Mimicry, Saltational Evolution, and the Crossing of Fitness Valleys

Abstract

The relative contribution of gradual and saltational change to evolution has been debated ever since Darwin (1859) emphasized gradualism in his theory of evolution by natural selection. The phenomenon of mimicry was an important example in this debate. In mimicry evolution, members of a population or species become similar in appearance to an aposematic model species and thereby gain increased protection from predation. The number of steps in the approach to mimicry could be few or many and their sizes either large or small. In 1915, Punnett published an influential book on mimicry in butterflies, in which he summed up his strong opposition to gradualistic accounts of mimicry evolution. He dismissed previous suggestions by Poulton (e.g. Poulton 1912, 1913) that mimicry could emerge in a sequence of steps, beginning with the appearance of a rough likeness of the would-be mimic to its model, followed by further improvement in resemblance. Punnett’s main argument was representative of the thinking of the early Mendelians, who often pointed to a lack of intermediates between existing variants, inferring that the variants had originated as mutants in a single step, as opposed to being molded by natural selection through successive replacements of intermediate forms. As supporting evidence, Punnett used examples of female-limited polymorphic mimicry, for instance the one found in the butterfly Papilio polytes, where the developmental switching between female morphs was known to be controlled by a small number of Mendelian factors, later shown to be alleles at an autosomal locus (Clarke and Sheppard 1972). Punnett’s position was thus a radical saltationism. In terms of Adaptive Landscapes, if the mimic-to-be resides on one adaptive peak and the model on another, mimicry evolves in a single mutational leap, and the issue of natural selection only enters through the constraint that the peak jumped to should be higher than the starting peak. This would apply both to Mullerian mimicry, where the starting point is an aposematic species, and to Batesian mimicry, where the starting adaptive peak of the palatable mimic-to-be is determined by functions other than aposematism, for instance crypsis or other protective coloration, like flash coloration (Cott 1940; Ruxton et al. 2004), or partner choice. In response to claims like those by Punnett (1915), and as part of his efforts to unify gradualism and Mendelian genetics, Fisher (1927, 1930) presented a fully gradualistic alternative. He envisaged a genetically variable population of mimics-to-be and proposed that individuals with trait values deviating from the population mean in the direction of the traits of the model species would be slightly favored over deviations in the opposite direction, because of a slightly higher probability of being mistaken for the model by predators. The outcome could be a gradual shifting of the mean trait values in the direction of improved resemblance. It appears that Fisher intended his gradualistic scenario to apply both to Batesian and Mullerian relations, but it gained more attention in the latter case. In terms of (frequency-dependent) Adaptive Landscapes of Mullerian mimicry evolution, Fisher’s process would be a gradual shifting of two peaks, until they overlap and, approximately, merge to a single peak. Fisher’s (1927, 1930) proposition was not generally accepted. The strongest opposition came from