When environmental variation is spatially continuous, dispersing individuals move among nearby sites with similar habitat conditions. But as an environmental gradient becomes steeper, gene flow may connect more divergent habitats, and this is predicted to reduce the slope of the adaptive cline that evolves. We compared quantitative genetic divergence of Rana temporaria frog populations along a 2,000-m elevational gradient in eastern Switzerland (new experimental results) with divergence along a 1,550-km latitudinal gradient in Fennoscandia (previously published results). Both studies found significant countergradient variation in larval development rate (i.e., animals from cold climates developed more rapidly). The cline was weaker with elevation than with latitude. Animals collected on both gradients were genotyped at 2,000 single-nucleotide polymorphism markers, revealing that dispersal distance was 30% farther on the latitudinal gradient but 3.9 times greater with respect to environmental conditions on the elevational gradient. A meta-analysis of 19 experimental studies of anuran populations spanning temperature gradients revealed that countergradient variation in larval development, while significant overall, was weaker when measured on steeper gradients. These findings support the prediction that adaptive population divergence is less pronounced, and maladaptation more pervasive, on steep environmental gradients. abstract: When environmental variation is spatially continuous, dispersing individuals move among nearby sites with similar habitat conditions. But as an environmental gradient becomes steeper, gene flow may connect more divergent habitats, and this is predicted to reduce the slope of the adaptive cline that evolves. We compared quantitative genetic divergence of Rana temporaria frog populations along a 2,000-m elevational gradient in eastern Switzerland (new experimental results) with divergence along a 1,550-km latitudinal gradient in Fennoscandia (previously published results). Both studies found significant countergradient variation in larval development rate (i.e., animals from cold climates developed more rapidly). The cline was weaker with elevation than with latitude. Animals collected on both gradients were genotyped at ∼2,000 singlenucleotide polymorphism markers, revealing that dispersal distance was 30% farther on the latitudinal gradient but 3.9 times greater with respect to environmental conditions on the elevational gradient. A meta-analysis of 19 experimental studies of anuran populations spanning temperature gradients revealed that countergradient variation in larval development, while significant overall, was weaker when measured on steeper gradients. These findings support the prediction that adaptive population divergence is less pronounced, and maladaptation more pervasive, on steep environmental gradients.

A pattern of population structure called isolation-by-environment (IBE) evolves when gene flow connecting populations in different habitats is lower than expected. Although IBE is widespread, there is limited information on its magnitude compared with other factors influencing gene flow. We estimated the relative importance of IBE in the frog Rana temporaria in the Swiss Alps, a geographic context in which IBE should be relatively pronounced. The environmental factor potentially causing IBE was the length of the growing season, which is highly correlated with elevation. A sample of 992 individuals from 82 breeding sites were genotyped at 1827 single-nucleotide polymorphism markers; gene flow was estimated in four ways related to F ST , genetic distance, allele sharing, and distance on a population graph. Gravity modeling and random forest regression evaluated the importance of six at-site covariates, 10 between-site covariates, and geographic distance. There was broad agreement among analysis methods and measures of gene flow: isolation-by-distance (IBD) and habitat quality between sites were of highest importance, the elevation and ruggedness of the dispersal path were about half as important, and IBE was about 10-20% as important as IBD. These results combine with other evidence to suggest that population divergence across elevational gradients is underway in amphibians. K E Y W O R D S : Altitude, connectivity, ddRAD, isolation-by-adaptation, topography. Clearly, the alternative habitats must impose divergent selection on traits that influence gene flow (Widmer et al. 2009; Wang and Bradburd 2014; Butlin and Smadja 2018). Selection imposed by phenological differences between habitats has attracted particular 9 6 2

Sex chromosomes in vertebrates range from highly heteromorphic (as in most birds and mammals) to strictly homomorphic (as in many fishes, amphibians, and nonavian reptiles). Reasons for these contrasted evolutionary trajectories remain unclear, but species such as common frogs with polymorphism in the extent of sex chromosome differentiation may potentially deliver important clues. By investigating 92 common frog populations from a wide range of elevations throughout Switzerland, we show that sex chromosome differentiation strongly correlates with alleles at the candidate sex‐determining gene Dmrt1. Y‐specific Dmrt1 haplotypes cluster into two main haplogroups, YA and YB, with a phylogeographic signal that parallels mtDNA haplotypes: YA populations, with mostly well‐differentiated sex chromosomes, occur primarily south of the main alpine ridge that bisects Switzerland, whereas YB populations, with mostly undifferentiated (proto‐)sex chromosomes, occur north of this ridge. Elevation has only a marginal effect, opposing previous suggestions of a major role for climate on sex chromosome differentiation. The Y‐haplotype effect might result from differences in the penetrance of alleles at the sex‐determining locus (such that sex reversal and ensuing X‐Y recombination are more frequent in YB populations), and/or fixation of an inversion on YA (as supported by the empirical observation that YA haplotypes might not recombine in XYA females).