here; length (mean 80 cm, range 20-180, n = 155), width (mean 60cm, range = 20-110, n = 157) and depth (mean 40 cm, range 20-80, n = 155). Although Baldwin [1979] included the nest thickness in estimating the depth; depth as measured here cannot generally be seen from the ground. Fruth [1995] described the shape and architecture of bonobo nests, and Rayadin and Saitoh [2009] measured the size of orangutan nests. Bonobo nests are on average a diameter of 90 cm (range 65-130, n = 19) and a depth of about 18 cm (range 10-30, n = 5), which are very similar to length and depth of chimpanzee nests reported here [Fruth 1995]. Orangutan nests were on average larger, with a mean diameter of 115 cm (range 50-170, n = 92), but differed significantly across age and sex classes from an average of 64 cm for juveniles to 139 cm for adult flanged 76 males [Rayadin and Saitoh 2009]. The size of ape nests may scale with body size, as Rayadin and Saitoh [2009] found for age and sex classes of orangutans. They measured ape nests from the ground using two long poles, which may not be comparable to direct measurements reported here. Also, previous studies have measured only the diameter of nests and the finding here that nests are on average oval in shape with differences between sites in width, but not length, highlights the need for close up and detailed examination of nests in future comparisons across species and communities. Nest archi tec ture In contrast to shape, overall nests in Fongoli and Issa differed architecturally. This is likely due to the differences between tree and ground nests in Fongoli and between seasons in Issa. Ground nests are frequently made of structurally different materials to trees, e.g. grass, and do not require the formation of support, or weight bearing, branch configurations. Given that the location of nests within trees and the sizes of the trees used for building differed between Fongoli and Issa, that the number of branches of different diameters did not differ between the two sites may indicate that chimpanzees are constrained in the sizes of branches used for building regardless of substrate selected. For example, as tree nests in Fongoli are built lower in the tree and closer to the trunk, I expect that the material available for building would be of larger diameter in Fongoli trees than in Issa trees. However, when trees are smaller, there may be no difference in the material available, or the chimpanzees in Fongoli may select similar-sized branches to chimpanzees in Issa from the material available around the nest spot. This is supported by the similarity in selected support diameter between the two sites. However, constructed support diameter was larger in Fongoli. As discussed above regarding shape differences, chimpanzees in Fongoli may be larger, and so the first supporting stages were built using larger branches than in Issa. However, similar numbers of large branches (2-3cm and >3cm) were used in nest building at both sites. I also recorded the type of material used for building, i.e. main branch, lone side branch, side branch, or twig. The numbers of pieces and the mean diameters of each different material type were the only other measurements of nest architecture to differ between Fongoli and Issa nests. Despite this difference in the mean size of each type of material (MB, SB, etc.) between Fongoli and Issa, there was no difference in the number of branches of each size category (<1cm, 1-2cm, etc.) between Fongoli and Issa. That branch types differ in diameter size, but the number of branches of different sizes used to build a nest does not differ, between Fongoli and Issa, suggests that my definition of the difference between a twig and a side branch (a twig 77 terminates with leaves, whilst a side-branch has subsequent forks into twigs or side-branches) may not be salient to a chimpanzee. Branches might be selected according to other features, e.g. size, leafiness, tensile strength, or brittleness, rather than shape. The branches used may also depend on the nest spot selected within the tree, or may be constrained in that only branches up to a certain size can be broken or bent by a chimpanzee. In this study a main branch was not defined by its size, but by whether or not it contributed more side-branches or twigs to the nest. This is important in considering how a nest is constructed; if successive lone branches are pulled into the structure the result is a loose piling of branches that remains insecure, whilst if several main branches are pulled in to form the base, followed by successive incorporation of side-branches from each of these main branches then the result is a tightly connected and stable structure (pers. obs.). Previous researchers have referred to the ‘inter-weaving’ of great ape nests [Goodall 1962; Groves and Sabater Pi 1985; Russon et al. 2007]. Yet in other animal built structures, ‘weaving’ has a specific meaning where by a material is passed in and out of the fabric of another material, e.g. weaving of grass nests by weaver-birds [Hansell 2000]. Chimpanzees (and likely other great apes) rarely use a weaving action to build their nests; instead branches are most frequently held together as side-branches and twigs are folded over each other. This technique of interlocking branches is called ‘entangle’ in the study of bird nest architecture [Hansell 2000]. I was able to discern tucking of material in and out of the nest structure on average in approximately 10% of manipulations. Such weaving actions seemed more frequent in the initial stages of building, which may need more complex building techniques to get the structure started. Chimpanzees do therefore interweave some branches in nest building, which may function to secure the structure, but this technique may be distinct from that of ‘entangle’ used in the construction of most of the mattress of the nest. However, due to the historic use of the term ‘inter-weaving’ in ape nest studies, I will continue to use the term in this thesis. Fongoli chimpanzees built tree nests with fewer main branches, but of larger diameter. That these main branches were larger could influence the use of a fewer number by provision of more support or more subsequent side-branches and twigs. Thus, the total number of branches of different sizes manipulated does not differ between the two sites. These are the first quantitative descriptions of nest architecture in chimpanzees. Fruth [1995] measured the architecture of bonobo nests using similar methods, so I compare my results to hers in Table 3-11. Earlier studies of nest architecture include Mackinnon [1974], who cut down seven orangutan nests for detailed study. He distinguished branches differently again to this study, according to their function as “rimming”, “hanging”, “pillaring”, or “loose”, and characterised all branches as bent. Schaller [1961] examined 13 orangutan nests and found that 78 nests had on average 4.7 broken branches (range 3-9) and on average 10 twigs added as detached lining to each nest (range 0-34). Overall, bonobo and chimpanzee nest architecture seems very similar, although bonobos seem to use fewer branches in support and more branches than chimpanzees in Fongoli. Fruth [1995] saw no use of large branches over 3 cm in diameter in bonobo nest building. Table 3-11. Bonobo and chimpanzee tree nest architecture, compared (P. paniscus nest measurements are taken from [Fruth 1995]). P. t . v e rus P. t . s chwe in fur th i i P . panis cus Architecture measurement Fongoli, dry season Issa, overall Lomako x ̄ Range n x ̄ Range n x ̄ Range n Branches used cm 2.7 0-7 89 3.6 0-11 70 3.9 2-9 22 Supporting branches cm 3.3 1-4 70 3.2 0-4 70 1.6 1-3 23 Bends <1cm n 10.1 0-34 86 12.7 0-37 70 7.7 2-16 10 1-2cm n 6.5 0-17 86 5.6 0-16 70 4.3 2-11 9 2-3cm n 1.6 0-7 86 1.2 0-4 70 1.6 1-3 5 >3cm n 0.4 0-3 86 0.3 0-3 70 - - - Breaks <1cm n 4.2 0-17 86 3.5 0-23 70 3.7 1-7 13 1-2cm n 4.0 0-17 86 3.2 0-14 70 3.8 1-7 12 2-3cm n 1.4 0-10 86 1.0 0-6 70 1.7 1-3 11 >3cm n 0.6 0-5 86 0.4 0-4 70 - - - Detachments <1cm n 8.3 0-61 86 7.0 0-24 70 8.7 1-28 15 1-2cm n 2.1 0-19 86 1.6 0-19 70 1 1-1 5 2-3cm n 0.2 0-5 86 0.2 0-2 70 - - - >3cm n 0.1 0-2 86 0.1 0-3 70 - - - Summary In summary, nest and nesting tree characteristics differed across almost all variables recorded in Fongoli and Issa. Nests in Issa were more frequently built on outer tree forks, on branch-cup supports, higher, higher within tree crowns, and further out from the trunk and the main stem of the trees. That nests were also built further from the tree crown edge in Issa indicates that these differences in nest position are due in part to differences in tree size between the two sites. 79 Nesting trees in Issa were larger across all measures: DBH, height, crown base height, tree crown height, and tree crown diameter. Trees also differed in morphology with a greater amount of crown cover and leaf amount in Issa, but more stems at breast height and greater canopy connectivity in Fongoli. Some similarities also emerged: The proportion of nests in both sites that were integrated from multiple trees or other sources of building material, in addition to those with overhead cover was similar. Finally, shape and architecture were overall similar between the two sites. This suggests that cross-site differences in the nesting trees and positions within trees selected do not influence shape or architecture of nests. Studies of captive chimpanzees revealed that nest building is partly learned and must be acquired early in life [Bernstein 1967; Videan 2006a]. It is therefore surprising that chimpanzees in two populations separated by >5000 km build overnight structures in very similar ways. Perhaps there are simply few ways in which to construct an arboreal platform, in order to provide support to such a large-bodied ape. Seasonal differences in Issa Nest character i s t i c s There is much fluctuation in group sizes over time within Fongoli and within seasons in Issa. Many different factors influence fission and fusion of chimpanzee parties over time [Anderson et al. 2002], and similarly, seasonal differences in group size in Issa likely reflect factors such as food availability and presence of oestrous females that are beyond the scope of this study. Nest types did not differ seasonally in Issa. Although the proportion of different nest types built by gorillas differ seasonally [Mehlman and Doran 2002; Tutin et al. 1995], nest types recorded in this study, and frequently recorded in other chimpanzee nest studies, are not defined by their structural properties (e.g. amount of material), but by their position within a tree or incorporation of multiple trees. This is also reflected in the lack of difference in the proportion of nests built from multiple trees between seasons in Issa. Data were collected in Fongoli only in the dry season, but it is likely that ground nests are a seasonal occurrence. Several previous studies have found that chimpanzees build more open nests during the wet season, and it is hypothesised that open building sites are preferred during the wet season due to increased risk of rain [Baldwin et al. 1981]. A chimpanzees’ night’s sleep may be more comfortable, even if soaked through by a deluge versus prolonged dripping of water from branches overhead even after the rain has stopped [Fruth 1995]. The result here that nests are not more open in the wet season could also be a result of different vegetation structure. 80 Hernandez-Aguilar [2006] also found no difference in the proportion of open or covered nests in Issa in the wet or dry season. The above lack of seasonality in the proportion of nests built under cover is mirrored by no difference in the position of the nests in number of decisions, height, or absolute distance from the trunk, main stem, or tree crown edge in the wet compared to the dry season. This result contrasts with that of Baldwin et al. [1981], who found that the chimpanzees of Assirik and Equatorial Guinea nested higher in the wet seasons. Seasonal differences may be driven by a number of weather variables, e.g. rainfall, temperature or wind, and detailed comparison of how nests vary under different conditions may reveal effects that are masked in gross seasonal comparisons (see Chapter 6). Nest ing trees Fruiting trees are rarely used for nesting by chimpanzees at a number of study sites. That a greater proportion of fruiting trees are used for nesting in the dry season may reflect this paucity of material. However, this is speculative without further data on food and leafing nesting tree availability in Issa in the dry and wet season Trees did not differ in size or canopy connectivity between seasons. However in the dry season trees with a greater number of main stems and stems at breast height were used for nest building. There was also no difference in the cumulative DBH or the DBH of the largest tree stems used between seasons. Interestingly crown cover and leaf amount of nesting trees was greater in the dry season, which is contrary to what would be expected if deciduous nesting trees lose leaves during the dry season. However, these data are from selected nesting trees only, and it is possible that nesting trees have more leaves in the dry season, because only trees that have lost and re-leafed are selected for building. It would be necessary to compare selected nesting trees to available trees to determine whether trees with greater crown cover or leaf amount are selected during the dry season. These data were collected along phenology transects and vegetation plots described in Chapter 2 and will be analysed at a later date. Nest shape and archi tec ture No previous study has analysed variation in shape or architecture of nests across sites or seasons. That nests in Issa are thicker in the wet season suggests that intra-site variation may be in part in response to environmental conditions; such variation may reflect an insulatory function of nests in wetter or colder conditions (see Chapter 6). Nest support is greater in the wet season, both in the number of steps and in the diameter of constructed support branches. This pattern is 81 investigated further in Chapter 6, and could be influenced by weather conditions, e.g. greater wind speeds. Also, although there was no difference in the total number of steps, nests in the wet season appear to be more complex, as they took longer to deconstruct. A greater number of breaks, large diameter branches, and twigs used in the wet season may also result in a more insulated nest structure. Summary In summary, nests in Issa were similar in almost all nest and tree characteristics. Nest group size, type, support, integration, cover, and position were similar in the dry and wet season in Issa. Nesting trees were also similar in size, but differed in morphology, i.e. in number of main stems, crown cover, or leaf amount. These differences may result from selection of available trees; however, further study is necessary to determine if selectivity varies in different seasons in Issa. Nest shape and architecture differed between seasons in Issa. Nests were larger and thicker, built with greater support, more breaks, large bends, and twigs in the wet season. These differences likely result from variation in weather conditions like temperature, humidity, or rainfall and may reflect the thermoregulatory function of nests. These hypotheses will be tested further in Chapters 4 and 6. Conclusion Cultural variation in nest building is difficult to determine due to the large amount of variation in structures within sites. The structure of a nest is also more complex to define and measure, than for example an ant-dipping tool. Perhaps this is why few studies have attempted to investigate nest-building in light of possible cultural variation. I have shown here that gross measures of nest characteristics are likely due mostly to variation in the trees selected for building. Future cross- site comparisons must control for variation in tree size and morphology. In addition, nest shape and architecture is remarkably similar across sites and is influenced by season. Seasonal and cross-site variation described here suggests possible functions of nests. In the following Chapters, I will focus on the function of simple chimpanzee shelters. Chapter 4 presents an experimental approach, whilst the data described here form the basis for further investigation of the anti-predator (Chapter 5) and thermoregulatory (Chapter 6) function of nests. However, as described for the case of ant-dipping tool variation, these results do not preclude potential cultural variants in nest-building techniques. It remains for further study of nuanced variation in nest building techniques to determine if there are possible cultural variations on this ubiquitous behaviour. 82 Chapter 4 Why sleep in a nest? Empirical testing of the function of simple shelters made by wild chimpanzees Musings from a night as a chimpanzee: “We arrived at the nest site at le bonne heure de chimpanzee, 5:30 p.m., and I climbed up to a suitable location (that likely of Karamoko, the previous night). This infrequent chimpanzee act of nest re-use was followed by a distinctly chimpy act of nest modification whereby I made my sleeping place as comfortable as possible by pulling back in sprung branches and adding some more leafy lining over the dry leaves of the night before, although Karamoko had used most of the material available around the nest the night before. Despite his efforts this nest was not very complex, or comfortable! I settled down for my night’s sleep after filling my belly (no, not with Spondias – although there was a lovely feeding tree directly above my nest as we shall see later – but with, mmm... rice and fish balls). In my leafy cocoon, the air smelled of freshly broken branches and leaves, and surprisingly few mosquitoes came to disturb my rest (only eight bites the whole night – well, those that left a mark anyway). I was joined by a small party of my fellows, who nested not far from me, and accompanied me in the first couple hours with pant hoots and grunts, and the huh huh noise of their nest grunts. As darkness fell, their vocals were a comfort and the night not so lonely. A refreshing breeze meant I actually had 4 of my 5.5 hours sleep before 2 a.m. when it began to rain. Just speckling at first, I toughed it out, as the good chimp that I am, but at 4:30 a.m. the deluge began and being the good human that I am – I put on my raincoat! I then lay, as my fellow slumbering primates, in my nest with the rain falling all over and around – needless to say soaking me through despite my additional plastic cover! I am filled with appreciation for the trials and tribulations of nest sleep, although my leafy support provided much warmth as I began to feel very cold (later revealed temperature was still over 20 °C – but in Senegal that feels almost frosty!). Dozing through the heavy rain splatter – as my nest was built for a dry night and hence underneath a covering of overhead branches from which the rain fell in great drops – until 6 a.m. when I began to make my escape from my trip into chimpland and back to the humanland of Fongoli. My fellow nesters had other plans. As I began to descend from “my” nest, I was - for wont of a better word – busted! I don’t know who was more astonished at the soggy human (yes, a human!) clambering out of a chimp nest 6m up in a tree. At the same level we looked at one another, someone not quite belonging in this arboreal world – yes, the one dangling in shock surrounded by wraah barks on the end of a rope! As I slowly descended they watched (and made clear their disapproval) and once I was safely on the ground Karamoko sealed his disapproval by launching a great branch into the centre of my... er… his?... nest, as though trying to destroy the evidence of this embarrassing theft! The separation of worlds was complete as I watched from the ground as they munched on the Spondias in the tree above, which I sadly denied as my dinner the night before.” 83 INTRODUCTION Great apes build a new nest at least once a day from weaning onwards by bending, breaking and inter-weaving branches into a platform. Nest-building is ubiquitous across great apes, and was likely present before pongine and hominine lines separated, making it an important ancestral trait to model early hominin behaviour and evolution of shelter-construction in humans. Humans in all cultures on every continent make shelters [Brown 1991], and their global distribution may depend on this basic trait for environmental protection. Nest-building by early hominins has been hypothesized by several researchers [McGrew 1992; Sabater Pi et al. 1997; Sept 1992]. Sabater Pi et al. [1997] outlined factors supporting early hominin nest-building: an ecological transition from forest to wooded savanna, poor nocturnal and crepuscular vision common to diurnal primates, greater predator density and pressure in an open environment, lack of evidence of fire use (likely necessary to deter predators) by early hominins preceding late H. erectus, and a likely physiological requirement for recumbent relaxed sleeping postures. The noun ‘shelter’ refers to a structure, or feature, that provides protection from environmental challenges, or refuge from danger, and as such ape nests are shelters. The primary function of ape shelters is arguably for sleep. Fruth and Hohmann [Fruth and Hohmann 1996] proposed that the ability to sleep safely in a recumbent posture likely increased REM (rapid eye movement) sleep, which may have aided memory consolidation and enabled cognitive evolution. McGrew [2004] framed three additional hypotheses of nest function: anti-predation, anti- pathogen, and thermoregulation. I tested experimentally these simple shelter functions by sleeping overnight in chimpanzee nests and on the bare ground to test differences in sleep quality, bites from possible vectors (as a proxy for pathogens), and insulation (thermoregulation). Predation was not tested directly (due to the risk involved), but likely influence from the threat of predation is discussed. Sleep quality Sleep is important for brain and body restorative processes, but amount and type of sleep varies across species with variables such as age, weight, and ecology, e.g. diet or sleeping site safety [Siegel 2005]. Sleep architecture, the distribution of slow-wave sleep stages in the beginning and REM sleep towards the end of the sleep period, is suggested to reflect an evolutionary trade-off between the benefits of sleep versus risk of predation [Lima et al. 2005]. Great ape nest building may be a solution to this trade-off allowing a large-bodied primate to have relaxed sleeping postures and greater REM sleep in the safety of a platform more, or similarly, removed from large accessible branches as smaller roost-sleeping monkeys. 84 In this study I aimed to test through self assessment whether ground sleep provides a greater quality of sleep than arboreal nest sleep. This is not directly comparable to what a chimpanzee may experience, but provides a comparative experience across conditions. Anti-predation Across primate species, predator avoidance underlies sleeping site selection and sleep-related behaviours [reviewed in Anderson 1984; 1998; 2000]. Observations of predation on great apes are rare, but have been reported from several long-term field sites [Boesch 1991; Hiraiwa- Hasegawa et al. 1986; Tsukahara 1993]. Pruetz et al. [2008] found chimpanzees nested higher and closer together in Assirik, a predator-rich site of similar vegetation physiognomy to Fongoli, a site lacking predators. This suggests arboreal sleep to be a function of predator pressure, but not the behaviour of nest-building itself. Yet, nest-building may be a solution to increased body size if nests permit apes to sleep in locations within trees that would not otherwise support their weight. Nest-building may function in part to reduce predation, but other hypothesized functions are likely; e.g. even when sleeping on the ground (3% of nests) Fongoli chimpanzees continue to build nests, and continue to sleep arboreally despite no predator pressure [Pruetz et al. 2008]. Anti-pathogen Primates adjust sleeping or feeding site usage to avoid gastro-intestinal parasite infection from build-up of faecal contamination [Cercocebus albigena: Freeland 1980; Papio cynocephalus: Hausfater and Meade 1982]. It is unknown if great apes also do so, but gastro-intestinal infection is unlikely to influence nest function. Less is known in primates about non-faecal-oral transmitted pathogens, e.g. vector transmission. Of 408 parasite species infecting primates 32% are transmitted by arthropod vectors, mostly biting insects [Pedersen et al. 2005]. Protozoa are most common (e.g. malaria, Plasmodium sp.), but viruses and helminths may also be vector-transmitted [Pedersen et al. 2005]. Recent work revealed several Plasmodium species infect wild apes including the most virulent species in humans, P. falciparum [Krief et al. 2010; Liu et al. 2010; Prugnolle et al. 2010]. The extent to which Plasmodium, or other vector-borne parasites, are pathogenic in wild apes remains unknown. Although influence of nest versus arboreality cannot be determined here I conduct a preliminary test (using bite frequency as a proxy) of vulnerability to vector- transmitted pathogens on the ground, or in an arboreal nest. In theory, nests might reduce exposure to pathogens in several non-exclusive ways: nests may shield parts of the body from bites by forming a physical barrier and changing position in the nest may dislodge biters from the 85 unshielded portion of the body; and aromatic substances released from vegetation may act as repellents or mask odours used by vectors to locate their prey. Thermoregulation Hypothesized thermoregulatory function of nests is suggested in nature and captivity where great apes adjust building techniques and site selection in response to climate. Western gorillas (Mondika, Republic of Congo) made full, presumably insulating, nests in the wet season, and slept on the bare earth or minimal constructions in the dry season [Mehlman and Doran 2002]. Similarly, lowland gorillas (Lope, Gabon) make fuller constructions in the rainy season [Tutin et al. 1995]. Captive chimpanzees may select cooler sleeping sites in conditions of greater temperature and humidity [Videan 2006b]. These correlations are indicative of thermoregulatory function, but here I aimed to directly test insulation. Proximate functions suggested from this study may inform our reasoned explanations for the evolution of shelter, its persistence in all great apes, and its proliferation in Homo sapiens. METHODS Study site The experiment was done within the 63km2 home range of the Fongoli chimpanzee community. Fongoli has few large mammals and no evidence of predatory species (see Chapter 5). Data collection I spent 12 nights from 28th October 2007 to 23rd March 2008 sleeping out at least once per month, alternating tree versus ground sleep either in a chimpanzee or self-constructed arboreal nest, or on the bare ground. Twice, I re-used nests that chimpanzees had built the night before; I built nests for sleep on the other four nights using similar building techniques to chimpanzees. Nest height ranged from 1.5-8.8m (mean: 5.2m); within the range of Fongoli chimpanzees [Pruetz et al. 2008] and were accessed either free-climbing or with climbing equipment. I tried to sleep for 12 hr overnight based on findings from the only study to measure the inactive period of wild chimpanzees [Lodwick et al. 2004]. From November 2007 to January 2008, only two sleep- outs were done given extremely low overnight temperatures (Figure 4-1). Thus data were analyzed for five night’s on the bare ground, and six in a nest, which varied in duration from 9.25-12 hrs. (Such experiments done directly on wild chimpanzees are inadvisable for both ethical and logistical reasons, and using oneself as an experimental subject allows for control of variables not possible through observational study.) 86 Sleep qual i ty I dictated all waking events and external (vocalizations, noises, or weather) and internal (subjective fears or concerns) sources of disturbance into a recorder (Memory stick ICD-MS1 recorder). Time slept, and length of sleep bouts, was measured to the nearest 15 min by extrapolating time awake versus asleep from recorded waking events. Sleep disturbance was measured as the number of waking events/hr, to control for variation in experiment duration. Sleep quality was measured as the total time slept/time in ‘bed’ [sensu Videan 2006b]. Mean sleep bout length was calculated from all sleep bouts per night. Anti -pathogen I counted all visible bites on my person, with aid of a mirror, prior to experiment and again the next morning. I divided the difference by experiment duration to calculate bites/hr. Thermoregulat ion Overnight mean and minimum temperature was calculated for the overnight duration of each experiment (approx. 1800-0600hrs, see above) from two data-loggers (HOBO pendant UA-001- 08) deployed, one at ground-level and one at nest-height, to record ambient temperature every 15 min for the duration of each experimental night. As lower temperatures are hypothesized to influence nest insulation more I also calculated mean daily minimum temperatures (1st-14th, and 15th-end of each month) from six data loggers (Hobo H8 Pro series), which were permanently deployed, to log every 30 min, in representative vegetation types. I wore two data loggers (pendants), one on my front midriff, one on my back, whilst lying supine in both conditions. The same clothing was worn, and the midriff, where loggers were attached, was kept bare for each experimental night’s sleep. The differential temperature (back sensor minus midriff sensor temperature) was taken as a measure of insulation, as it should control for ambient temperature variation. I excluded data points from analysis if I was not lying supine. RESULTS Sleep quality Disturbed nights were roughly equal across conditions, but arboreal animal noises were more salient arboreally (5/6 vs. 1/5 nights; Table 4-1). I heard animals terrestrially including domestic cattle, porcupine (Hystrix cristata), jackal (Canis sp.), spotted hyena (Crocuta crocuta), and chimpanzee (Table 4-1). Arboreal animals likely included fruit bats (Pteropodidae). Arboreal and terrestrial animal movements and vocalizations contributed to internal sources of disturbance, 87 e.g. terrestrial animals, including hyenas, were more concerning during ground sleep (3/5 vs. 1/6 nights; Table 4-1), although snakes were always a concern. There was no difference in amount of sleep or sleep quality between nest and ground nights (Mann-Whitney, U=20.5, p=0.31; Table 4-2). However, sleep was more disturbed on the ground as more waking events/hr occurred terrestrially (U=3, p=0.03), and the mean duration of sleep bouts was longer in a nest than on the ground (U=28.5, p=0.01; Table 4-2). Table 4-1. External and internal sources of disturbance during sleep experiments under the two conditions of nest or ground sleep. Location (n nights) External sources of disturbance (number of nights disturbed) chimpanzees nearby chimpanzee vocals hyena vocals other vocals arboreal animal noise terrestrial animal noises rain Nest (n=6) 2 3 1 2 5 5 1 Ground (n=5) 2 3 1 2 1 4 1 Location (n nights) Internal sources of disturbance (number of nights disturbed) falling unsafe from terrestrial animals unsafe from arboreal animals uncomfortable temperatures snakes Nest (n=6) 1 1 1 3 5 Ground (n=5) 0 3 1 2 4 Table 4-2. Comparison of median sleep amount (measured in total hrs of sleep and sleep quality) and sleep disturbance (measured by waking events and mean sleep bout length) between nest and ground conditions. Median sleep amount Median sleep disturbance Total [hr] (range) Sleep quality (range) Total frequency waking events (range) Frequency waking events/hr (range) Mean sleep bout length/night [min] (range) Nest 5.63 (4-7) 0.49 (0.41-0.61) 7 (4-8) 0.6 (0.43-0.74) 47 (38-60) Ground 4.75 (3.75-5.5) 0.44 (0.38-0.52) 7 (7-11) 0.7 (0.62-1.05) 30 (28-41) 88 Anti-pathogen I was bitten more during ground (median 28, range 13-30) than nest sleep (median 1, range 1- 16). The number of bites per hr was also fewer during nest (median 0.1, range 0.1-1.4) than ground sleep (Median 2.7, range 1.2-2.8; Mann-Whitney, n1=6, n2=5, U=1, p=0.01). Of bites that were observed directly (n=64), most were from mosquitoes (83%), few from tsetse flies (6%), and although some bites were from ants (11%) this is likely not sufficient to explain differences between conditions. Figure 4-1. Mean minimum overnight ambient temperature recorded during bi-weekly periods from October to March with permanently deployed loggers (x symbol) and mean differential temperature recorded in a nest (closed circles) and on the ground (open circles) below. Error bars indicate max and min. Thermoregulation There was no significant difference in mean differential temperature during consecutive nest and ground sleep (Wilcoxon’s matched pair analysis; n=5, z=-0.94, p=0.34), however for four of five pairs mean differential temperature is greater in a nest. Mean ambient temperature was also no -­‐2   3   8   13   18   23   28   Te m pe ra tu re  C °   Mean  differen2al  (Ground)   Mean  differen2al  (Nest)   Mean  minimum  ambient   89 different between loggers at nest or ground height (n=10, z=1.27 p=0.20). There was a tendency for maximum differential temperature to be higher in nests when the minimum ambient temperature at nest height was lower (spearman’s, n=6, r=0.75, p=0.07), but no relationship between maximum differential temperature on ground and minimum ambient temperature at ground height (n=5, r=0.20, p=0.75). No relationship was found between mean differential temperature and mean ambient temperature during nest or ground sleep (ground n=5, r=0.60, p=28; nest n=6, r=-0.43, p=0.40). However, as mean biweekly minimum ambient temperature falls the mean differential temperature in nest sleep rises (r=-0.83, p=0.04), whereas the mean differential temperature on ground falls (r=0.9, p=0.04). More nights were spent sleeping-out from February to March when mean minimum temperatures increased (Figure 4-1). DISCUSSION Sleep quality All experimental nights’ sleep in this study were uncomfortable and characterized by low sleep quality (<0.50), compared to mean sleep quality recorded for captive chimpanzees (0.86) and human societies [Videan 2006b]. However, that nest sleep was less disturbed with longer bouts of sleep provides support for the improved sleep function of nests. Sleep quality is affected by temperature and humidity in humans and chimpanzees [Okamoto-Mizuno et al. 1999; Videan 2006b], and so is likely affected in wild chimpanzees in response to environmental conditions. Sleep stages, quality, or duration are difficult to measure in wild primates, and future research could use videography for proxy measures such as eye movement, posture changes, and nocturnal behaviours in both captive and wild studies of primate sleep [Anderson 1998]. Greater disturbance of ground sleep seemed to be due to anxiety and alertness, revealed by internal sources of disturbance. For example, the lowest quality of sleep (0.38) was twice recorded during ground sleep when either hyena or jackal vocalizations were heard, whilst the most disturbed sleep (11 waking events) was recorded during ground sleep in presence of domestic cattle. Although anthropogenic, cattle are similar to large herbivorous mammals such as buffalo found in less disturbed chimpanzee habitats. Large terrestrial mammal movements may be an important factor in nest-site choice for apes; e.g. western gorillas may nest arboreally near some feeding trees to avoid disturbance by elephants [Tutin et al. 1995]. Coolidge and Wynn [2009] proposed a transition from tree to ground sleep by Homo erectus that had important effects on sleep quality and cycles. They highlight the role of REM sleep and dreaming on memory consolidation, enhanced skills via ‘priming’, and cognition. Although functions of all sleep stages remain unclear [Siegel 2005], the role of REM sleep on 90 memory consolidation has been supported in recent research [Cai et al. 2009; Roth et al. 2010]. REM sleep across mammals is associated with relative brain size, with humans exhibiting the most, but is also negatively associated with predation risk [Lesku et al. 2006]. Although H. erectus was large-bodied, this species was certainly vulnerable to predation, so changes in sleep quantity and quality may not have directly followed the transition from tree to ground. Across all primates the most important factors influencing sleeping site selection and sleep related behaviours includes protection from predators, in addition to thermoregulation, or parasite avoidance [Anderson 1984; 1998; 2000]. The relationship between inactivity and predation risk [Lima and Rattenborg 2007; Lima et al. 2005], also reflected by observations here, makes it unlikely that the transition from tree to ground sleep would have permitted lower vigilance necessary to increase sleep quality without other factors to ameliorate predation risk [e.g. structural protection, Kortlandt 1980; or fire, e.g. Wrangham 2009]. Anti-pathogen The hypothesis that arboreal nest sleep reduces bites from possible vectors was supported with an order of magnitude more bites received on the ground. However, it is not known which mosquito species were responsible, nor what diseases transmitted through biting vectors in Fongoli. Other studies have investigated this hypothesized function of nest-building; Koops et al. [in prep] found no difference in mosquito density at different heights or altitudes (Nimba Mountains, Guinea). In contrast, Largo et al. [2009] found orangutans build preferentially in tree species with known anti-mosquito properties during periods of high mosquito density and also carry these species’ twigs as covers for other nests. If insect density does not vary with height (sensu Koops et al. [in prep]), yet number of bites does, then the freshly broken branches of nests may chemically or aromatically deter biting insects. Perhaps this is what causes a great ape to build a new nest nightly. Some bird species select aromatic nesting materials to reduce nest bacterial load [Starlings, Gwinner and Berger 2005] or repel biting insects [Blue-tits, Lafuma et al. 