Computational protocol: Microgeographic morphological variation across larval wood frog populations associated with environment despite gene flow

Similar protocols

Protocol publication

[…] To quantify the extent of gene flow among ponds, I evaluated genetic population structure among the 16 ponds by using a clustering analysis to detect genetic clusters and by calculating pairwise genetic differentiation. Both measures of genetic differentiation were based on data from nine microsatellite loci from approximately 20 individuals per population published in a previous study (Zellmer & Knowles, ).To test for genetic clusters, I used STRUCTURE v 2.3.4 (Pritchard, Stephens, & Donnelly, ). I tested whether there was detectable genetic structure across the 16 ponds and if so whether that structure was related to selection regime (open‐ versus closed‐canopy ponds) or to individual ponds. Three analyses were performed using 1) no prior information, 2) individual ponds as sampling location prior, 3) open‐ and closed‐canopy as a sampling location prior. For each analysis, all other default settings were used with a burnin of 5 × 106 and 1 × 106 iterations to assure alpha converged in each run. For each set of analyses, K was set from 1–17 and was replicated three times for each K. Following the STRUCTURE analyses, I evaluated the data using Structure Harvester (Earl & vonHoldt, ) and CLUMPP (Jakobsson & Rosenberg, ) via CLUMPAK (Kopelman, Mayzel, Jakobsson, & Rosenberg, ). The highest likelihood and the max ΔK value (Evanno, Regnaut, & Goudet, ) in addition to histograms of population assignment values were each used to identify the number of clusters with the best support.Genetic differentiation was calculated as D est (Hedrick, ; Jost, ) using the R “diveRsity” package (Keenan, Mcginnity, Cross, Crozier, & Prodöhl, ), since D est is less susceptible to gene variation resulting in a better estimator of allelic differentiation among populations as compared to G ST values (Jost, ). Populations were considered diverged if the 95% confidence intervals for Jost's D did not overlap 0.To determine whether gene flow is limited by environmental difference, I conducted an Isolation by Environment (IBE) analysis (Wang, ). IBE tests whether gene flow is limited by differences in the environment controlling for geographic distances separating ponds. To test for IBE, I used a multiple matrix regression technique (MMRR; Wang, ). MMRR uses permutation tests to evaluate correlations of multiple predictor matrices (geographic distance and environment) with the response variable matrix (genetic distance). Gene flow was measured as pairwise Jost's D among ponds. Geographic distances were calculated as Haversine distances using the R “geosphere” package (Karney, ). MMRR was conducted in R (Wang, ). […]

Pipeline specifications

Software tools Structure Harvester, CLUMPP, Clumpak
Application Population genetic analysis
Organisms Drosophila melanogaster, Rana sylvatica