Repository logo

Visual Adaptations and Behavioural Strategies to Detect and Catch Small Targets



Change log


Fabian, Samuel Timothy  ORCID logo


Predatory behaviours are ideal for studying the limits of performance and control within animals. Predation naturally creates a competition between the sensors and physiology of predator and prey. Aerial predation demonstrates the greatest feats of physical performance, demanding the highest speeds and accelerations whilst both predator and prey are free to pitch, yaw, and roll. These high speeds and degrees of rotational freedom make control a complex problem. However, from the perspective of the researcher attempting to decipher the control laws that underpin predator guidance, the question is made more soluble by the predator’s fixation on its target. The goal of the pursuer is clear, to contact the target, and thus their systems are focused on the optimization of that action. This is as opposed to more mundane activities, where conflicting interests compete for the attention and behavioural response of the animal. In order to study the necessary trade-offs that underpin aerial predation, this thesis will focus on the hunting behaviour of two fly species. The first is a robber fly, Holcocephala fusca, on which the majority of the first two chapters focus. Secondarily, work with the killer fly Coenosia attenuata will be included in the latter two chapters as a direct contrast to results from Holcocephala. Both are miniature dipteran predators, but not closely related. The structure of this thesis is broken into six chapters, summarised in the following list:

  1. Thecompoundeyeofinsectsgenerallyhasmuchpoorerresolutionthanthatofcameratype eyes. Poor resolution is exacerbated in smaller insects that cannot commit the resources required for eyes with large lenses that facilitate high spatial resolution. Holcocephala has developed a small number of facets into a forward-facing acute zone where the spatial acuity is reduced to ~0.28°, rivalling the very best resolution of any compound eye. The only compound eyes with a comparable spatial resolution belong to dragonflies, in excess of an order of magnitude larger than Holcocephala.
  2. Numerous potential targets may be airborne within the visual range of a predator. Not all of these may be suitable. Chasing unsuitable targets may waste energy or result in direct harm should they turn out to be larger than the predator can overcome. It is thus a strong imperative for a predator to filter the targets it takes after. Targets silhouetted against the sky display a paucity of cues that a predator could use to determine their size. Holcocephala displays acute size selectivity towards smaller targets. This selectivity goes beyond heuristic rules and size/speed ratios. Instead, Holcocephala appears able to determine absolute size and distance of targets.
  3. Both Holcocephala and Coenosia intercept targets, heading for where the target is going to be in the future rather than its current location. Both species plot trajectories in keeping with the guidance law of proportional navigation, an algorithm derived for modern guided missiles. There are key differences evident in the internal physiological constants applied to the control system between the species. These differences are likely linked to the specific environmental conditions and visual physiologies of the flies, especially the range at which targets are attacked.
  4. Stemming from the use of the proportional navigational framework, this chapter dives into the intricacies of gain and the weighting of the navigational constant, and the geometric factors that underpin the control effort and eventual success of the control system.
  5. “Falcon-diving” can be found in killer flies dropping from their enclosure ceiling, in which they miss targets after diving towards them. Through proportional navigation, it can be demonstrated that the navigational system combined with excessive speed results in acceleration demands the body cannot match.
  6. Holcocephala is capable of evading static obstacle whilst intercepting targets. Application of proportional navigation and a secondary obstacle-evasive controller can demonstrate where the fly is combining multiple inputs to guide its heading.





Gonzalez-Bellido, Paloma Tamara


Insect, Flight, Predation, Navigation, Interception, Steering, Vision, Behaviour


Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
This work was funded by the United States Airforce Office of Scientific Research.