2001], although it is unknown if these materials chemically repel, or aromatically mask heat and odour signals attractive to, biting insects. Thermoregulation Support was found for the hypothesis that nests function to provide insulation as greater differential temperatures correlated with decreased overnight temperatures during nest but not ground sleep. That the difference between differential temperatures of nest and ground sleep was not significant may have been influenced by the extreme temperatures in Fongoli. Temperatures 91 reached a mean monthly maximum of 43 C°, when overnight temperatures correspondingly increased, and experimental nights were biased towards these warmer conditions due to discomfort of cold. Fongoli chimpanzees are subject to these extreme temperatures (7-45 C°) and an alternative test of the thermoregulatory hypothesis will investigate variation in nest structure in response to environmental conditions (see Chapter 6). Conclusions Some evidence is provided here that arboreal nests reduce risk of insect bites and improve sleep through decreased disturbance. Results also indicate nests serve a thermoregulatory function through provision of greater insulation in colder conditions, although there is likely inter-play between microclimatic conditions, predation risk, arboreality, and sleep quality. Future research should investigate the relative importance of the functions of these simple shelters, which can inform on factors likely involved in the transition from tree to ground sleep and the proliferation of shelter construction in Homo. 92 Chapter 5 Do chimpanzee nests have an anti-predator function? Leopard (Panthera pardus) caught by trip camera deployed in Issa (Photo credit: UPP/MPI EVA) 93 INTRODUCTION All animals, vertebrate and invertebrate, spend proportions of their lives in a vulnerable state of sleep or sleep-like behaviour [reviewed in Lima et al. 2005]. Although the functions of sleep are still debated, researchers agree that the primary function of sleep involves neural maintenance [Siegel 2005]. However, sleep compromises the ability to detect predators [Anderson 1998; 2000; Lima and Rattenborg 2007; Lima et al. 2005; Rattenborg et al. 1999], and early research into sleep function highlighted the role predation pressure may have played in the evolution of sleep [Meddis 1975]. Some species have the ability to sleep only one half of the brain at a time, which enables the sleeper to detect predators [Rattenborg et al. 1999]. In an effort to address why more species do not engage in such unihemispheric sleep, Lima and Rattenborg [2007] developed a theoretical model that suggests complete behavioural shutdown (or unconsciousness) may be the safest way for animals to achieve necessary neural maintenance functions of sleep. Similar to trade-offs in predation risk versus energy intake in animal foraging strategies, there are likely trade-offs in predation risk versus sleeping site selection, sleep architecture, and sleep duration [Lima et al. 2005]. The evolution of shelter construction in the great apes may thus have been a solution to the trade-off between sleep and predation-risk, which allowed a large bodied ape to sleep recumbent in a safe, comfortable spot. REM (rapid eye movement) sleep in large mammals is accompanied by loss of muscle tone and so may require a stable platform and recumbent posture. Increased REM sleep has been proposed to have enabled cognitive evolution in great apes and Homo, through increased memory consolidation and enhancement [Coolidge and Wynn 2009; Fruth and Hohmann 1996]. In the previous Chapter 3, I described and compared nest and nesting tree characteristics between Fongoli and Issa. Many differences emerge, most which can be explained by environmental differences between the two sites. Here I address in greater detail the question of the influence of predation-risk on chimpanzee nesting behaviour. I begin with a review of the evidence of the selective pressure of predation on sleep in apes and other primates and then compare variation in nests and nesting sites. Comparison of the two sites is pertinent to the question of whether or not chimpanzee nests function to reduce risk of predation. Predator avoidance in other primates Most primates live arboreally; however even more terrestrial species such as baboons (Papio spp.), macaques (Macaca spp.) and chimpanzees (Pan spp.) return to the refuge of elevated sleeping sites at night. Safety from predators is the primary factor underlying sleep-related behaviours and sleeping site selection in primates [Anderson 1984; 1998; 2000]. Sleep related 94 behavioural adaptations include huddling [Li et al. 2010], adjustment of sleeping site locations [Matsuda et al. 2010] and differential selection of sleeping sites by younger or female individuals [Fan and Jiang 2008]. Many primates select locations for sleeping that are difficult to access, for example by height, distance from the trunk of the tree, and presence of an alternate escape route [Anderson 1984; 1998; 2000]. There are numerous examples of the influence of predator avoidance on sleeping site selection across a range of primates [see Anderson 1984; 1998; 2000 for comprehensive review]. Hamilton [1982] found that baboon’s sleeping site preference ranged from cliff edges, through emergent trees, to open woodland and proposed that these decreasing preferences reflected the increasing danger of attack by predators. Gibbons select trees difficult for predators to access, e.g. larger, taller trees with few low branches or attached lianas or vines [e.g. Nomascus concolor, Fan and Jiang 2008; Hylobates lar, Reichard 1998]. Many primates also sleep towards the terminal ends of branches, which may allow early detection of an arboreal predator and permit escape [Anderson 2000]. Alternatively, early detection may give the primates time to mob predators as a counter-strategy [Busse 1980]. Are chimpanzees and other large apes at risk of predation? Predation is a major cause of mortality in non-human primates [Cheney et al. 2004], but its selective influence is often difficult to study, given that direct observations of predation are rare and few studies approach the question through direct study of sympatric predators [e.g. Hawkins and Racey 2008; Jenny and Zuberbühler 2005; Mitani et al. 2001]. The extent to which great apes are at risk from predation is debated, given that predation rate decreases with increasing primate body size across taxa [Cheney and Wrangham 1987]. Predator-prey interactions have been studied at only one site, the Tai forest, where chimpanzees live sympatrically with predatory species [Jenny and Zuberbühler 2005; Shultz et al. 2004; Zuberbühler and Jenny 2002]. Through radio-tracking data of individual leopards and subsequent faecal analysis Zuberbühler and Jenny [2002] determined that leopards more often preyed upon larger and more abundant prey. Despite this general trend, the authors found chimpanzees to be an exception, as they were rarely preyed upon, and so they suggested that apes may be large enough to be beyond the prey range of leopards [Zuberbühler and Jenny 2002]. Shultz et al. [2004] re-analysed these same data to control for prey abundance by calculating predation rates for all predators as a proportion of available prey. They report a leopard predation rate of 10% of available chimpanzees, which is similar to the cumulative predation rate (including all predators) on other sympatric primates and ungulates in the park [Shultz et al. 2004]. Hayward et al. [2006a] analysed leopard prey preferences across their range 95 and found that leopards preferentially prey on species within the weight range of 10-40kg and the most preferred prey body weight is 23kg. Chimpanzees fall very close to this preferred body weight because body weights were adjusted to take into account leopard bias towards hunting juveniles and females; adjusted weight for chimpanzees is 22.5kg [Hayward et al. 2006a]. In their analysis chimpanzee and gorilla are taken by leopards in accordance to their abundance in the habitat. However, Hayward et al. [2006a] and Shultz et al. [2004] use data collected by Zuberbühler and Jenny [2002] to analyse predation rates of leopards, but Zuberbühler and Jenny report evidence of predation on chimpanzee in only 1 of 200 faecal samples. Prey preferences of other potential ape predators have not been studied where they live sympatrically with great apes. Prey preferences of lions differ dramatically from that of leopards, with preferred prey body size falling within the range 190-550kg [Hayward and Kerley 2005]. However, Hayward and Kerley [2005] highlight the lack of studies of lions living in ecosystems other than open grassland savanna, such as the wooded savannas where lions live sympatrically with chimpanzees. Studies of spotted hyena predatory behaviour are similarly biased towards open grassland savanna habitats and in contrast to leopard and lion, spotted hyena preferentially prey upon few species; most prey are taken according to their abundance [Hayward 2006]. African wild dog (Lycaon pictus) prefer prey of either 16-32kg or 120-140kg depending on their group size [Hayward et al. 2006b]. Indiscriminate predation on available mammals by spotted hyena, and the preferred prey size of African wild dog, suggests that a large and often terrestrial primate like the chimpanzee could be at risk from predation by these species. Risk of predation by spotted hyena and wild dog is intuitively greater when chimpanzees are terrestrial, and although this may put pressure on arboreal nesting and pre- and post-sleep behaviours, it is unlikely to select for further inaccessibility of nests within trees. However, lion and leopard may put pressure on location of ape nests within trees because they are able climb adeptly [Makacha and Schaller 1969] and leopards are reported to take other primates from their sleeping trees during the night [Busse 1980; Cowlishaw 1994]. Direct observation of predation on apes is extremely rare and most evidence is indirect. The largest data set of evidence of leopard predation on chimpanzees (over five years) is reported in the Tai forest and includes nine likely attacked individuals, but none directly observed and only one leopard scat with chimpanzee bones and hair [Boesch 1991; Zuberbühler and Jenny 2002]. Leopards in Tai were once reported to scavange on a chimpanzee carcass, and of two radio-collared individuals, one leopard actively avoided vocalising chimpanzee parties [Jenny and Zuberbühler 2005]. Evidence of leopard predation on chimpanzees (6/196 leopard scats) has also been recorded in Lope, Gabon [Henschel et al. 2005]. Evidence of leopard 96 predation on other great apes is yet more rare, although Fay et al. [1995] described a probable leopard attack on lowland gorilla including a digit in leopard scat. Recent systematic studies of predator scats in areas sympatric with other great apes have revealed evidence of leopard predation on gorillas [9/196 scats: Henschel et al. 2005] and bonobos [1/5 scats: D'Amour et al. 2006]. Thus, rates of leopard predation on apes where sufficient data have been collected range from less than 1% to just 5% of scat samples. Evidence of lion predation has been described in Mahale Mountains National Park and the deaths of four individuals were attributed to lion consumption over a four month period [Tsukahara 1993]. However, the lions did not remain in the area long and Tsukahara’s observations are the only reports of large carnivore predation on chimpanzees in Mahale National Park in more than 45 years of research. The limited evidence available suggests that great apes are at low risk of predation by large carnivores, whether due to their low densities [Hayward et al. 2006a], large body size [Zuberbühler and Jenny 2002], or predominant distribution to closed forest habitat [Lehmann and Dunbar 2009]. Campbell et al. [2011] recently reported severe declines in primate and duiker species within Tai National Park, but outside the research study area. One possibility for more evidence of leopard predation on the habituated community at Tai could be reduction in availability of more preferred prey species, like duikers or terrestrial monkeys [Jenny and Zuberbühler 2005]. Finally, great apes may have evolved behavioural counter-strategies to predation, resulting in low observed rates of predation. There are several reports of chimpanzees’ aggression towards leopards and lions, including aimed throwing of sticks, which may be a response to a recognised threat [Boesch 1991; Goodall 1968; Hiraiwa-Hasegawa et al. 1986; Izawa and Itani 1966; Kano 1972; Nishida 1989]. Nest-building could be another such behaviour that limits risk of predation. Although observed rates of predation are currently low, this does not preclude a significant selective pressure from predation during hominid evolution. Shelter construction and predator defence One of the primary functions of animal construction behaviour is defence against predators. Anti-predatory function of shelters takes two forms; concealment and repulsion of attack following detection [Hansell 2005]. Crypsis could be a possible function of nest building in great apes, as a leafy mass may be less detectable than the silhouette of a large-bodied primate on a branch. Such a hypothesis deserves further investigation but is beyond the scope of this study. Nests are relatively large structures and so less likely to function to avoid detection [Hansell 2005]. Two other large mammals build convergent structures to great apes: Brown nosed coatis, 97 Nasua nasua [Olifiers et al. 2009], and Andean (spectacled) bears, Tremarctos ornatus [Goldstein 2002]. In Costa Rica Cebus capucinus have been observed to raid Coatis’ nests to prey upon the young [Perry and Rose 1994]. Coatis’ nests are also hypothesised to function for predator defence against four large felid species; Panthera onca, Puma concolor, Leopardus pardalis, and Puma yagouaroundi [Olifiers et al. 2009]. Andean bear nests are built next to areas of possible conflict with humans (the only predators of bears) where domestic cattle have been killed and fed upon by the bears [Goldstein 2002]. Great ape nests are hypothesised to discourage attack by predators even after detection, by making the builders inaccessible. Many studies have examined great ape nesting in detail and reveal preferences for particular species of trees, areas of the landscape, and morphological characteristics of trees [Ancrenaz et al. 2004; Brownlow et al. 2001; Fruth 1995; Hernandez- Aguilar 2006; Tutin et al. 1995; Wrogemann 1992]. Some of the physical characteristics of trees selected for nesting are in accordance with anti-predatory function of arboreal nesting; for example chimpanzees in Issa select trees that are taller, larger, and have higher lowest branch height than other suitable trees in the vicinity [Hernandez-Aguilar 2006]. However, few studies have compared nesting behaviour across sites that differ in characteristics hypothesised to influence nest function, such as predator presence [Baldwin et al. 1981; Pruetz et al. 2008]. Baldwin et al. [1981] found that chimpanzees in a predator-rich, dry savanna site in Senegal (Mt Assirik) nested higher, in larger groups, and more often within the same trees as conspecifics than chimpanzees in the predator-poor, tropical forest in Equatorial Guinea. However, without controlling for differences in tree height, community size, and available nesting trees it is not possible to determine whether variation in nesting was due to predator pressure or other cross-site differences. Pruetz et al. [2008] addressed this issue in cross-site comparisons by contrasting nesting behaviour in Mt Assirik and predator-poor Fongoli in Senegal, which are separated by only ~50km and have similar vegetation. Chimpanzees in Mt Assirik nested at greater heights and closer together than chimpanzees in Fongoli, supporting the hypothesis that nesting in great apes protects against predation [Pruetz et al. 2008]. Here I aim to further test this hypothesis by comparing two communities of chimpanzees that differ in the presence of predators and in vegetation by controlling for other sources of variation where possible. Specifically, I hypothesise: a) Chimpanzees in Issa, a predator-rich site, prefer to nest in forest patches more frequently than do chimpanzees in Fongoli, a site devoid of predators, as forest may provide greater opportunity for arboreal escape. 98 b) Nesting group size is larger in Issa and on average a greater proportion of the community nests together. c) Nesting sites in Issa more often have an arboreal, alternative escape route than descent to the ground. d) Nests are built more frequently in trees with conspecifics in Issa than in Fongoli. e) Chimpanzees at Issa nest comparatively higher within trees and further out from the tree trunk, controlling for tree size. METHODS Predator presence The study sites of Fongoli and Issa are described in full in Chapter 2. At both study sites opportunistic data were collected on presence of large mammal species. In Issa, we also walked fortnightly 60km of transects, in order to record prevalence and density of sightings and traces of large mammal species. These data are not included in this chapter beyond total frequency of observations during the study period, as they form part of a concurrent study conducted by Alex Piel (UCSD), not yet analysed. Fongoli had little evidence of potential predatory species, whilst to date Issa has a full complement of predators (Table 5-1). Evidence of predators in Issa was also described in Chapter 2. Table 5-1. Evidence of carnivore species in Fongoli and Issa. Species Fongoli Frequency (6 months) Issa Frequency (12 months) Crocuta crocuta Vocalisations 2 Scat & foot prints 7 Lycaon pictus No evidence 0 Seen 1 Panthera leo No evidence 0 Vocalisations, scat, & possible roan antelope kill 5 Panthera pardus No evidence 0 Seen, vocalisations, scat & footprints 12 Previous research also reported no evidence of large predators in Fongoli, although one hyena faecal was observed during nine months of study between 2000 and 2004 [Pruetz et al. 2008]. Two spotted hyena vocalisations were recorded during six months in Fongoli, but these were heard during nest sleep experiments (described in Chapter 4), and no evidence would have 99 been recorded without sleeping out within the study area. This suggests that all potential predators except for hyenas are absent from Fongoli, and even hyenas occur only in very low numbers. Pruetz et al. [2008] reported that predators in Fongoli were common within the last century and so may have been present during the lifetimes of some of the older members of the Fongoli community. In Issa, the only previous long-term study, in 2001-2003, reported frequent observations of potential predators. Hernandez-Aguilar [2006] found on average evidence of leopard every two weeks, spotted hyena every month, lion every two months, and African wild dogs every three months over the course of a 22-month study. She also saw lion and leopard once each during her study period, and when I visited the area of her camp site in 2005 I saw a lion nearby. Hernandez-Aguilar [2006] also reported an apparent link between chimpanzee vocalisations in response to leopard or lion vocalisations. Kano [1972] found the highest density of all four carnivore species in Ugalla, compared to other areas he surveyed in western Tanzania. Less frequent observations of predators during this study, compared to Hernandez-Aguilar’s [2006] observations, may reflect a decrease in predator density in the area, but alternatively predatory species may differentially use areas of the habitat. During my study period in Issa, 2008-2009, the main camp site was based in mountainous terrain at 1600m elevation whilst the main camp site of Hernandez-Aguilar was at 1100m elevation within a broad open flat valley beside a swamp. Open valleys and swamps in Ugalla typically have higher densities of large mammals, and therefore predators, due to availability of water throughout the year in these locations, but data on mammal density in different vegetation types and topographic levels are not yet analysed. The Issa community of chimpanzees ranged from the mountainous terrain over 1600m elevation near the current camp to large open valley flats around the swamp near Hernandez-Aguilar’s 2001- 2003 camp and, despite discrepancies in frequency of observations, were exposed to four types of potential predators. Nest data Only nests measured in Issa in the dry season are included in these analyses, in order to be comparable to Fongoli, where data were collected only in the dry season. In addition, only tree nests are compared. However, 12% of confirmed night nests were built on the ground in Fongoli in this study and Pruetz et al. [2008] suggested that a high prevalence of ground nests may be due to absence of predators in the area. Here I compare only fresh nests built in trees for overnight sleep. The following variables were compared: 100 Vegetation type: I recorded the vegetation type in which each nest was built and compared the frequency of use of open (woodland, wooded grassland, and grassland) versus closed (forest) vegetation types for nesting. The number of groups built per vegetation type was compared (where a nest group spans more than one vegetation type, both were counted) in order to control for pseudo- replication. Fongoli and Issa had similar availability of closed versus open vegetation types [Issa: 1.5% forest, Hernandez-Aguilar 2006; Fongoli: 2% forest, Pruetz 2006]. Group size: Nesting group size was measured as the number of fresh nests found on the same day within 100m from the nearest fresh nest. The size of Issa’s community was at least twice that of Fongoli, so I compared the size of nesting groups recorded in each site as a proportion of total community size. Escape route: I noted the presence or absence of an alternative escape route out of the nesting tree, other than descending to the ground. Nest tree sharing: A proxy measure of proximity, number of nests per tree, was used to assess whether chimpanzees in Issa nest closer together than in Fongoli. Position of nest within the tree crown: In order to control for differences in tree size between Issa and Fongoli the following variables of nest position within trees were calculated post hoc (see Figure 5-1): 1) Nest position as a proportion of tree crown height (a): The height of the nest above the tree crown base (b) was calculated by subtracting the crown base height (c) from the nest height above trunk base (d). I then divided the height of the nest above the tree crown base (b) by the height of the tree crown (a) to give the nest position as a proportion of the tree crown height. Nest position ranges from 0 (bottom of the tree crown) to 1 (top of the tree crown). 2) Nest position as a proportion of tree crown radius: The sum of the distance of the nest to the trunk (e) and the nest to the tree crown edge (f) is taken as the radius of the tree crown (g) (specifically along the horizontal axis of the nest position relative to the tree 101 trunk). I divided the horizontal distance of the nest to the trunk (e) by this tree crown radius (g) to calculate the position of the nest as a proportion of the tree crown radius. Nest position ranges from 0 (exactly above the trunk base) and 1 (at the periphery of the tree crown). Figure 5-1. Nest position as a proportion of tree crown height and nest position as a proportion of tree crown radius were calculated using the depicted measurements. 1) Nest position = c/a (c = d-b). 2) Nest position = e/g (g = e + f). Analyses All data were analysed in PASW. Comparisons of non-normal data were made with Mann- Whitney tests. Where multiple comparisons were used, all p-values were Bonferroni corrected [Field 2005]. Normal data were analysed using one-way analysis of variance, (ANOVA). a b f e g c d 102 RESULTS In Fongoli, 58 nest groups were found, compared to 40 during the dry season in Issa. The proportion of nest groups built in open versus closed vegetation did not differ between Fongoli and Issa (χ2 = 1.65, df = 1, p > 0.05). At Issa in the dry season, 47% of nest groups were built in forest and 53% in woodland, whilst in Fongoli 32% were built in forest and 68% in open vegetation types (Figure 5-1). If low availability of forest vegetation types is taken into account, nests were more frequently found in forest than expected by chance in both sites (Fongoli: χ2 = 322.4, df = 1, p <0.001; Issa: χ2 = 279.1, df = 1, p <0.001), but open habitat (woodland, wooded grassland, and grassland) was used randomly in Fongoli (χ2 = 2.7, df = 1, p >0.05) and avoided in Issa (χ2 = 6.0, df = 1, p < 0.05; Figure 5-1). The median number of nests per group in Fongoli (median = 6, range = 1 – 18) did not differ from Issa (median 4.5, range = 1-26; Mann-Whitney, n1 = 58, n2 = 40, z = 1.07, p = 0.29). However, the proportion of the community that nested together in one group was greater in Fongoli (median = 21%, range = 3-53%) than in Issa (median = 12%, range = 1-39%; Mann- Whitney, n1 = 58, n2 = 40, z = 3.64, p < 0.001). In both Fongoli and Issa, most nests (>80%) had an alternative escape route (Figure 5- 2). There is no significant difference in the proportion of nests built in forest vegetation types which have alternative escape routes in Issa versus Fongoli (χ2 = 0.23, df = 1, p = 0.63). However, in open vegetation types a greater proportion of nests in Fongoli had an alternative escape route than in Issa (χ2 = 21.26, df = 1, p <0.001). The number of nests per tree tended to be greater in Issa (median = 1, mean = 1.4, range = 1-7, n = 226) than in Fongoli (median = 1, mean = 1.2, range = 1-6, n = 317; Mann-Whitney, z = 1.93, p = 0.054). More nests were built in the same trees as other nests in Issa (χ2 = 3.87, df = 1, p 0.049); in Fongoli 36% (n = 411) of nests and in Issa 43% (n = 307) of nests shared a tree with one or more nests. 103 Figure 5-2. Frequency of fresh nest groups recorded in closed (Issa, 1.5%; Fongoli, 2%) or open (Issa, 98.5%; Fongoli, 98%) vegetation in Fongoli and Issa. Figure 5-3. Nests with alternative escape routes in closed (Issa, 1.5%; Fongoli, 2%) or open (Issa, 98.5%; Fongoli, 98%) vegetation in Fongoli and Issa. n  =  21   n  =  20   n  =  44   n  =  23   0   10   20   30   40   50   60   70   80   Fongoli   Issa   %  o f  n es t  g ro up s   Closed   Open   n  =  131   n  =  182  n  =  223   n  =  64   0   10   20   30   40   50   60   70   80   90   100   Fongoli   Issa   %  o f  n es ts  w ith  a lte rn a@ ve  e sc ap e   ro ut e   Closed   Open   104 There was no difference in proportional nest height within the tree crown among tree nests built in forest, woodland, or grassland vegetation types in Fongoli (F = 1.82(2, 331), p = 0.17). In Issa, there was a tendency for nests to be built higher within the tree crown in forest than in woodland (t = 1.93, n1 = 186, n2 = 99, p = 0.055). A one-way ANOVA shows differences between sites and vegetation types (F = 10.49(3,568), p < 0001). Post hoc comparisons confirm that no difference existed between nests built in woodland versus forest vegetation in Fongoli (p = 0.73) or Issa (p = 0.30). Nests in Issa were built proportionately higher within the tree crown than nests in Fongoli (forest, p = 0.002; woodland, p = 0.046; Figure 5-3). Figure 5-4. Nest position in tree crown as a proportion of tree crown height in open and closed vegetation types in Fongoli versus Issa. (┌─*─┐indicates significant difference). Bars outside boxes indicate range. 105 The proportional distance of the nest from the tree trunk did not differ across vegetation types used for nesting in Fongoli (Kruskal-Wallis, H = 0.64(2, 297), p = 0.72), but in Issa, nests built in forest vegetation types were proportionately closer to the trunk than nests built in woodland (Mann-Whitney, n1 = 194, n2 = 100, z = 2.44, p = 0.015). There was no difference in the proportional distance of the nest from the trunk in Issa versus Fongoli in forest vegetation (Mann-Whitney, n1 = 108, n2 = 194, z = 1.81, p = 0.140), although nests were built proportionately farther away from the tree trunk in Issa than in Fongoli in woodland vegetation (Mann-Whitney, n1 = 147, n2 = 100, z = 2.83, p = 0.010; Figure 5-4). Figure 5-5. Nest position in the tree crown as a proportion of tree crown radius in different vegetation types in Fongoli and Issa. (┌─*─┐indicates significant difference). Bars outside boxes indicate range, excluding outliers, which are indicated by circles. 106 DISCUSSION The results of this cross-site comparison do not provide a consensus on how predation influences nesting behaviour. Although relative to availability, forest vegetation types are preferred for nesting in both sites, Issa chimpanzees avoided nesting in open vegetation relative to its availability, whilst Fongoli chimpanzees nested randomly in open vegetation. Pruetz et al. [2008] found the same in comparing use of closed and open vegetation types relative to availability in Assirik (predators present) versus Fongoli. However, my comparison here includes data only from the dry season in both sites and year-round data from Issa show that avoidance of nesting in woodland is seasonal; in the wet season selection of vegetation types for nesting was reversed with a greater proportion of nests groups built in woodland (75%, χ2 = 2.0, df = 1, p >0.05). Across an annual cycle and including nests of all stages of decay, Pruetz et al. [2008] found that Fongoli chimpanzees nested in forest six times less frequently than chimpanzees in Assirik, which is a total frequency of 8% of nests in forest. In Issa, Hernandez-Aguilar [2009] found that only 7% of nests were built in forest compared to a total proportion of 36% of nest groups (or 49% of nests) in this study. In both Senegal and Tanzania, higher proportions of forest nests could be influenced by several factors. In the current study, sample size was lower, only fresh nests were included, and nests were located by tracking chimpanzees compared to systematic survey of nests of all stages of decay by Pruetz et al. [2008]and Hernandez-Aguilar [2009]. It may be that their results were influenced by decay and re-use rates of nests, as nests in forest disintegrate and are re-used at faster rates than nests in woodland [Stewart et al. in press; Chapter 8]. Thus, the relationship between predator presence and increased use of forest vegetation types found in this study and by Pruetz et al. [2008] may be due to other factors. The seasonal difference in use of forest patches for nesting may be more influenced by seasonal availability of food resources or to greater availability of leafy vegetation in forest in the dry season. Issa and Fongoli are both highly seasonal environments. In Issa, woodland feeding tree species fruit less in the wet than in the dry season, but yet fruit more than forest vegetation in the wet and dry season [Hernandez-Aguilar 2006]. In Fongoli, open vegetation types also provide the majority of all chimpanzee food species, and availability of fruits is greatest in the late dry season [Pruetz 2006]. Most trees in Issa and Fongoli lose their leaves from the beginning to the end of the dry season, with very few trees in leaf towards the end of the dry season [Issa, Hernandez-Aguilar 2006; Fongoli, pers. obs.]. Thus, in both of these dry sites the chimpanzees are likely foraging and ranging in open vegetation in the dry seasons, but returning to leafy evergreen forest patches for night nesting. 107 No difference in nesting group size was found in this study. However, Baldwin et al. [1981] found that chimpanzees in Senegal (Mt Assirik) nested in larger groups than chimpanzees in Equatorial Guinea and hypothesised that this difference was influenced by greater predation pressure in Mt Assirik. The median and range of group sizes found at Fongoli and Issa fall within the range of other sites where chimpanzee nest groups have been studied [Fruth and Hohmann 1996]. Tutin et al. [1983] suggested that savanna chimpanzees in Assirik were more cohesive, due in part to increased threat of predation in an open habitat as a high proportion of the community remained together compared to other populations. In contrast, a greater proportion of the community in Fongoli was found to make up nest groups than in Issa. This would be expected in Issa if chimpanzees were aggregating at night due to predation pressure. However, a large number of factors in addition to predation-risk influence sub-group sizes within chimpanzee communities, including food availability and oestrous females [Anderson et al. 2002]. My results could also have been influenced by differing methods of finding nest groups in Fongoli versus Issa; data collection in Fongoli was biased towards nest groups containing males as only males are followed as focal subjects [Pruetz 2006]. However, data collection in Issa may have been similarly biased to mixed and large parties, because most fresh nest groups were detected via the chimpanzees’ vocalisations. Baldwin et al. [1981] also found no difference in inter-nest distance between the two sites, but did find that more nests were built in the same trees in Mt Assirik than Equatorial Guinea. A similar tendency was found in this study, and a greater proportion of nests were built in the same tree as other nests in Issa versus Fongoli, suggesting that the presence of predators may influence proximity of the nesters. However, tree sharing may also reflect the size of the tree or availability of suitable trees, which was not controlled for in this study. Predator presence may fluctuate in Issa and a concurrent study, which deployed an acoustic monitoring system across the study area [Piel and Moore 2007], may provide a method to measure predator presence over time, allowing an intra-site analysis of variation in nesting behaviour in response to increased predator presence. Most nests in Assirik were found to have at least one alternative escape route [Baldwin 1979], which was interpreted as a possible solution to predation pressure from lions and leopards that may be able to climb nesting trees. However, this proportion has not previously been compared to other chimpanzee study sites regardless of predator guild. Contrary to the hypothesis that a greater proportion of nests would have an alternative escape route in a predator-rich (Issa) habitat, results show a greater proportion of nests in open vegetation types in a predator-less (Fongoli) habitat had escape routes. This suggests that the presence of an escape 108 route may be influenced more by the vegetation density, but data on the density of trees in Fongoli and Issa are not yet available for analysis. Potential for escape from a site may be a less important characteristic of a good sleeping site than inaccessibility to predators. For example two studies have demonstrated reluctance of baboons to leave a sleeping tree during the night: Bert et al. [1967 cited in Anderson 1998] found that baboons refused to flee when the authors shone lamps and created disturbance at the base of the tree. Busse [1980] watched baboons harass a leopard in their sleeping tree from the refuge of the terminal branches. Similarly, chimpanzee nests may allow these large apes to take refuge in the terminal branches where leopards cannot get to them. Only by comparing nest position within trees was it possible to control for cross-site difference in other variables that influence nest characteristics; e.g. nest height correlates with tree height (Chapter 3). My results support the hypothesis that a function of nests may be to allow chimpanzees refuge in locations more peripheral and inaccessible in trees to protect against predator attack. Assirik chimpanzees nest higher within trees than both those of Fongoli [Pruetz et al. 2008] and Equatorial Guinea [Baldwin et al. 1981]. However, no study had previously examined the distance of the nest from the trunk, an alternative indicator of inaccessibility, and my results show that chimpanzees in Issa also nest more peripherally within trees than chimpanzees in Fongoli. Other large-bodied primates like baboons have been suggested to sleep in peripheral locations within the tree crown as a protective measure against predation, resulting in males nesting closer to the trunk than females or juveniles [Anderson 1984]. Yet other large primates, like gibbons, sleep leaning up against the trunk of smaller sleeping trees [Reichard 1998]. Gibbons are vulnerable to predation by several sympatric species, including several felids, birds of prey, pythons and humans. They may have to rely on selecting inaccessible trees for protection from predators [Reichard 1998]; especially as they lack nest-building technology that might allow larger -bodied apes to sleep more peripherally. Further investigation is needed to test whether limb locations selected for nesting are too thin to support the weight of the builder without the construction of a nest combining multiple supporting branches. I had to access many nest locations using a more centrally located point of rope attachment, or an attachment within a neighbouring tree, because the branches around the nest were too small to support my weight. Pruetz et al. [2008] posed a provocative hypothesis in their discussion of the anti- predatory function of arboreal nesting whereby they predicted that arboreal nesting in Fongoli will decrease as the chimpanzees continue to live in a predator-free environment. Mean nest height recorded at Fongoli in this study was only 4.9m compared to a mean of 7.1m recorded 109 during the dry seasons in 2000 to 2001 by Pruetz et al. [2008]. The proportion of ground nests in this study (12%) was also a factor of four times greater than that found by Pruetz et al. [2008; 3%]. These results accord with their hypothesis. However, habituation efforts began at the same time as Pruetz et al. [2008] collected data on nests in Fongoli. Hicks [2010] found that a higher prevalence of ground nesting was associated only with lower human density and evidence of poaching across 160km of transects through the northern Democratic Republic of Congo. In contrast, Hicks [2010] found no association between density of signs of other non-human predators and the prevalence of ground nests. The Fongoli chimpanzees live in close habitation with local humans, but the sudden increased human interest in chimpanzees (for habituation) during Pruetz et al.’s [2008] study might have increased nesting height and decreased ground nesting temporarily in this community. Also, as only fresh nests were included in this study, differences in proportion of ground nests and nest heights may have been influenced by a differential decay or detection rate of ground nests. A large number of factors influence nest site selection in chimpanzees, of which danger from terrestrial predators is likely to be one. The anti-predator function of nests is supported by relatively more peripheral nests built within trees where there is high predator presence. 