Abstract Understanding how environmental variation influences population genetic structure can help predict how environmental change influences population connectivity, genetic diversity, and evolutionary potential. We used riverscape genomics modelling to investigate how climatic and habitat variables relate to patterns of genetic variation in two stonefly species, one from mainstem river habitats (Sweltsa coloradensis) and one from tributaries (Sweltsa fidelis) in 40 sites in northwest Montana, USA. We produced a draft genome assembly for S. coloradensis (N50 = 0.251 Mbp, BUSCO > 95% using “insecta_ob9” reference genes). We genotyped 1930 SNPs in 372 individuals for S. coloradensis and 520 SNPs in 153 individuals for S. fidelis. We found higher genetic diversity for S. coloradensis compared to S. fidelis, but nearly identical genetic differentiation among sites within each species (both had global loci median FST = 0.000), despite differences in stream network location. For landscape genomics and testing for selection, we produced a less stringently filtered data set (3454 and 1070 SNPs for S. coloradensis and S. fidelis, respectively). Environmental variables (mean summer precipitation, slope, aspect, mean June stream temperature, land cover type) were correlated with 19 putative adaptive loci for S. coloradensis. but there was only one putative adaptive locus for S. fidelis (correlated with aspect). Interestingly, we also detected potential hybridization between multiple Sweltsa species which has never been previously detected. Studies like ours, that test for adaptive variation in multiple related species are needed to help assess landscape connectivity and the vulnerability of populations and communities to environmental change.

Environmental variables used in GEA tests for loci in Sweltsa with high genetic differentiation (F_ST). Land cover for 30m pixels were collected from the National Land Cover Database 2016 (https://www.mrlc.gov/data/nlcd-2016-land-cover-conus). Annual, spring (SpPrecip) and summer (SumPrecip) precipitation data come from Dayment (Thornton et al. 1997, 2012). Aspect (northness), slope and elevation data come from NED 10 m DEM (http://seamless.usgs.gov). Net primary productivity (NPP) come from https://www.ntsg.umt.edu/files/modis/MOD17UsersGuide2015_v3.pdf. Day to last snow cover (DLS) was calculated after determining the first snow free 8-day period (Modis 2017). Mean modeled June (STJune) and July (STJuly) stream temperature data were from mean monthly stream temperature predictions for baseline period 1986-2005 (see Jones et al. 2014, 100 m pixels). Daily minimum air temperatures (DMT) were collected from Daymet (Thornton et al. 1997, 2012) and mean degree days (DegDays) for the period from Jan 1st - June 7th for 2012-2017 were calculated from the same data set.