Dietary variation within and across species drives the eco-evolutionary responsiveness of genes necessary to metabolize nutrients and other components. Recent evidence from humans and other mammals suggests that sugar-rich diets of floral nectar and ripe fruit have favored mutations in, and functional preservation of, the ADH7 gene, which encodes the ADH class 4 enzyme responsible for metabolizing ethanol. Here we interrogate a large, comparative dataset of ADH7 gene sequence variation, including the amino acid residue located at key site 294 that regulates the affinity of ADH7 for ethanol. Our analyses span 171 mammal species, including 59 newly sequenced. We report extensive variation, especially among frugivorous and nectarivorous bats, with potential for functional impact, and widespread variation in the retention and probable pseudogenization of ADH7. However, we find little statistical evidence of a broad impact of diet on putative ADH7 function or tuning at site 294 across mammals, suggesting that the evolution of ADH7 is shaped by complex factors. Our study reports extensive new diversity in a gene of longstanding ecological interest, offers new sources of variation to be explored in functional assays in future studies, and advances our understanding of the processes of molecular evolution.

We used two complementary approaches to examine ADH7 sequence variation: 1) a targeted-capture massively-parallel sequencing approach, and 2) amplification of exon 7 followed by a Sanger sequencing approach. Sanger sequencing was completed on an Applied Biosystems 3730xl (96 capillary) Genetic Analyzer using BigDye Terminator chemistry. For the target capture approach, we used a custom set of ​​biotinylated RNA probes (myBaits®, Arbor Biosciences, Ann Arbor, MI) to capture the ADH7 coding region, followed by sequencing on an Illumina NextSeq 500 using a 2x150 paired end NextSeq 500/550 mid-output v2.5 (300 Cycles) kit. We removed adapters and trimmed Illumina short-reads using BBDuk from the BBTools suite (Bushnell 2018). We removed bases lower than q=10 from the left and right ends of reads and only retained reads that were at least 25 bases long after trimming. We used the pipeline HybPiper (https://github.com/mossmatters/HybPiper/wiki) to map, extract, and assemble gene sequences from the targeted enrichment sequencing reads.