Our paper “Visual complexity of egg patterns predicts egg rejection according to Weber’s Law” has just been published in the journal Proceedings of the Royal Society B. This research was led by Tanmay Dixit, and carried out together with Andrei Apostol, Kuan-Chi Chen, Tony Fulford, Chris Town and Claire Spottiswoode, in a collaboration between biologists and computer scientists. We used machine learning to compute a biologically-relevant measure of egg pattern complexity, and combined this with field experiments in Zambia to investigate how complexity evolves in an arms race between host egg signatures (by tawny-flanked prinias) and parasitic egg forgeries (by cuckoo finches).
Specifically, we quantified the complexity of egg patterns of tawny-flanked prinias (hosts of cuckoo finches) using a machine learning algorithm that optimised the complexity measure such that complexity differences between eggs best predicted egg rejection, according to field data. This means that complexity was quantified in a biologically-relevant manner, and such an algorithm could be used to quantify complexity in other systems.
We also showed that complexity predicts rejection according to Weber’s Law (c.f. our earlier paper on Weber’s Law and how to test it, https://onlinelibrary.wiley.com/doi/full/10.1111/evo.14290; also see the news post at https://www.africancuckoos.com/2021/07/new-paper-published-on-webers-law-and-mimicry). Finally we showed that cuckoo finch eggs have simpler patterns than prinia eggs (see image, with a prinia egg on the left and cuckoo finch egg on the right), suggesting that high complexity in egg patterns has evolved to make forgery of these ‘signatures’ difficult.