Geographic isolation and larval dispersal shape seascape genetic patterns differently according to spatial scale

Abstract

Genetic variation, as a basis of evolutionary change, allows species to adapt and persist in different climates and environments. Yet, a comprehensive assessment of the drivers of genetic variation at different spatial scales is still missing in marine ecosystems. Here, we investigated the influence of environment, geographic isolation, and larval dispersal on the variation in allele frequencies, using an extensive spatial sampling (47 locations) of the striped red mullet (Mullus surmuletus) in the Mediterranean Sea. Univariate multiple regressions were used to test the influence of environment (salinity and temperature), geographic isolation, and larval dispersal on single nucleotide polymorphism (SNP) allele frequencies. We used Moran's eigenvector maps (db‐MEMs) and asymmetric eigenvector maps (AEMs) to decompose geographic and dispersal distances in predictors representing different spatial scales. We found that salinity and temperature had only a weak effect on the variation in allele frequencies. Our results revealed the predominance of geographic isolation to explain variation in allele frequencies at large spatial scale (>1,000 km), while larval dispersal was the major predictor at smaller spatial scale (<1,000 km). Our findings stress the importance of including spatial scales to understand the drivers of spatial genetic variation. We suggest that larval dispersal allows to maintain gene flows at small to intermediate scale, while at broad scale, genetic variation may be mostly shaped by adult mobility, demographic history, or multigenerational stepping‐stone dispersal. These findings bring out important spatial scale considerations to account for in the design of a protected area network that would efficiently enhance protection and persistence capacity of marine species.

Asymmetric dispersal is a critical element of concordance between biophysical dispersal models and spatial genetic structure in Great Barrier Reef corals

“…Relatively few studies have evaluated the effect of asymmetric larval dispersal on spatial genetic patterns in marine species (Riginos et al, ). Perhaps not surprisingly, these studies, like our results here with A. tenuis and A. millepora, consistently uncover substantial asymmetries in inferred dispersal (Benestan et al, ; Dalongeville et al, ; Xuereb et al, ). Clearly, conventional analyses, especially those based on summary statistics like F ST where all directional information is obscured, will fail to uncover important elements regarding relationships among populations (Kool et al, ; Riginos et al, ).…”

“…Geographic surveys of intraspecific genetic variation can provide important insights regarding dispersal (Hellberg, ; Riginos & Liggins, ; Selkoe, D'Aloia, et al, ) and indeed several studies have considered predictions arising from biophysical dispersal models alongside observed spatial genetic patterns, typically for coastal marine taxa. Although some individual studies report correlations (Benestan et al, ; Dalongeville et al, ; Foster et al, ; Galindo, Olson, & Palumbi, ; Schunter et al, ; Thomas et al, ; Truelove et al, ; White et al, ; Xuereb et al, ), a recent review of the field found only moderate to low concordance between biophysical predictions and empirical genetic patterns (Selkoe, Scribner, & Galindo, ).…”

“…Comparing all sampled populations holistically, Crandall, Treml, and Barber () constructed a series of coalescent models of gene flow and contrasted the likelihood of observed DNA haplotype distributions from the biophysically informed gene flow models against various other null geographic models; they demonstrated greater likelihood of biophysically informed gene flow models across three species of nerite snails in the South Pacific. Recently, three studies turned to asymmetric eigenvector mapping (AEM: Blanchet, Legendre, Maranger, Monti, & Pepin, ) where asymmetric processes (such as biophysical migration probabilities) are statistically modelled as spatial autocorrelation structures; inclusion of AEMs substantially improved predictions of spatial genetic structure for American lobster (Benestan et al, ), California sea cucumbers (Xuereb et al, ), and Mediterranean striped red mullet (Dalongeville et al, ). Thus, these first few studies that quantitatively incorporate asymmetric biophysical predictions suggest that directions of larval dispersal are important elements of marine population connectivity.…”

Host plant associations and geography interact to shape diversification in a specialist insect herbivore

“…Spatial predictor variables were extracted as dbMEM variables following Borcard and Legendre (). Dalongeville et al () provides an example detailing the process. To construct dbMEM variables, first a pairwise geographic distance matrix was created, using haversine distances between all pairs of the 58 sample locations, with the package geosphere version 1.5‐5 in r (Hijmans, ).…”

“…We apply a method proposed by Legendre, Fortin, and Borcard (2015) to SNP data obtained by genotyping-by-sequencing to isolate signals of spatial autocorrelation from HAD. The method uses distancebased Moran's eigenvector mapping (dbMEM), formally known as principal coordinates of neighbour matrices (Borcard & Legendre, 2002;Borcard, Legendre, Avois-Jacquet, & Tuomisto, 2004;Dray, Legendre, & Peres-Neto, 2006), to account for spatial autocorrelation associated with the distribution of sample sites and ultimately construct regression variables summarizing spatial structure (Dalongeville et al, 2018). These explanatory spatial variables then are used in canonical analysis, such as redundancy analysis (RDA), along with other environmental predictors to identify the proportion of variation due to spatial and environmental factors (Borcard et al, 2004).…”