Ten years of demographic modelling of divergence and speciation in the sea

Abstract

Understanding population divergence that eventually leads to speciation is essential for evolutionary biology. High species diversity in the sea was regarded as a paradox when strict allopatry was considered necessary for most speciation events because geographical barriers seemed largely absent in the sea, and many marine species have high dispersal capacities. Combining genome-wide data with demographic modelling to infer the demographic history of divergence has introduced new ways to address this classical issue. These models assume an ancestral population that splits into two subpopulations diverging according to different scenarios that allow tests for periods of gene flow. Models can also test for heterogeneities in population sizes and migration rates along the genome to account, respectively, for background selection and selection against introgressed ancestry. To investigate how barriers to gene flow arise in the sea, we compiled studies modelling the demographic history of divergence in marine organisms and extracted preferred demographic scenarios together with estimates of demographic parameters. These studies show that geographical barriers to gene flow do exist in the sea but that divergence can also occur without strict isolation. Heterogeneity of gene flow was detected in most population pairs suggesting the predominance of semipermeable barriers during divergence. We found a weak positive relationship between the fraction of the genome experiencing reduced gene flow and levels of genome-wide differentiation. Furthermore, we found that the upper bound of the 'grey zone of speciation' for our dataset extended beyond that found before, implying that gene flow between diverging taxa is possible at higher levels of divergence than previously thought. Finally, we list recommendations for further strengthening the use of demographic modelling in speciation research. These include a more balanced representation of taxa, more consistent and comprehensive modelling, clear reporting of results and simulation studies to rule out nonbiological explanations for general results.

Keywords: demographic inference; grey zone of speciation; heterogeneous gene flow; marine speciation; primary and secondary contact; reproductive isolation.

FIGURE 1

An overview of demographic modelling in the sea. (a) Increase in the number of publications (including hetM, and sometimes also hetN) in recent years (n = 30). (b) Distribution of the demographic methods used across publications (n = 30)

FIGURE 2

Preferred demographic scenarios in the sea. Distribution of the preferred demographic scenarios across all lineage pairs. Colours correspond to: (a) taxonomic phylum (n = 116), (b) hetM preference (n = 114), (c) hetN preference (n = 116)

FIGURE 3

Preferred demographic models in the sea and genome‐wide heterogeneity in effective size. Distribution of the preferred demographic scenarios across all lineage pairs in studies where hetN was tested (n = 65, b) or not (n = 49, a)

FIGURE 4

Distribution of (a) F _ST (n = 108) and (b) p (n = 62) across the preferred demographic models. For (b), only lineage pairs for which hetM was preferred are considered. p is the fraction of the genome with reduced gene flow

FIGURE 5

Correlation between F _ST and p (n = 56). p is the fraction of the genome with reduced gene flow. Colours correspond to: (a) preferred demographic scenarios, (b) taxonomic phylum, (c) categories of reduced gene flow based on quantiles of the ‘m _e/m’ distribution, (d) categories of propagule sizes (i.e. the size of the dispersal stage individuals) based on the quantiles of the size distribution, (e) categories of adult mass based on the quantiles of the mass distribution and (f) geographical context of the populations studied. For the (c) panel: High reduction if the value is below the 25% quantile (0.021), low reduction if the value is above the 75% quantile (0.19) and medium otherwise. For (d) and (e) panels: Small if the value is below the 25% quantile (0.019 cm, 5.2 g), big if the value is above the 75% quantile (3.35 cm, 5 kg) and medium otherwise. In all panels, the x axis is in log scale

FIGURE 6

Preferred demographic models along the speciation continuum in the sea. Distribution of the preferred demographic scenarios across all lineage pairs. Colours correspond to three categories of divergence, defined in Roux et al. (2016), either from Da values (a: N = 65, c: N = 56) or F _ST values (b: N = 110, d: N = 56): (i) populations: Da below 0.5% or F _ST below 0.19; (ii) species: Da above 2% or F _ST above 0.56; (iii) grey zone: Values in‐between. Panels (a) and (b) refer to the dataset we analysed (Table S1), while panels (c) and (d) refer to Roux et al. (2016)