110 Chapter 6 Thermoregulatory nest function: variation in nest characteristics, shape, and architecture in response to weather Above, thick nest built at beginning of dry season; below, thin nest built at end of dry season, Fongoli 111 INTRODUCTION Across species, animal-built structures most frequently function as shelters that provide protection against environmental conditions including extremes of climate and threat of predation [Hansell 2005]. Great ape nests are often classed apart from ‘shelters’ or ‘homes’ of other building species because they are used individually for short periods of rest, subsequently abandoned, and never function as a fix-point for repeated use or caching of offspring [Hediger 1977; Kappeler 1998]. However, in Chapter 4 I argued that great ape nests function as temporary shelters that provide protection against environmental conditions and outlined several hypothesized functions of ape shelters: anti-predation, anti-parasite, and thermoregulation. This chapter provides a further test of the thermoregulatory hypothesis by investigating whether nests vary with overnight weather conditions in ways that reflect hypothesised thermoregulatory function. Most studies of variation in great ape nest building have focussed on dichotomous seasonal comparisons. Baldwin et al. [1981] found that chimpanzees at Mt. Assirik in Senegal and in Equatorial Guinea nested higher and in more open sites during the rainy than the dry season; the authors suggested that this may allow chimpanzees to dry faster at dawn after wet nights in the rainy season. This pattern was also seen in chimpanzees in central Africa [Wrogemann 1992], and whilst bonobos also nested higher in the wet season, cover over nests remained constant throughout the year [Fruth and Hohmann 1994a]. However, bonobos tended to build more open nests on rainy nights [Fruth 1995]. In contrast, gorillas in Equatorial Guinea nested in more covered locations in the rainy season [Groves and Sabater Pi 1985]. Studies of gorilla nesting patterns in Gabon, Central African Republic, and more recently in Cameroon revealed that gorillas build more arboreal nests during the wet season and rainy months [Mehlman and Doran 2002; Remis 1993; Sunderland-Groves et al. 2009; Tutin et al. 1995] and arboreal nests are presumed to provide greater insulation in addition to elevating the builder off the wet ground. Nest complexity has been quantified in gorilla nesting studies through the use of an ordinal scale of five types of increasing complexity, defined by the level of structure and type and amount of plant materials incorporated [Tutin et al. 1995]. Mehlman and Doran [2002] investigated variation in nests beyond seasonality, and found independent significant relationships between the construction of fuller, more elaborate nest types in colder conditions and on nights following rainfall. In warmer conditions the gorillas often slept on the bare earth or minimal constructions [Mehlman and Doran 2002]. Groves and Sabater Pi [1985] conducted a preliminary study to quantify this difference by counting leaves and found more leaves were used in gorilla nest construction in Equatorial Guinea during the wet season. In Chapter 3, I described 112 some seasonal variation in chimpanzee nest characteristics, shape, and architecture of nests built in Issa, Tanzania. However, as can be seen in the study site section of Chapter 2, intra-seasonal variation can be greater than inter-seasonal variation in weather conditions; even within the dry season in Fongoli temperature, relative humidity, and wind speeds are more variable than in Issa across a whole year cycle. By analysing variation in nests across a range of weather conditions, it may be possible to determine which aspects of weather are influencing seasonal differences, i.e. wind, precipitation, humidity, or temperatures, and perhaps discern causes of variation that may be masked in a seasonal comparison. Since preliminary descriptions of ape nest architecture [chimpanzee: Bolwig 1959; Goodall 1962; bonobo: Horn 1980; Kano 1979; orangutan: MacKinnon 1974; gorilla: Schaller 1963], few studies have sought to measure and quantify nest shape and architecture beyond classification of nests into types [e.g. Tutin et al. 1995], which often precedes more detailed quantitative measures of technology. Fruth [1995] conducted a pioneering study in which she measured and described the shape and architecture of 24 wild bonobo nests. These methods were expanded for the current study resulting in a large number of architectural measurements (detailed in Chapter 2) of nests across two chimpanzee study sites; Fongoli and Issa. As these measurements record different variables of the same nest, measures are often inter-related. The first aim of this study was therefore to investigate nest structure using a principal components analysis to extract the key components of nest shape and architecture. Principal components analysis is frequently used to investigate traits underlying behaviour, e.g. relationship quality [Fraser et al. 2008], and in analysis of variation of complex shapes, e.g. primate crania [Fleagle et al. 2010]. This method results in composite quantitative measures of components of shape and architecture across all nests, which can be meaningfully compared between sites and nest types. The second aim of this study was to investigate variation in nest characteristics and in principal components of shape and architecture across overnight weather conditions. I aimed to test the hypothesis that tree nests will be built in locations that facilitate rapid drying in wetter conditions. Specifically, I tested the hypotheses that chimpanzees will nest higher, higher within the tree crown, further out from the trunk, and closer to the periphery of the tree crown in conditions of greater relative humidity and rainfall. Shape and architecture of nests are also hypothesised to vary in ways that reflect insulatory function. In particular, nests are hypothesised to provide greater insulation in colder and wetter conditions, and to provide stronger structural support during windy conditions. Experimental studies of bird and mammal nests have repeatedly shown that thickness, lining, amount and density of material used in shelter construction increases insulation [Gedeon et al. 2010; Pinowski et al. 2006; Redman et al. 1999; 113 Skowron and Kern 1980]. Structural support through the size and shape of the nest-cup correlates with body size across a number of bird species, and a recent analysis revealed that structural support is the primary factor driving design of nest shape in birds [Heenan and Seymour 2011]. Thus, insulation of chimpanzee nests is likely provided with greater nest thickness, more nest lining, mattress material, and overall amount of material and building steps; structural support of chimpanzee nests is likely stronger in nests with more support steps, main branches, and support branch diameter. This chapter investigates whether these measures of chimpanzee nest insulation and support vary with weather conditions in ways that reflect their function. Finally, resultant patterns of variation in shape and architecture with weather conditions in Fongoli versus Issa are discussed. Do chimpanzees in Fongoli and Issa modify their nests in similar ways in response to weather conditions? Differences in how nests vary may reveal possible cultural variation in nest building, not explained by other factors. If nests are stereotypically built, variation should follow similar patterns, and differences in patterns of variation in nest shape and architecture between the two sites be largely explained by differences in climate. METHODS Study sites and climate This chapter investigates variation in nest characteristics, shape, and architecture, with overnight weather conditions in Fongoli and Issa. As described in Chapter 2, six data loggers were deployed for the duration of the study periods, in different vegetation types in Fongoli, and in different vegetation types and topographic levels in Issa. Data loggers recorded temperature, relative humidity, and wind gust speed every 30 min at both sites, but also recorded leaf wetness in Issa. There was little to no rainfall in Fongoli, but in Issa an event data-logging rain gauge recorded every 2mm of rainfall. In Fongoli, climatic data were averaged across pairs of loggers deployed in the vegetation types described in Table 2-1, which represent different vegetation types used for nesting: forest (ecotone and gallery), woodland (woodland and wooded grassland), and grassland (bamboo and grassland; see Chapter 2). These data were taken to be representative of microclimate of these vegetation types across the study site. In Issa, climatic data from each logger were taken to be representative of the microclimate of each vegetation type (forest or woodland) and topographic level (plateau, valley, and slope) across the study area. An overnight period of 12 hrs, from 18:00 to 06:00 hrs, was selected for analyses following the only study to measure the inactive period of chimpanzees in the wild [Lodwick et 114 al. 2004], and because all nests included in this study were built for overnight sleep. For each night, and in each vegetation type, on which a nest was built in Fongoli and Issa, I calculated the mean overnight temperature, relative humidity (plus gust speeds and leaf wetness in Issa), using data recorded every 30 min on data loggers deployed in each vegetation type and topographic level (between 18:00-06:00 hrs). In Issa, because data were collected also during the rainy season, I calculated the total volume of rainfall (mm) during the overnight period on the night each nest was built. However, as not all wind loggers functioned for the whole study period in Fongoli, mean overnight gust speeds were pooled across data loggers deployed in open (woodland, wooded grassland, plateau, and bamboo) versus closed (gallery and ecotone) vegetation types. During the Issa study period, only valley forest, slope forest, valley woodland and slope woodland were used for nesting, which is why here only the data from these representative loggers were described and used to represent the conditions in the vegetation type and on the night on which the nest was built. In Fongoli data were collected during only the dry season from October 2007 to March 2008. In Issa, data loggers were not deployed until January 2009 so data on nests are included here only during the wet season from January to April 2009 and the dry season from May through September 2009. Within-site differences in microclimate across vegetation types were analysed with a repeated measures analysis of variance (ANOVA) or the non-parametric equivalent, Friedman’s ANOVA, each followed post hoc by Bonferroni corrected pairwise comparisons. Mean overnight climatic measures were averaged across vegetation types to compare between sites using student’s t-test or Mann-Whitney test. Data collected Data on nest characteristics, shape, and architecture are included in this chapter to investigate whether nests vary with overnight weather conditions in ways that reflect nest function. Data collection methods were described in Chapter 2 and characteristics, shape, and architecture of nests in Fongoli and Issa were described and compared in full in Chapter 3. Nest character i s t i c s In this chapter I analyse variation in night nest type and position in response to overnight weather conditions. Nest types include: Ground, 1st fork, Latter fork, Tree crux, Tree top, Tree branch to tree branch, Tree top to tree branch, Tree top to tree top, and Liana. Variables of nest position included nest height, nest distance from trunk, nest distance from tree-crown-edge, nest 115 position within the tree crown, and proportional distance from the trunk (see Chapter 2 for further description of these measurements). Nest shape and archi tec ture A large number of variables were recorded in order to fully describe nest shape and architecture (see Chapter 2). However, as many of these variables are inter-correlated, I used a principal components analyses (PCA) to obtain composite measures of nest shape and architecture. PCA is a statistical technique that can be used to identify clusters of variables, or patterns of correlations within sets of variables, which represent sets of underlying factors, or principal components [Tabachnick and Fidell 2007]. The PCA provides coefficients of correlation between each nest structure variable and each extracted component; correlated variables which could be measuring aspects of the same underlying factors. PCA reduces these groups of interrelated variables into underlying dimensions or components and the cumulative amount of variance explained by the components. As extracted components are by definition uncorrelated with each other, the total variance explained is the sum of the variance explained by each extracted component. I considered coefficients of correlation greater than 0.5 or less than -0.5 to be high loadings. A varimax rotation was used, which is an orthogonal rotation method that minimizes the number of variables that have high loadings on each component and is most commonly used to simplify the interpretation of the components [Tabachnick and Fidell 2007]. I used a minimum eigenvalue of 1.00, which is standard for PCA using similar sample sizes, in order to determine the number of components extracted from the PCA [Tabachnick and Fidell 2007]. I entered 25 of the variables of nest shape and architecture described in Chapter 2 into the PCA (Table 6-1). These variables were selected for their comparability between nest types and study sites. For example, in Chapter 3, I described in greater detail the number of main branches, side branches, twigs, and leaflets used in nest construction. However, these variables depend on definitions of ‘branch’ versus ‘twig’ versus ‘leaflet’, which in actuality encompass a range of branch diameter sizes, and these definitions may not be functionally relevant to chimpanzee nest building. This cline can be seen in Chapter 3 where differences were found between the number of different types of branches used in nests in Fongoli and Issa, but not between the numbers of branches of different sizes (measured as diameter at point of impact). Therefore here I included the similar measure of the number of branches of different diameter [sensu Fruth 1995] and not the type of branch used. In addition this permits comparison across all types of nest and nesting material, whether grass, liana, shrub, or tree. Variables included in the 116 PCA were also determined by a measure of sampling adequacy proposed by Kaiser [1970; termed the Kaiser-Meyer-Olkin measure of sampling adequacy (KMO)]. I used a KMO value greater than 0.7, which indicates that the variables included in the principal components analysis are sufficiently inter-correlated to yield distinct and reliable principal components for a sample size of about 100 [Hutcheson and Sofroniou 1999]. Two relevant measures of nest architecture that did not correlate sufficiently are therefore analysed separately to investigate their variation in response to overnight weather conditions; these are sum of selected support branch diameters, and sum of constructed support branch diameter. Table 6-1. Variables in the Principal Components Analysis of nest shape and architecture. Variable Variable description (see Chapter 2) Shape measurements Length Longest length, or diameter, across the upper surface of the nest Width Perpendicular width at the midpoint of nest length Misshapenness Total difference between the minimum and maximum lengths of nest radii Mean thickness Mean thickness of measures taken through the nest edges and centre Central thickness Distance vertically through the centre of the nest Depth Vertical height from the surface of the nest bowl to the height of nest edges Depth un-sprung Depth measured as above after depressing the nest centre Architectural measures Lining steps Number of building steps of detached material placed on nest surface Mattress steps Number of building steps providing leafy material, but little support Support steps Number of building steps providing support, but little leafy material Branches construct support Number of manipulated branches providing weight bearing support Main material Main branches, or grass clumps, which contribute other material to nest All material Total number of branches, twigs, grass clumps, lianas, etc., manipulated Bends <1cm diameter Number of different types of impacts on building material of different diameter sizes (with diameter measured at point of return or proximal end). Bend is defined as less than 50% severed, whilst break is more than 50% severed, and detachment is completely severed. Bends 1-2cm diameter Bends 2-3cm diameter Bends >3cm diameter Breaks <1cm diameter Breaks 1-2cm diameter Breaks 2-3cm diameter Breaks >3cm diameter Detachments <1cm diameter Detachments 1-2cm diameter Detachments 2-3cm diameter Detachments >3cm diameter 117 Analyses Variables of nest position are non-normally distributed and so a series of bivariate Spearman rank correlations was done to investigate how each variable of nest position varies with mean overnight temperature, relative humidity, and gust speed on the night following nest construction. All reported p-values were Bonferroni corrected, to control for Type-1 error. Nest type is categorical, so the proportion of nests of each type was correlated with mean monthly overnight weather conditions (two-tailed), which I limited only to those that had a significant effect on nest position. The resultant principal components of nest shape and architecture were rotated to be orthogonal, and so it was necessary to analyse each individually rather than entering components into a multivariate multiple regression. Therefore a series of multiple regressions was done with dependant variables of each principal component, and independent variables of mean overnight climatic measures, in order to investigate the influence of weather conditions on nest shape and architecture for each study site separately. Backward stepwise multiple linear regression models were generated using the following independent variables of weather conditions in Fongoli: mean overnight temperature, mean overnight relative humidity, and mean overnight gust speed. In Issa climatic variables were collinear, so I did a second PCA, in order to extract principal components of weather conditions for use as independent variables in the multiple linear regression models. When necessary, I applied a negative inverse transformation to the data, to correct for skewness and kurtosis. If data could not be normalised, then non-parametric bivariate correlations were used and a Bonferroni correction applied to control for multiple comparisons. RESULTS Weather conditions Mean overnight climatic measurements differed between each vegetation type in Fongoli and Issa (Table 6-2). Mean overnight temperatures in Fongoli were lowest in forest, followed by woodland and highest in grassland, whilst mean overnight relative humidity was greatest in forest and lowest in grassland (temperature, F (2, 173) = 355.9, p < 0.001; relative humidity, Friedman’s two-way analysis of variance, n = 175, df = 2, χ2=334.1, p < 0.001; Bonferroni corrected pairwise comparisons significant to p < 0.001; Table 6-2). Gust speeds were higher in open than closed habitats (Wilcoxon’s matched pairs, n = 175, z = 11.5, p < 0.001). In Issa, mean overnight temperatures were colder in forest than woodland, and on woodland slope than valley, although forest slope and valley did not differ (temperature, F (2, 173) = 355.9, p < 0.001, post hoc Bonferroni corrected pairwise comparisons significant to p < 0.001; 118 Table 6-2). Relative humidity was highest in forest valley, followed by forest slope and woodland valley (which did not differ), and lowest in woodland slope (Friedman’s two-way analysis of variance, n = 251, df = 3, χ2 = 390.6, p < 0.001; post hoc Bonferroni corrected pairwise comparisons significant to p<0.01). Leaf wetness varies similarly to relative humidity across vegetation types (Forest valley: 1.8-100%, mean 34.1%; Forest slope: 1.6-76.6%, mean 13.1%; Woodland valley: 1.1-100%, mean 22.6%; Woodland slope: 0.3-97.4%, mean 9.3%). Mean overnight gust speeds were lowest in forest valley, greater in woodland valley, and greatest in woodland slope, followed by forest slope (Friedman’s two-way analysis of variance, n = 253, df = 3, χ2 = 558.2, p < 0.001; post hoc Bonferroni corrected pairwise comparisons significant to p<0.01; Table 6-2). Table 6-2. Mean overnight weather conditions compared at Fongoli and Issa across vegetation types. Site Vegetation type Mean overnight temperature (°C) Mean overnight relative humidity (%) Mean overnight gust speed (m/s) Min Max Mean Min Max Mean Min Max Mean Fongoli Forest 13.9 31.6 21.3 20.3 100.0 56.8 0.00* 1.54* 0.15* Woodland 15.1 32.8 22.3 14.7 100.0 52.9 0.04* 5.18* 0.89* Grassland 17.5 33.2 23.8 13.0 98.7 43.0 Mean across vegetation types 15.7 32.3 22.5 16.0 99.6 50.7 0.02 3.36 0.52 Issa Forest slope 14.8 21.9 18.4 38.2 100.0 76.6 0.22 2.74 1.52 Forest valley 14.7 22.4 18.5 46.8 100.0 83.8 0.00 1.05 0.09 Woodland slope 15.1 23.1 19.2 35.6 100.0 74.5 0.00 5.66 0.94 Woodland valley 15.5 24.9 19.7 30.3 100.0 76.0 0.00 2.97 0.57 Mean across vegetation types 14.9 22.9 18.9 39.7 100.0 77.7 0.08 2.50 0.79 * Mean overnight gust speed in Fongoli is calculated for closed (forest) versus open (woodland and grassland) vegetation types. Table 6-2 compares mean overnight weather conditions for the duration of the study periods in Fongoli and Issa. As data were collected during the rainy season in Issa, I compared climate also between the dry seasons in Fongoli and Issa. Mean overnight temperatures are greater in Fongoli than in Issa in the dry season (t-test, n1 = 175, n2 = 153, t = 10.6, p < 0.001), but have a larger range (Figure 6-1). Mean overnight relative humidity is greater in Issa in the dry season than in Fongoli (Mann-Whitney test, n1 = 175, n2 = 153, z = 6.0, p < 0.001), and does not drop as low as in Fongoli (Figure 6-2). Mean overnight gust speeds are greater in Issa in the dry season than in Fongoli (Mann-Whitney test, n1 = 175, n2 = 153, z = 7.8, p < 0.001), but the range in Fongoli is similar (Figure 6-3). 119 Figure 6-1. Mean overnight temperature compared between Fongoli dry and Issa dry and wet season (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by circles. Figure 6-2. Mean overnight relative humidity compared between Fongoli dry and Issa dry and wet season (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range. ┌─────*─────┐ ┌─────*─────┐ ┌─────*─────┐ 120 Figure 6-3. Mean overnight gust speed compared between Fongoli dry and Issa dry and wet season (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by circles and stars. Variation in nest position and type with weather conditions Fongol i In Fongoli, nests were built higher (r = 0.15, n = 409, p = 0.006), and higher within the tree crown (r = -0.14, n = 336, p = 0.036) in conditions of greater relative humidity. There was no other relationship between nest position and wind speed or temperature. Proportion of nests open to the sky also does not correlate with mean monthly weather conditions. The proportion of different nest types also varied over time with decreased relative humidity (Figure 6-4). Overall, the proportion of nests built with multiple trees was greater in conditions of greater relative humidity (r = 0.83, n = 6, p = 0.042). The proportion of ground nests built per month tended to increase (r = -0.75, n = 6, p =0.084), and the proportion of nests built on outer forks of trees (r = 0.82, n = 6, p = 0.046) and integrating two tree-side branches increased (r = 0.81, n = 6, p = 0.050) as relative humidity decreased. Despite nest height decreasing with relative humidity, nests were more frequently built in single tree tops as relative humidity decreased through the dry season (r = -0.83, n = 6, p = 0.042). ┌─────*─────┐ ┌─────*─────┐ ┌───────────*──────────┐ 121 Figure 6-4. Proportions of nest types built per month with mean monthly overnight relative humidity. Issa In Issa nests were built higher (r = 0.14, n = 470, p = 0.010) and tended to be built higher within the tree crown (r = 0.12, n = 448, p = 0.075), in conditions of greater rainfall. Higher nests were also built in conditions of greater leaf-wetness (r = 0.13, n = 424, p = 0.035), and tended to be built in conditions of greater humidity (r = 0.11, n = 425, p = 0.10). No other relationship exists between nest position and gust speed or temperature. Proportion of nests open to the sky also does not correlate with mean monthly weather conditions. The proportions of nest types built appear similar across time and monthly rainfall volume (Figure 6-5), although the number of nests found per month was often very small and varied. Only months where more than 15 nests were found were included in the analyses (Figure 6-5). The proportion of integrated nests built per month increased as rainfall decreased (r = - 0.78, n = 8, p = 0.023). The only nest type that varied with rainfall was the proportion of integrated nests of type tree top to tree top, which was greater in months of greater rainfall (r = 0.78, n = 8, p = 0.022). However, if the proportion of integrated nests was compared on rainy versus dry nights there was no difference (χ2 = 0.00, df = 1, p = 0.95). n  =  32   n  =  52   n  =  101   n  =  56   n  =  100   n  =  68   0   10   20   30   40   50   60   70   80   90   100   0%   10%   20%   30%   40%   50%   60%   70%   80%   90%   100%   Oct-­‐07   Nov-­‐07   Dec-­‐07   Jan-­‐08   Feb-­‐08   Mar-­‐08   Re la @v e   hu m id ity  (R H)  %   Pr op or @o n   of  n es t  t yp e   Lianas   Tree  top  to  tree  top   Tree  top  to  side  branch   Tree  to  tree  side  branches   Tree  top   Tree  crux   Outer  fork   1st  fork   Ground   Mean  overnight  RH   122 Figure 6-5. Proportions of nest types built per month across months of varying rainfall. Principal components of shape and architecture Eight components were extracted from a PCA, including variables in Table 6-3 for tree and ground nests across both sites. Together, all components explained 73% of the overall variance in nest shape and architecture; variance explained by each component and loadings for each variable on each extracted component are presented in Table 6-3. Each variable loaded strongly onto one component and loadings of over 0.5 were interpreted as high loadings. The first component consisted of high loadings from bends and breaks <1cm diameter, mattress steps and main material. This suggests that the mattress of a nest is made of small twigs and branches <1cm in diameter, but that the number of such small branches manipulated is also associated with the number of pieces of main material or main branches. Together, these variables represent an overall measure of the amount of material and manipulations of the nest mattress, so I labelled this first extracted component ‘mattress.’ n = 11 n = 71 n = 0 n = 4 n = 16 n = 31 n = 70 n = 29 n = 56 n = 6 n = 151 n = 59 0 50 100 150 200 250 300 350 400 450 500 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% M on th ly r ai nf al l ( m m ) P ro po rt io n of n es t t yp e Tree top to tree top Tree top to tree side branch Tree to tree side branches Tree top Tree crux Outer fork 1st fork Monthly rainfall 123 The variables with high loadings on the second component were detachments < 1cm diameter, line steps, and all material. Two types of detached material seem to be used in nest building; small twigs and branches <1cm in diameter which correlate here with lining steps, and larger diameter detached material that is integrated into the nest structure rather than placed on top as lining. I therefore labelled this second component ‘lining’. That the total amount of all material also correlates with lining suggests that lining is an addition to the main nest mattress, rather than a substitute or alternative building technique used to achieve create a similar nest shape and structure. The third component extracted has the highest loading on bends of 2-3cm and 1-2 cm in diameter, branches constructed support, and support steps. The fourth component extracted had the highest loading from breaks 2-3 cm in diameter, but also had high loadings from breaks 1-2 cm, and support steps. With the exception of support steps, these variables are all related to the diameter of branches used for building. As larger diameter branches are always manipulated at the beginning of nest building, these branches are necessarily related to the number of support steps. Thus components 3 and 4 may be described as different support types, the first is labelled ‘bend support’, as this component is associated more with number of bends than breaks, and the latter ‘break support’. The fifth component consists of detached branches 1-2 cm, 2-3 cm, and >3 cm in diameter, in addition to large breaks >3cm in diameter. These large detached branches and breaks do not correlate with ‘lining’, but rather may be another dimension of nest architecture, so I labelled this component ‘fractures’. The last three components extracted relate to different dimensions of nest shape. The sixth component is labelled ‘thickness’ as the variables of central thickness and mean thickness are strongly loaded. The number of bends >3 cm in diameter is also strongly loaded onto the thickness component, suggesting that large bent branches are most likely to contribute to increased thickness of nests. The seventh component is labelled ‘area’, as length, width, and misshapenness are loaded onto this component and reflect two-dimensional measurements, rather than three-dimensional measures of depth or thickness. Lastly, the eighth component extracted is labelled ‘depth’, as it had high loadings only from depth and from depth un-sprung. 124 Table 6-3. Varimax rotated component matrix of nest shape and architecture. Loadings greater than 0.5 of each variable onto components of shape and architecture are highlighted in bold. Component labels and percentage of variance are highlighted in italics. Variables Component 1 2 3 4 5 6 7 8 Bends <1cm diameter (n) .868 -.054 -.018 -.109 -.117 .041 .028 .081 Breaks <1cm diameter (n) .785 -.038 -.080 .180 .008 -.053 .049 .120 Mattress steps (n) .664 .352 .302 .275 .218 .045 .100 .051 Pieces of main material (n) .622 -.003 .224 .396 .138 .102 .119 .194 Detachments <1cm diameter (n) .026 .958 .021 .012 .026 .078 -.058 .098 Lining steps (n) -.069 .872 .053 .019 .037 .145 -.046 .149 Pieces of material (n) .496 .613 .284 .372 .127 .133 .061 .052 Bends 2-3cm diameter (n) .006 .043 .775 .118 -.038 -.035 .233 -.095 Branches construct support (n) -.049 .050 .706 .343 .054 .217 -.130 .178 Bends 1-2cm diameter (n) .323 .006 .616 -.206 -.027 .335 .151 -.008 Support steps (n) .073 .177 .598 .510 .132 .248 -.132 .035 Breaks 2-3cm diameter (n) .131 .101 .073 .806 .048 .027 -.019 .085 Breaks 1-2cm diameter (n) .295 -.035 .270 .583 .241 .047 .060 .043 Detachments >3cm diameter (n) .066 -.019 .002 .039 .838 -.013 -.019 .003 Detachments 2-3cm diameter (n) -.025 -.047 .049 .090 .826 .081 -.011 .126 Detachments 1-2cm diameter (n) -.060 .421 -.089 .106 .632 .176 .020 .171 Breaks >3cm diameter (n) .108 .134 .082 .453 .524 .054 .084 -.116 Mean thickness (cm) .162 .192 .134 -.009 .082 .882 .131 .063 Central thickness (cm) -.038 .185 .202 .081 .058 .865 .079 -.022 Bends >3cm diameter (n) -.081 -.102 -.003 .414 .120 .524 .259 .031 Nest length (cm) .107 -.060 .138 -.040 -.046 .042 .859 .118 Nest width (cm) .111 .034 .194 -.072 -.062 .186 .803 .108 Misshapenness (cm) -.019 -.051 -.201 .221 .152 .107 .622 -.182 Nest depth (cm) .146 .132 -.053 .093 .064 -.094 -.019 .890 Nest depth un-sprung (cm) .176 .161 .087 .024 .111 .159 .111 .861 Component labels Mattress Lining Bend support Break support Fractures Thickness Area Depth Variance explained 11.2% 10.2% 9.3% 9.2% 9.2% 8.8% 8.1% 7.2% 125 Cross-s i t e comparison o f pr inc ipal components o f nest shape and archi tec ture Further support for the validity of the principal components analysis of shape and architecture was shown by repeating comparison of shape and architecture across study sites and seasons. Results of Chapter 3 were confirmed in a one-way analysis of variance to test differences in shape and architecture across ground and tree nests in Fongoli and dry and wet season nests in Issa. There was no difference in the principal components of mattress (F (3,187) = 1.08, p = 0.36), lining (F (3,187) = 0.89, p = 0.45), or depth (F (3,187) = 1.79, p = 0.15) across study sites, seasons in Issa, or between ground and tree nests in Fongoli. There were significant differences across groups in bend support (F (3,187) = 19.54, p < 0.001), break support (F (3,187) = 6.20, p < 0.001), thickness (F (3,187) = 11.78, p < 0.001), and area (F (3,187) = 5.09, p = 0.002). Post hoc tests revealed that bend support (p < 0.001) and break support (p = 0.045) values were lower in ground than tree nests in Fongoli, and in Issa dry versus wet season nests (bend support, p = 0.007; break support, p < 0.001). There was no difference in bend support or break support between tree nests in Fongoli or Issa dry or wet seasons. Ground nests are thinner than tree nests in Fongoli and Issa (p < 0.001). Tree nests in Fongoli tend to be thicker than dry season nests in Issa (p = 0.093) but there is no difference in thickness between tree nests built in Fongoli and in Issa in the wet season (p = 0.22). However, nests built in Issa in the wet season are thicker than those built in the dry season (p = 0.009). Fongoli ground and tree nests are similar in area (p = 1.00), as are Issa dry and wet season nests (p = 0.99), but Fongoli tree nests are larger in area than dry season nests in Issa (p = 0.002). The principal component fractures was analysed using a Kruskal-Wallis test, which revealed no difference in nests across sites or seasons (H (3, 187) = 6.17, n = 191, p = 0.10). Variation in shape and architecture with weather conditions More principal components of nest shape and architecture correlate with mean overnight weather conditions in Fongoli than Issa (Tables 6-4, 6-5 and 6-7). Fongol i Results of backward stepwise linear regression models including all Fongoli nests are reported in Table 6-4. The amount of mattress material increased with greater relative humidity, which explained 9% of the variation in ‘mattress’. Nest lining, bend support, thickness, and depth were all greater in colder overnight conditions in Fongoli, although the amount of variance explained by mean overnight temperature is small; 10%, 9%, 11%, and 6% respectively. However, thickness also correlated negatively with gust speeds, and tended to decrease with greater relative 126 humidity. Thus, nests were thicker in colder conditions, yet thinner in windier conditions. Increased break support correlated with greater gust speeds and relative humidity, which explains more total variance (21%) because gust speed and relative humidity were inversely correlated. There was no correlation between the size of the nest area and weather conditions; nor was there a relationship between fractures and weather conditions, although there was a tendency towards more fractures in warmer conditions (temperature: r = 0.24, p = 0.060, Bonferroni corrected). Ground nests were included in the above analysis as they are part of the repertoire of typically built nest types in Fongoli. However, materials used for building ground nests are often structurally different from tree nests, e.g. elephant grass (Pennisetum purpureum) is weaker, smaller in diameter, and longer than tree branches. If ground nests were excluded from analyses (Table 6-5), the relationships between weather conditions and principal components of nest architecture and shape remained similar, although a greater amount of variance was explained in mattress (21%), lining (12%), break support (24%) and depth (18%). The amount of mattress material was positively correlated also with increased gust speeds. In tree nests, bend support and thickness were no longer correlated significantly with any weather variables, but thickness approached a tendency to still correlate with decreased temperature. Greater diameter of constructed support also correlates with greater wind speeds, but sum of selected support diameter does not correlate with any weather conditions (Table 6-5). 