Surprising genetic architecture for colour pattern diversity in guppies

Colour patterns of the Trinidadian guppy (Poecilia reticulata) are amongst the most variable traits observed in nature, and have fascinated biologists for a hundred years. We sought to understand the genetic basis of these highly variable colour traits.
Surprising genetic architecture for colour pattern diversity in guppies

Male guppy colour patterns are amongst the most variable traits observed in nature but, beyond the observation that sons inherit similar colour patterns to their fathers, the genes that control variation remain unknown. The observation of many diverse colour patterns within a species or population represents a classic example of evolution maintaining variation at a trait (balancing selection). To what extent is the genetic basis of colour linked to the guppy sex chromosome? Are there other regions of the genome that contribute towards colour? Are these maintained as a large supergene as seen in other species?

Colour traits of the Trinidadian guppy present an exciting system for investigating the evolution of colour polymorphism. Strong sexual dimorphism exists, with males displaying a mosaic of complex and diverse colouration patterns, made up of blacks, oranges, and iridescent blues, which vary in number, shape, size and position of spots. These extraordinary colour patterns have been taken to extremes by domestic breeders - “fancy guppies” - popular in many home aquaria. Female guppies on the other hand are a tan, uniform colour.

A typical male (top) and female (bottom) guppy. From Deacon et al 2015, Photo: Sean Earnshaw

Female guppies show a strong mate choice preference for male colour patterns, in particular for large spots and/or intensity of orange colour. Despite this strong directional selection, the levels of segregating variation in individual components of guppy colour pattern are some of the highest reported in the literature. There is growing evidence that this remarkably high amount of phenotypic variance is maintained by negative frequency dependent selection (NFDS), driven by female mate preference of rare morphs (a fresh face) and also frequency-dependent survival (predators can more easily recognise common males). In many guppy populations, colour variation is largely Y-linked, meaning sons resemble their fathers. Therefore, we know that colour variation is heritable and under selection, but beyond that, the underlying genetic architecture remains largely unknown.

Variation in guppy colour pattern. (A) female; (B,C,D) male guppies sampled from different rivers in Northern Trinidad. From Valvo et al 2019. Photos: Jennifer Valvo 

Due to the extremely high variation in colour pattern traits, we used an innovative approach to identify genomic regions associated with colour polymorphism.

Schematic representation of the Iso-Y line breeding design

We created four inbred “Iso-Y” lines. Each Iso-Y line was founded by one male showing distinct colouration patterns. Every 2-3 generations, for ~40 generations, the sons with colour patterns most closely resembling that of the original father were crossed with daughters from the other Iso-Y lines. This unique breeding strategy was designed to homogenise regions unrelated to colour, and thus allowed us to better delineate regions of the genome related to the heritability of Y-linked colour pattern. Our phenotype analysis of the Iso-Y lines showed that they were significantly distinct from one another across nearly 10,000 colour measurements (per fish)!

What might we predict about the architecture and location of colour pattern loci? We predicted that the Iso-Y males would differ in parts of the Y chromosome, perhaps forming a supergene with the sex-determining locus. Our results surprised us.

Looking at genetic differentiation between four Iso-Y lines (using Pool-seq), we found that indeed the whole chromosome containing the sex-determining locus (LG12) was differentiated between the lines. However, we were surprised to find that the most consistent region of differentiation was a 7.4 Mb section in the middle of an autosome (LG1). Not only was this region the most differentiated, we also found that it was incredibly diverse, comprising up to 4 haplotypes, some of which co-occurred within Iso-Y lines. The region also contained promising colour gene candidates.

But of course, these analyses were conducted on inbred Iso-Y lines so our next step was to explore what LG1 looks like in the natural population. Using whole genome sequencing of 26 wild-caught individuals, we found that indeed, the LG1 region is maintained in nature, where it shows high linkage disequilibrium and again, multiple maintained haplotypes. We found no evidence that the diversity of the region was the result of structural variants, repeat elements, duplications or translocations. Honing in on the region further, we pinpointed conserved breakpoints within the LG1 region, including a section where males are out of Hardy Weinberg Equilibrium, and a section of high differentiation between males and females. This suggests that selection is operating differently between the sexes.

So how is an autosomal region (LG1) associated with sex-linked colour polymorphism? We hypothesise through Y-autosome epistasis, which has been shown to be an important mechanism for maintaining male-specific fitness traits in other species. We think that Y-autosome epistasis may be a well-suited architecture to maintain diversity in a Y-linked trait, such as NFDS for colour pattern in guppies. This would allow divergent colour haplotypes on autosomes to be regulated by sex-linked regions on the Y-chromosome, masking phenotypes in females and providing an additional mechanism to generate further diversity. Our ongoing research is directly testing the role of Y-autosome epistasis versus Mendelian inheritance by conducting focussed crosses between Iso-Y lines with different LG1 haplotypes.

You can read the full paper at: