Explaining the evolution of imperfect mimicry
Mimicry research has played a crucial role in the development of evolutionary theory and the modern synthesis. Imperfect mimicry, however, was largely ignored, or even assumed to not exist, by many early theoreticians who postulated that imperfect mimics exist in an “adaptive valley” between crypsis and accurate mimicry. In recent decades, the widespread existence of imperfect mimics has gained attention as researchers recognised the need to explain this apparent evolutionary puzzle. The last three decades have seen many explanations suggested for imperfect mimicry, resulting in a plethora of hypotheses, but limited understanding of general principles. The aim of this thesis is to increase our understanding the evolution of imperfect mimicry. The first Chapter reviews our current state of knowledge. Hypotheses are classified along several orthogonal dimensions, identifying similarities and differences between them, and their assumptions and underlying evidence are discussed. Methods for quantifying similarity are reviewed, and desirable properties for subjects of imperfect mimicry studies are discussed.
A challenge for imperfect mimicry research is how to quantify variation in mimetic accuracy. Chapter two introduces an open source R package, trajr, that aids in the analysis and numerical characterisation of paths followed by moving animals. Chapter three describes methods for quantifying visual resemblances, and provides R implementations to simplify the use of these methods by future researchers.
Hypothesis testing is essential for understanding the evolution of imperfect mimicry. Chapter four tests for behavioural mimicry in the flight paths of clearwing moths that mimic wasps and bees, and finds that accurate locomotor mimics may compensate for poor morphological mimicry. Chapter five uses Australian ant mimics to test a number of hypotheses that were explored in chapter one. We find evidence that limited sampling of mimics by naïve predators leads to decreased selection for accurate mimicry, and also prevents further sampling that would allow them to learn to discriminate mimics: the information limitation hypothesis. We find no evidence that an accurate mimetic trait can either compensate for – or constrain – another, or that rapid movement can reduce selection pressure for good mimicry.
It is widely assumed that predators exert strong selective pressure on their prey, which is expressed by the life-dinner principle: rabbits run faster than foxes because the rabbit runs for its life, while the fox only runs for its dinner. This assumption can also be applied to a mimicry context, where mimics should ‘run’ faster to be indistinguishable from their models than predators ‘run’ to distinguish mimics from models. In chapter six, we develop models to show that, since prey typically outnumber their predators, the life-dinner principle is not as generally applicable as usually assumed, and selection may be stronger on predators than on prey, so predators may “outrun” their prey.
This thesis tests a variety of possible general mechanisms for maintaining imperfect mimicry, and finds evidence supporting two of them: information limitation and reduced selection pressure on abundant prey. It documents a possible example of accurate behavioural mimicry compensating for poor morphological mimicry in clearwing moths. Perhaps most importantly, it aims to help guide future research by imposing some order on imperfect mimicry hypotheses and by providing tools for analysing locomotor behaviour and quantifying visual resemblances.