127 Table 6-4. Fongoli, all nests: Model statistics of stepwise backward multiple regression of each principal component of nest architecture against possible predictors of mean overnight temperature, mean overnight relative humidity, and mean overnight gust speed. R2 value represents the amount of variation explained by the model (e.g. in model with R2 value of 0.09, the significant variables explain 9% of variance). Constant and significant variables of the best model are presented through b values, standard error of b and standardised β. The value and the sign (+/-) of Beta indicate the strength and direction of the proportional relationship between the variables. Dependant variable Independent variable& Regression model b b Std. Error Standardized β R2 F-value df1, df2 p Mattress Constant 0.09 5.46 1, 96 0.022 -0.48 0.20 Relative humidity 0.01 0.00 0.23* Lining Constant 0.10 7.67 1, 96 0.007 0.72 0.07 Temperature -0.01 0.00 -0.27** Bend support Constant 0.09 8.07 1, 96 0.005 1.19 0.51 Temperature -0.07 0.02 -0.28** Break support Constant 0.21 12.03 2, 95 <0.001 1.08 0.12 Relative humidity 0.01 0.00 0.58*** Mean gust speed 0.24 0.06 0.50*** Thickness Constant 0.11 4.90 3, 94 0.003 2.23 0.70 Temperature -0.07 0.02 -0.31** Mean gust speed -0.42 0.20 -0.27* Relative humidity -0.01 0.01 -0.24+ Area - 0.01 0.17 3, 94 0.91 Depth Constant 0.06 5.61 1, 96 0.020 1.00 0.46 Temperature -0.05 0.02 -0.24* &Only variables present in the model are shown, β , *** p<0.001, ** p<0.01, * p<0.05, +p<0.10 128 Table 6-5. Fongoli, ground nests excluded: Model statistics of stepwise backward multiple regression of each principal component of nest architecture, in addition to further models including selected and constructed support diameter, against possible predictors of mean overnight temperature, mean overnight relative humidity, and mean overnight gust speed. R2 value represents the amount of variation explained by the model (e.g. in model with R2 value of 0.09, the significant variables explain 9% of variance). Constant and significant variables of the best model are presented through b values, standard error of b and standardised β. The value and the sign (+/-) of Beta indicate the strength and direction of the proportional relationship between the variables. Dependant variable Independent variable& Regression model b b Std. Error Standardized β R2 F-value df1, df2 p Mattress Constant 0.21 8.51 2, 70 <0.001 -1.33 0.33 Relative humidity 0.02 0.01 0.59*** Mean gust speed 0.37 0.17 0.32* Lining Constant 0.12 6.94 1, 71 0.010 0.69 0.09 Temperature -0.01 0.00 -0.30** Bend support - 0.00 0.09 3, 69 0.97 Break support Constant 0.24 13.19 2, 70 <0.001 1.01 0.14 Relative humidity 0.01 0.00 0.64*** Mean gust speed 0.35 0.07 0.67*** Thickness Constant 0.08 2.63 1, 71 0.11 1.14 0.63 Temperature -0.05 0.03 -0.19 Area - 0.01 0.36 3, 69 0.85 Depth Constant 0.18 14.79 1, 71 <0.001 1.91 0.53 Temperature -0.10 0.03 -0.42*** Sum selected support diameter - 0.04 0.04 3, 77 0.99 Sum constructed support diameter Constant 0.10 7.03 1, 68 0.010 0.80 0.02 Mean gust speed 0.08 0.03 0.31** &Only variables present in the model are shown, *** p<0.001, ** p<0.01, * p<0.05, +p<0.10 129 Issa Variables of mean overnight temperature, mean overnight relative humidity, and mean overnight gust speed were multi-collinear in Issa, and so could not be used together in multiple regression analysis without first using a data-reduction analysis. Nests were also sampled in Issa during the wet season, so it was possible to include the influence of rainfall and dew, using leaf wetness data. Thus, these five variables were entered into a second principal components analysis in order to generate principal components of weather conditions to enter into backward stepwise linear regression models, to investigate the influence of weather conditions on shape and architecture in Issa. Two components were extracted from a PCA that explained 73% of the variance in weather conditions (Table 6-6). The first component had high loadings from volume of rainfall, leaf wetness, and relative humidity; these variables all reflected moisture and so the component was termed ‘wetness’. Temperature, and gust speeds were strongly positively correlated, so they cannot be differentiated; temperature and wind had high positive loadings onto the second component, whilst relative humidity was negatively loaded. This component was termed ‘temperature/wind’ for analyses. Table 6-6. Varimax rotated component matrix of mean overnight weather conditions in Issa. Loading scores >0.5 are highlighted in bold. Component labels and variance explained are highlighted in italics. Variables Component 1 2 Overnight rainfall volume .874 -.024 Mean overnight leaf wetness .856 -.188 Mean overnight relative humidity .595 -.564 Mean overnight temperature -.072 .891 Mean overnight gust speed -.140 .804 Component labels Wetness Temperature/wind Variance explained 37.5% 35.9% There was no relationship between most principal components of nest shape and architecture and principal components of weather conditions in Issa (Table 6-7). Nest thickness correlates negatively with temperature/wind, as wind and temperature were collinear, the influence of each could not be separated, but it is intuitively more likely that nest thickness is most influenced by colder conditions rather than slower gust speeds. Nest thickness and break support also correlated with wetter conditions. Finally, there was a tendency for greater depth with greater temperature/wind. There was no correlation between fractures and wetness or temperature/wind (r = -0.08, p =0.47; r = -0.15, p = 0.17 respectively). 130 Table 6-7. Issa: Model statistics of stepwise backward multiple regression of each principal component of nest architecture, in addition to further models including selected and constructed support diameter, against possible predictors of principal components of weather conditions; wetness and temperature/wind. R2 value represents the amount of variation explained by the model (e.g. in model with R2 value of 0.09, the significant variables explain 9% of variance). Constant and significant variables of the best model are presented through b values, standard error of b and standardised β. The value and the sign (+/-) of Beta indicate the strength and direction of the proportional relationship between the variables. Stepwise backward multiple regression of each principal component of nest architecture against principal components of weather conditions; wetness and temperature/wind. Dependant variable Independent variable& Regression model b b Std. Error Standardized β R2 F-value df1, df2 p Mattress - 0.00 0.15 2, 81 0.86 Lining - 0.02 0.93 2, 81 0.40 Bend support Constant 0.04 1.85 2, 81 0.16 Break support Constant 0.07 5.83 1, 82 0.018 1.76 0.03 Wetness 0.07 0.03 0.26* Thickness Constant 0.13 5.88 2, 81 0.004 0.06 0.10 Temperature/wind -0.21 0.10 -0.22* Wetness 0.28 0.10 0.28** Area - 0.03 1.19 2, 81 0.31 Depth Constant 0.04 3.03 1, 82 0.085 0.11 0.12 Temperature/wind 0.20 0.12 0.19+ Sum selected support diameter - 0.01 0.52 2, 81 0.60 Sum constructed support diameter - 0.01 0.21 2, 81 0.81 &Only variables present in the model are shown, *** p<0.001, ** p<0.01, * p<0.05, +p<0.10 131 DISCUSSION Consistent with previous seasonal comparisons of nest height in chimpanzees [Baldwin et al. 1981; Wrogemann 1992], I found that greater moisture (rainfall, relative humidity), is the main influence on chimpanzee nest height and height within the tree crown in Issa and Fongoli. Correlation between nest height and rainfall is found in Issa, although no difference was found in nest heights in the wet versus dry season in Issa (see Chapter 3). Baldwin et al. [1981] suggested that nesting higher in the wet season and so building nests open to the sky may allow individuals to dry out more rapidly after rain stops, or prevent prolonged dripping from overhead vegetation, even after the rain has stopped. This may be especially influential in dry areas like south-eastern Senegal, where showers are usually brief but heavy. That relative humidity also influences nest height in Fongoli during the dry season suggests that higher moisture in the air, or dew, rather than rainfall alone also influences nesting height. Few studies have reported the proportion of nests integrated with multiple trees or plant materials. As described in Chapter 3, previous comparisons have reported bonobos to integrate trees for nest building for 37% of nests, whilst the only data available on chimpanzees nest integration suggests fewer than 10% of nests are integrated [Fruth and Hohmann 1994a]. However, no studies have reported variation in nest types between seasons or across weather conditions. In Issa, more integrated nests were built in conditions of less rainfall; however, this relationship is not clear, as there was no difference in nest integration on rainy versus dry nights. In contrast, in Fongoli, more nests were integrated in conditions of greater relative humidity. Nesting higher within the smaller trees and sparser vegetation of Fongoli may require integration of multiple trees, but testing this hypothesis would require more detailed information about nesting tree morphology and vegetation structure. Koops et al. [2007] investigated the ecological influences on ground-nesting in the Nimba mountains chimpanzees, concluding that social and not environmental factors may influence the greater frequency of ground nesting in this population; they found no seasonal or climatic variation in the proportion of ground nests built. In Fongoli, ground nests are built only seasonally (Pruetz pers. comm.), and a greater proportion of ground nests found tended to be built in conditions of low relative humidity later in the dry season. My study was done only in the dry season and recorded a much greater proportion of ground nests: 12% of (n=411) fresh nests versus 3% reported by Pruetz et al. [2008]. Several factors may have influenced this disparity (e.g. predation or habituation progress, discussed in Chapter 5); these data are from the dry season only, compared to Pruetz et al. [2008] who collected data over several seasons. Also, only fresh nests were used in this study, and often nest-building was observed, so if ground nests disappear 132 more quickly or are more difficult to see from transects than tree nests, then fewer ground nests may have been found in the initial surveys by Pruetz et al. (2008). Although the most frequently used primary building material in ground nests are small trees (24% of ground nests), elephant grass is second most frequently used (20% of ground nests) and it is used only after the grass is dead and dry. Yet, towards the very end of the dry season in March, this grass burns throughout the study area. The availability, or absence, of suitable vegetation could influence ground nesting behaviour in this population. In this study, ground nests were found to be structurally thinner and made with fewer large branches than tree nests; this variation which may result from overall trends for nests to use fewer branches, less support and be less thick and elaborate during dry, warm conditions. Pruetz et al. [2008] noted that although a greater frequency of ground nesting in Fongoli could be due to a lack of predators in the area, Fongoli chimpanzees continue to nest arboreally much more often than ground nesting. Thus, although ground nesting may be proximately influenced by ecological conditions such as predation pressure, available materials, or climatic conditions, ultimately the presence of the behaviour in certain populations may still be social, as suggested by Koops et al. [2007]. The primary goal of this study was to analyse structural variation in chimpanzee nests in response to environmental conditions, across two dry habitat study sites, with the aim being to further test the thermoregulatory function of nests. In contrast to gross characteristics of nests (e.g. nest position, height, nest tree selection), analyses of variation in nest shape or architecture across seasons or weather conditions have not yet been reported for great apes. Although evidence in support of the thermoregulatory function of nests has been found in studies investigating variation in proportions of different nest-types built by gorillas, in which fuller and more arboreal structures are built in the wet season and when rainfall is heavy [Mehlman and Doran 2002; Remis 1993; Sunderland-Groves et al. 2009; Tutin et al. 1995]. Different relationships of variation in shape or architecture with weather conditions were found across the two study sites, with nests in Issa found to vary less with weather conditions than those in Fongoli. However, both sites provide evidence that nest construction likely serves a thermoregulatory function; nest variation in both sites revealed nests to be thicker in cooler conditions and to use more broken branches 1-3cm in diameter and support steps (break support) in conditions of greater moisture (relative humidity and ‘wetness’). These shape and architectural measures reflect greater insulation of nest structure. In Fongoli alone, results support the hypothesis that nests with greater structural support are built in windier conditions. 133 All measures of overnight weather conditions influenced different components of shape and architecture and, although multi-collinearity of climatic variables in Issa complicates interpretation of these influences, these can be inferred. Greater correlation of variation in shape and architecture of Fongoli chimpanzees’ nests with weather conditions may reflect the greater variability of climatic conditions in south-east Senegal; minimum and maximum mean overnight temperatures in Fongoli differed by a mean annual range of 19.3°C compared to 10.6°C in Issa. Yet, minimum mean overnight temperatures and maximum mean overnight gust speeds are similar across the two sites (Table 6-2; Figures 6-1 and 6-3). Chimpanzees in Fongoli and Issa may achieve the same functional goal of thermoregulation or structural stability of the nest by adjusting nest shape and architecture in different ways. For example, in Fongoli the amount of lining added correlated with colder temperatures, and there was no difference overall in the use of lining in nest construction between Fongoli and Issa. In Issa, the amount of lining added to nests did not correlate with weather conditions. Similarly, in Chapter 3, I found no difference in lining between seasons in Issa. In Fongoli, the amount of mattress material used to construct nests increased with colder conditions, and, similarly to nest lining, there was no difference in mattress material between the two sites, but in Issa the amount of mattress material also does not correlate with weather conditions. Issa chimpanzees may achieve the same functional goal of greater insulation in colder temperatures by increasing the amount of breaks 1-3cm in diameter, which also increases nest support (resulting in a negative correlation with wind speed as temperature and wind are collinear), whereas Fongoli chimpanzees increase mattress material <1cm in diameter, add more lining, and adjust the depth of the nest. In Fongoli, nest depth seems to be of greater importance than thickness in cold conditions, or perhaps building a deeper nest may result in a thinner nest as a by-product. Yet in Issa, nest depth is greater in conditions of greater wind speeds, but also greater temperatures; this is most likely influenced by wind speeds, in contrast to Fongoli where temperature influences depth. Although greater wind speeds were recorded in Issa, only in Fongoli did greater wind speeds correlate with greater amount and diameter of support material. Presumably, both depth and amount of material can provide greater structural support and safety in windy conditions. Correlation between support and wind in Fongoli may again be due to greater variability of weather conditions in Fongoli, where overnight wind gust speeds increase suddenly, but predictably in the mid-late dry season from January to March (these strong winds are an annual phenomenon in south eastern Senegal). In Issa, although wind speeds are higher, they are often high throughout the year and may be unpredictable. Issa chimpanzees thus may build year-round with sufficient support. 134 Wetness also appeared to be of greater influence over aspects of nest structure (bend support and thickness) in Issa, than Fongoli. Relative humidity provides a proxy for ‘wetness’ in Fongoli (as data were collected only in the dry season), and more 1-3cm diameter breaks and mattress material were used in conditions of greater humidity in Fongoli. Further data are needed across wet and dry seasons in Fongoli to fully address these cross-site differences. It is surprising that in Fongoli nests overall tend to be thicker than nests built in Issa in the dry season, especially as mean overnight temperatures are higher in Fongoli. One explanation may be the possible difference in body size between chimpanzee subspecies [Morbeck and Zihlman 1989]. This may be because temperatures in Issa also increase to their highest during the dry season, and in Issa nest thickness is also strongly influenced by wetness. Data in rainy conditions in Fongoli are needed to investigate this further. The only weather condition to vary with nest position was wetness (relative humidity and rainfall in Fongoli and Issa, respectively). An alternative hypothesis as to why chimpanzees nest higher in the wet season is that there may be a greater availability of leafy mattress material and support branches for building higher within the tree crown. The relationship between higher nest positions and greater wetness (relative humidity or rainfall) could be due to the apes’ selection of building sites with enough suitable material to build an insulating nest, rather than seeking a position in which to be open to the sky as proposed by Baldwin et al. [1981]. Across-site climatic differences do not seem to be sufficient to explain divergent relationships between nest shape and architecture with weather conditions in the two sites; Issa is windier, yet nests are not more supported, Issa has similarly cold nights to Fongoli, yet use of lining does not increase on these nights. Such differences in nest shape and architecture present possible measures to investigate cultural variation in nest construction in the future. Previous study of behavioural variation in chimpanzees across Africa did not include nest-building in analyses, as this behaviour is universal [Whiten et al. 1999]. However, van Schaik et al. [2003] in a similar analysis of orangutan behavioural variation found vocalisations made at nesting time to be possible cultural variants. Detailed analysis of nest-building techniques may indicate possible socio-cultural variation similar to some tool-using behaviours that are influenced by, but are not explained by, environmental conditions [Schöning et al. 2008]. In addition to differential variation of nest structure with weather conditions, measures of shape and architecture found here to be invariant with weather conditions may be useful for cross-site comparisons. Nest area does not vary with overnight weather conditions and is smaller in Issa than Fongoli, but perhaps this reflects the smaller body size of the east African subspecies (Pan troglodytes schweinfurthii) compared to other subspecies of P. troglodytes [Morbeck and Zihlman 1989]. Amount of fractures 135 is also an interesting component of nest architecture; in some cases nests were constructed almost entirely of large detached branches (pers. obs.), perhaps this results from the brittleness of the tree used, or an individual building style that may be learnt through observation? Additionally, despite the functional variation in shape and architecture described in this chapter to correlate with weather conditions, a large proportion of variation remains unexplained. Some variation could be individual, age, or sex specific, or due to the substrates used for building. I will address further variation across known builders in Chapter 7. This study has shown that principal components analysis is a useful tool for the study of ape nest architecture, and that detailed analysis of shape and architecture can indicate possible ultimate functions of these ape shelters. Further specific tests are needed to determine causality definitively, but this study has shown that there is a relationship between nest structure and weather conditions that reflects the functions of nest building tested here: thermoregulation and a supported and safe platform for sleep. 136 Chapter 7 Sex-bias and social influences on nest-building techniques and behaviour Adult female, Lingua, travelling with Juvenile, Jino. 137 INTRODUCTION Chapter 3 discussed the potential for cultural variation in this ubiquitous material skill, nest- building, in chimpanzees. Few cross-community differences emerged, but there is variation within the two sites studied, Fongoli and Issa. In Chapter 5 and 6, I focussed on how some of this variation may be functional, resulting in the construction of more insulating nests in colder or wetter conditions. However, few studies have investigated social influences on variation in nest building. The aims of this study are twofold: first, I aim to describe the nesting of the savanna chimpanzees in a social context and to provide comparative data on nesting times and duration of building. Second, I aim to investigate possible sex, age, or individual differences in nest building and nest characteristics, shape, and architecture. All weaned great apes build a fresh nest each night, and early reports of the behaviour of nest builders suggested similarities across all species [reviewed in Fruth and Hohmann 1996]. In all hominid species, individuals build nests just before or at sunset [Fruth 1995; Fruth and Hohmann 1993; Goodall 1962; Groves and Sabater Pi 1985]. Duration of the group’s building varies with group size, weather, and species [Fruth 1995; Groves and Sabater Pi 1985]. Lodwick et al. [2004] reported that chimpanzees spent on average 12 hrs per night in their nests, but no study to date has observed great apes overnight in the wild. In captivity, chimpanzees sleep for only about 8 hrs per night, and their sleep patterns are influenced by temperature, relative humidity, age and sex [Videan 2006b]. In the wild the active period was also influenced by sex and by sexual receptivity in females, with non-receptive females remaining in their nests longer than males or receptive females [Lodwick et al. 2004]. Thus, they concluded that the social dynamics of the group (e.g. dominance and mating interactions) and the trade-off with daily foraging requirements influenced the length of the active period in the wild. Seasonal data on nesting are not available from Fongoli, but this community of chimpanzees lives in one of the most extreme environments where chimpanzees survive today [cf. McGrew et al. 1981]. Several differences are expected in the behaviours of chimpanzees living in these dry, hot habitats [Moore 1996]. Following Lodwick et al.’s [2004] conclusions, Fongoli chimpanzees may be expected to adjust their nesting to extend the length of the day, in order to accomplish required social and foraging activities. In this study, I therefore describe the night nesting of groups in Fongoli, including time of onset and duration of building. Although my observations were few and limited, I provide some descriptions of nocturnal behaviour in Fongoli, in relation to possible influences of extreme environment and seasonality on nesting. Nest-building duration ranges from one to seven minutes across all the great apes [Fruth and Hohmann 1996]. Known details of ontogeny and sex differences are few, but are similar 138 across all species [reviewed in Fruth and Hohmann 1996]. The first attempts at nest building occur during the daytime at 8-12 months of age; however, individuals do not begin to make night nests, until after weaning at 3-5 years of age. The development of this skill resembles other technological skills; e.g. termite fishing is mastered between 2.5 and 5.5 years [Lonsdorf 2005], or nut-cracking between 3-7 years [Biro et al. 2006]. Ontogeny of nest building was not a focus of this study and so only snap-shots of building by sex and age classes is described and discussed in terms of likely development of this universal technological skill in the great apes. Previous studies have focused on sex differences in foraging tool use, which has implications for the evolution of sexual division of labour in humans [nut-cracking: Boesch and Boesch 1984; termite fishing: McGrew 1979]. Lonsdorf et al. [2004] found these sex differences began early during development of termite fishing skills. Most studies have discussed differences in tool use for extractive foraging, but sex differences also occur in the use of sticks in play in wild juvenile chimpanzees [Kahlenberg and Wrangham 2010]. A female bias has also been described in a wide-range of tool using behaviours, from self-maintenance to social play, in chimpanzees and bonobos in captivity [Gruber et al. 2010]. Sex differences in nest-building have been described in some form across all of the great apes. In all species, females nest higher than males, and in orangutan, chimpanzee and bonobo females build more day nests [Brownlow et al. 2001; Fruth 1995; Fruth and Hohmann 1996]. Bonobo females build nests earlier in the evening [Fruth and Hohmann 1993] and silverback male gorillas initiate group nest building [Groves and Sabater Pi 1985]. No studies have reported sex differences in initiation of building in chimpanzees, or sex differences in the architectural details of nests in great apes. I hypothesise that the sex differences described for chimpanzee tool use extend to nest-building, with females investing more time and effort (in terms of structural details) in constructing a safe structure for sleep. Nest building is learned in chimpanzees [Bernstein 1967; Videan 2006a] and an initial goal of this study was to investigate whether offspring built similar nests to their mothers, peers, or frequent associates. I hoped to disentangle the relative contributions of environment and social influences on nest building as a material skill. I hypothesised that offspring would build using similar techniques to their maternal model as has been shown for termite fishing and ant- dipping [Humle et al. 2009; Lonsdorf 2005; Lonsdorf 2006]. However, without another means of identifying individual nest builders [e.g. genetic identification, sensu Bradley et al. 2008], such a detailed analysis was impossible with the current sample size. The relatedness between builders is unknown, so instead I conducted a preliminary test to see if there is more inter-individual variation in nests than between individual variation. Individuals would be expected to differ in 139 technique if nest building is socially learned from different maternal models. Alternatively, nest- building techniques may vary in response to the climate, or substrate, while similarities could reflect convergent problem-solving, regardless of kinship. METHODS Only data from Fongoli are included in this chapter, as the Issa chimpanzees are unhabituated to observers. Nesting behaviour Chapter 2 describes observational methods. Here I detail methods of behavioural data collection relevant to individual and group nesting. I also summarise some nocturnal observations. Observations often were limited, as individuals could not be identified after nightfall, and on some occasions, the party or individuals were lost before building as the chimpanzees were harder to see and to follow at dusk and after nightfall. Except for a few night-time follows my observations were made only during the afternoon, until after the last individual in a party had nested. Group behaviour I recorded the following variables whenever possible for each observed group nesting event: Group composition: All individuals within sight were recorded during 15-minute scans during daytime follows. The last scan before nesting was taken as the nest group composition. Group size: Group composition was used to summarise the total number of individuals in the party, as total number of nest-builders, and the number of adult/subadult, adolescent/juvenile males and females. Group nest-building duration: Time from the start of the first nest’s construction, to the completion of the last nest, by all individuals in the party. If after dark, duration was recorded from the beginning to the end of all nest-building sounds. Initiator: Age/sex class and identity of the initiator of nest-building was recorded whenever possible. 140 Last builder: Age/sex class and identity of the last nest builder was recorded whenever possible. Activities: Activities of individuals in the party were recorded ad libitum as feed, rest, groom, play, or travel. These observations recorded individual activities immediately before nesting (below), because focal individuals were switched to record the behaviour of nest-builders who built during daylight. Individual behaviour Focal individuals were followed until nesting, but I switched to record the first individual within sight to begin building, as it was difficult to record behaviour once night had fallen. Thus, individual behavioural observations are biased towards individuals who built earlier in the evening. Where possible I recorded focal nest-builders on film or recorded the following variables with a note-book or a voice-recorder if night fell: Builder identity: Sex/age class or individual identity were recorded wherever possible Activities: I noted the activity immediately before nesting and after reclining in the nest as feed, rest, groom, play, or travel. Start time: Time the individual started nest-building. Building duration: Time from the first branch broke to the last manipulation in building the nest. Pauses greater than one minute were deducted from the whole duration. Building duration invariably ended with the builder lying down in the nest. Duration of use: Although all nests recorded were night nests, some nests were abandoned that evening. When nests were seen to be abandoned the duration of use of the nest was noted. Rest position: Posture of rest was recorded as supine, right-side, left-side, or prone. This position was recorded only for the first position adopted after nest-building, as no observations were made overnight of sleeping posture changes. 141 Nest and tree characteristics, shape and architecture Over the study period in Fongoli, 71 nests of known builder identity were measured for nest and nesting tree characteristics. A further 19 nests of known sex and age class were seen to be built, but individual identity could not be recorded with certainty, due to failing light at or after dusk. Thus, only 71 nests built by 25 different individuals were analysed. The following variables were compared across sex/age classes and between individuals (see Chapter 2 for detailed description of data collection method): Nest characteristics: Nest height, nest height above lowest branch, number of decisions, and distance of nest to trunk. Nest position within the tree crown and the proportional distance of the nest from the trunk were also compared in order to control for selected tree size. Nesting tree characteristics: Tree height, lowest branch height, and DBH. Nest shape: Length, width, misshapenness, circularity, depth (unsprung), central thickness. Nest architecture: Building steps: Total number of steps, number of lining steps, number of mattress steps, number of support steps, proportion of steps interwoven; Support: Sum of constructed support branch diameters, mean constructed support diameter, number of constructed support branches, sum of selected support branch diameters, mean selected support diameter, number of selected support branches; Manipulations: Total number of breaks, bends, and detachments, and number of breaks bends and detachments <1cm, 1-2cm, 2-3cm and >3cm in diameter; Material diameters: Mean diameter of all material, main branches, lone side branches, and side branches Material: Number of main branches, lone side branches, side branches, twigs, and leaflets or leaves 142 Age sex class comparison In order to control for pseudo-replication error, the mean value for each continuous variable (behaviour and nest) was calculated for each individual. Each individual was observed from one to seven times (Table 7-1). Not all nests observed were measured or accessed, and sometimes builders were identified, but their building behaviour was not recorded. Given the small sample size of each age/sex class, adult and subadult age classes were grouped as adult, whilst adolescent and juvenile were classed as immature. If differences exist between adults and immatures in nest building, then this division is most likely to reflect differences; i.e. subadults are fully grown and so likely to build as adults, whilst adolescents are still smaller and so may be expected to build differently. Analyses compared mean nest measurements (nest and tree characteristics) for 10 adult males, 8 adult females, and 6 immature males. Fewer nests were climbed into for shape and architectural measurements, so analyses of shape were compared between only 10 adult males, 5 adult females, and 3 immature males, whilst analyses of architecture were compared between only 10 adult males, 4 adult females, and 3 immature males. Nelly (NE) was the only juvenile female in the community, and her nest was measured only once, so although her nest is described here, it was not included in analyses. Mean measures per individual of behaviour, nest characteristics, shape, and architecture of each sex/age class were compared using Kruskal-Wallis tests of difference. If the result was significant, multiple Mann-Whitney comparisons were used post hoc to determine whether there were differences in nest or tree characteristics between adult males and females, immature and adult males, and adult females and immature males. No Bonferroni correction was applied in this study because although the Bonferroni correction controls for Type 1 error, it can increase Type 2 error [Moran 2003; Perneger 1998] and this effect is greater with very small sample sizes [Garamszegi 2006]. Therefore I report significant results here as preliminary trends that require further study with a greater sample size. 143 Table 7-1. Frequency of nest building recorded and nests measured of known individuals of each sex and age class. ♂ Age Adult Subadult Adol. Juvenile Total Individual KL MM DF BI KM SI FO YO BN LP BO LT DW MI FR LX Behaviour 6 3 6 5 2 7 3 2 4 5 3 0 0 2 0 0 48 Nest characteristics 5 5 5 3 3 3 3 2 4 5 2 2 2 4 3 1 52 Architecture 5 3 2 1 2 1 2 2 3 4 1 1 0 0 1 0 28 Shape only 0 0 0 0 0 0 1 0 1 1 0 0 0 0 1 0 4 Total nests 6 5 6 5 3 7 3 2 4 5 3 2 2 4 3 1 61 ♀ Age Adult Subadult Adol. Juvenile Total Individual LI NI NN DA FA LU TI TM NE Behaviour 0 3 2 0 1 3 0 1 0 10 Nest characteristics 1 1 2 1 2 4 2 6 1 20 Architecture 0 1 2 1 0 1 0 0 1 6 Shape only 0 0 0 0 1 0 0 0 0 1 Total nests 1 3 2 1 2 4 2 6 1 22 Inter-individual comparison The Kruskal-Wallis test can be used with groups differing in sample size, as applied in the above sex and age class comparisons. However, when comparing three groups, a sample size of an average of five per group (or 15 in total across three groups) is recommended [Fowler et al. 1998]. I therefore made comparisons only when there were more than an average of five nests sampled per individual. Nesting onset and building duration were not recorded five or more times for more than one individual, once abandoned nests were excluded, so no individual comparison was made of behaviour. Four adult males were sampled five times (KL, MM, DF, and LP; Table 7-1) so I made a preliminary analysis of individual differences in nest and tree characteristics comparing these males. The nests of two adult females were measured six and four times respectively (TM and LU; Table 7-1), so I used a Mann-Whitney test to make a preliminary case study of difference in nest and tree characteristics of these two females. No immature individual’s nests were measured often enough to compare across individuals. Too few repeated measures of shape and architecture were taken for each individual to thoroughly investigate individual differences. Only two males were sampled enough to meet the minimum criteria for a Kruskal-Wallis test to compare multiple groups, therefore a Mann- 144 Whitney test was used to conduct a preliminary case study comparing the shape and architecture of two males (KL and LP). RESULTS Behaviour Nest group Complete group nest building was observed on 55 nights. On 5 of these occasions, the whole nest group was later abandoned, either while observers were present or at some time during the night, as when we arrived to the nest site the following morning, the nests were discovered not to have been slept in. On six other occasions complete data could not be collected on nest group behaviour; once the group was lost after dark when a gunshot was heard at 19:20 hr and the group travelled quickly away and was lost. On three other occasions, the travelling groups were lost after dark but before nesting, at 19:55, 19:45, and 20:07 hrs, respectively. Finally, twice I left the group before nesting was finished due to the late time of night, as the chimpanzees had not finished or started nesting by 22:30 and 20:30 hrs respectively. Group composition Most groups were of mixed sex composition. Of 49 nest groups for which a complete party count was made, all contained adult males (mean 9, range 1-10, males per group) and 41 contained adult females (mean 2, range 1-5, females per group). In 30 groups, an average of 2 infants and 2 immatures were present (range 1-6 and 1-4 respectively). Time of construction On average, chimpanzee nest groups in Fongoli began building on average at 18:48, and ended building at 19:17. Onset of nesting ranged from 17:09 to 22:11 (Figure 7-1) and nest building began later as the dry season progressed (r = 0.46, n = 55, p < 0.001). Sunset during the study ranged from 18:20 to 19:01 from the beginning to the end of the study, and time of beginning nest building was on average 27 min after sunset (ranging from 53 min before sunset to 195 min after sunset). Although time of sunset was later as the dry season progressed (r = 0.89, n = 50, p < 0.001), the amount of time after sunset that chimpanzees nested was also greater into the dry season as chimpanzees went to bed later than expected by increased hours of daylight (r = 0.31, n = 50, p = 0.030). Larger groups were also more frequent towards the end of the dry season in this study and so onset of nesting also correlated with group size (r = 0.40, n = 55, p = 0.006). 145 Figure 7-1. Onset of nesting for nest groups, at intervals of 10 min (n = 55). Duration of construction The duration of nest group construction ranged from 4-90 min and lasted on average 29 min. However, observed group sizes ranged from 1 to 23 (average = 9) individuals and building duration of nest groups also correlated with group size (r = 0.65, n = 49, p < 0.001). Figure 7-2 shows the relationship between group size and building duration. If group sizes are split into categories of small (<4 individuals), medium (4-9 individuals), and large (>9 individuals), then no correlation was found between building duration and time after sunset of onset of building for small groups (r = 26, n = 10, p = 0.47) or medium groups (r = 0.28, n = 19, p = 0.24). However, large groups build more synchronously if building started later after sunset (r = 0.52, n = 21, p = 0.016). 