Using neutral, selected, and hitchhiker loci to assess connectivity of marine populations in the genomic era

Abstract

Estimating the rate of exchange of individuals among populations is a central concern to evolutionary ecology and its applications to conservation and management. For instance, the efficiency of protected areas in sustaining locally endangered populations and ecosystems depends on reserve network connectivity. The population genetics theory offers a powerful framework for estimating dispersal distances and migration rates from molecular data. In the marine realm, however, decades of molecular studies have met limited success in inferring genetic connectivity, due to the frequent lack of spatial genetic structure in species exhibiting high fecundity and dispersal capabilities. This is especially true within biogeographic regions bounded by well-known hotspots of genetic differentiation. Here, we provide an overview of the current methods for estimating genetic connectivity using molecular markers and propose several directions for improving existing approaches using large population genomic datasets. We highlight several issues that limit the effectiveness of methods based on neutral markers when there is virtually no genetic differentiation among samples. We then focus on alternative methods based on markers influenced by selection. Although some of these methodologies are still underexplored, our aim was to stimulate new research to test how broadly they are applicable to nonmodel marine species. We argue that the increased ability to apply the concepts of cline analyses will improve dispersal inferences across physical and ecological barriers that reduce connectivity locally. We finally present how neutral markers hitchhiking with selected loci can also provide information about connectivity patterns within apparently well-mixed biogeographic regions. We contend that one of the most promising applications of population genomics is the use of outlier loci to delineate relevant conservation units and related eco-geographic features across which connectivity can be measured.

Keywords: connectivity; gene flow; marine conservation; population genomics; population structure.

Figure 1

The demographic parameters values behind weak F_ST values. (A) Because of the nonlinear relationship between F_ST and N_em in the island model, genetic differentiation (F_ST, in color scale) rapidly shrinks as the per-generation number of migrants (N_em) increases. (B) At equilibrium, weak to null genetic differentiation is expected for small (N_e = 10³) and highly connected (m = 10⁻¹) populations, but also for large (N_e = 10⁷) and demographically independent (m = 10⁻⁵) populations.

Figure 2

The chromosomal signature of local selection acting on a de novo mutation in panmixia. We consider a two habitats Levene's model (Levene 1953) represented in the left box, with random mating (in the dotted circle) and random dispersal (arrows) across two habitats of equal size (rectangles). A new selected mutation (allele a, red star) appears in habitat 1 on a haplotype bearing rare neutral variants (in green) at variable recombination distances (the initial frequency is 1/2N_e). The selected mutation has symmetrical antagonistic effects on the fitness of genotypes with respect to habitat (Habitat 1: ω_AA/ω_Aa = 0.5, ω_aa/ω_Aa = 2; Habitat 2: ω_AA/ω_Aa = 2, ω_aa/ω_Aa = 0.5). At equilibrium, varying selection among genotypes and habitats results in differentiation between habitats at the selected locus (in this example Δp ≈ 0.3). During the progress toward equilibrium, neutral variants hitchhike with the selected allele, transiently producing a narrow chromosomal region where genetic differentiation is increased around the selected locus (green line). As the selected allele progressively recombines away from its haplotypic background, differentiation at neutral alleles rapidly vanishes (green arrows). After a few thousands of generations, differentiation is almost limited to the selected locus (dashed green line).

Figure 3

Using the inflow of foreign alleles to reveal within-species connectivity patterns. At generation zero, two partially reproductively isolated species meet on a linear stepping-stone model between demes 10 and 11 and start to exchange genes. The auto-recruitment rate is 1 − m, and migration to adjacent demes is m/2 (with m = 0.5). A weak barrier to gene flow (m = 0.01) was set between demes 20 and 21, in the middle of the range of the species localized on the right side. Strong selection (s = 0.5) acts against heterozygote genotypes at a reproductive isolation locus, which is linked to neutral markers located at variable recombination distances (from closely linked to unlinked). A recombination rate of 1 cM per Mb was used to convert genetic into physical distances. (A) The step size, calculated as the difference in allele frequency between demes 20 and 21 (Δp), as a function of the number of generations postcontact. (B) Spatial allele frequency patterns after 10 000 generations of introgression showing the frequency step between demes 20 and 21. (C) The step size between demes 20 and 21 as a function of the physical distance to the reproductive isolation locus. (D) The step size between demes 20 and 21 as a function of the difference in allele frequency between species.

Figure 4

Genomic islands of differentiation and the information therein. (A) A genomic island of differentiation between Atlantic and Mediterranean sea bass lineages (Dicentrarchus labrax) on chromosome 7 (RAD-sequencing data from Tine et al. 2014). (B) Geographic clines between two partially reproductively isolated species of sole, Solea senegalensis (Sp. 1, left side) and Solea aegyptiaca (Sp. 2, right side) assessed by RAD-Sequencing (A. Souissi, P.-A. Gagnaire, L. Bahri-Sfar, F. Bonhomme, unpublished). Red and orange clines correspond to expectations near reproductive isolation loci (i.e. at the center of a genomic island, where there is no introgression), for a diagnostic locus (red) and a locus only polymorphic in S. senegalensis (orange) due to incomplete lineage sorting. The green cline shows a gradient (or a tail) of introgression due to the inflow of S. aegyptiaca alleles in the S. senegalensis background. At this locus, the shared allele is a consequence of secondary introgression instead of incomplete lineage sorting. Such gradients of introgression are expected to be found at loci showing intermediate degrees of linkage with reproductive isolation loci (i.e. located in the periphery of a genomic island, where introgression is reduced but not zero). Introgression tails may be used to reveal cryptic genetic structure where freely recombining neutral loci remain uninformative (black dashed line), as it is the case in S. senegalensis.