The Metapopulation Bridge to Macroevolutionary Speciation Rates: A Conceptual Framework and Empirical Test

Abstract

Whether large-scale variation in lineage diversification rates can be predicted by species properties at the population level is a key unresolved question at the interface between micro- and macroevolution. All else being equal, species with biological attributes that confer metapopulation stability should persist more often at timescales relevant to speciation and so give rise to new (incipient) forms that share these biological traits. Here, we develop a framework for testing the relationship between metapopulation properties related to persistence and phylogenetic speciation rates. We illustrate this conceptual approach by applying it to a long-term dataset on demersal fish communities from the North American continental shelf region. We find that one index of metapopulation persistence has phylogenetic signal, suggesting that traits are connected with range-wide demographic patterns. However, there is no relationship between demographic properties and speciation rate. These findings suggest a decoupling between ecological dynamics at decadal timescales and million-year clade dynamics, raising questions about the extent to which population-level processes observable over ecological timescales can be extrapolated to infer biodiversity dynamics more generally.

Keywords: biodiversity; comparative demography; diversification; emergent traits; extinction biology; macroevolution; metacommunity; microevolution; persistence; range size.

FIGURE 1

Persistence syndromes and their possible relationship to macroevolutionary speciation rate. (A) Four general types of persistence syndromes (i.e., metapopulation dynamic modes, represented by colours grey, black, blue and red). The persistence space maps those syndromes as a combination of two axes: Resilience (x‐axis) and commonness (y‐axis). All else being equal, each persistence syndrome would have its characteristic stable structure (e.g., occupancy is illustrated, but abundance could be an alternative), and its average time to recover the stable structure (given by the turnover rate) after a disturbance that decreases the whole metapopulation to a few cells. This is shown both in the temporal series of lattices and in the plots of occupancy through ecological time. Species located in the commonness‐prone region of the persistence space (e.g., species A.I, blue) persist mainly by being widespread in the geographic space, as their recovery after perturbations is relatively slow when compared to other species. Species located in the resilience‐prone region of the persistence space (e.g., species A.IV, red) would rapidly recover their at‐equilibrium spatial occupancy, although would not necessarily be common. Species may be able to maximise the values of both axes (e.g., species A.II, black), meaning such species would be both geographically widespread and rapid in recovering their equilibrium states. Finally, it is possible that some species could have relatively low values of both persistence axes (e.g., species A.III, grey)—and so are, at least theoretically, less prone to persist in the long term, at least when compared to the other species in the space. If a specific persistence syndrome is what confers a higher speciation rate, as shown in (B), then the phylogenetic patterns of persistence at the metapopulation level track the diversification of the clade, as shown in the hypothetical phylogeny of this group ((C), speciation rates shown at tips). One can visualise the phylogenetic signal of persistence within this framework by painting branches that belong to the same persistence syndrome in similar colours. In this plot, if persistence has phylogenetic signal, closely related lineages will have the same branch colours (as shown in (C)).

FIGURE 2

Geographic variation in metapopulation‐level persistence of ray‐finned fishes. Squares show the spatial positioning of all cells included in this study, coloured by Biogeographic region. Subpanels show examples of region‐specific persistence space (i.e., τ and ε estimates for each species–region pair, before the mixed model was employed). Every point within subpanels refers to a species within a region (species–region pair), and different point clouds amongst subpanels show the region‐specific trends of persistence within a specific sampled biogeographic region. Dashed grey lines in subplots show the median value on both persistence axes across all species–region pairs (see all regions in Figure S23). τ unit is log₂ metapopulation turnover events per cell per year, and ε unit is log‐odds of colonisation.

FIGURE 3

Phylogenetic signal in metapopulation‐level persistence of ray‐finned fishes. Branch colours and tips are coloured based on their position in the persistence space (details in Section S3.3). Clades with similarly coloured branches emphasise the phylogenetic signal of persistence syndromes (like hypothesized in Figure 1C), while state changes on branches (e.g., multiple colours on single branches) reflect a linear interpolation between node states, projected into the 9‐colour persistence space. Consequently, larger or faster colour transitions emphasise branches or clades with more macroevolutionary lability in persistence syndromes. Many of the older branches of the phylogeny are grey, likely because the diffusive nature of the Brownian motion model tends to assign ancestral traits to central parameter values. Labelled nodes and sets of tips grouped by vertical black bars emphasise the species within our dataset that belong to some fish clades with relatively more representation in our dataset (legend: A = Clupeiformes, B = Gadiformes, C = Gadidae, D = Sciaenidae, E = Scorpaeniformes, F = Sebastidae, G = Zoarcidae, H = Pleuronectiformes). All silhouette credits: Nathan Hermann (CC0 1.0).

FIGURE 4

Variation in metapopulation‐level persistence. (A) shows the proportion of variance explained by regional and taxonomic factors in the Markov model estimates that only include multi‐region species (34 species, 79 region‐species pairs), as well as factors related to biogeographic regions, phylogenetic components (here represented by taxonomic levels) and sampling issues (see Section S2 for explanations on h_rel), as well as residual variation. (B) shows the empirical persistence space for all species of fishes, with the y‐axis showing ε_raw. (C–H) show different fish families as filled points (colour‐coded as in B), their estimated ‘phylo‐persistence‐space’, while the rest of the species are not shown. (B–H) Points representing species are colour‐coded by quadrant occupied by that species in the full space. Note black lines connecting edges in (C–H) show within‐family phylogeny topology. In this phylomorphospace projection, phylogenetic signal is shown by fewer crosses amongst distant branches, together with slow diffusion in this space as the phylogeny bifurcates (note the difference between (C–E) and (F–H)). τ unit is log₂ metapopulation turnover events per cell per year, and ε_raw unit is the proportion of turnover due to colonisation. All silhouette credits: Nathan Hermann (CC0 1.0).

FIGURE 5

Relationship between phylogenetic speciation rate and ray‐finned fish persistence syndromes. (A) Phylogenetic distribution of speciation rates across the ray‐finned fish phylogeny pruned to the 189 species in our empirical dataset. Tip colours denote log‐transformed tip speciation rates (λ_DR). Note that focal set of taxa spans many evolutionarily‐independent shifts in speciation rate and in persistence syndromes (for the later, see also Figure 3). Labelled nodes and sets of tips grouped by vertical black bars emphasise the species within our dataset that belong to some fish clades with relatively more representation in our dataset (legend: A = Clupeiformes, B = Gadiformes, C = Gadidae, D = Sciaenidae, E = Scorpaeniformes, F = Sebastidae, G = Zoarcidae, H = Pleuronectiformes—same clades as Figure 3). (B) Tip speciation rates are expressed as residual deviations from the dataset median; colours denote the corresponding persistence syndrome of each tip. Note that persistence syndromes appear randomly scattered across both fast‐ and slow‐speciating lineages. (C) Persistence space (i.e., same as Figure 2B), with species (points) colour‐coded according to their values of λ_DR. (D) persistence space, but only the 20% slower speciating lineages are colour‐coded by their phylogenetic speciation rate, and other lineages are shown as grey points. (E) same as panel D, but in it only the 20% faster speciating lineages are colour‐coded. If speciation rate was strongly correlated with persistence, same‐coloured points would appear only in specific quadrants in panels C, D or E (see also Figure 1B). Speciation rate unit is log₂ events per lineage per million years. τ unit is log₂ metapopulation turnover events per cell per year, and ε_raw unit is the proportion of turnover due to colonisation. Figure S25 shows scatterplots between each individual persistence component and all speciation rate estimates (λ_DR and λ_BAMM).