0   1   2   3   4   5   6   7   8   9   10   Fr eq ue nc y   of  n es t  g ro up s   Time  of  day   146 Figure 7-2. Duration of nest building in small-, medium- and large-sized nest groups (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by stars or circles. Initiator and final builder of group nest-building Nest building was initiated by a large number of individuals, with no individual markedly more frequently than others (although this does not control for group composition). Eighteen individuals initiated building during the study, 6 females and 12 males. Males initiated nest building on 72% (n = 29) of occasions, but initiation of nest building by males was no different than expected by their proportion in the community (χ2 = 0.92, df = 1, p > 0.05). The last builder in a group usually was a male; on 88% (n = 16) of occasions a male built last, but this also was no different than expected by their proportion in the community (χ2 = 2.11, df = 1, p > 0.05). Individual nest -bui lding behaviour summary I saw 64 nests built during the study. However, two of these were day nests, 5 were re-used, and 14 were abandoned shortly afterwards. Figure 7-3 shows the variation in building duration of all overnight, abandoned and re-used nests. On average nests built for overnight use were built in 3 ┌──────*──────┐ ┌──────*──────┐ ┌─────────────*─────────────┐ 147 min and 54 sec. Abandoned nests were used for 2 min and 20 sec (range 12 sec to 8 min), and were built in 3 min on average (range 2-4 min). Individuals tended to spend more time building nests that were used overnight than nests that were abandoned shortly afterwards (Wilcoxon’s matched pairs, z = 1.86, n = 7, p = 0.063). Addition of fresh material when chimpanzees re-used a nest took only 1 minute and 15 sec on average (range: 30-143 sec). The two day nests were also built in less time: 66 and 75 sec. Figure 7-3. Duration of building in seconds of all overnight, abandoned, and re-used nests (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, which are indicated by stars. Activity in night nests immediately following construction was most frequently rest (89%, n = 38). Feeding was thrice observed in night nests (6%) and play once between a mother and infant. The behaviour immediately preceding nest building was equally often feed or travel (49%, n = 45). Often the chimpanzees travelled several hundred meters from their last feeding tree before nesting. During rest, supine posture was adopted in almost all cases observed (90%, n = 39). Prone posture and lying on the right-side were seen only twice each. ┌─────*─────┐ ┌─────*─────┐ ┌───────────*───────────┐ 148 Differences between age and sex classes Nesting behaviour Age and sex classes did not differ in the start time of nest building (H = 0.90, df = 2, p = 0.63), but duration of construction did (H = 8.50, df = 2, p = 0.014). Females spend significantly longer building their nests than males (Figure 7-4; Mann-Whitney, n1 = 10, n2 = 4, U = 18.5, p = 0.005). Median building duration of adult females was 5 min and 55 sec, of adult males was 3 min and 16 sec, and of immature males was 3 min and 13 sec. There was no difference in duration of nest building between adult and immature males (n1 = 10, n2 = 2, U = 9, p = 0.83), however only two immature males were observed building. Figure 7-4. Building duration of males (10 individuals), females (4 individuals), and immature males (2 individuals) (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers. Nest and tree character i s t i c s Median measures of nest and tree characteristics differed across sex and age classes (Table 7-2). Kruskal-Wallis tests of difference revealed significant differences across adult male, adult female, and immature male nests in nest height, number of decisions, and DBH of nesting trees, but not ┌─────*─────┐ Immature  male   149 in the nest height above the lowest branch, distance to the trunk, position within the tree crown, or proportional distance to the trunk (Table 7-2). Post hoc Mann-Whitney comparisons reveal that adult females nested significantly higher than adult males (U = 10.5, p = 0.006), with a greater number of decisions (U = 11.5, p = 0.009). Females also nested higher above the lowest branch, although the Kruskal-Wallis test was not significant (U = 17, p = 0.043). Adult females built nests in larger trees than males, as measured by DBH (U = 9, p = 0.003), but not tree height or lowest branch height (Table 7-2). Immature males nested higher (U = 12, p = 0.050) and tended to nest in locations of more decisions than adult males (U = 15, p = 0.064). Immature males also nested further out from the trunk than adult males (U = 11, p = 0.039), although the Kruskal-Wallis test was not significant. No other measures were significantly different in immature than adult males (DBH, U = 15, p = 0.22; Table 7-2). Adult females and immature males did not differ significantly across all measures (nest height, U = 24, p = 1.00; number of decisions, U = 18.5, p = 0.43; DBH, U = 14, p = 0.38). Table 7-2. Nest and nesting tree characteristics of each age/sex class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). Nest measurement Adult male Adult female Immature male Kruskal- Wallis test Immature female M Range M Range M Range H p Value Nest height m 4.6 1.5-9.7 7.9 5-10.5 7.9 3-11 7.79 0.020 7 Decisions n 3.9 1.3-6.7 6.8 3-12 4.8 3.3-10 6.76 0.034 4 Nest height above LBH m 3.1 0-6.0 4 3-7.5 5 1.5-9 3.80 0.15 3 Nest to trunk m 1.1 0.3-4.3 1.5 0-3.8 2.9 1-6 3.85 0.15 4 Position in tree crown ratio 0.45 0.22-0.89 0.51 0.34-0.74 0.47 0.30-0.63 0.39 0.82 0.20 Proportion nest to trunk ratio 0.48 0.17-0.81 0.64 0-1.0 0.58 0.47-1.0 1.47 0.48 0.80 Nest tree measurement Tree height m 8.8 4.3-16 12.4 9-20.5 11.7 6-20 3.88 0.14 18 Lowest branch height m 2.2 1-4.5 3.3 2-5 2.9 1.5-6 3.51 0.17 3 DBH cm 19.7 10.5-37.5 36.2 22.3-48.6 24 15.3-53.4 7.44 0.024 40.2 150 Nest shape Nest shape differed among sex and age classes (Table 7-3). Kruskal-Wallis tests showed significant differences in length, circularity, and thickness between the adult males, females, and immature males, but no differences in width, misshapenness, or depth (Table 7-3) Post hoc Mann-Whitney comparisons revealed that adult females built nests of smaller length (U = 7, p = 0.028), but not width, than nests built by adult males. The circularity ratio was therefore closer to 1.00 in females than in males (U = 3, p = 0.005). Females also built thicker nests than males (U = 8, p = 0.040). Other shape measures did not differ between adult females and males (Table 7-3). Adult males did not differ from immature males across all shape measurements compared (length, U = 7, p = 0.22; circularity, U = 7.5, p = 0.20; central thickness, U = 11, p = 0.57. Nests built by adult females differed in some similar ways from immature males as from adult males. Female nests were thicker (U = 0.0, p = 0.036) and more round (U = 0.0, p = 0.036) than immature male nests. There was no difference in other shape measures (length, U = 6, p = 0.79; Table 7-3). Table 7-3. Nest shape for each sex and age class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). Nest measurement Adult male Adult female Immature male Kruskal- Wallis test Immature female M Range M Range M Range H p Value Length cm 90.5 82-105.2 76 70-100 78 74-96 5.54 0.063 49 Width cm 68.9 60-83.3 70 60-100 56 51-70 3.23 0.20 30 Misshapenness cm 11.8 7-30 8 2-22 9 4.5-9 2.61 0.27 8 Circularity ratio 0.77 0.67-0.89 0.98 0.79-1.00 0.73 0.65-0.76 9.51 0.009 0.86 Depth unsprung cm 13.5 0-25 18.5 8-20 18.7 11-25 2.64 0.27 16 Central thickness cm 12.6 1-25 20 13-30 8 7-12.5 6.35 0.042 30 Nest archi tec ture Kruskal-Wallis tests revealed significant differences across age and sex classes in the total number of steps and number of selected support branches (Table 7-4). There was also a tendency for difference in number of lining steps, mattress steps, and proportion of steps interwoven (Table 7-4). 151 Post hoc Mann-Whitney tests showed that adult females used more lining steps in construction than males (U = 3, p = 0.014), but there was no difference in the total number of steps (U = 10, p = 0.19), or mattress steps (U = 12, p = 0.30) Immature males used more total steps in nest building than adult males (U = 1, p = 0.014). There was no difference in the number of lining steps between immature and adult males (U = 9.5, p = 0.37), but immature males tended to build nests with a greater number of mattress steps (U = 3, p = 0.049). Complexity, as measured by the proportion of steps interwoven, did not differ between adult males and females (U = 16, p = 0.64), but was lower in immature than adult male nests (U = 2, p = 0.028). Both selected and constructed support was similar between adult males and females (Table 7-4), although females tended to use a greater number of selected support branches (U = 6.5, p = 0.054). Immature male nests were also built upon selected supports of a greater number of branches (U = 0, p = 0.030). Table 7-4. Summary of nest architecture and support of each sex and age class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). Nest measurement Adult male Adult female Immature male Kruskal-Wallis test Immature female M Range M Range M Range H p Value Number of steps Total n 33.8 11.5-55 56.5 13-78.5 71 44-79 6.23 0.044 84 Lining n 1.8 0-6.8 12.5 3-22 20 0-32 5.06 0.080 9 Mattress n 24.7 10-50 41 4-51 45.5 45-46 4.73 0.093 68 Support n 2 0-6.5 3 3-8.5 3.5 2-5 2.07 0.36 7 Complexity Weave ratio 0.13 0.03-0.24 0.18 0.03-0.46 0.01 0.00-0.08 5.37 0.068 0.05 Constructed support Sum diameter cm 6.8 3.6-17.3 6.9 4.5-7.2 6.6 5.2-8 0.05 0.98 8.3 Mean diameter cm 1.9 1.5-5.8 1.9 1.5-2.3 1.9 1.7-2 0.07 0.97 2.1 Branches n 1.8 0-4 3.3 3-4 3.5 3-4 2.97 0.23 4 Selected support Sum diameter cm 19.7 4.2-25.8 8.1 4-33.8 14.4 7.6-21.1 1.17 0.56 18.5 Mean diameter cm 5.6 1.1-24 2.8 1.4-11.3 3.6 1.9-5.3 2.04 0.36 4.6 Branches n 2 0-3 3 2-4 4 4-4 7.63 0.022 4 152 Table 7-5. Detailed measures of nest architecture of each age and sex class. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). Nest measurement Adult male Adult female Immature male Kruskal- Wallis test Immature female M Range M Range M Range H p Value Bends n 20.6 4.5-41 14.3 6-30 21 21-22 1.73 0.42 16 Breaks n 9.9 2-18 10.3 2-34 2 1-15 0.38 0.83 10 Detachments n 3.2 0-13.33 15 4-42.5 35 21-58 8.3 0.016 59 Bends <1cm n 11 2-65 7 3-11.5 16 2-18 1.58 0.46 1 1-2cm n 5.6 1-35 3.8 0-17 4 4-15 0.34 0.85 11 2-3cm n 1.3 0-3 1.5 1-2 1 0-4 0.13 0.94 3 >3cm n 0.13 0-1.6 0.5 0-2 0 0-0 2.41 0.30 0 Breaks <1cm n 3.5 0.5-8.7 3 0-14 2 0-9 0.23 0.89 1 1-2cm n 3.2 0.5-11 2.3 2-13 0 0-4 1.81 0.40 7 2-3cm n 1.1 0-5 2.3 0-7 1 0-1 0.38 0.83 2 >3cm n 0 0-0.8 0 0-3.5 0 0-1 0.03 0.98 0 Detachments <1cm n 2.7 0-8 8.5 3-34 33 21-51 8.40 0.015 44 1-2cm n 1 0-4.3 1 1-8 1 0-3 1.34 0.51 15 2-3cm n 0 0-1 0 0-0 0 0-0 2.37 0.31 0 >3cm n 0 0-0.7 0 0-0.5 0 0-0 0.89 0.64 0 Mean diameter All material cm 0.94 0.68-1.11 1.13 0.98-1.20 0.59 0.48-0.86 8.02 0.018 0.93 Main branch cm 2.33 2.03-3.20 3.04 1.95-3.23 2.35 1.90-2.52 1.20 0.55 1.93 Lone side-branch cm 1.53 0.88-4.23 1.68 1.47-2.01 0.93 0.60-2.17 1.16 0.56 1.91 Side-branch cm 1.45 0.89-2.46 1.82 1.26-1.93 0.76 0.67-1.96 1.73 0.42 1.82 Number of Main branch n 2.3 0-3 2.5 0-4 2 1-5 0.24 0.89 3 Lone side-branch n 5.5 0-10 3 2.5-11 14 3-15 2.26 0.32 10 Side-branch n 3.4 0-11.5 3.5 0-10.5 3 2-6 0.01 0.99 3 Twig n 12.3 0-25.3 24.5 6-35 26 16-35 3.53 0.17 41 Leaves or leaflet n 0.6 0-15.7 15 1-26.5 23 17-35 8.76 0.013 27 153 Detailed comparison of nest architecture between age and sex classes revealed few differences (Table 7-5). Kruskal-Wallis tests show significant differences only in total number of detachments, number of detachments <1cm diameter, mean diameter of all nest material, and the number of leaves/leaflets (Table 7-5). The number of detachments tended to be greater in female than male built nests (U = 7, p = 0.076), and in immature than adult male built nests (U = 0, p = 0.007), although this difference resulted from use of more detachments < 1cm diameter than adult males (adult females, U = 7.5, p = 0.076; immature males, U = 0, p = 0.007). Detachments of larger sizes did not differ (Table 7-5). The mean diameter of all material used in nest building tended to be greater in adult females’ (U = 5, p = 0.050), but smaller in immature males’ (U = 3, p = 0.064), than in adult males’ nests. Leaves or leaflets tended to be used more in nest building by females than by males (U = 6, p = 0.054), and more by immature than by adult males (U = 0, p = 0.007). There was no difference in amount of all other types of material used by adult males and females or immature males (Table 7-5). Individual differences Nest and tree character i s t i c s No significant differences emerged in comparison of nests built by four adult males, or in the trees in which they built their nests (Table 7-6). Between the two females, TM and LU, most measurements of nest characteristics and nesting trees used were not significantly different; however, TM built her nests closer to the trunk than LU (Table 7-6). Nest shape and archi tec ture There were few significant differences in shape or architecture between KL and LP. KL built more misshapen nests than LP and tended to use less support construction steps. LP also tended to use more side-branches (Table 7-7). 154 Table 7-6. Comparison of individual nest and nesting tree characteristics between males and females. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). Adult males Kruskal-Wallis test Adult females Mann- Whitney test Median nest measurement DF KL LP MM H p TM LU U p Nest height m 5 3.5 5 5 2.8 0.43 7 6.5 14 0.66 Decisions n 5 4 5 4 1.9 0.59 7 6.5 11.5 0.91 Nest height above LBH m 3 3 3 2 2.7 0.44 4.5 4 10.5 0.69 Nest to trunk m 2 1 0 2 3.2 0.37 1 3.5 1 0.016 Position in tree crown ratio 0.34 1 0.46 0.44 5.1 0.17 0.65 0.67 7 0.60 Proportion nest to trunk ratio 0.55 0.67 0.42 0.42 2.4 0.49 0.19 0.93 0 0.010 Median nesting tree measurements Tree height m 13.5 4 8 10 4.6 0.20 12 8 16 0.39 Lowest branch height m 1.5 1 2 2 2.1 0.56 2 2 9 1.00 DBH cm 13.2 11.8 11.9 21.2 2.3 0.51 27.5 17.3 10 0.80 155 Table 7-7. Comparison of nest shape and architecture between two adult males. Bolded p values indicate tendency (p < 0.10) or significance (p < 0.05). Median nest measurement: Shape Adult males Mann-Whitney test KL LP U p Length cm 105 110 12 0.92 Width cm 70 73 11 0.75 Misshapenness cm 14 8 0 0.008 Circularity ratio 0.78 0.75 12 0.92 Depth unsprung cm 0 6.5 7 0.42 Central thickness cm 18 14 11 0.75 Nest architecture Number of steps Total n 30 33 10 0.69 Lining n 2 3 8 0.33 Mattress n 27 26 10 0.60 Support n 1 6 3 0.080 Complexity Weave ratio 0.10 0.08 7 0.15 Support Branches constructed n 1 4 5 0.17 Branches selected n 2.5 3 7 0.22 Manipulation Bends n 21 23 10.5 0.68 Breaks n 9 8 7.5 0.30 Detachments n 2 5 10.5 0.67 Mean diameter All material cm 1.2 1.00 10.5 0.68 Main branches cm 2.9 2.2 4 0.25 Lone side branches cm 1.4 1.3 8 0.35 Side-branches cm 1.4 1.4 9 0.81 Number of Main branches n 3 2 11.5 0.83 Lone side branches n 1 6 4 0.072 Side-branches n 5 4 9 0.46 Twigs n 10 18 8 0.34 Leaves or leaflets n 0 0 10 0.69 156 DISCUSSION Nesting behaviour of Fongoli chimpanzees Time and durat ion o f group nest bui lding Few studies have reported the time of nesting of chimpanzees. Fruth [1995] reported bonobos to have two distinct peaks in the distribution of times of initiation of night nesting, which correspond to nesting earlier on wetter days compared to dry days. However, the time of onset of building in bonobos was much earlier than recorded here for chimpanzees, with a median start time of 16:06 – 16:18 hr, and no groups began after 18:00 hr when the sun set [Fruth 1995]. Chimpanzees in Fongoli rarely nested before 18:00 hr, but data were not collected during the wet season when a similar pattern of early nesting in wetter conditions may occur. Goodall [1962] reported chimpanzees to nest earlier in the wet season in Gombe, but usually between 18:00 and 19:15. Chimpanzees may therefore nest later than bonobos, although on average also before sunset [Goodall 1962]. Chimpanzees and gorillas in Equatorial Guinea nested between 18:00 and 18:50 hrs and 17:30 and 18:35 hrs respectively, and nesting times appeared to be influenced by group size, composition and weather [Groves and Sabater Pi 1985]. Duration of group nest building is similar in chimpanzees and gorillas [Groves and Sabater Pi 1985] and bonobos [Fruth 1995]. Similar to this study, Fruth [1995] found that bonobos take from 4 to 90 min for the group to complete nesting and that the group nests more synchronously if nesting is initiated later. She proposed that individuals use the ‘extra’ time for foraging or travelling before nesting, compared to cold days when energy may be conserved by nesting earlier [Fruth 1995]. The most frequently observed activity before night nesting in this study was travel as individuals nesting after sunset often travelled from the feeding tree together to a separate site for nest-building, whereas when nesting was initiated before sunset individuals often left a feeding tree at intervals and nested close by. Lodwick et al. [2004] recorded the active period of chimpanzees over an 18-year period, and found that on average chimpanzees spent ~12 hrs overnight in nests. However, they did not report the nesting times in this study, but focussed on the duration of the active period between chimpanzees arising from their nests in the morning and retiring to their nests to sleep at dusk. It was not possible to record here the time individuals spent in nests in order to compare directly to Lodwick et al. [2004], as the chimpanzees frequently nested and arose in the dark, so often it was not possible to record individual’s nesting times. That the chimpanzees in Fongoli built their nests on average 30 min after sunset suggests that their active period may be longer than that of the chimpanzees in Gombe. However, Lodwick et al. [2004] demonstrated that males and oestrous females have longer active periods (and so later nesting times) than non-receptive 157 females, and during my study data collection was biased towards nest groups containing males and a single female, Tumbo, was frequently in oestrous during the study period. The heightened social activities resulting from the presence of an oestrous female may have influenced the late nesting times found here. Alternatively, chimpanzees in Fongoli activity patterns may have been influenced by the extreme temperatures in the latter dry season, which discourage sufficient foraging or mating during the heat of the day and result in increased activity during the night. Pruetz and Bertolani [2009] showed that adult males in Fongoli are more active in the early morning during the dry season than other chimpanzee communities. They ascribe this to the extreme conditions in Fongoli, and also note that the chimpanzees sometimes travelled up to 1km on moonlit nights [Pruetz and Bertolani 2009]. On three nights, I observed male chimpanzees to rest on the ground long after sunset, whilst Tumbo continued foraging in the tree above and, although data were too few to test this, I hypothesise that the presence of an oestrous female influences males abandonment of nests or movement of the nest group during the night. Once, I saw Tumbo nest arboreally above seven males, four of whom built ground nests and three built tree nests in between. Many vocalisations were heard throughout the night (as the chimpanzees had nested close to camp) and on my return the following morning Tumbo’s nest had been entirely destroyed. More nests had been added to the group during the night, suggesting that individuals had arisen during the night and built new nests; multiple nest construction during the night is seen also in western gorillas [Bradley et al. 2008]. Nesting and nocturnal behaviours described above, support the mate-guarding hypothesis of the function of ground-nest building proposed by Koops et al. [2007]. Although ground nesting did not permit the males’ immediate access to the oestrous female above, it may have permitted them to sleep closer to her and to the other males than otherwise possible in an arboreal nest. On another occasion the group did not nest until after 22:00 hrs. I spent the night out with this group and although I believe all individuals nested at this time, two hours later individuals arose and continued foraging in a nearby tree until about 02:00 hrs. New nests were constructed when these individuals returned to bed. The nocturnal foraging of these individuals coincided with the moons’ rising. In contrast, I spent one night out with a large group of chimpanzees, including an oestrous female, but during a new moon, and the group made no sounds or movements between 19:00 and 05:00 hrs. Further research is needed into the nocturnal behaviour of chimpanzees across a range of habitats and the influence of group composition, female receptivity, and moonlight on activity patterns. Foraging, travelling, and construction of multiple nests during the night might not be that uncommon across populations. Alternatively, less time may be spent sleeping at night in 158 Fongoli, due to environmental constraints of a dry habitat with extreme temperatures, less available water and food, and may be permitted by a lack of predators that would otherwise make nocturnal activity dangerous. Ini t iator o f nes t bui lding Previous studies have reported sex differences in the initiator of construction in great apes. Groves and Sabater Pi [1985] reported that the silverback male gorilla selected the nesting site, and Fruth [1995] found that bonobo females initiated nest building more than expected by their proportion in the community. She proposed that decision-taking about nesting may be dominance-related, in which case, we may expect chimpanzee males to nest first. However, no relationship was found in this study between sex and initiation of nest construction, and as described above, there may be several influences on timing of onset of nest building. More data may permit a more detailed comparison with regard to female receptivity and group composition and initiator of nest-building. Duration o f construct ion Chimpanzees re-use nests at greater frequency during the dry season, perhaps due to a paucity of available materials [Wrangham 1992]. Thus, savanna-dwelling populations of chimpanzees may also be hypothesised to re-use nests at a greater frequency, as these habitats have fewer trees and more extreme seasonality [Fruth and Hohmann 1996; McGrew et al. 1981; Moore 1996]. Re-use of nests may also be an energy-saving measure when the active period of chimpanzees is extended for foraging or social activity. Re-use here was shown to take significantly less time than building a new nest. With more data, one could test whether or not frequency of re-use increases with the lateness of time of onset of nest building. When I observed abandonment of nests, it appeared to me that individuals were building nests for the night and perhaps to initiate group nesting. However, that abandoned nests were also built in less time than nests used for overnight sleep suggests that individuals built them only for temporary rest similar to day nests. Brownlow et al. [2001] found that chimpanzee day nests were less well-made structures than night nests in Budongo, and Fruth and Hohmann [1993] that bonobo day nests were built in less time than night nests. Activi t i es Few day nests were observed in this study as data collection focussed on night nests of chimpanzees in both Fongoli and Issa. Bonobos use day nests not just for rest, but also for other 159 social activities like grooming and play [Fruth and Hohmann 1993]. They also observed that day nests were used as social ‘tools’ to create and symbolise “personal space” [Fruth and Hohmann 1993]. In this study, nests were used almost exclusively for rest during the observation period, although feeding remains were found in and beneath many night nests (pers. obs.), suggesting that individuals either travelled to their nests with food or foraged during the night and returned to their nests with food. I observed social juvenile play within old nests (1-2 days old), but I saw no social interaction in a night nest. However, other researchers have observed the use of day nests for social play, deposition of infants by mothers, and grooming (Pruetz, pers. comm.). Few observations were made overnight, but as described above, chimpanzees are not necessarily static in their nests overnight and other social activities may occur. There are no overnight studies of chimpanzees in the wild, however, in captivity Videan [2006b] found that chimpanzees slept most of the night on either their right or left side, compared to supine, and least amount of time prone. In this study chimpanzees almost exclusively adopted a supine posture immediately after finishing nest construction. Supine posture may be a more comfortable position in an arboreal nest, which tends to have a cup- shape, in comparison to captive conditions where individuals make nests from different materials and on a flat support structure [Videan 2006a; Videan 2006b]. That chimpanzees often travelled just before nesting suggests that although chimpanzees ranging is influenced by distribution of fruit trees [Basabose 2005; Furuichi and Hashimoto 2004], nest sites at Fongoli are not always built near feeding trees. Chimpanzees are highly selective of the tree species and morphology used for nesting, and in Chapter 8 I discuss selection for specific locations within trees. Sex differences Several sex differences in nest-building emerged in this study: Females nested higher than males, but not higher within trees [as found in bonobos, Fruth 1995]. Rather, females selected larger trees than males for building in Fongoli. Females also took longer to build nests and built nests that were shorter in diameter, more circular, and thicker than males’ nests. Architecturally, females used more lining, leaves/leaflets, detachments <1 cm diameter, material of greater diameter overall, and constructed support branches in nest building upon more selected support branches than males. Chimpanzee females are known from several study sites to nest higher than males [Brownlow et al. 2001; Pruetz et al. 2008]. However, no other studies have reported similar sex differences in time of construction or in architectural details in great apes. Fruth [1995] found 160 that bonobo females not only built higher, but also earlier than males. No such sex differences were found in this study in the time of nest building and there was also no indication that females initiated nest building more often than males. Fruth and Hohmann [1993] proposed that because males in both Pan species compete for access to receptive females [Gerloff et al. 1999; Wroblewski et al. 2009], males may select locations beneath females in order to remain close to them and prevent competing males from approaching them. This was also proposed as an explanation for ground nesting by Koops et al. [2007]. If males mapped their nest positions onto the females’, I would have expected to see a difference in onset of nesting between males and females, with males starting later. Anecdotally, males appeared to concentrate their nests around and below the single oestrous female observed in this study, Tumbo, but whether or not such a difference would emerge with a larger sample size of observed nesting events over longer duration, remains to be seen. Higher nesting sites within the canopy may receive more sunlight and so afford a greater volume of thinner twigs, leaves, and leaflets than sites lower in the canopy. However, I could find no studies of the distribution of leaves throughout the canopy in open environments. One study cited by Fruth [1995; Hladik 1978] reports a greater density of leaves in the middle canopy of primary rainforest in Gabon. In more open environments there may be no influence of canopy height on leaf density because light penetrates throughout the canopy. Also, females build nests with a greater mean diameter of all material than males despite also using more leaves/leaflets and detached twigs of <1cm diameter, which suggests that females include a greater number of larger branches than males. Building higher in the canopy might afford more leaves/leaflets and small twigs to females, but also require the use of more relatively large branches to provide support in less stable locations than those used by males. More research is needed into the morphology of nesting sites selected by chimpanzees (although see Chapter 8 for how nesting sites may be pre-fabricated by previous use). Females may invest more in nest building due to their having dependent offspring and so having to build larger nests for two occupants for most of their adult lives. Females’ nests may therefore need to be more structurally sound and safe in order to provide a stable platform for infants. If this were the case, the two females without infants during the study would be expected to invest less in nest construction, yet they constructed nests that were similar to other females with infants (unpub. data.). Female chimpanzees may be pre-disposed to proficiency in material skills such as nest-building or tool use. Previous studies have reported sex differences in termite fishing; for example female chimpanzees in Gombe more frequently termite fished and ant dipped for longer bouts than males [McGrew 1979]. He also proposed that females may show 161 more sophisticated tool-using skills, due to greater nutritional demands from pregnancy and infant care. Or males may be less proficient because their time spent observing and learning is constrained by time spent maintaining social hierarchy and relationships [Boesch and Boesch 1984]. This study, and Fruth’s [1995] study of bonobos, shows that Pan females are also more proficient nest-builders, which could result from selection pressure of infant care or a pre- disposition to learn material skills. Differences in material skills likely emerge during ontogeny (see below). Age differences The ontogeny of nest building by any of the great apes has not yet been studied in detail. Nest- building is known to be a learned behaviour as determined by comparisons of skill in wild-born versus captive-born chimpanzees in captivity [Bernstein 1967; Videan 2006a]. Some building behaviour, or simply the disposition to build a nest, may be innate as many captive individuals gather materials for sleep, but only individuals with appropriate experience construct full and complex nests [Bernstein 1967; Videan 2006a]. Lonsdorf et al. [2004] demonstrated sex differences in the ontogeny of termite-fishing skills and suggested that males’ termite fishing technique and tools had less fidelity with the mothers’ than juvenile females’, because males invest more time in play and social activities. Lonsdorf [2005; Lonsdorf et al. 2004] followed the development of termite fishing skills of individual juveniles over a four-year period. A similar study of the ontogeny of nest-building is necessary to determine whether sex differences in nest building emerge early in development, but some differences in nest-building between adults and immature individuals suggest that this may also be the case in nest-building skills. Fruth [1995] compared duration of construction and use of day nests by individuals of different age and sex classes and found that as age-class increased, duration of construction decreased and duration of use increased. She concluded that immature infants and juvenile bonobos practice nest-building during the day even before beginning night nest construction, and their building skills improve over time. My comparisons in this study of nest-building by different age and sex classes included only overnight nests and only immatures compared to adult builders, rather than infants, juveniles and adolescents separately. In some aspects, nests of immature males resemble those of adult males, but in others more closely resemble adult females. Immature males spent a similar amount of time building and built nests that were similarly oval shaped and thin compared to adult male nests. However, immature males built nests higher in the canopy than did adult males, and further out from the trunk than adult males and females. Architecturally, immature males did not use more lining 162 steps than adult males, but used more total steps in building. Immature males used more constructed support branches, but complexity, in terms of proportion of steps interwoven, was lower in immature than in adult male nests. Selection of nesting position within trees may be related to body size, as immature males nested higher than adult males, and further out from the trunk where branches were likely thinner. As discussed in Chapter 5, this may be because immatures are more vulnerable to predation, or alternatively there are more branches of thinner diameter that are more easily manipulated by smaller individuals. Other aspects of building by immature males appear to be typically male, such as oval shape, thinness, lack of lining steps. The only nest measured of a juvenile female (Nelly) was found to be very similar to adult females and to her mother. She incorporated lining, a large number of building steps, support branches, a round shape and large thickness. Do males learn to build a ‘typical male’ nest, or do females pay more attention to the skills of their mothers as seen in termite-fishing ontogeny [Lonsdorf 2005; 2006]? This question can only be addressed through longitudinal study of the development of nest-building skills. Individual differences The relative contribution of observational social learning and trial-and-error practice to the development of nest building skill in chimpanzees remains unknown. The above sex and age differences suggest that social learning probably does influence development of nest building techniques in chimpanzees. I found no significant differences among the nests of four individual males. However, two females did differ in the selected location within the tree for nesting. From analyses conducted in Chapter 6, it is clear that nests were adjusted according to the environmental conditions. Data presented here for individuals are preliminary, and analyses with a greater sample size in the future, controlling for environmental conditions, may elucidate individual differences. For example, the sample size in this study would be more than sufficient to investigate such individual differences if genetic analyses were completed on faecal samples that were collected beneath all nests. In Chapter 2, I outlined a rapid genetic screening method for identifying individual nest-builders; this method was devised through preliminary analyses I completed on 27 individuals in Fongoli. The two females who had enough nests for comparison represented two extremes of the female chimpanzee condition. Tumbo was a cycling female frequently in oestrous, but without dependant offspring, whilst Lucille was unusual in having two dependant infants and one juvenile. Perhaps females with offspring nest further away from the tree trunk, in order to nest close to their juveniles, or more remotely from the activity of the rest of the group. A greater 163 sample size is necessary to investigate further these hypotheses, but although this thesis has analysed nests as single entities, chimpanzees nest in a social group and the relationships and social situations of the group during the day likely influences selection of nest sites during the night [as hypothesised by Fruth 1995]. Genetic analyses described in Chapter 2 would also allow investigation of variation in individual nest site selection in relation to the social context and in turn, how this may influence individual nest architectural variation. In lieu of a long-term study of the ontogeny of nest building, comparing individual variation in nest-building techniques and site selection may be informative for understanding to what extent techniques are socially transmitted. For example, does nest building vary more with the substrate or species used for building than between individuals? If there is consistency in individuals’ building techniques, regardless of substrate used, then such variation present target variables for the investigation of similarity or difference between related individuals, frequently associating peers, or between communities to investigate possible social transmission of nest- building skills. Conclusions Data presented in this study are preliminary, due to the small numbers of individuals and numbers of nests measured per individual. However, results indicate strong sex differences in building behaviour between males and females. Further study is needed both to investigate whether females modify their building behaviour dependant on the presence of a dependant offspring, and when during development these sex differences emerge. No individual differences were found in this study, but this does not preclude the existence of patterns and the hypothesis that specific building techniques are socially transmitted remains to be tested rigorously. 