Genome-enabled discovery of evolutionary divergence in brains and behavior

Chinar Patil,Jonathan B Sylvester,Patrick T McGrath,Jeffrey T Streelman,Kawther Abdilleh,Karen Pottin,Ryan F Bloomquist,Milan Malinsky,Michael W Norsworthy,Zachary V Johnson

doi:10.1038/s41598-021-92385-8

Abstract

Lake Malawi cichlid fishes exhibit extensive divergence in form and function built from a relatively small number of genetic changes. We compared the genomes of rock- and sand-dwelling species and asked which genetic variants differed among the groups. We found that 96% of differentiated variants reside in non-coding sequence but these non-coding diverged variants are evolutionarily conserved. Genome regions near differentiated variants are enriched for craniofacial, neural and behavioral categories. Following leads from genome sequence, we used rock- vs. sand-species and their hybrids to (i) delineate the push–pull roles of BMP signaling and irx1b in the specification of forebrain territories during gastrulation and (ii) reveal striking context-dependent brain gene expression during adult social behavior. Our results demonstrate how divergent genome sequences can predict differences in key evolutionary traits. We highlight the promise of evolutionary reverse genetics—the inference of phenotypic divergence from unbiased genome sequencing and then empirical validation in natural populations.

Highlights

Lake Malawi cichlid fishes exhibit extensive divergence in form and function built from a relatively small number of genetic changes
We compared the genomes of 8 rock dwellers and 14 sand dwellers to uncover the genomic signature of rock- versus sand-evolutionary diversification
We found that 0.06% of Single Nucleotide Polymorphisms (SNPs) and 0.44% of InDels are alternately fixed between rock- and sand-groups

Summary

Introduction

Lake Malawi cichlid fishes exhibit extensive divergence in form and function built from a relatively small number of genetic changes. Our understanding of how the genome encodes natural variation in form and function is still limited This is the case for almost any trait, from height to behavior to complex disease[1]. The reasons for this are manifold, but they include an underappreciated role of non-coding genetic variants linked to differences in traits. This is apparent in our assumptions and in syntheses of data. The take home message from this work has been that a small number of genes from recognizable pathways explain a considerable proportion of phenotypic variance These studies may be biased in interpretation and limited in inference space. Approach is similar to genome-first examples in mammals[17,18] wherein unbiased analysis of genome data selects the traits rather than a trait-first model like QTL mapping

Objectives

Methods

Results

Conclusion