Natural selection shapes variation in genome-wide recombination rate in Drosophila pseudoobscura

While recombination is widely recognized to be a key modulator of numerous evolutionary phenomena, we have a poor understanding of how recombination rate itself varies and evolves within a species. Here, we performed a comprehensive study of recombination rate (rate of meiotic crossing over) in two natural populations of Drosophila pseudoobscura from Utah and Arizona, USA. We used an amplicon sequencing approach to obtain high-quality genotypes in approximately 8000 individual backcrossed offspring (17 mapping populations with roughly 530 individuals each), for which we then quantified crossovers. Interestingly, variation in recombination rate within and between populations largely manifested as differences in genome-wide recombination rate rather than remodeling of the local recombination landscape. Comparing populations, we discovered individuals from the Utah population displayed on average 8% higher crossover rates than the Arizona population, a statistically significant difference. Using a QST-FST analysis, we found that this difference in crossover rate was dramatically higher than expected under neutrality, indicating that this difference may have been driven by natural selection. Finally, using a combination of short and long read whole-genome sequencing, we found no significant association between crossover rate and structural variation at the 200-400kb scale. Our results demonstrate that (1) there is abundant variation in genome-wide crossover rate in natural populations, (2) at the 200-400kb scale, recombination rate appears to vary largely genome wide, rather than in specific intervals and (3) interpopulation differences in recombination rate may be the result of local adaptation.

Methods
- Primers were designed as described in Samuk et al. 2020

- We followed the GT-seq library prep protocol in Campbell et al. (2017) https://onlinelibrary.wiley.com/doi/full/10.1111/1755-0998.12357

- We sequenced the libraries in two NextSeq 500 lanes at the Duke Genomics Core

- We demultiplexed the libraries using fastq-multx, as shown here https://github.com/ksamuk/samuk_et_al_curr_biol_2020/blob/master/analysis_GATK/scripts/01_demultiplex_one_lane.sh