164 Chapter 8 Living archaeology: Artefacts of specific nest-site fidelity in wild chimpanzees Unusually low nest in Issa; only frame remains and new growth has regenerated from the main branch 165 INTRODUCTION All great apes build at least one nest daily for their entire (post-weaning) lives, as sleeping platforms serve for both day-time rest and over-night sleep. In a lifetime, an individual may build more than 19,000 nests [Fruth and Hohmann 1994a], making this behaviour the most pervasive form of technology in the great apes. Both across and within species, construction behaviour is more frequent than tool manufacture and use, and the study of nest building may reveal insights into the evolution of cognition [Hansell and Ruxton 2008]. Nest building is a learned behaviour [Videan 2006a], which requires sequential combination of branches into a secure platform over highly variable arboreal substrates. As an example of problem-solving ability, nest-building may demand similar or greater cognitive complexity to some tool-use behaviours [Hansell and Ruxton 2008]. Fruth and Hohmann [1996] proposed nest-building to have facilitated the evolution of cognition and technological skill in hominids. Coolidge and Wynn [2009] suggested that a major jump in cognitive evolution may have occurred during the transition from tree to ground sleeping, with Homo erectus. Determining factors involved in nest site selection, construction, and nest function in existing hominids is thus important in modelling the behaviour of pre-erectus hominins. Nests, plus other remnants of chimpanzee material culture, are evidence of chimpanzee habitat use and site-selection and can thus be used to interpret evidence of early hominin ranging behaviour and archaeological sites [Hernandez-Aguilar 2009; McGrew 1992; 2004; McGrew et al. 2003; Moore 1996; Sept 1992; 1998]. Although chimpanzees produce an archaeological record [Mercader et al. 2007; 2002], and a new field of (non-hominin) primate archaeology has begun to study these phenomena [Carvalho et al. 2009; 2008a; Haslam et al. 2009], most ape material culture is ephemeral [McGrew 1992]. Because of this, studies of nests as “artefacts” on the landscape [sensu Hernandez-Aguilar 2009; Sept 1992] have considered their decay rate. However, Fruth and Hohmann [1994b] found that traces from bonobo nest building last much longer than the time taken for the nest to decay. They observed deformed but healed branches growing at nest height in the canopy, sometimes even beneath fresh nests, which they interpreted as evidence of past nest construction events. Theoretically, the lifetime of such scars could be as long as the lifetime of the trees containing the nests, thus increasing the time-depth of nests as “living artefacts” and suggesting them as identifiable archaeological signs. Fruth and Hohmann observed re-use of nest sites, nest trees, and sometimes even nest-spots within a tree, which led them to hypothesise that great apes may “inadvertently alter their favourite trees in such a way as to create branch structures prefabricated for future nest construction” [1994b; p. 311]. Such a feedback system of environmental modification occurs in a range of species, forming a part of evolutionary theory 166 termed “niche construction” [Odling-Smee et al. 2003]. Using an archaeological framework, Biro et al. [2010] highlighted the importance of “use-wear” and “functional analysis” in studying the effectiveness of technologies, and the same will be applied here to nests. The remnants of nest building described by Fruth and Hohmann [1994b] can be defined as macro-scale use-wear of nest construction, and can be used to study these nests from an archaeological perspective, as functional artefacts in specific locations within trees to reveal information about chimpanzee nest site re-use and formation through niche construction. During study of nest architecture in Issa, Tanzania, I aimed to test Fruth and Hohmann’s [1994b] hypothesis that nests are built in locations that have been pre-fabricated by previous building events for future nest construction through investigation of: a) greater prevalence of scars in nest versus non-nest locations; b) proportion of old nest locations re-growing new nesting material; and c) frequency of specific nest location re-use within trees over time. If specific nesting locations are re-used artefacts, then I predict that more scars or “use-wear traces” (these terms are used inter-changeably in this paper) will occur around the nest than around random non-nest locations within the same trees. Prefabrication of nest sites could occur through moulding of branch structures or increased growth at sites of damage, which I use as a qualitative measure of nest site improvement, or prefabrication. Finally, monitoring of specific nesting locations in trees sought to measure and predict the rate of specific nest artefact re-use over time-scales beyond the lifetime of a nest. Woody plant growth is a feedback system primarily under apical control and is influenced by water, nutrients, hormones, and branch biomass [Wilson 2000]. Apical control means that the main stem of the plant controls the growth of side stems or branches. These side branches then control growth of further side branches, and so on. Thus, when an apex at any level within the tree breaks, apical control ceases and new branch growth occurs. Re-growth varies due to various influences, including tree-species, climate, light-level, and natural growth- rate. Humans have exploited natural regenerative abilities of woody plants for millennia, exemplified by such practices as pruning, pleaching, training, moulding, coppicing, and pollarding [Harris 1983]. Pruning, pollarding, and coppicing produce denser re-growth around sites of damage, whilst pleaching, training, and moulding rely on branches continuing to grow in a new, desired direction. Other primate species, such as Cebus capucinus, benefit from the natural regenerative properties of plants, by removing the terminal buds of branches when feeding and so increasing branching and thus food availability [Oppenheimer and Lang 1969]. Such a feed- back system of woody plant re-growth may improve or shape nest-sites, resulting from damage caused by manipulation of branches during chimpanzee nest building. 167 Fruth and Hohmann [1994b; p310] described three types of scars, likely resulting from damage during nest building, with associated re-growth of branches: “(1) broken but healed and continuing growth in a new direction, (2) dead at the site of fracture with a young shoot perpendicular to the branch, either vertical or horizontal in relation to the living part, and (3) a combination of the two.” Similar deformations of branches occur in trees often used for nest building by orang-utans [Ancrenaz et al. 2004] and around nests built by chimpanzees in Kibale National Park, Uganda [Llorente Caño 2004]. During the study period in Fongoli, similar scars of what seemed to be previous nesting events were found within and around fresh nests. These scars near fresh nests were found in 77% of 104 nests [Stewart et al. 2010]. Re-use of recent nests occurs in all great apes, albeit at low and variable frequencies across seasons and study sites [Ancrenaz et al. 2004; Fruth and Hohmann 1996; Iwata and Ando 2007]. For example, male chimpanzees in Gombe National Park, Tanzania, were never seen to re-use nests during the wet season, yet during the dry season they re-used nests in 30% of direct observations [Wrangham 1992]; in Budongo Forest, Uganda, 16% of night nests were re-used [Plumptre and Reynolds 1997], and in Kibale National Park, Uganda, 12% of recent nests were re-used [Llorente Caño 2004]. Re-use of nests usually occurs over short periods when nests are still cushioned with green leaves, prior to substantial nest decay [Llorente Caño 2004; Wrangham 1992; pers. obs.]. Re-use of a nest versus re-use of a specific nest-spot differs, as the former occurs through modification of an intact structure by adding new material, whilst the latter has not been investigated before and could occur long after a nest has disappeared. Re-use of recent nests is hypothesised to occur at higher rates in drier habitats because nesting material is scarcer in areas with few semi-deciduous trees [Fruth and Hohmann 1996; Wrangham 1992]. By studying nests and nest scars archaeologically, when use-wear of nests is classified and measured, re-use beyond the nest’s lifetime can be determined. Chimpanzees differentially use areas of the landscape for feeding and nesting and prefer specific tree species and structures, vegetation types, proximity to food resources, and topography [Baldwin et al. 1981; Brownlow et al. 2001; Furuichi and Hashimoto 2004; Ghiglieri 1984; Goodall 1968; Goodall 1962; Hernandez-Aguilar 2009; Jones and Sabater Pi 1971; Sept 1992; Suzuki 1969]. However, these factors alone cannot account for nest site selection at a landscape scale [Hernandez-Aguilar 2006]. Sept [1992] found nesting patterns of the chimpanzees in Ishasha, Democratic Republic of Congo, to be “spatially redundant” as the chimpanzees re-used some areas in the landscape but not others, which did not seem to be due only to resource distribution. Sept’s [1992] research using chimpanzee nests as a proxy to interpret debris or artefacts at early archaeological sites, was part of an innovative 168 interdisciplinary approach seeking to understand archaeological site formation processes and functions. Hernandez-Aguilar [2009], expanding on the work of Sept [1992], found that re-use of nesting sites by chimpanzees in Ugalla occurs across seasons and even years. Sept [1992] and Hernandez-Aguilar [2009] challenged earlier interpretations of early archaeological sites as being temporary home-bases where division of labour, central place foraging, and food sharing were hypothesised to account for the spatial patterning of concentrations of stone tools (such as cores and flakes) and bones [Isaac 1978]. Hernandez-Aguilar [2009] elaborated that re-use of chimpanzee nesting sites may represent a pre-curser to full home-base use requiring significant changes in social structure. Such interpretation is part of a broader perspective using archaeological evidence from extant species to understand hominin behaviour [Carvalho et al. 2009; 2008a; 2008b; Haslam et al. 2009; Hernandez-Aguilar 2009; McGrew 1992; 2004; McGrew et al. 2003; Sept 1992; 1998]. This study investigates re-use of specific nest locations, which are long-lasting artefacts in the landscape, and whether these artefacts may be prefabricated for future use. Such environmental modification through construction behaviour may have played a role in both chimpanzee and early hominin ranging and patterns of artefact deposition across the landscape. METHODS Study site This chapter describes a study conducted only in Issa, Tanzania. During the first study period in Fongoli, I noticed a high prevalence of scars around fresh nests (77% of nests), which formed the impetus for studying this behaviour in more detail at Issa. I collected data on nest scars, nest decay, re-use, and branch regeneration from October 2008-2009 and excluding nest scars, data collection was continued by two field assistants from October 2008 to May 2010. Data collection I initially monitored 517 fresh nests. I marked each nesting tree with a metal numbered tag, and the specific location of the nest within the tree with a numbered metal stake in the ground below the nest; for some trees with large crowns, a branch map recorded the location of the nest within the tree. Nest groups were added to the monitoring sample as they were found, from October 2008 to August 2009, so the duration of monitoring varies. We monitored nests for 9-19 months. Fifteen groups with 88 nests, were too far away from camp for monthly visits. Thirteen other groups with 153 nests were excluded as I could not confirm reliable monitoring of the 169 correct location. One nest was dropped as the tree died prior to nest decay. Thus, I report here on 275 nests monitored monthly. During shape and architectural measurements I recorded presence or absence of scars for 112 nests in Issa. As nests were selected for safe accessibility, using climbing equipment, some nests were therefore inaccessible, introducing a bias into the nest sample; however, comparative data collected from random locations within nesting trees is similarly biased, which controls for differences in scarring propensity due to location within trees. I counted each type of scar within the specific nest site – a 1.5 m radius area from a nest’s centre (as most material used in nest building is from within this distance; Stewart, pers. obs.). I recorded four types of scars: (A) old breaks, with the branch more than 50% broken but still attached, and new growth sprouting at the point of fracture (while the branch has often died); (B) old bends, when the unbroken branch has healed and continues to grow in the new, aberrant direction; (C) dead ends, when characteristic tails of stripped bark show where a branch has been detached and new shoots sprout from the break; and (D) old frame branches, where dead, detached branches remain from a nest that has decayed (see Figure 8-1). Old frame branches are often invisible from the ground, as just one or two branches may remain, but this scar type can be classified as re-use of a stage 4 nest, see categories described below), as sometimes several branches forming an intact frame remain. Signs of previous nests were rarely visible from the ground, as the frame and scars are usually hidden by fresh leaves used for nest-building. I randomly selected 10 fresh nest groups and re-visited nesting trees that I had climbed for fresh nests (n = 32 of 89 trees and 117 nests) and recorded scar frequency. I randomly selected a suitable potential nesting site within each tree and counted the scars within a 1.5 m radius around the site. I used a random number generator to select the potential site to access for control measures. I defined a suitable potential nesting site as having both available building material and a supporting branch structure, as outlined in Chapter 3. We monitored nests weekly for the first four weeks, then monthly after that. Periodically, I confirmed age categories and nest locations within trees and groups with each of the field assistants. Each month we recorded the state of nest decay [following Plumptre and Reynolds 1997] and the presence of twigs or branches re-growing from breaks and bends in the nests. State of decay was categorised as stages 1–4: 1) leaves green and nest structure intact; 2) some leaves brown, but nest structure intact; 3) nest rotting and structure disintegrating; and 4) only frame and <5% of leaves remain. Nests were considered decayed from stage 4, following Plumptre and Reynolds [1997] and recorded as gone when no nest frame remained. Time to complete disappearance is used in some studies to calculate decay rate [Kouakou et al. 2009; 170 Tutin and Fernandez 1984]. With these broad definitions of decay stage, I recognised re-use if the nest “reverted” in age from an older to a younger state (amount of re-use may be underestimated, as re-use may occur before decay to stage 3, and a stage 3 nest could be re-used and return to stage 3 within the one month interval between checks). We recorded the numbers of new branches sprouting from the nest site on a scale of “few” (0-5), “some” (5-10), and “many” (>10). Of 275 nests monitored, I had climbed into and fully deconstructed 57 for architectural study. Although I re-built the nest as it had originally been made, this deconstruction process may affect rate of decay so I excluded these nests from calculations of decay rates. As the goal was to calculate decay rate, I excluded 17 nests that were re-used when fresh, and 10 nests that were re-used before stage 4. As not all nests decayed fully during the study period decay rate was calculated to stage 4 for 191 nests using the exponential method [Plumptre and Reynolds 1996], which uses the decay curve to calculate the time for 50% of nests to decay. No difference in proportion of nests, or specific nest locations re-used within the first nine months of monitoring, was found between deconstructed versus all other nests (n1 = 57, n2 = 218, df = 1, χ2 = 1.83, p>0.05), so I included all nest locations in analysis of re-use. I included all 275 nests in analysis of branch regeneration. RESULTS Deconstruction of nests in situ revealed both scars and construction pattern. Nests were made from branches that were broken or bent radially into a central structure to form an interwoven, stable, and comfortable circular platform. Most (98.3%) nests had visible supports as outlined in Chapter 3, over which the first branches manipulated often formed a triangular shaped support, overlain by a mattress of overlapping branches. “Old bends” were usually found beneath the nest and formed triangular support structures. Manipulated support and mattress branches usually remained attached and pliable, whether broken (defined as >50% severed) or bent (<50% severed), and so later sprang out undamaged or retained their previous shape when healed. Fully detached branches and twigs were often added as a lining. Fresh use-wear traces around the nest were seen where branches or twigs had been freshly detached and were characterised by a tail of bark. As nests decayed, broken branches died and sprouted new growth, as opposed to bent branches, which continued growing. Whether the characteristic tails of “dead end” scars were the use-wear traces of detached lining material or the remnants of broken branches is unknown (Figure 8-1). 171 Figure 8-1. Four types of scar, or macro-use-wear, described as follows: A) old break – note the new shoots sprouting around the break. Above, shoot continues growing vertically, and to left, shoot freshly broken into new nest; B) old bend – normal growth continues (horizontally) in direction of main branch. Two side branches continue growing aberrantly perpendicular to main branch; C) dead end – note characteristic tail at old detachment point and two new shoots have sprouted. Above, one continues to grow vertically, and to left, other freshly detached and used in nest building; D) old frame branch - several dead branches remain, some detached and others broken, forming an old frame. Scars Of 112 nests studied architecturally, 79% had scars (Figure 8-1). The median number of scars found at a nest site was 5 (n = 112, range = 0-21; Figure 8-2). Of these, dead ends were most frequent (73% of nests, median per nest = 3, range = 0-14), then frame branches (50% of nests, median per nest = 2, range = 0-7), old breaks (47% of nests, median per nest = 1, range = 0-5), and old bends (44% of nests, median per nest = 1, range = 0-6). The proportion of nests with scars did not differ by season (χ2 = 3.1, df = 1, p>0.05), nor did the most frequently used tree species (including species used on more than six occasions: Brachystegia stipulata [plus B. pubulara and B. utilis, which could not be discerned in the field and so are combined for analyses], B. bussei, B. spiciformis, Combretum molle, and Julbernardia unijugata; χ2 = 2.7, df = 4, p>0.05). 172 Figure 8-2. Number of scars around fresh nests and around random control suitable nest spots (┌─*─┐indicates significant difference). Boxes indicate first and third quartile of range and horizontal bar indicates median. Bars outside boxes indicate range, excluding outliers, indicated by stars and circles. Fewer control locations had scars than nesting sites (19%, n1 = 112, n2 = 32, df = 1, χ2 = 38.2, p<0.01; Figure 8-2). The median number of scars per site was 0 (range = 0-11; Figure 8-2). Dead ends were most frequent (16% of sites, median per site = 0, range = 0-4), followed by old breaks (12% of sites, median per site = 0, range = 0-1), old bends (9% of sites, median per site = 0, range = 0-5), and potential old frame branches (6% of sites, median per site = 0, range = 0-1). In three cases, the randomly selected location appeared to be a previously used nest location. In order to control for possible effects of tree species on scarring propensity, a paired-comparison was made between the number of scars found in nesting versus random locations within the same tree. More scars were found within a 1.5 m radius of a fresh nest than at a randomly selected location suitable for nesting within the same tree (n = 32, z = 4.6, p<0.001, Wilcoxon’s matched pairs). Some natural dead ends were found at random locations, however, these did not meet the criteria of dead ends counted around nest sites, as they lacked characteristic “tails” of bark that occur whenever a branch is detached or broken by hand. If, however, these data are ┌────────*────────┐ 173 conservatively included as scars, the relationships remain significant (56% of random sites, n1 = 32; 79.5% of nests, n2 = 112; df = 2, χ2 =5 .9, p<0.05; n = 32, z = 4.2, p<0.001, Wilcoxon’s matched pairs). Branch re-growth Of 275 nests monitored, 181 (or 66%) had re-growth of branches within the first nine months of monitoring. New twigs sprouted in a few nests as soon as one month after nest construction, although most re-growth began three (24% of nests) to four (44% of nests) months later (see Figure 8-3). I counted the number of branches re-grown per nest for some nests within four to eight months of monitoring. A mean maximum number of 7.5 branches or twigs re-grew per nest, ranging from 1-35 (n = 70). More nests built during the rainy season had new branch growth than nests built during the dry season in both forest (rainy: 62%, n1 = 33, vs. dry: 16%, n2 = 38; df = 1, χ2 = 14.28, p<0.01) and woodland (rainy: 83%, n1 = 135, vs. dry: 59%, n2 = 69; df = 1, χ2 = 12.05, p<0.01) vegetation. Also, more nests built in woodland (75%, n1 = 204) regenerated during the study period than nests built in forest (38%, n2 = 71; df = 1, χ2 = 29.3, p<0.01). Figure 8-3. Percentage of nests with re-growth and amount of re-growth at each nest increases over time. Re-growth measured in categories of few (0-5), some (5-10), or many (>10) branches or twigs. 0   10   20   30   40   50   60   70   0   1   2   3   4   5   6   7   8   9   %  o f  n es ts   Months  monitored   MANY   SOME   FEW   174 Nest decay Time to decay of nests in Issa was highly variable, ranging from 7 to >427 days. Comparisons of the median time to decay for nests that decayed fully revealed differences between vegetation types and seasons (Kruskal-Wallis, χ2 = 13.2, p = 0.004). Thus, mean time to decay was calculated separately for nests built in each vegetation type and season. Mean time to decay of nests built in woodland was longer in the dry season (185.5 days, n = 49, range = 14 to >357) than in the rainy season (139.2 days, n = 91, range = 7 to >427). Nests built in forest in the dry season decayed faster (83.3 days, n = 26, range = 14-224) than nests built in forest in the rainy season (118.9 days, n = 25, range = 14-350). Decay curves are shown in Figure 8-4. Excluding nests that were re-used prior to complete disappearance, mean time for a nest to disappear completely is 243 days for nests built in forest (n = 33) and 432 days for nests built in woodland (n = 101). Figure 8-4. Nest survival over time. Nests built in forest in dry season (n = 26) decay faster than nests built in wet season (n = 25), whilst nests built in woodland in dry season (n = 49) take longer to decay than nests built in rainy season (n = 91). Nests built in woodland overall take longer to decay than those in forest. 0   10   20   30   40   50   60   70   80   90   100   0   50   100   150   200   250   300   350   400   450   %  o f  n es ts  re m ai ni ng   Days   Woodland/Rainy   Forest/Rainy   Woodland/Dry   Forest/Dry   175 Re-use Of 517 fresh nests, 32 (6%) were seen from the ground to be episodes of re-use, with fresh material added to existing older nests. Only four of these cases were stage 3 nests (0.8%), whilst 28 (5%) were stage 1 or 2. Fresh nests not classed as re-use were targeted for architectural measurements. Despite this, of 112 fresh nests that were accessed, 7 (6%) were found to be re- used at stage 1 or 2. On these occasions, I found dried or wilted leaves beneath fresh new material on the surface of the nest. Thus, nests classed as “fresh” from the ground may show re- use that cannot be determined without close-up inspection. During nine months of monitoring of 275 nests, 24% of nest locations were re-used: 4% of nests were re-used prior to decay (stage 2-3), 15% after the nest had decayed to stage 4, and 5% following complete disappearance of the nest. Thus, most re-use occurs at the same nesting position within the tree once the nest has decayed. Four specific nest locations were re-used more than once. More specific nest locations were re-used in forest (41%, n1 = 71) than in woodland (19%, n2 = 204, df = 1, χ2 = 12.81, p<0.01). There was no difference in re-use of locations built in the wet or dry season (forest: χ2 = 0.01, df = 1, p>0.05; woodland: χ2 = 1.41, df = 1, p>0.05), or between frequently used tree species ([Brachystegia stipulata, B. pubulara, B. utilis], B. bussei, B. spiciformis, Combretum molle, Julbernardia unijugata, Pterocarpus tinctorius, Isoberlinia tomentosa, B. microphylla, Lannea schimperi, and Parinari curatellifolia; χ2 = 11.9, df = 9, p>0.05). Over nine months of monitoring in forest, and 17 months of monitoring in woodland, the proportion of specific nest positions that had been re-used at least once increased exponentially (Figure 8-5). Data beyond nine months of monitoring in forest were not included, as the sample size of nests monitored dropped too low, from 71 to 35. After nine months of monitoring in woodland, the sample size of nests dropped gradually, and data beyond 17 months of monitoring were excluded, as the sample size dropped below 50. The proportion of nest positions re-used increased in both vegetation types over time. I hypothesise that all nest positions eventually may be re-used, but longer term monitoring is needed to test this. These data did not consider more than one re-use of specific nesting locations. Of 56 nests in woodland and 15 nests in forest, all monitored for 19 months, 12 specific nest locations were re-used twice and one site thrice. Mean time between bouts of re-use of the same nest position was 6 months, including all nest positions re-used on multiple occasions (n = 24, range = 3-12). 176 Figure 8-5. Percentage of specific nest locations not re-used over time in woodland and forest trees. Specific nest sites in forest were re-used at faster rate than in woodland. Polynomial trend lines indicate possible continued re-use in forest (y = -0.08x3 - 1.51x + 100, r2= 0.998) and woodland (y = 0.011x3 - 0.29x + 100, r²= 0.989). DISCUSSION Since Fruth and Hohmann’s [1994b] observations that nest-building by bonobos leaves long- lasting living artefacts in the form of misshapen tree branches, there has been little further study of such artefacts. Given that such scars occur in the nesting trees of bonobos [Fruth and Hohmann 1994b], orangutans [Ancrenaz et al. 2004], and are found in fresh chimpanzee nest sites in Fongoli, Senegal [77%; Stewart et al. 2010], in Kanyawara in Kibale [58%; Llorente Caño 2004], and in Issa, Tanzania (79%), such identifiable types of use-wear traces in these ape artefacts probably are widespread. However, without corresponding data on type and abundance of traces or scars from control sites, as collected here, it is not possible to know whether apes or other natural processes (e.g., feeding activities, plant disease, or climate) are responsible for their creation, or whether or not macro-use-wear is distributed evenly throughout trees frequently 0   20   40   60   80   100   0   2   4   6   8   10   12   14   16   18   %  o f  n es t  s ite s  w ith in  tr ee s  n ot  re -­‐u se d   Time  in  months   woodland   forest   177 selected by chimpanzees for nesting. My results show that use-wear traces of nest building are distributed unevenly throughout trees, in greater numbers around nests. This provides the first evidence that nests are built preferentially at specific positions within trees where nests were built before. Following Fruth and Hohmann [1994b], I hypothesised that nest sites may be improved through branch re-growth at sites of damage. They tested the potential for bonobo nesting sites to regenerate new growth by monitoring cut and broken saplings and trees along transects at Lomako, Democratic Republic of Congo. They found that 97% of cuts or breaks regenerated and 39% healed, much higher rates than in the current study (although results presented here are from a shorter monitoring period). They also hypothesised that regeneration rate may be influenced by climate and vegetation type, where trees in forests may regenerate more often than those in woodlands. Issa nests more often regenerated if built during the rainy season, suggesting climatic influence, yet more nests built in woodland regenerated than nests in forests. This incongruity may be due to better visibility of re-growth in woodland, or may result from variation in growth rates of different tree species, e.g., those in forest may take longer to regenerate. Nest sites may also become “prefabricated” through the healing of support branches. Fruth and Hohmann [1994b] noted that beneath some nests the typical triangular structure of a nest was formed by previously broken but healed branches growing in a new formation. These residual structures may be as important in specific nest site re-use as re-growth of supple new materials. I found similar triangular structures of healed branches at Issa and Fongoli. Although most nests (66%) re-grew branches within nine months of monitoring, few new branches seemed to reach sufficient size and number to be the sole material used in construction of a fresh nest in that time. However, I often observed that around fresh nest sites less than half of the new growth sprouting from scars was incorporated into fresh nests (See Figure 1, A and B) for new growth used in building and left to grow). Therefore new branches may continue growing even whilst the nest is re-used, leading to more frequent opportunities for re-use than is reflected in the rate of re-growth of new branches. A specific nesting location may also have ample available material in order for multiple nesting events to occur over a short period. Re-use of specific nest positions within trees, which occurs after the nest has decayed, should be distinguished from re-use of nests, which usually occurs while the nest is fresh or recent. Re-use of monitored nests supports this, as only 4% of nests were re-used before decaying to stage 4. However, this measure may underestimate nest re-use during the four weeks between successive monitoring checks or may not be discernable whilst nests are at stage 3. The 178 proportion of all fresh nests found that were fresh cases of re-use (6%) also may be underestimated, as re-use cannot always be seen from the ground. This proportion is low, similar to that observed at other study sites [Fruth and Hohmann 1996; Plumptre and Reynolds 1996; Plumptre and Reynolds 1997; Wrangham 1992]. That only nests thought to be freshly built were accessed using climbing equipment, and that 6% of these nests were actually fresh episodes of re-use, emphasises the importance of climbing in field studies of chimpanzee nest-building. Houle et al. [2004] also emphasised the importance of tree-climbing in primate ecological study; here, I add its utility for non-hominin primate archaeology. No other study of nest decay has reported continued monitoring of nests beyond the point of decay [Marchesi et al. 1995; Plumptre and Reynolds 1996; Tutin and Fernandez 1984; Tutin et al. 1995]. By doing so, I found that re-use of specific locations occurs well beyond nest lifetime. Following habituated individuals at long-term research sites would allow a validity test, and also see if the same individuals re-use locations over time. Chimpanzees at Fongoli often nested in what appeared to be virgin positions, but upon deconstruction of nests the following day, scars were found (pers. obs.). Hernandez-Aguilar [2009] found that, although 93% of 5354 nests were built in woodland, chimpanzees preferred building in forest, which represents only 1.5% of the present study area. However, Hernandez-Aguilar [2009] also found that forest trees are larger and occur at higher densities, so they may afford more specific nest sites than woodland. This study reveals a preference for re-use of nests in forest (41% in forest versus 19% in woodland), supporting the hypothesis that forest is a preferred nesting habitat, but specific locations may be more heavily re-used due to greater availability of good building sites and forest vegetation. Faster return to re- use nests in forest may reflect ranging patterns (forest foods are patchily distributed and may be visited frequently to monitor food patches), quicker nest site re-usability (because forest nests decay faster), or greater density of available branches within the specific building area. Lack of seasonal difference in scar prevalence suggests that re-use did not occur as a result of paucity of material; fresh nests continued to be built in woodland habitat throughout the year at Issa. Chimpanzees living in dry, open habitats, such as Ugalla, are estimated from nest counts to have home ranges ranging in size from 278–560 km2 [Baldwin et al. 1982; Kano 1972; Ogawa et al. 2007]. At the only dry study site where chimpanzees are currently habituated, so that home range is known from direct observation, the community has an estimated minimum home range of 65 km2 [Pruetz and Bertolani 2009]. A high proportion of nests in Issa (19%) were re-used before the nest completely disappeared, which could slightly underestimate chimpanzee density and overestimate home range size. Accurate rates of decay, re-use, and construction of nests in 179 the landscape are known to be crucial parameters for censusing chimpanzees using nest counts in varied ecological conditions [Plumptre 2003; Plumptre and Cox 2006; Plumptre and Reynolds 1996; Plumptre and Reynolds 1997]. The repeated use of specific nesting locations may be due to the benefits of building a nest atop a prefabricated support structure, using supple new growth from repeated “pruning” of building positions. Human niche construction is a defining characteristic driven by cultural transmission processes, which form a feedback system that likely influenced the rate and processes of hominin evolution [Laland et al. 2000]. Kimura [1999] argued that patterns of tool production in Oldowan sites reveal a flexible strategy that adapts to the type of available raw material, keeping energy costs low to achieve desired goals. Similarly, chimpanzees may select tool composites for nut-cracking based on the quality, or efficacy, of specific combinations of tools [Carvalho et al. 2009]. Biro et al. [2010, p.150] proposed the utility of an archaeological definition of tool [from Karlin and Pelegrin 1988, p.823] for integrating study of human and nonhuman tools: “Intentionally (or purposefully) made objects or any natural object or knapping debris which shows use wear at the macro- or micro-scale”. In this way a “tool construction” has affinities to nest construction as “two or more objects (whether transformed or not) that must be used in combination in order to function and achieve a specific goal” [Carvalho et al. 2008a, p.159]. Thus, a nest can be seen as a tool construction formed from several branches used in combination to achieve a unique goal (to make one nest). Carvalho et al. [2009] suggested that such preferential and frequent re-use of tools may lead to augmentation of use-wear traces: this appears to occur in re-use of nests as “composite tools” of combined branches. Hernandez-Aguilar [2009] found that although food abundance and distribution influenced the ranging of chimpanzees at Issa, it did not explain the patchy distribution of nests. Topographical features of the landscape and morphological features of trees influence site selection as well, but these environmental features alone do not explain why certain areas are selected over others for nesting [Hernandez-Aguilar 2006; 2009]. Density of high quality specific building sites within trees, and stages of re-growth and healing of support structures, may also influence landscape scale nest site selection. Such environmental modification is common among many species of animal builders and burrowers (e.g., nests and burrows may be millennia- or centuries-old [Hansell 2005; 2007]). To date, only at the dry study sites of Issa [Hernandez-Aguilar 2009] and Ishasha [Sept 1992] have chimpanzee nest distributions been systematically studied to determine if the patterning of debris (using nests as a proxy) produced through chimpanzee ranging behaviour is distinguishable from that of early hominins as seen in the archaeological record. Both Sept 180 [1992] and Hernandez-Aguilar [2009] found redundant re-use of nesting sites resulting in a patchy, clumped distribution, similar to the distribution of archaeological materials in early hominin sites. Their results demonstrate that Isaac’s [1978] home-base model with associated social changes such as food-sharing and division of labour is not necessary to produce this spatial patterning of materials. Hernandez-Aguilar [2009] additionally proposed that preferential use of favourite sleeping sites by chimpanzees may be analogous to transport of food to preferred and tree-shaded places offering refuge to early hominins. She thus hypothesised that ape nesting behaviour may have been a precursor to such hominin-specific behaviour as carcass processing sites. The number of nests within sites monitored by Hernandez-Aguilar [2009] may be underestimated, as specific nest locations were counted once (unless signs of nest re-use were observed). My results indicate that these locations show multiple episodes of use. Hernandez- Aguilar [2009] and Sept [1992] showed that chimpanzees return repeatedly to use the same geographic areas for nesting. Combined with hominin discard and deposition of archaeological materials as proposed by Sept [1992], the result would be scatters and patches of debris over large areas, e.g., Hernandez-Aguilar [2009] monitored nest sites up to 1.5 km long. Specific artefact (i.e., nest) re-use shown here, within this larger geographic scale, if combined with archaeological debris, could account for micro-structuring of archaeological sites. Blumenschine and Masao [1991] found through “landscape archaeology” that artefacts were scattered across wide areas “off-site,” but scatters were not homogenous and higher densities of artefacts correlated with densities of bones. Potts [1984] proposed concentrations of artefacts could occur by deliberate caching of stone tools, whilst Schick [1987] proposed an inadvertent build-up through a feed-back system of artefact re-use, transport, and discard. However, few archaeological sites provide densities of artefacts measured in a way directly comparable to analogous chimpanzee archaeological sites [Carvalho and Mcgrew in press]. Despite the longevity of chimpanzee artefacts (nests and use-wear traces) seen in this study, the specificity of sleeping site re-use is a behaviour that is absent in the archaeological record, but which provides clues to factors such as niche construction that may have influenced early hominin artefact use and discard in the landscape. 