Extensive genome-wide homozygosity tracts reveal micro-environment population structure in Drosophila populations.
Authors: Peter Andolfatto 1; Clair Han 2,3; Patrick Reilly 2,4; Andrew Taverner 2; Ana Pinharanda 1; Sheel Chandra 1; Kevin Deitz 5; Daniel Matute 6
Affiliations: 1) Columbia University; 2) Princeton University ; 3) HHMI Janelia ; 4) Yale University; 5) American Museum of Natural History; 6) University of North Carolina Chapel Hill
Keywords: b. population variation; a. genome evolution
Drosophila population samples are usually established as descendent populations of a single wild-caught female (“isofemale lines”). As a result, whole genome sequencing (WGS) data has generally been collected either from isogenized versions of isofemale lines or from pools of wild-caught individuals. Both methods represent a departure from the approach used for most other organisms, which is to sequence outbred diploid individuals. In a population genomic survey of Drosophila santomea, an island endemic sister-species of D. yakuba, we re-sequenced the genomes of 34 wild-caught individuals widely distributed across the island of Saõ Tomé. Plots of population-level genome-wide diversity reveal nothing particularly remarkable. Despite this, closer inspection reveals unusually long runs of homozygosity (ROH) within individuals. Remarkably, a substantial proportion of these individuals (>75%) harbor more ROH than expected for offspring of matings at the level of 1st cousins or closer. The level of diversity between individuals greatly exceeds that of within individuals (FIT = 0.194), strongly suggesting non-random mating. This is particularly surprising given the small geographic scale over which these samples were collected and indicates the existence of a large number of extremely small micro-environments on the island. The general lack of WGS of outbred diploid individuals in Drosophila prevents the easy identification of such ROH patterns. Among the rare exceptions is D. sechellia, for which wild collected individuals exhibit non-random mating patterns similar to that of D. santomea. Simulations reveal that the most likely demographic model for both species is one of many extremely small populations (resembling isofemale lines) with low but non-zero migration between them. Our findings have important implications for the fine-scale structure of Drosophila populations and downstream population genomic inference of demographic and selection parameters.