When species interact antagonistically, ‘arms races’ played out on coevolutionary battlegrounds can drive the rapid evolution of intricate and complex adaptations in each species. However, evolution in one species may prevent optimality in these adaptations from being realised in the other.

In the Choma District of southern Zambia, the brood-parasitic cuckoo finch Anomalospiza imberbis parasitises four species of cisticolid warbler. In response to parasitic egg mimicry, these four hosts have evolved inter-individual variation in egg colours and patterns: so-called egg signatures. Such individual signatures, used to convey identity, are in many ways analogous to human signatures, passwords, and codes. This means we can use principles from fields such as computer science and cryptography to ask (1) how could signature traits be optimised in principle; and (2) do we observe such optimality in nature? The aim of this thesis, alongside arguing for a broader definition of coevolution (Chapter 2), is to answer these two questions using a combination of theoretical perspectives from the physical sciences, and field experiments.

In Chapter 3, I generate hypotheses about how different forms of perception influence coevolution. In Chapter 4, I test these ideas in my study system. Using an optimisation algorithm and field experiments, I show that egg pattern complexity predicts egg rejection, determine how complexity is perceived, and predict its evolutionary trajectory. In Chapter 5, I show that while this evolutionary trajectory has been followed, traits in hosts and parasites are nevertheless suboptimal. In Chapters 6 and 7, I use mathematical tools to quantify phenotypes in hosts and parasites, showing that hosts do not use traits optimally, likely due to perceptual constraints or inherent trade-offs in signature production. Finally, in Chapter 8 I show that whether host signature variation is categorically or continuously distributed has important consequences for hosts.

Overall, I conclude that while rapid evolution can occur due to the strong selection pressures inherent to arms races, mechanistic factors may constrain the evolution of optimal traits. Even under strong coevolutionary selection pressures, animals may not exhibit traits as optimal as an engineer might design them.