181 Chapter 9 Conclusion Introduction In this thesis I have investigated variation in the most pervasive material skill in the great apes, nest-building, under a frame-work of the evolution of shelter. Evidence for shelter construction in early hominins is scarce, but in modern humans, shelter use and construction is a defining characteristic [Brown 1991]. That nest-building is absent in all anthropoid primates except the great apes [Kappeler 1998] suggests that this behaviour was present before the hominid lines separated, making it an important ancestral trait in seeking to model the behaviour of early hominids. Shelter construction across the animal kingdom primarily functions to protect the builders from parasites, predators, or climatic stress [Hansell 2005]. I tested these basic proximate functions of great ape shelters, in order to reveal possible ultimate reasons for the evolution of this behaviour and its persistence across all the great apes. Variation in shelter construction by Homo sapiens across the globe is indisputably both functional and cultural. In comparing nest-building between two communities (Chapter 3), and within one community (Chapter 7), I sought to reveal further insights into variation in nest building and to see if the behaviour is more homogeneous within than between communities, as would be expected for cultural behaviour. Finally, animal builders may be more successful species, buffered against extinction processes by their niche construction capabilities [Hansell 2005]. In humans, this ability has contributed to our global distribution and success, to the detriment of other species. To what extent are great apes also environmental engineers, and was hominid shelter construction the platform of future constructivity in hominins? Proximate functions of nests The extent to which the large goals of this thesis (the evolution of shelter and cultural variation in nest building) could be addressed was limited. Use of chimpanzees to model early hominin behaviour is currently strongly debated (e.g. White et al. [2009] versus McGrew [2010]). I have hypothesised that nest building was present in the LCA and in many early hominins (cf. Homo erectus) and that the proximate functions of nest building by chimpanzees and other great apes may also have influenced nest building in early hominins. However, these hypotheses cannot be tested, only inferred, because nests are made of ephemeral materials and so not found in the archaeological record. In this thesis, I found some support for all hypothesised proximate 182 functions of nests. In Chapter 4, I highlighted the utility of self-experimentation for field study. Although the results of this experiment are preliminary and limited by small sample size, some support was found for anti-parasite, thermoregulation, and anti-predation (if subjective fears and concerns are interpreted as predation risk) functions of nests. The experiment in Chapter 4 revealed that great ape nests function to reduce bites from insects, which might be accomplished in three ways; first, nests may create a partial, physical barrier between the builder and vector. Secondly, nests might be made in trees with anti- mosquito properties [e.g. Largo et al. 2009]. Finally, aromatic compounds (without particular anti-mosquito phyto-chemicals) released from breaking branches may mask the heat and odour signals that attract biters. These hypotheses require further testing using complementary methods (e.g. phytochemical analysis or insect-trapping experiments, see below). Although a difference was found in the number of bites received on the bare earth compared to within a leafy arboreal nest, it is not possible to determine whether this difference was due to the height within the tree or the barrier of a freshly built nest. Finally, complementary methods are necessary because although it is valid to use myself as a subject for comparison between experimental conditions, it is not possible to know how a chimpanzee is affected by biting insects, for example does their thick hair influence number of bites received? Only recently have non-invasive methods of isolating virus pathogens, anti-bodies, and blood borne parasites, allowed us to link illness in wild apes to specific pathogens [e.g. Bermejo et al. 2006; Keele et al. 2009; Leendertz et al. 2006; Liu et al. 2010]. Recent research of epidemiology of wild apes has focussed on respiratory disease outbreaks, which are human introduced, kill rapidly and in large numbers [Kaur et al. 2008; Köndgen et al. 2008; 2010; Williams et al. 2008]. Many studies have investigated faecal-oral transmitted parasites, although symptoms of disease may be more subtle than from viruses [Chapman et al. 2005; Gillespie et al. 2008; Howells et al. 2010]. The extent to which great apes suffer from diseases caused by vector- borne parasites is unknown, e.g. malaria [Liu et al. 2010]. Future research in long-term study populations could investigate the health impacts of such pathogens on individual and population health. Chimpanzees frequently include medicinal plants in their diet and the relationship between feeding on such plants followed by a reduction in parasite-load is well-documented [reviewed in Huffman 2001]. Chimpanzees are highly selective in the tree species used for nesting, but one characteristic not yet studied is possible aromatic compounds that reduce exposure to pathogen vectors. Initially, as described in Chapter 2, I aimed to test the above hypotheses through experiments in which I deployed traps within and without experimental nests and in preferred and non-preferred tree species. Similar, but successful, field experiments 183 should be conducted in the future to address the anti-vector hypothesis, in combination with phyto-chemical analysis of the leaves of preferred compared to non-preferred tree species. Such studies could reveal more about the possible anti-pathogenic benefits of great apes building a fresh nest each night. Limited support was found for the anti-predatory function of nests in experimental study (Chapter 4). Subjectively, during my overnight experiments, I slept less on the ground, due to discomfort, concerns, fears, and anxiety. However, this experiment was conducted in Fongoli where there are no predators, and chimpanzees frequently sleep on the ground. Ground-nesting was thought to occur only in populations that have no predators [Koops et al. 2007; Maughan and Stanford 2001; Pruetz et al. 2008], but Hicks [2010] reports ground nesting in chimpanzees of the Democratic Republic of Congo, in areas where evidence of non-human predators was also found. He proposed that ground nesting was influenced only by risk of human predation. At neither Issa nor Fongoli do chimpanzees appear to be at risk of human predation, but in Chapter 5, I showed that Issa chimpanzees do nest higher, and more peripherally, within trees than Fongoli chimpanzees. Another test of this hypothesis may be conducted in collaboration with A. Piel in the future; the influence of predation might be tested through comparison of nests built at random times throughout the year to nests built following the recording of predator vocalisations on the remote acoustic surveillance system that Piel deployed during my study period in Issa. Use of such a passive monitoring system in the future primatological studies may thus allow ‘natural experiments’ to be conducted, without the use of potentially unethical predator playback experiments. More research is necessary to determine if selection of peripheral spots within nesting trees in Issa is due to factors other than the presence of predators, e.g. the distribution of leaves within the tree, which might not scale with tree size. For example, in Chapter 6 chimpanzees of both sites were found to nest higher in conditions of greater atmospheric moisture (relative humidity and rainfall). However, to address both of these questions thoroughly in the future phenology data described in Chapter 2 should be included in order to control for the influence of the availability of leaves within nesting trees, and the availability of leafing trees, on nest position, shape, or architecture. More research is also needed on the diet of leopards living sympatrically with chimpanzees and in a variety of habitats. The best data available to date are from a single research site (Tai) [Jenny and Zuberbuhler 2005], and rates of the presence of ape remains within leopard faeces appear to be low (see Chapter 5). However, the impact of even low rates of predation on long-lived, slowly reproducing species like the great apes is likely to be high. 184 Through experimental study limited support was found for the thermoregulatory function of nests. The hypothesis that insulation (differential temperature) would differ between ground and nest sleep was not supported. However, that insulation increased on colder nights in nest sleep, but not ground sleep, suggests that nests do provide some thermoregulatory function. Fongoli also has an extreme temperature range and I biased the experiment to nights that were warm; an experiment with a greater sample size, on colder nights, and also in habitats with less extreme conditions, is necessary to test the thermoregulatory hypothesis more thoroughly. As described above another limit of this experiment is that I, as the experimental subject, am not a chimpanzee. A chimpanzee’s full hair coat may influence how much insulation a nest provides and how this may differ from sleeping on an open branch. Studies of bird and mammal nests measure the conductance of materials in order to calculate insulation properties of nests [Heenan and Seymour 2011; Pinowski et al. 2006; Redman et al. 1999]. Alternatively, the same could be done with ape nests, which would allow a greater sample size and control of other variables such as the conditions on the experimental night. In both Fongoli and Issa, I found much variation in nest characteristics, shape, and architecture. The principal components analysis used in this study may be a useful tool for future research into ape nest building. It was not possible to measure all variables of nest architecture, thus some variation remains unexplained (~25%) by the resultant components and future research into nest architecture may reveal other useful variables to quantify nest structure. Some of the variation in the resultant components was accounted for by different weather conditions, which supports the thermoregulatory function of nests. The chimpanzees built with more material, lining, thickness, or depth in colder or wetter conditions, and in Fongoli, nests had greater support in windier conditions. This study in Chapter 6 does not control for the influence of the environment, e.g. availability of leaves, or variation in the trees selected, e.g. species, leaf size, branch brittleness or tensile strength, etc. These data have been collected (described in Chapter 2) and could be included in future analyses. If the builders of more nests are identified using rapid genetic screening outlined in Chapter 2, it will be possible to investigate intra- versus inter-individual, sex, or age-class, variation in nest building with environmental conditions. Some proximate functions of nests were neglected in this study: sleep and social functions. The importance of sleep is discussed in Chapter 4 and below. Fruth and Hohmann [1993; 1996] proposed that nests may function as a social tool. Day nest building was not observed in this study, but Fruth and Hohmann [1993] saw bonobos build 147 nests during the day, most of which were used for rest, but some were used also for feeding, social grooming and play. A small proportion (7%) they termed “taboo” nests, which seemed to function to prevent 185 the approach of other more dominant individuals and allowed the builder more time feeding undisturbed. This possible function for the monopolisation of food sources led them to theorize a possible scenario for the evolution of nest building (see below). Fruth and Hohmann [1996] also highlighted that nest-building has social aspects, and so nest construction is influenced by factors including inter-individual relationships and dynamics within the group. They hypothesised also that nest groups may function as information centres. Variation in the composition of the nest group, as a unit, was neglected in this study, but data were collected on individual nest positions using three-dimensional mapping described in Chapter 2. Future identification of individual builders through the use of genetics may provide information on social influences of nest building, and particularly in a savanna environment, how chimpanzees maintain their social structure across potentially huge ranges [Baldwin et al. 1982; Moore 1996]. Great ape nests and the evolution of shelter All nest functions investigated in this thesis are proximate. Fruth and Hohmann [1996] proposed a plausible evolutionary scenario for the original invention of nest-building in hominids described below. [Hominoid here refers to all living and extinct taxa of gibbons, great apes, and humans. Hominid refers to all great apes and humans, which can be split in to the pongines (orangutans and extinct asian great apes) and hominines (all African great apes and humans)]. Hominoids evolved in Africa during the early Miocene, but later radiated out of Africa across Eurasia [Begun 2007]. There is much debate regarding whether or not the ancestral hominine evolved in Africa, requiring separate dispersal events of a hominoid ancestral to hylobatidae (gibbons and siamang), and a hominid ancestral to pongines out of Africa, or if the ancestral hominid evolved in Eurasia followed by dispersal of the ancestral pongine to Asia and hominine to Africa [Begun 2005; 2010; Harrison 2010; Stewart and Disotell 1998]. There is much evidence supporting major climatic and vegetation changes in Europe at ~9.6 Mya (termed the mid-Vallesian crisis). During this time, the climate became drier, and more seasonal, and vegetation changed from tropical forest to deciduous woodland, which some authors suggest led to the extinction of Eurasian hominoids [Agustí et al. 2003], and others suggest led to dispersal of hominoids south, tracking suitable forested habitats into Asia and Africa [Begun 2010]. This time of climatic change in the mid to late Miocene led to expansion of drier, more open habitat, adapted species of cercopithicines (in some areas), accompanied by the reduction in great apes, which is mirrored by changes in other mammalian fauna, including replacement of old predator guilds with new (machairodontine) felids [Agustí et al. 2003; Andrews 1996]. However, this 186 dramatic turnover in fauna is also accompanied by dispersals of large mammalian fauna into Africa, which supports a possible concurrent dispersal of hominids into Africa [Begun 2010]. Fruth and Hohmann [1996] describe how colobines and cercopithecines may have outcompeted many great apes during their expansion in the middle Miocene through evolving a greater tolerance of plant secondary compounds, which enabled them to feed on leaves, or on unripe fruits. This may have contributed to the hominoid’s extinction, but a few taxa survived. Fruth and Hohmann [1996] describe a scenario in which the evolution of nest-building aided the survival of the ancestors of extant great apes. They suggest that the first ‘proto-nests’, may have been nests inadvertently constructed during feeding, which became useful as a stable and secure platform for feeding, forming the precursor to sleeping nests. Sleeping nests may have become beneficial as body sizes increased, initially functioning to enable these great apes to monopolise, and guard, feeding sources, but ultimately providing protection from predators at night and the physiological benefits of a good night’s sleep [Fruth and Hohmann 1996]. Given that all four species of extant apes build nests, but not all hominoids (including the Hylobatidae: gibbons and siamang), it is likely that the behaviour evolved before the pongine and hominine lines separated. Evidence of hominoids is sparse in Africa between 13 and 7 Mya [Harrison 2010]. Therefore, it is possible that nest-building evolved not in Africa, but in Eurasia, in an ancestral hominid. This would place the evolution of nest-building in the middle Miocene, at which time there is evidence for the beginnings of major climatic changes [Agustí et al. 2003; Andrews 1996; Begun 2010]. Fruth and Hohmann’s [1996] scenario for the evolution of nest- building is highly plausible. That there are no fossil nests means that any hypotheses regarding the timing of the evolution of nest-building cannot be tested at present, but their conclusion that “nest-building is a phylogenetically conservative behaviour that must have evolved in the Miocene” [Fruth and Hohmann 1996; p.238] is logical. Although the original nests may have evolved through feeding competition, perhaps the lineages of surviving apes that dispersed into Africa and Asia during periods of climatic change were able to expand their range due to these primitive shelters. Hansell [2005; 1993] proposed that through their building behaviour, animal architects make their environment more stable, which may allow them to expand their range and be buffered against extinction processes. For example, the only ants with a world-wide distribution are the weaver ants [Hansell 2005], and the evolution of high-quality paper nests in wasps (Polistinae) allows these species to attach nests to a greater variety of building sites, which may have permitted their greater pantropical range compared to more primitive mud nest- builders [Hansell 1993]. As the climate cooled, and habitat dried, from the middle to late Miocene, nests may have served as shelters to buffer apes against the environmental conditions, 187 including changing climate, vegetation, and predator guilds; thus, ultimately aiding the dispersal and survival of a branch of hominids represented by extant shelter-making species of great apes and humans. This in turn may have improved type and quantity of sleep through better thermoregulation, and a recumbent posture in a safe and secure location, which may in turn have facilitated the evolution of greater cognitive abilities in hominids [cf. Fruth and Hohmann 1996]. A culture of nest-building Few studies have attempted to investigate cultural variation in nest-building across populations [cf. Baldwin et al. 1981]. An initial goal of this thesis was to compare nest-building between two populations in enough detail to assess whether or not any parts of the behaviour represent putative cultural variants. I originally sought to control for environmental influences through detailed and comprehensive measurements of trees and nests. However, it is difficult to control for all environmental influences on variation, even in geographically close communities [e.g. Pruetz et al. 2008]. In addition, the two sites selected for study were on opposite sides of the chimpanzees’ range in Africa and different sub-species. Thus, for any differences found a genetic explanation cannot be ruled out, although other studies argue that genetic co-variance with behavioural variation does not rule out a cultural explanation [Lycett et al. 2011]. In retrospect, cultural variation may have been more thoroughly addressed in comparison of two neighbouring or nearby communities of chimpanzees inhabiting the same eco-type. Comparisons of two disparate communities here rather reveal the extent of similarity in nest building, despite large differences in the vegetation and climate. Most differences in the characteristics of nests (height, distance from the trunk, etc.) described in Chapter 3 could be explained by the environment. For example, Fongoli has much shorter, smaller trees than Issa, and although chimpanzees nest higher within trees in Issa, this may be due to the potential risk of predation (Chapter 5). Baldwin et al [1981] aimed to determine possible cultural variants in nest-building in their earlier cross-site comparison; similarly, they found no differences unexplained by the environment. In Chapter 2, I described data collected on random trees in plots, and on the phenology of trees, which could not be included in this thesis due to time constraints. Availability of trees will influence variation in the trees used, and in turn the nesting spots selected; in the future I hope to use these data to compare how selection of nesting trees and spots differs between the two sites. Some nesting variables used in this study could also have been controlled for other environmental differences between sites differences (e.g. tree height and crown diameter in Chapter 5); for example the number of decisions to the nesting spot is not informative without data on the number of 188 remaining decisions within the nesting tree. Future studies that compare nesting across sites should take care to collect data that can be controlled for differences in the environment (e.g. available vegetation, tree species, tree morphology, etc.). Another possible method to address cultural variation in nest building, using the data collected in this study, would be a case study comparing the trees and nests of tree species common to both sites. Remarkable similarity was also found in nest shape and architecture between Fongoli and Issa. This is at first surprising, considering the distance separating these populations. However, as discussed above, nest-building likely has deep evolutionary roots, and although it is known to have a considerable learned component [Bernstein 1967; Videan 2006a], it is also likely to be invariant because it is phylogenetically conservative and, due to its functions, may not have been characterised by many innovations since its origin. Chapter 8 describes how shaping of specific nest building spots may result in repeated re-use of these spots over time; such niche construction also makes future nest-building more predictable. There may also be few ways in which to construct an arboreal platform. This question could be addressed through cross-taxa comparison, for example, between the nests built by coatis [Olifiers et al. 2009], Andean bears [Goldstein 2002] of South America, and species of great ape may reveal whether convergent evolution of the behaviour results in as much similarity of construction. Experimentally, it would be interesting to recruit architects and engineers to design and build an arboreal platform, given a remit of using only material available in the tree, which must support up to 40kg in weight. Would such structures converge on the size, shape and architecture of great ape nests? Nests at both sites also exhibited a large amount of variation. This variation suggests possible methods for investigating social transmission of nest-building techniques in the future. For example, in Chapter 6, I found that Issa and Fongoli chimpanzees vary in use of lining or depth of nests in response to overnight weather conditions. If these phenomena are socially learned, and spread through the group, such variation may be cultural, despite the influence of environment. In the future, in order to further understand the influence of the environment, I hope to expand the study in Chapter 6 to analyse variation in nest shape and architecture in response to tree-species or tree-morphology; thus, controlling for other environmental influences in addition to climate. However, a major difficulty of the ‘method of exclusion’ to study culture in wild populations, is that it will tend to highlight maladaptive or neutral behaviours, and ‘miss’ socially transmitted behaviours that are influenced by the environment. Culture is beneficial in a species that must rapidly adapt to changing environmental conditions, and cultural traits will then spread, which aid the survival of individuals [Laland et al. 2000]. 189 In Chapter 7, I found that some variation in nest building within a community was accounted for by age or sex and there was variation across individuals. Byrne’s [2003; 2007] framework of learning through ‘behaviour parsing’ provides a possible method to study social transmission of complex skills, like nest-building, within wild living communities, and to compare homogeneity of techniques between communities. In the future I hope to use a rapid screening genetic method (outlined in Chapter 2) to identify individual builders of nests deconstructed in this study, and also apply Byrne’s [2003; 2007] method to analyse the sequence and techniques of nest-construction between individuals and communities. This method may also reveal more about the cognitive complexity of nest building and variation (individual, sex, or community level) in the pattern of behaviour during construction, rather than the detailed summary of components that has been analysed to date. The ontogeny of nest-building has been a neglected field of study and future research in this field is necessary. Some part of nest-building may be innate. Social-learning is likely involved, as individuals in captivity will not build without early observation of a skilled model [Bernstein 1967; Videan 2006a]. However, trial and error learning is also likely involved: Fruth [1995; p.149] describes the first attempts of nest-building by infants as “struggling with the ‘tree-devil’ itself”. Individual learning, aided by social observation, may be important in acquisition of skills, and such observational learning is critical in individual’s attainment of population-specific norms [e.g. proposed mechanisms of cultural spread of nut-cracking, Biro et al. 2003]. When do males learn to build a typically male nest? Or conversely, do females learn their mother’s skills, whilst males pay less attention, as found in termite fishing [Lonsdorf 2005; 2006]? Previous longitudinal studies have investigated learning in termite-fishing, nut-cracking, and leaf-sponging skill acquisition [Biro et al. 2006; Lonsdorf 2005]; such studies of nest-building skill acquisition are long overdue. Niche construction The discovery of a greater density of scars around nest spots within trees described in Chapter 8 was supplementary to the original aims of this thesis. All animal architects, through their building behaviour, exert an influence over the environment, which makes it more predictable [Hansell 2005]. Laland et al. [2000] propose that genetic transmission of behaviour is most favoured by natural selection when environmental conditions are stable, whilst learning is favoured in unpredictable conditions of frequent environmental change. Above, I outlined how the evolution of nests as shelters might have aided survival of great apes during a period of environmental change. The niche construction of hominid nest-building behaviour may have created also a 190 feed-back loop of socially transmitted information, which in turn influenced future building, ranging, and habitat change (e.g. through seed-dispersal [Wrangham et al. 1994]). Future research should investigate how regeneration and shaping of these specific nest spots influences ranging behaviour of apes in different habitats, in particular to comparative data from forested sites. In both Fongoli and Issa, most nest spots had scars. Previous research in the forested site of Kanyawara, in Kibale National Park, Kenya, found that only 58% of nests had scars [Llorente Caño 2004]. If there are less suitable nesting sites available in a savanna environment, this may mean re-use of specific sleeping spots is greater in these drier, more open habitats, which has implications for hominin evolution and behaviour in similar environments. Chimpanzees in dry habitats may be constrained by the availability of nesting spots, leading to high levels of re-use and a greater influence of the niche construction effects of building. Alternatively, in forests, nests spots may regenerate faster and permit more frequent re-use than in dry habitats; if re-use is similarly rapid or frequent in forested habitats, then this would provide strong evidence of the importance of ‘pre-fabrication’ of nest-spots to chimpanzee nest building. Implications for hominin evolution Nest building by early hominins has been hypothesised by several researchers [McGrew 1992; Sabater Pi et al. 1997; Sept 1992; 1998].Wrangham [1987] highlighted the importance of the phylogenetic method for modelling behaviour of early hominins, and as described above, the presence of nest-building in four extant hominids suggests that the behaviour evolved during the middle Miocene and was present in the last ancestral hominid. Many early hominins, prior to the appearance of Homo erectus, exhibited some ape-like anatomical adaptations to arboreality (Ardipithecus ramidus: [White et al. 2009b]; Australopithecus afarensis: [Alemseged et al. 2006]; A. africanus: [Berger and Tobias 1996]; Homo habilis: [Richmond et al. 2002]; cf. A. sediba [Berger et al. 2010]). Due to the dangers of overnight sleep on the ground (discussed in Chapters 4 and 5), even H. erectus seems unlikely to have slept terrestrially, without some protection from predation. Wrangham [2009] proposed that the use of fire might have appeared in H. erectus, and the benefits of cooked food may have contributed to the rapid increase in brain size in this species. Fire also may have provided additional protection from terrestrial predators. Early hominin use of fire is controversial [James 1989]. Some authors have proposed that regular use of fire was common over 1.5 Mya [Clark and Harris 1985; Gowlett et al. 1981], whilst others argued that there is insufficient evidence until 3-400,000 years ago [Roebroeks and Villa 2011]. Kortlandt [1980] proposed that the use of thorny corrals, which may have provided protection during overnight terrestrial sleep in early hominins. Whichever species began first to sleep terrestrially 191 was likely to already exhibit nest-building skills, and nests may have served to buffer the builders from adverse conditions and predation. How were these primitive shelters first adapted for terrestrial sleep? Evidence of shelter is scarce in the archaeological record. The earliest form of shelter dates from approximately 380,000 years ago in Europe [Gamble 1999]. Temporary shelters of nomadic hunter-gatherers are often simple constructions made of few materials, including small trees, saplings, grass and bark [Lee and Daly 1999], and even wet season shelters may take only up to a few hours to build (e.g. Wahadzabe of Tanzania;[Marlowe 2010]). These shelters are built of ephemeral materials and similar structures are not visible in the archaeological record. It seems likely that such protective shelters were necessary for the first dispersals out of Africa (H. erectus, [Anton and Swisher 2004]), in order to survive in more temperate conditions and expand ranges into such a wide variety of habitats. 192 References Agustí J, Siria AS, Garcés M. 2003. Explaining the end of the hominoid experiment in Europe. Journal of Human Evolution 45:145-153. Alemseged Z. 2003. An integrated approach to taphonomy and faunal change in the Shungura Formation (Ethiopia) and its implication for hominid evolution. Journal of Human Evolution 44:451-478. Alemseged Z, Spoor F, Kimbel WH, Bobe R, Geraads D, Reed D, Wynn JG. 2006. A juvenile early hominin skeleton from Dikika, Ethiopia. Nature 443:296-301. Ancrenaz M, Calaque R, Lackman-Ancrenaz I. 2004. Orangutan nesting behavior in disturbed forest of Sabah, Malaysia: implications for nest census. International Journal of Primatology 25:983-1000. Anderson DP, Nordheim EV, Boesch C, Moermond TC. 2002. Factors influencing fission- fusion grouping in chimpanzees in the Tai National Park, Cote d'Ivoire. In: Boesch C, Hohmann G, Marchant LF, editors. Behavioral Diversity in Chimpanzees and Bonobos. Cambridge: Cambridge University Press. p 90-101. Anderson JR. 1984. Ethology and ecology of sleep in monkeys and apes. Advances in the Study of Behavior 14:165-229. Anderson JR. 1998. Sleep, sleeping sites, and sleep-related activities: Awakening to their significance. American Journal of Primatology 46:63-75. Anderson JR. 2000. Sleep-related behavioural adaptations in free-ranging anthropoid primates. Sleep Medicine Reviews 4:355-373. Andrews P. 1996. Palaeoecology and hominoid palaeoenvironments. Biological Reviews 71:257- 300. Anton SC, Swisher CC. 2004. Early dispersals of Homo from Africa. Annual Review of Anthropology 33:271-96. Baldwin PJ. 1979. The natural history of the chimpanzee (Pan troglodytes verus) at Mt. Assirik, Senegal [Ph.D dissertation]. Stirling: University of Stirling. Baldwin PJ, McGrew WC, Tutin CEG. 1982. Wide-ranging chimpanzees at Mt. Assirik, Senegal. International Journal of Primatology 3:367-385. Baldwin PJ, Sabater Pi J, McGrew WC, Tutin CEG. 1981. Comparisons of nests made by different populations of chimpanzees (Pan troglodytes). Primates 22:474-486. Basabose AK. 2005. Ranging patterns of chimpanzees in a Montane forest of Kahuzi, Democratic Republic of Congo. International Journal of Primatology 26:33-54. 193 Begun DR. 2005. Sivapithecus is east and Dryopithecus is west, and never the twain shall meet. Anthropological Science 113:53-64. Begun DR. 2007. Fossil record of Miocene hominoids. In: Henke W, Tattersall I, Begun DR, editors. Handbook of Paleoanthropology. Berlin: Springer Berlin Heidelberg. p 921-977. Begun DR. 2010. Miocene hominids and the origins of the African apes and humans. Annual Review of Anthropology 39:67-84. Berger LR, de Ruiter DJ, Churchill SE, Schmid P, Carlson KJ, Dirks PHGM, Kibii JM. 2010. Australopithecus sediba: A new species of Homo-like Australopith from South Africa. Science 328:195-204. Berger LR, Tobias PV. 1996. A chimpanzee-like tibia from Sterkfontein, South Africa, and its implications for the interpretation of bipedalism in Australopithecus africanus. Journal of Human Evolution 30:343-348. Bermejo M, Rodriguez-Teijeiro JD, Illera G, Barroso A, Vile C, Walsh PD. 2006. Ebola outbreak killed 5000 gorillas. Science 314:1564. Bernstein IS. 1967. Age and experience in chimpanzee nest building. Psychological Reports 20:1106. Biro D, Carvalho S, Matsuzawa T. 2010. Tools, traditions, and technologies: Interdisciplinary approaches to chimpanzee nut cracking. In: Lonsdorf EV, Ross SR, Matsuzawa T, editors. The Mind of the Chimpanzee: Ecological and Experimental Perspectives. Chicago: University of Chicago Press. p 141-155. Biro D, Inoue-Nakamura N, Tonooka R, Yamakoshi G, Sousa C, Matsuzawa T. 2003. Cultural innovation and transmission of tool use in wild chimpanzees: Evidence from field experiments. Animal Cognition 6:213-223. Biro D, Sousa C, Matsuzawa T. 2006. Ontogeny and cultural propagation of tool use by wild chimpanzees at Bossou, Guinea: Case studies in nut cracking and leaf folding. In: Matsuzawa T, Tomonaga M, Tanaka M, editors. Cognitive Development in Chimpanzees. Tokyo: Springer. p 476-508. Blumenschine RJ, Masao FT. 1991. Living sites at Olduvai Gorge, Tanzania? Preliminary landscape archaeology results in the basal bed-II lake margin zone. Journal of Human Evolution 21:451-462. Boesch C. 1991. The effects of leopard predation on grouping patterns in forest chimpanzees. Behaviour 117:220-242. Boesch C, Boesch H. 1984. Possible causes of sex differences in the use of natural hammers by wild chimpanzees. Journal of Human Evolution 13:415-440. 194 Bolwig N. 1959. A study of nests built by mountain gorilla and chimpanzee. South African Journal of Science 55:286-291. Bradley BJ, Boesch C, Vigilant L. 2000. Identification and redesign of human microsatellite markers for genotyping wild chimpanzee (Pan troglodytes verus) and gorilla (Gorilla gorilla gorilla) DNA from faeces. Conservation Genetics 1:289-292. Bradley BJ, Chambers KE, Vigilant L. 2001. Accurate DNA-based sex identification of apes using non-invasive samples. Conservation Genetics 2:179-181. Bradley BJ, Doran-Sheehy DM, Vigilant L. 2008. Genetic identification of elusive animals: Re- evaluating tracking and nesting data for wild western gorillas. Journal of Zoology 275:333-340. Brown DE. 1991. Human Universals. New York: McGraw-Hill. 220 p. Brown F, Harris J, Leakey R, Walker A. 1985. Early Homo erectus skeleton from west Lake Turkana, Kenya. Nature 316:788-792. Brownlow AR, Plumptre AJ, Reynolds V, Ward R. 2001. Sources of variation in the nesting behavior of chimpanzees (Pan troglodytes schweinfurthii) in the Budongo Forest, Uganda. American Journal of Primatology 55:49-55. Busse C. 1980. Leopard and lion predation upon Chacma baboons living in the Moremi wildlife reserve. Botswana Notes and Records 12:15-21. Byrne RW. 1993. Cultural learning: Hierarchical levels of limitation. Behavioural and Brain Sciences 16:516-517. Byrne RW. 2003. Imitation as behaviour parsing. Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 358:529-536. Byrne RW. 2007. Culture in great apes: using intricate complexity in feeding skills to trace the evolutionary origin of human technical prowess. Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 362:577-585. Byrne RW, Byrne JME. 1993. Complex leaf-gathering skills of mountain gorillas (Gorilla g. beringei): Variability and standardization. American Journal of Primatology 31:241-261. Byrne RW, Corp N, Byrne JME. 2001. Estimating the complexity of animal behaviour: How mountain gorillas eat thistles. Behaviour 138:525-557. Cai DJ, Mednick SA, Harrison EM, Kanady JC, Mednick SC. 2009. REM, not incubation, improves creativity by priming associative networks. Proceedings of the National Academy of Sciences of the United States of America 106:10130-10134. 