Abstract: Climate change is having profound effects on the distributions of species globally. Trait-based assessments predict that specialist and range-restricted species are among those most likely to be at risk of extinction from such changes. Understanding individual species’ responses to climate change is therefore critical for informing conservation planning. We use an established Species Distribution Modelling (SDM) protocol to describe the curious range-restriction of the globally threatened White-tailed Swallow (Hirundo megaensis) to a small area in southern Ethiopia. We find that, across a range of modelling approaches, the distribution of this species is well described by two climatic variables, maximum temperature and dry season precipitation. These same two variables have been previously found to limit the distribution of the unrelated but closely sympatric Ethiopian Bush-crow (Zavattariornis stresemanni). We project the future climatic suitability for both species under a range of climate scenarios and modelling approaches. Both species are at severe risk of extinction within the next half century, as the climate in 68–84% (for the swallow) and 90–100% (for the bush-crow) of their current ranges is predicted to become unsuitable. Intensive conservation measures, such as assisted migration and captive-breeding, may be the only options available to safeguard these two species. Their projected disappearance in the wild offers an opportunity to test the reliability of SDMs for predicting the fate of wild species. Monitoring future changes in the distribution and abundance of the bush-crow is particularly tractable because its nests are conspicuous and visible over large distances.

Methods: We collated all available geo-referenced records of White-tailed Swallows and their nests, and Ethiopian Bush-crows and their nests, collected by various observers between 2005 and 2011. Between 2012 and 2015, we conducted 266 walked 1-km transects at locations across, and outside, the Ethiopian Bush-crow's core range, additionally surveying for White-tailed Swallows on 255 of them. Bush-crows or their nests were recorded on 102 of these 1-km transects. White-tailed Swallows were recorded on 19 transects. Additionally, in 2014, swallow nest records were documented in the north-west of the species’ range as part of an intensive breeding study. We also collected GPS locations for all ad-hoc Ethiopian Bush-crow and White-tailed Swallow observations throughout this period. This yielded a total of 2,921 presence locations for the bush-crow, and 574 presence locations for the swallow.

We repeatedly drove the road network in and around the Bush-crow’s range between 2008 and 2014 with at least two experienced observers in the vehicle searching for nests. Bush-crow nests are large, robust and conspicuous structures, made from thick twigs, often projecting above the canopy of the bush or tree in which they are built, and can last for several years. They can be detected at distances of up to one kilometre from a moving vehicle and most stretches of road were driven several times by different observers. Hence, we assumed that stretches of roads with no records of birds or nests on any survey represented Bush-crow absences. We divided the driven road network into 1-km segments, and took the mid-point of each segment without a bird or nest record within one kilometre of the route as an absence point, thus all absence points had a precision of one kilometre. In addition to the absences from walked transects, this gave us a total of 1,905 absence locations for the bush-crow.

For the White-tailed Swallow, we took the mid-point of the 236 1-km transect legs on which swallows had not been recorded. Unlike for the bush-crow, we did not use the lack of observations from stretches of road transects as true absence points because, unlike the bush-crow and its highly visible nests, White-tailed Swallows are much more difficult to detect reliably from a moving vehicle.

To increase the range of environmental variables on which the models were built, we took a random sample of 4,764 pseudo-absences (for the swallow) and 3,095 pseudo-absences (for the bush-crow) from a region stretching from 1.86°–6.87°N and 33.17°–43.67°E. We restricted these points to be at least 10 km from any presence location for the species concerned. When combined with the known absence records for each species, this gave a total of 5,000 points treated as absences in the analysis.

For model validation purposes, the White-tailed Swallow’s range was divided radially into five geographic segments, and the Ethiopian Bush-crow's range was divided radially into eight geographic segments.

The data files are as follows:

"Bush-crow+nest_records_2005-2015" contains data on all verified records of the Ethiopian Bush-crow Zavattariornis stresemanni and its nests, collected by various observers between 2005 and 2015. Data comes from a mix of walked and driven transects across the species' range, and all ad-hoc records made during fieldwork.

"White-tailed_Swallow+nest_records_2005-2015" contains data on all verified records of the White-tailed Swallow Hirundo megaensis and its nests, collected by various observers between 2005 and 2015. Data comes from a mix of walked and driven transects across the species' range, and all ad-hoc records made during fieldwork.

"Bush-crow_true_absences" contains data on known locations within the range of the Ethiopian Bush-crow Zavattariornis stresemanni where the species does not occur. This includes the centre points of walked 1-km transects where no birds or nests were recorded, and the centre point of each 1-km stretch of regularly driven road on which no birds or nests were recorded within 1 km.

"White-tailed_Swallow_true_absences" contains data on known locations within the range of the White-tailed Swallow Hirundo megaensis where the species does not occur. These are derived from the centre points of walked 1-km transects where no birds or nests were recorded.

"Bush-crow_background_points" contains a set of randomly sampled locations drawn from within a tile stretching from 1.86°–6.87°N and 33.17°–43.67°E, but restricted to be at least 10 km away from the nearest record of an Ethiopian Bush-crow Zavattariornis stresemanni bird or nest.

"White-tailed_Swallow_background_points" contains a set of randomly sampled locations drawn from within a tile stretching from 1.86°–6.87°N and 33.17°–43.67°E, but restricted to be at least 10 km away from the nearest record of an White-tailed Swallow Hirundo megaensis bird or nest.

"Bush-crow_segments" contains a set of polygons covering a tile stretching from 1.86°–6.87°N and 33.17°–43.67°E which were used for dividing up the Bush-crow presence, absence and background data for model validation purposes.

"White-tailed_Swallow_segments" contains a set of polygons covering a tile stretching from 1.86°–6.87°N and 33.17°–43.67°E which were used for dividing up the swallow presence, absence and background data for model validation purposes.