195 Campbell G, Kuehl H, Diarrassouba A, N'Goran PK, Boesch C. 2011. Long-term research sites as refugia for threatened and over-harvested species. Biology Letters published online in advance of print, 10.1098/rsbl.2011.0155. Carvalho S, Biro D, McGrew WC, Matsuzawa T. 2009. Tool-composite reuse in wild chimpanzees (Pan troglodytes): archaeologically invisible steps in the technological evolution of early hominins? Animal Cognition 12:S103-S114. Carvalho S, Cunha E, Sousa C, Matsuzawa T. 2008a. Chaines operatoires and resource- exploitation strategies in chimpanzee (Pan troglodytes) nut cracking. Journal of Human Evolution 55:148-163. Carvalho S, Mcgrew WC. in press. The origins of the Oldowan: Why chimpanzees (Pan troglodytes) are still good models for technological evolution in Africa. In: Domingues- Rodrigo M, editor. Theoretical debates in Early Pleistocene Archaeology. Cambridge: Cambridge University Press. Carvalho S, Sousa C, Matsuzawa T. 2008b. The nut-cracker: Bridging archaeology and primatology - Chimpanzee stone tool use in Bossou and Diecke, Guinea. Folia Primatologica 79:319-319. Chapman CA, Gillespie TR, Goldberg TL. 2005. Primates and the ecology of their infectious diseases: How will anthropogenic change affect host-parasite interactions? Evolutionary Anthropology 14:134-144. Cheney DL, Seyfarth RM, Fischer J, Beehner J, Bergman T, Johnson SE, Kitchen DM, Palombit RA, Rendall D, Silk JB. 2004. Factors affecting reproduction and mortality among baboons in the Okavango Delta, Botswana. International Journal of Primatology 25:401- 428. Cheney DL, Wrangham RW. 1987. Predation. In: Smuts BB, Cheney DL, Seyfarth RM, Wrangham RW, Struhsaker TT, editors. Primate Societies. Chicago: University of Chicago Press. p 227-239. Clark JD, Harris JWK. 1985. Fire and its roles in early hominid lifeways. African Archaeological Review 3:3-27. Collias EC, Collias NE. 1964. The development of nest-building behavior in a weaverbird. The Auk 81:42-52. Coolidge FL, Wynn T. 2009. The Rise of Homo sapiens: The Evolution of Modern Thinking. Malden, MA: Wiley-Blackwell. 308 p. Corp N, Byrne RW. 2002a. Leaf processing by wild chimpanzees: Physically defended leaves reveal complex manual skills. Ethology 108:673-696. 196 Corp N, Byrne RW. 2002b. The ontogeny of manual skill in wild chimpanzees: Evidence from feeding on the fruit of Saba florida. Behaviour 139:137-168. Cowlishaw G. 1994. Vulnerability to predation in baboon populations. Behaviour 131:293-304. D'Amour DE, Hohmann G, Fruth B. 2006. Evidence of leopard predation on bonobos (Pan paniscus). Folia Primatologica 77:212-217. Dawkins R. 1982. The Extended Phenotype: The Long Reach of the Gene. Oxford: Oxford University Press. 307 p. Fan P-F, Jiang X-L. 2008. Sleeping sites, sleeping trees, and sleep-related behaviors of black crested gibbons (Nomascus concolor jingdongensis) at Mt. Wuliang, Central Yunnan, China. American Journal of Primatology 70:153-160. Fay JM, Carroll R, KerbisPeterhans JC, Harris D. 1995. Leopard attack on and consumption of gorillas in the Central African Republic. Journal of Human Evolution 29:93-99. Field A. 2005. Discovering Statistics using SPSS. London: SAGE Publications. 771 p. Fleagle JG, Gilbert CC, Baden AL. 2010. Primate cranial diversity. American Journal of Physical Anthropology 142:565-578. Fowler J, Cohen L, Jarvis P. 1998. Practical Statistics for Field Biology. Chichester, England: John Wiley. 257 p. Fragaszy DM, Perry S, editors. 2003. The Biology of Traditions: Models and Evidence. Cambridge: Cambridge University Press. 445 p. Fraser ON, Schino G, Aureli F. 2008. Components of relationship quality in chimpanzees. Ethology 114:834-843. Freeland WJ. 1980. Mangabey (Cerocebus albigena) movement patterns in relation to food availability and fecal contamination. Ecology 61:1297-1303. Fruth B. 1995. Nests and nest groups in wild bonobos (Pan paniscus): Ecological and behavioural correlates. Aachen, Shaker: Ph.D, Ludwig-Maximilians-Universität München. 187 p. Fruth B, Hohmann G. 1993. Ecological and behavioral aspects of nest building in wild bonobos (Pan paniscus). Ethology 94:113-126. Fruth B, Hohmann G. 1994a. Comparative analyses of nest building behavior in bonobos and chimpanzees. In: Wrangham RW, McGrew WC, de Waal FBM, Heltne PG, editors. Chimpanzee Cultures. Cambridge, MA: Harvard University Press. p 109-128. Fruth B, Hohmann G. 1994b. Nests: Living artefacts of recent apes? Current Anthropology 35:310-311. 197 Fruth B, Hohmann G. 1996. Nest building behaviour in the great apes: The great leap forward? In: McGrew WC, Marchant LF, Nishida T, editors. Great Ape Societies. Cambridge: Cambridge University Press. p 225-240. Furuichi T, Hashimoto C. 2004. Botanical and topographical factors influencing nesting-site selection by chimpanzees in Kalinzu Forest, Uganda. International Journal of Primatology 25:755-765. Furuichi T, Hashimoto C, Tashiro Y. 2001. Extended application of a marked-nest census method to examine seasonal changes in habitat use by chimpanzees. International Journal of Primatology 22:913-928. Gagneux P, Wills C, Gerloff U, Tautz D, Morins PA, Boesch C, Fruth B, Hohmann G, Ryder OA, Woodruff DS. 1999. Mitochondrial sequences show diverse evolutionary histories of African hominoids. Proceedings of the National Academy of Sciences, USA 96:5077- 5082. Gamble C. 1999. The Paleolithic Societies of Europe. Cambridge: Cambridge University Press. 498 p. Garamszegi LZ. 2006. Comparing effect sizes across variables: generalization without the need for Bonferroni correction. Behavioral Ecology 17:682-687. Gedeon CI, Markó G, Néméth I, Nyitrai V, Altbäcker V. 2010. Nest material selection affects nest insulation quality for the European ground squirrel (Spermophilus citellus). Journal of Mammalogy 91:636-641. Gerloff U, Hartung B, Fruth B, Hohmann G, Tautz D. 1999. Intracommunity relationships, dispersal pattern and paternity success in a wild living community of bonobos (Pan paniscus) determined from DNA analysis of faecal samples. Proceedings of the Royal Society of London Series B: Biological Sciences 266:1189-1195. Ghiglieri MP. 1984. The Chimpanzees of the Kibale Forest: A Field Study of Ecology and Social Structure. New York: Columbia University Press. 226 p. Gibson A. 2005. Orangutan nesting preferences in a disturbed tropical deep-peat swamp forest, Central Kalimantan, Indonesia [M. Sc dissertation]. Ipswich: Otley College. 131 p. Gillespie TR, Nunn CL, Leendertz FH. 2008. Integrative approaches to the study of primate infectious disease: Implications for biodiversity conservation and global health. Yearbook of Physical Anthropology 51:53-69. Goldberg TL, Ruvolo M. 1997. Molecular phylogenetics and historical biogeography of east African chimpanzees. Biological Journal of the Linnean Society 61:301-324. 198 Goldstein IR. 2002. Andean bear-cattle interactions and tree nest use in Bolivia and Venezuela. Ursus 13:369-372. Goodall J. 1968. The behaviour of free-living chimpanzees in the Gombe Stream Reserve. Animal Behaviour Monographs 1:165-311. Goodall JM. 1962. Nest building behavior in free ranging chimpanzee. Annals of the New York Academy of Sciences 102:455-467. Goodwin D. 1976. Crows of the World. Ithaca, NY: British Natural History Museum Publications. 299 p. Goren-Inbar N. 2004. Evidence of hominin control of fire at Gesher Benot Ya'aqov, Israel. Science 304:725-727. Gowlett JAJ, Harris JWK, Walton D, Wood BA. 1981. Early archaeological sites, hominid remains and traces of fire from Chesowanja, Kenya. Nature 294:125-129. Groves CP, Sabater Pi J. 1985. From ape's nest to human fix-point. Man 20:22-47. Gruber T, Clay Z, Zuberbühler K. 2010. A comparison of bonobo and chimpanzee tool use: evidence for a female bias in the Pan lineage. Animal Behaviour 80:1023-1033. Gwinner H, Berger S. 2005. European starlings: nestling condition, parasites and green nest material during the breeding season. Journal of Ornithology 146:365-371. Hamilton WJ. 1982. Baboon sleeping site preferences and relationships to primate grouping patterns. American Journal of Primatology 3:41-53. Hansell M. 1984. Animal Architecture and Building Behaviour. London: Longman. 324 p. Hansell M. 2005. Animal Architecture. Oxford: Oxford University Press. 314 p. Hansell M. 2007. Built by Animals: the Natural History of Animal Architecture. Oxford: Oxford University Press. 258 p. Hansell M, Ruxton GD. 2008. Setting tool use within the context of animal construction behaviour. Trends in Ecology & Evolution 23:73-78. Hansell MH. 1993. The ecological impact of animal nests and burrows. Functional Ecology 7:5- 12. Hansell MH. 2000. Bird Nests and Construction Behaviour. Cambridge: Cambridge University Press. 273 p. Harris R. 1983. Arboriculture: Care of Trees, Shrubs, and Vines in the Landscape. Englewood Cliffs, NJ: Prentice-Hall. 656 p. Harrison T. 2010. Apes among the tangled branches of human origins. Science 327:532-534. Hashimoto C. 1995. Population census of the chimpanzees in the Kalinzu Forest, Uganda: Comparison between methods with nest counts. Primates 36:477-488. 199 Haslam M, Hernandez-Aguilar A, Ling V, Carvalho S, de la Torre I, DeStefano A, Du A, Hardy B, Harris J, Marchant L et al. 2009. Primate archaeology. Nature 460:339-344. Hausfater G, Meade BJ. 1982. Alternation of sleeping groves by yellow baboons (Papio cynocephalus) as a strategy for parasite avoidance. Primates 23:287-297. Hawkins CE, Racey PA. 2008. Food habits of an endangered carnivore, Cryptoprocta ferox, in the dry deciduous forests of western Madagascar. Journal of Mammalogy 89:64-74. Hayward MW. 2006. Prey preferences of the spotted hyena (Crocuta crocuta) and degree of dietary overlap with the lion (Panthera leo). Journal of Zoology 270:606-614. Hayward MW, Henschel P, O'Brien J, Hofmeyr M, Balme G, Kerley GIH. 2006a. Prey preferences of the leopard (Panthera pardus). Journal of Zoology 270:298-313. Hayward MW, Kerley GIH. 2005. Prey preferences of the lion (Panthera leo). Journal of Zoology 267:309-322. Hayward MW, O'Brien J, Hofmeyr M, Kerley GIH. 2006b. Prey preferences of the African wild dog Lycaon pictus (Canidae: Carnivora): Ecological requirements for conservation. Journal of Mammalogy 87:1122-1131. Hediger H. 1977. Nest and home. Folia Primatologica 28:170-187. Heenan CB, Seymour RS. 2011. Structural support, not insulation, is the primary driver for avian cup-shaped nest design. Proceedings of the Royal Society B: Biological Sciences published online before print February 16, 2011, doi:10.1098/rspb.2010.2798. Henschel P, Abernethy KA, White LJT. 2005. Leopard food habits in the Lopé National Park, Gabon, Central Africa. African Journal of Ecology 43:21-28. Herbinger I, Boesch C, Rothe H. 2001. Territory characteristics among three neighboring chimpanzee communities in the Taï National Park, Côte d'Ivoire. International Journal of Primatology 22:143-167. Hernandez-Aguilar R. 2006. Ecology and nesting patterns of chimpanzees (Pan troglodytes) in Issa, Ugalla, Western Tanzania [Ph.D dissertation]. Los Angeles: University of Southern California. Hernandez-Aguilar RA. 2009. Chimpanzee nest distribution and site reuse in a dry habitat: implications for early hominin ranging. Journal of Human Evolution 57:350-364. Hicks TC. 2010. A chimpanzee mega-culture? Exploring behavioral continuity in Pan troglodytes schweinfurthii across northern DR Congo [Ph.D dissertation]. Amsterdam: Universiteit van Amsterdam. 300 p. 200 Hiraiwa-Hasegawa M. 1989. Sex differences in the behavioral development of chimpanzees at Mahale. In: Heltne PG, Marquardt LA, editors. Understanding Chimpanzees. Cambridge, MA: Harvard University Press. p 104-115. Hiraiwa-Hasegawa M, Byrne RW, Takasaki H, Byrne JME. 1986. Aggression toward large carnivores by wild chimpanzees of Mahale Mountains National Park, Tanzania. Folia Primatologica 47:8-13. Hladik A. 1978. Phenology of leaf production in rain rainforest of Gabon: Distribution and composition of food for folivores. In: Montgomery GG, editor. The Ecology of Arboreal Folivores. Washington, DC: Smithsonian Institution Press. p 51-71. Horn AD. 1980. Some observations on the ecology of the bonobo chimpanzee (Pan paniscus, Schwarz 1929) near Lake Tumba, Zaire. Folia Primatologica 34:145-169. Houle A, Chapman CA, Vickery WL. 2004. Tree climbing strategies for primate ecological studies. International Journal of Primatology 25:237-260. Howells ME, Pruetz J, Gillespie TR. 2010. Patterns of gastro-intestinal parasites and commensals as an index of population and ecosystem health: the case of sympatric western chimpanzees (Pan troglodytes verus) and guinea baboons (Papio hamadryas papio) at Fongoli, Senegal. American Journal of Primatology 73:173-179. Huffman MA. 2001. Self-medicative behavior in the African great apes: An evolutionary perspective into the origins of human traditional medicine. BioScience 51:651-661. Humle T, Matsuzawa T. 2002. Ant-dipping among the chimpanzees of Bossou, Guinea, and some comparisons with other sites. American Journal of Primatology 58:133-148. Humle T, Snowdon CT, Matsuzawa T. 2009. Social influences on ant-dipping acquisition in the wild chimpanzees (Pan troglodytes verus) of Bossou, Guinea, West Africa. Animal Cognition 12:S37-S48. Hunt KD, McGrew WC. 2002. Chimpanzees in the dry habitats of Assirik, Senegal and Semliki Wildlife Reserve, Uganda. In: Boesch C, Hohmann G, Marchant LF, editors. Behavioural Diversity in Chimpanzees and Bonobos. Cambridge: Cambridge University Press. p 35- 51. Hutcheson G, Sofroniou N. 1999. The Multivariate Social Scientist. London: SAGE publications Ltd. 274 p. Isaac G. 1978. The food-sharing behavior of protohuman hominids. Scientific American 238:90- 108. Iwata Y, Ando C. 2007. Bed and bed-site reuse by western lowland gorillas (Gorilla g. gorilla) in Moukalaba-Doudou National Park, Gabon. Primates 48:77-80. 201 Izawa K, Itani J. 1966. Chimpanzee in Kasakati Basin, Tanganyika. I. Ecological study in the rainy season 1963-1964. Kyoto University African Studies 1:73-156. James SR. 1989. Hominid use of fire in the Lower and Middle Pleistocene: A review of the evidence. Current Anthropology 30:1-26. Jenny D, Zuberbühler K. 2005. Hunting behaviour in West African forest leopards. African Journal of Ecology 43:197-200. Jones C, Sabater Pi J. 1971. Comparative ecology of Gorilla gorilla (Savage & Wyman) and Pan troglodytes (Blumenbach) in Rio Muni, East Africa. Bibliotheca Primatologica 13:1-96. Jones CG, Lawton JH, Shachak M. 1994. Organisms as ecosystem engineers. Oikos 69:373-386. Kahlenberg SM, Wrangham RW. 2010. Sex differences in chimpanzees' use of sticks as play objects resemble those of children. Current Biology 20:R1067-R1068. Kaiser H. 1970. A second-generation little jiffy. Psychometrika 35:401-415. Kamilar JM, Marshack JL. 2011. Variation in chimpanzee 'culture' is predicted by local ecology, not geography. Biology Letters published online 4 May 2011 doi: 10.1098/rsbl.2011.0376 Kano T. 1972. Distribution and adaptation of the chimpanzee on the eastern shore of Lake Tanganyika. Kyoto University African Studies 7:37-129. Kano T. 1979. A pilot study of the ecology of pygmy chimpanzees, Pan paniscus. In: Hamburg DA, McCown ER, editors. The Great Apes. Menlo Park, CA: Benjamin/Cummings. p 123-135. Kano T. 1992. The Last Ape: Pygmy Chimpanzee Behavior and Ecology. Stanford, CA: Stanford University Press. 248 p. Kappeler PM. 1998. Nests, tree holes, and the evolution of primate life histories. American Journal of Primatology 46:7-33. Karkanas P, Shahack-Gross R, Ayalon A, Bar-Matthews M, Barkai R, Frumkin A, Gopher A, Stiner MC. 2007. Evidence for habitual use of fire at the end of the Lower Palaeolithic: Site-formation processes at Qesem Cave, Israel. Journal of Human Evolution 53:197- 212. Karlin C, Pelegrin J. 1988. Outil. In: Leroi-Gourhan A, editor. Dictionaire de la prehistoire Quadrige. Paris: Presses Universitaires de France. p 823. Kaur T, Singh J, Tong S, Humphrey C, Clevenger D, Tan W, Szekely B, Wang Y, Li Y, Alex Muse E et al. 2008. Descriptive epidemiology of fatal respiratory outbreaks and detection of a human-related metapneumovirus in wild chimpanzees (Pan troglodytes) at Mahale Mountains National Park, Western Tanzania. American Journal of Primatology 70:755- 765. 202 Keele BF, Jones JH, Terio KA, Estes JD, Rudicell RS, Wilson ML, Li YY, Learn GH, Beasley TM, Schumacher-Stankey J et al. 2009. Increased mortality and AIDS-like immunopathology in wild chimpanzees infected with SIVcpz. Nature 460:515-519. Kimura Y. 1999. Tool-using strategies by early hominids at Bed II, Olduvai Gorge, Tanzania. Journal of Human Evolution 37:807-831. Köndgen S, Kühl H, N'Goran PK, Walsh PD, Schenk S, Ernst N, Biek R, Formenty P, Mätz- Rensing K, Schweiger B et al. 2008. Pandemic human viruses cause decline of endangered great apes. Current Biology 18:260-264. Köndgen S, Schenk S, Pauli G, Boesch C, Leendertz F. 2010. Noninvasive monitoring of respiratory viruses in wild chimpanzees. EcoHealth 7:332-341. Koops K, Humle T, Sterck EHM, Matsuzawa T. 2007. Ground-nesting by the chimpanzees of the Nimba Mountains, Guinea: environmentally or socially determined? American Journal of Primatology 69:407-419. Kortlandt A. 1980. How might early hominids have defended themselves against large predators and food competitors? Journal of Human Evolution 9:79-112. Kouakou CY, Boesch C, Kuehl H. 2009. Estimating chimpanzee population size with nest counts: validating methods in Tai National Park. American Journal of Primatology 71:447-457. Krief S, Escalante AA, Pacheco MA, Mugisha L, Andre C, Halbwax M, Fischer A, Krief JM, Kasenene JM, Crandfield M et al. 2010. On the diversity of malaria parasites in African apes and the origin of Plasmodium falciparum from bonobos. PLoS Pathogens 6:1-12. Lafuma L, Lambrechts MM, Raymond M. 2001. Aromatic plants in bird nests as a protection against blood-sucking flying insects? Behavioural Processes 56:113-120. Laland KN, Janik VM. 2006. The animal cultures debate. Trends in Ecology & Evolution 21:542-547. Laland KN, Odling-Smee J, Feldman MW. 2000. Niche construction, biological evolution, and cultural change. Behavioral and Brain Sciences 23:131-175. Langergraber KE, Boesch C, Inoue E, Inoue-Murayama M, Mitani JC, Nishida T, Pusey A, Reynolds V, Schubert G, Wrangham RW et al. 2011. Genetic and 'cultural' similarity in wild chimpanzees. Proceedings of the Royal Society B: Biological Sciences 278:408-416. Largo CL, Bastian ML, van Schaik CP. 2009. Mosquito avoidance drives selection of nest tree species in Bornean orang-utans. Folia Primatologica 80:163-163. Lee RB, Daly R. 1999. The Cambridge Encyclopedia of Hunters and Gatherers. Cambridge: Cambridge University Press. 493 p. 203 Leendertz FH, Pauli G, Maetz-Rensing K, Boardman W, Nunn C, Ellerbrok H, Jensen SA, Junglen S, Christophe B. 2006. Pathogens as drivers of population declines: The importance of systematic monitoring in great apes and other threatened mammals. Biological Conservation 131:325-337. Lehmann J, Dunbar R. 2009. Implications of body mass and predation for ape social system and biogeographical distribution. Oikos 118:379-390. Lesku JA, Roth Ii TC, Amlaner CJ, Lima SL. 2006. A phylogenetic analysis of sleep architecture in mammals: the integration of anatomy, physiology, and ecology. American Naturalist 168:441-453. Li D, Ren B, Grueter CC, Li B, Li M. 2010. Nocturnal sleeping habits of the Yunnan snub-nosed monkey in Xiangguqing, China. American Journal or Primatology 72:1092-9. Lima SL, Rattenborg NC. 2007. A behavioural shutdown can make sleeping safer: a strategic perspective on the function of sleep. Animal Behaviour 74:189-197. Lima SL, Rattenborg NC, Lesku JA, Amlaner CJ. 2005. Sleeping under the risk of predation. Animal Behaviour 70:723-736. Liu WM, Li YY, Learn GH, Rudicell RS, Robertson JD, Keele BF, Ndjango JBN, Sanz CM, Morgan DB, Locatelli S et al. 2010. Origin of the human malaria parasite Plasmodium falciparum in gorillas. Nature 467:420-425. Llorente Caño M. 2004. Estudio comparado de la conducta nidificadora de los chimpancés (Pan troglodytes schweinfurthii) de la comunidad de Kanyawara (Parque Nacional de Kibale, Uganda) [PhD dissertation]. Barcelona, Spain: Universidad de Barcelona. Lodwick JL, Borries C, Pusey AE, Goodall J, McGrew WC. 2004. From nest to nest - influence of ecology and reproduction on the active period of adult Gombe chimpanzees. American Journal of Primatology 64:249-260. Lonsdorf EV. 2005. Sex differences in the development of termite-fishing skills in the wild chimpanzees, Pan troglodytes schweinfurthii, of Gombe National Park, Tanzania. Animal Behaviour 70:673-683. Lonsdorf EV. 2006. What is the role of mothers in the acquisition of termite-fishing behaviors in wild chimpanzees (Pan troglodytes schweinfurthii)? Animal Cognition 9:36-46. Lonsdorf EV, Eberly LE, Pusey AE. 2004. Sex differences in learning in chimpanzees. Nature 428:715-716. Lycett SJ, Collard M, McGrew WC. 2007. Phylogenetic analyses of behavior support existence of culture among wild chimpanzees. Proceedings of the National Academy of Sciences 104:17588-17592. 204 Lycett SJ, Collard M, McGrew WC. 2011. Correlations between genetic and behavioural dissimilarities in wild chimpanzees (Pan troglodytes) do not undermine the case for culture. Proceedings of the Royal Society B: Biological Sciences published online 13 April 2011 doi: 10.1098/rspb.2010.2599 MacKinnon J. 1974. The behaviour and ecology of wild orang-utans (Pongo pygmaeus). Animal Behaviour 22:3-74. Makacha S, Schaller GB. 1969. Observations on lions in the Lake Manyara National Park, Tanzania. African Journal of Ecology 7:99-103. Marchesi P, Marchesi N, Fruth B, Boesch C. 1995. Census and distribution of chimpanzees in Cote d'Ivoire. Primates 36:591-607. Marlowe F. 2010. The Hadza: Hunter-Gatherers of Tanzania. Berkeley: University of California Press. 319 p. Massawe ET. 1992. Assessment of the status of chimpanzee populations in western Tanzania. African Study Monographs 13:35-55. Mathewson PD, Spehar SN, Meijaard E, Nardiyono, Purnomo, Sasmirul A, Sudiyanto, Oman, Sulhnudin, Jasary et al. 2008. Evaluating orangutan census techniques using nest decay rates: Implications for population estimates. Ecological Applications 18:208-221. Matsuda I, Tuuga A, Higashi S. 2010. Effects of water level on sleeping-site selection and inter- group association in proboscis monkeys: why do they sleep alone inland on flooded days? Ecological Research 25:475-482. Maughan JE, Stanford CB. 2001. Terrestrial nesting by chimpanzees in Bwindi Impenetrable National Park, Uganda. American Journal of Physical Anthropology Suppl 32:104. McGrew WC. 1977. Socialization and object manipulation of wild chimpanzees. In: Poirer FE, Chevalier-Skolnikoff S, editors. Primate Bio-social Development: Biological, Social and Ecological Determinants. New York: Garland Press. p 261-288. McGrew WC. 1979. Evolutionary implications of sex differences in chimpanzee predation and tool use. In: Hamburg DA, McCown ER, editors. Perspectives on Human Evolution. Menlo Park, CA: Benjamin/Cummings. p 441-463. McGrew WC. 1992. Chimpanzee Material Culture: Implications for Human Evolution. Cambridge: Cambridge University Press. 277 p. McGrew WC. 2004. The Cultured Chimpanzee: Reflections on Cultural Primatology. Cambridge: Cambridge University Press. 248 p. McGrew WC. 2010. In search of the last common ancestor: new findings on wild chimpanzees. Philosophical Transactions of the Royal Society B: Biological Sciences 365:3267-3276. 205 McGrew WC, Baldwin PJ, Marchant LF, Pruetz JD, Scott SE, Tutin CEG. 2003. Ethoarchaeology and elementary technology of unhabituated wild chimpanzees at Assirik, Senegal, west Africa. Paleoanthropology 1:1-20. McGrew WC, Baldwin PJ, Tutin CEG. 1981. Chimpanzees in a hot, dry and open habitat: Mt. Assirik, Senegal, West Africa. Journal of Human Evolution 10:227-244. Meddis R. 1975. On the function of sleep. Animal Behaviour 23:676-691. Mehlman PT, Doran DM. 2002. Influencing western gorilla nest construction at Mondika Research Center. International Journal of Primatology 23:1257-1285. Mercader J, Barton H, Gillespie J, Harris J, Kuhn S, Tyler R, Boesch C. 2007. 4,300-year-old chimpanzee sites and the origins of percussive stone technology. Proceedings of the National Academy of Sciences of the United States of America 104:3043-3048. Mercader J, Panger M, Boesch C. 2002. Excavation of a chimpanzee stone tool site in the African rainforest. Science 296:1452-1455. Mitani JC, Sanders WJ, Lwanga JS, Windfelder TL. 2001. Predatory behavior of crowned hawk- eagles (Stephanoaetus coronatus) in Kibale National Park, Uganda. Behavioral Ecology and Sociobiology 49:187-195. Moore J. 1992. Savanna chimpanzees. In: Nishida T, McGrew WC, Marler P, Pickford M, de Waal FBM, editors. Topics in Primatology. Tokyo: University of Tokyo Press. p 99-118. Moore J. 1996. Savanna chimpanzees, referential models and the last common ancestor. In: McGrew WC, Marchant LF, Nishida T, editors. Great Ape Societies. Cambridge: Cambridge University Press. p 275-292. Moran MD. 2003. Arguments for rejecting the sequential Bonferroni in ecological studies. Oikos 100:403-405. Morbeck M, Zihlman A. 1989. Body size and proportions in chimpanzees, with special reference to Pan troglodytes schweinfurthii from Gombe National Park, Tanzania. Primates 30:369-382. Moyer D, Plumptre AJ, Pintea L, Hernandez-Aguilar RA, Moore JJ, Stewart FA, Davenport TRB, Piel AK, Kamenya S, Mugabe H et al. 2006. Surveys of Chimpanzees and other Biodiversity in Western Tanzania. Report to United States Fish and Wildlife Service. 65 p. Nishida T. 1989. A note on the chimpanzee ecology of the Ugalla Area, Tanzania. Primates 30:129-138. 206 Nsubuga AM, Robbins MM, Roeder AD, Morin PA, Boesch C, Vigilant L. 2004. Factors affecting the amount of genomic DNA extracted from ape faeces and the identification of an improved sample storage method. Molecular Ecology 13:2089-2094. Odling-Smee FJ, Laland KN, Feldman MW. 2003. Niche Construction: The Neglected Process in Evolution. Princeton, NJ: Princeton University Press. 472 p. Ogawa H, Gen'ich I, Moore J, Pintea L, Hernandez-Aguilar RA. 2007. Sleeping parties and nest distribution of chimpanzees in the savanna woodland, Ugalla, Tanzania. International Journal of Primatology 28:1397-1412. Okamoto-Mizuno K, Mizuno K, Michie S, Maeda A, Iizuka S. 1999. Effects of humid heat exposure on human sleep stages and body temperature. Sleep 22:767-773. Olifiers N, Bianchi RC, Mouro GM, Gompper ME. 2009. Construction of arboreal nests by brown-nosed coatis, Nasua nasua (Carnivora: Procyonidae) in the Brazilian Pantanal. Zoologia 26:571-574. Oppenheimer JR, Lang GE. 1969. Cebus monkeys: Effect on branching of Gustavia trees. Science 165:187-188. Pedersen AB, Altizer S, Poss M, Cunningham AA, Nunn CL. 2005. Patterns of host specificity and transmission among parasites of wild primates. International Journal for Parasitology 35:647-657. Perneger TV. 1998. What's wrong with Bonferroni adjustments. British Medical Journal 316:1236-1238. Perry S, Rose L. 1994. Begging and transfer of coati meat by white-faced capuchin monkeys, Cebus capucinus. Primates 35:409-415. Piel AK, Moore JM. 2007. Locating elusive animals: Using a passive acoustic system to study savanna chimpanzees at Ugalla, Western Tanzania. American Journal of Physical Anthropology Suppl. 44:189. Piel AP, Moore J. 2010. Monitoring movements: Tracking unhabituated chimpanzees in Ugalla, Western Tanzania using a novel method - real-time acoustic monitoring. International Primatological Society. Kyoto, Japan. Pinowski J, Haman A, Jerzak L, Pinowska B, Barkowska M, Grodzki A, Haman K. 2006. The thermal properties of some nests of the Eurasian tree sparrow Passer montanus. Journal of Thermal Biology 31:573-581. Pintea L, Plumptre AJ. 2006. Prediction of suitable habitat for chimpanzees using remote sensing and GIS. In: Moyer D, Plumptre AJ, Pintea L, Hernandez-Aguilar A, Moore J, Stewart FA, Davenport TRB, Piel AP, Kamenya S, Mugabe H and others, editors. Surveys of 207 chimpanzees and other biodiversity in Western Tanzania: unpublished report to US Fish and Wildlife. Plumptre AJ. 2003. Recommendations for censusing chimpanzee populations in forests. In: Kormos R, Boesch C, Bakarr MI, Butynski TM, editors. West African Chimpanzees: Status Survey and Conservation Action Plan. Gland, Switzerland: IUCN - World Conservation Union. p 179-182. Plumptre AJ, Cox D. 2006. Counting primates for conservation: primate surveys in Uganda. Primates 47:65-73. Plumptre AJ, Reynolds V. 1996. Censusing chimpanzees in the Budongo Forest, Uganda. International Journal of Primatology 17:85-99. Plumptre AJ, Reynolds V. 1997. Nesting behavior of chimpanzees: Implications for censuses. International Journal of Primatology 18:475-485. Potts R. 1984. Home bases and early hominids. American Scientist 72:338-347. Pruetz JD. 2006. Feeding ecology of savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal. In: Hohmann G, Robbins MM, Boesch C, editors. Feeding Ecology in Apes and Other Primates: Ecological, Physical and Behavioral Aspects. Cambridge: Cambridge University Press. p 161-182. Pruetz JD. 2007. Evidence of cave use by savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: implications for thermoregulatory behavior. Primates 48:316-319. Pruetz JD, Bertolani P. 2009. Chimpanzee (Pan troglodytes verus) behavioural responses to stresses associated with living in a savanna-mosaic environment: Implications for hominin adaptations to open habitats. Paleoanthropology 2009:252-262. Pruetz JD, Fulton SJ, Marchant LF, McGrew WC, Schiel M, Waller M. 2008. Arboreal nesting as anti-predator adaptation by savanna chimpanzees (Pan troglodytes verus) in southeastern Senegal. American Journal of Primatology 70:393-401. Pruetz JD, Marchant LF, Arno J, McGrew WC. 2002. Survey of savanna chimpanzees (Pan troglodytes verus) in southeastern Senegal. American Journal of Primatology 58:35-43. Prugnolle F, Durand P, Neel C, Ollomo B, Ayala FJ, Arnathau C, Etienne L, Mpoudi-Ngole E, Nkoghe D, Leroy E et al. 2010. African great apes are natural hosts of multiple related malaria species, including Plasmodium falciparum. Proceedings of the National Academy of Sciences of the United States of America 107:1458-1463. Rattenborg NC, Lima SL, Amlaner CJ. 1999. Half-awake to the risk of predation. Nature 397:397-398. 208 Rayadin Y, Saitoh T. 2009. Individual variation in nest size and nest site features of the Bornean orangutans (Pongo pygmaeus). American Journal of Primatology 71:393-399. Redman P, Selman C, Speakman JR. 1999. Male short-tailed field voles (Microtus agrestis) build better insulated nests than females. Journal of Comparative Physiology B: Biochemical, Systemic, and Environmental Physiology 169:581-587. Reed KE. 1997. Early hominid evolution and ecological change through the African Plio- Pleistocene. Journal of Human Evolution 32:289-322. Reed KE. 1998. Using large mammal communities to examine ecological and taxonomic structure and predict vegetation in extant and extinct assemblages. Paleobiology 24:384- 408. Reichard U. 1998. Sleeping sites, sleeping places, and presleep behavior of gibbons (Hylobates lar). American Journal of Primatology 46:35-62. Remis MJ. 1993. Nesting behavior of lowland gorillas in the Dzanga-Sangha Reserve, Central African Republic: Implications for population estimates and understandings of group dynamics. Tropics 2:245-255. Richmond BG, Aiello LC, Wood BA. 2002. Early hominin limb proportions. Journal of Human Evolution 43:529-548. Roebroeks W, Villa P. 2011. On the earliest evidence for habitual use of fire in Europe. Proceedings of the National Academy of Sciences 108:5209-5214. Roeder AD, Jeffery K, Bruford MW, Inprimat c. 2006. A universal microsatellite multiplex kit for genetic analysis of great apes. Folia Primatologica 77:240-245. Roth TC, Rattenborg NC, Pravosudov VV. 2010. The ecological relevance of sleep: the trade-off between sleep, memory and energy conservation. Philosophical Transactions of the Royal Society B: Biological Sciences 365:945-959. Rudicell RS, Piel AP, Stewart FA, Moore D, Learn GH, Li Y, Takehisa J, Pintea L, Moore J, Sharp PM et al. in press. High prevalence of Simian Immunodeficiency Virus Infection in a community of savanna chimpanzees. Journal of Virology. Ruff C. 2009. Relative limb strength and locomotion in Homo habilis. American Journal of Physical Anthropology 138:90-100. Russon AE, Handayani DP, Kuncoro P, Ferisa A. 2007. Orangutan leaf-carrying for nest- building: toward unraveling cultural processes. Animal Cognition 10:189-202. Sabater Pi J, Vea JJ, Serrallonga J. 1997. Did the first hominids build nests? Current Anthropology 38:914-916. Schaller GB. 1961. The orang-utan in Sarawak. Zoologica 46:73-82. 209 Schaller GB. 1963. The Mountain Gorilla: Ecology and Behavior. Chicago: University of Chicago Press. 431 p. Schick KD. 1987. Modeling the formation of early stone-age artifact concentrations. Journal of Human Evolution 16:789-807. Schoeninger MJ, Reeser H, Hallin K. 2003. Paleoenvironment of Australopithecus anamensis at Allia Bay, East Turkana, Kenya: evidence from mammalian herbivore enamel stable isotopes. Journal of Anthropological Archaeology 22:200-207. Schöning C, Humle T, Möbius Y, McGrew WC. 2008. The nature of culture: Technological variation in chimpanzee predation on army ants revisited. Journal of Human Evolution 55:48-59. Sept JM. 1992. Was there no place like home - a new perspective on early hominid archaeological sites from the mapping of chimpanzee nests. Current Anthropology 33:187-207. Sept JM. 1998. Shadows on a changing landscape: Comparing nesting patterns of hominids and chimpanzees since their last common ancestor. American Journal of Primatology 46:85- 101. Shultz S, No R, McGraw WS, Dunbar RIM. 2004. A community-level evaluation of the impact of prey behavioural and ecological characteristics on predator diet composition. Proceedings of the Royal Society, Series B: Biological Sciences 271:725-732. Siegel JM. 2005. Clues to the functions of mammalian sleep. Nature 437:1264-1271. Skowron C, Kern M. 1980. The insulation in nests of selected north American songbirds. The Auk 97:816-824. Stanford CB, O'Malley RC. 2008. Sleeping tree choice by Bwindi chimpanzees. American Journal of Primatology 70:642-649. Stewart C-B, Disotell TR. 1998. Primate evolution - in and out of Africa. Current Biology 8:R582-R588. Stewart F, Pruetz J, Hansell M. 2007. Do chimpanzees build comfortable nests? American Journal of Primatology 69:930-939. Stewart FA, Piel AP, Mcgrew WC. in press. Living archaeology: artefacts of specific nest-site fidelity in wild chimpanzees. Journal of Human Evolution 10.1016/j.jhevol.2011.05.005. Stewart FA, Piel AP, Pruetz JDP. 2010. Living archaeology: Evidence of specific bed-site re-use in wild chimpanzees. International Primatological Society XXIII Congress 2010. Kyoto, Japan. 210 Sunderland-Groves J, Ekinde A, Mboh H. 2009. Nesting behavior of Gorilla gorilla diehli at Kagwene Mountain, Cameroon: Implications for assessing group size and density. International Journal of Primatology 30:253-266. Suzuki A. 1969. An ecological study of chimpanzees in savanna woodland. Primates 10:103-148. Tabachnick BG, Fidell LS. 2007. Using Multivariate Statistics. Boston: Pearson/Allyn & Bacon. 980 p. Tsukahara T. 1993. Lions eat chimpanzees: The first evidence of predation by lions on wild chimpanzees. American Journal of Primatology 29:1-11. Tutin CEG, Fernandez M. 1984. Nationwide census of gorilla (Gorilla g. gorilla) and chimpanzee (Pan t. troglodytes) populations in Gabon. American Journal of Primatology 6:313-336. Tutin CEG, McGrew WC, Baldwin PJ. 1983. Social organization of savanna-dwelling chimpanzees, Pan troglodytes verus, at Mt. Assirik, Senegal. Primates 24:154-173. Tutin CEG, Parnell RJ, White LJT, Fernandez M. 1995. Nest building by lowland gorillas in Lope Reserve, Gabon: Environmental influences and implications for censusing. International Journal of Primatology 16:53-76. Vallet D, Petit EJ, Gatti S, Levrero F, Menard N. 2008. A new 2CTAB/PCI method improves DNA amplification success from faeces of Mediterranean (Barbary macaques) and tropical (lowland gorillas) primates. Conservation Genetics 9:677-680. van Schaik CP, Ancrenaz M, Borgen G, Galdikas B, Knott CD, Singleton I, Suzuki A, Utami SS, Merrill M. 2003. Orangutan cultures and the evolution of material culture. Science 299:102-105. Videan EN. 2006a. Bed-building in captive chimpanzees (Pan troglodytes): the importance of early rearing. American Journal of Primatology 68:745-751. Videan EN. 2006b. Sleep in captive chimpanzee (Pan troglodytes): the effects of individual and environmental factors on sleep duration and quality. Behavioral Brain Research 169:187- 192. Wasser S, Poole J, Lee P, Lindsay K, Dobson A, Hart J, Douglas-Hamilton I, Wittemyer G, Granli P, Morgan B et al. 2010. Elephants, ivory, and trade. Science 327:1331-1332. White TD, Ambrose SH, Suwa G, Su DF, DeGusta D, Bernor RL, Boisserie J-R, Brunet M, Delson E, Frost S et al. 2009a. Macrovertebrate paleontology and the Pliocene habitat of Ardipithecus ramidus. Science 326:67, 87-93. White TD, Asfaw B, Beyene Y, Haile-Selassie Y, Lovejoy CO, Suwa G, WoldeGabriel G. 2009b. Ardipithecus ramidus and the paleobiology of early hominids. Science 326:75-86. 211 Whiten A. 2011. The scope of culture in chimpanzees, humans and ancestral apes. Philosophical Transactions of the Royal Society B: Biological Sciences 366:997-1007. Whiten A, Goodall J, McGrew WC, Nishida T, Reynolds V, Sugiyama Y, Tutin CEG, Wrangham RW, Boesch C. 1999. Cultures in chimpanzees. Nature 399:682-685. Whiten A, Goodall J, McGrew WC, Nishida T, Reynolds V, Sugiyama Y, Tutin CEG, Wrangham RW, Boesch C. 2001. Charting cultural variation in chimpanzees. Behaviour 138:1481-1516. Williams JM, Lonsdorf EV, Wilson ML, Schumacher-Stankey J, Goodall J, Pusey AE. 2008. Causes of death in the Kasekela chimpanzees of Gombe National Park, Tanzania. American Journal of Primatology 70:766-777. Williams JM, Pusey AE, Carlis JV, Farm BP, Goodall J. 2002. Female competition and male territorial behaviour influence female chimpanzees' ranging patterns. Animal Behaviour 63:347-360. Wilson BF. 2000. Apical control of branch growth and angle in woody plants. American Journal of Botany 87:601-607. Wrangham RW. 1987. The significance of African apes for reconstructing human social evolution. In: Kinzey WG, editor. The Evolution of Human Behavior: Primate Models. Albany, NY: State University of New York Press. p 51-71. Wrangham RW. 1992. Comment on Sept, J. M. 1992. Was there no place like home - a new perspective on early hominid archaeological sites from the mapping of chimpanzee nests. Current Anthropology 33:201-202. Wrangham RW. 2009. Catching Fire: How Cooking Made Us Human. New York: Basic Books. 309 p. Wrangham RW, Chapman CA, Chapman LJ. 1994. Seed dispersal by forest chimpanzees in Uganda. Journal of Tropical Ecology 10:355-368. Wroblewski EE, Murray CM, Keele BF, Schumacher-Stankey JC, Hahn BH, Pusey AE. 2009. Male dominance rank and reproductive success in chimpanzees, Pan troglodytes schweinfurthii. Animal Behaviour 77:873-885. Wrogemann D. 1992. Wild chimpanzees in Lope, Gabon: Census method and habitat use [PhD dissertation]. Bremen: Bremen University. Zuberbühler K, Jenny D. 2002. Leopard predation and primate evolution. Journal of Human Evolution 43:873-886.