Contrasting impacts of competition on ecological and social trait evolution in songbirds

Abstract

Competition between closely related species has long been viewed as a powerful selective force that drives trait diversification, thereby generating phenotypic diversity over macroevolutionary timescales. However, although the impact of interspecific competition has been documented in a handful of iconic insular radiations, most previous studies have focused on traits involved in resource use, and few have examined the role of competition across large, continental radiations. Thus, the extent to which broad-scale patterns of phenotypic diversity are shaped by competition remain largely unclear, particularly for social traits. Here, we estimate the effect of competition between interacting lineages by applying new phylogenetic models that account for such interactions to an exceptionally complete dataset of resource-use traits and social signaling traits for the entire radiation of tanagers (Aves, Thraupidae), the largest family of songbirds. We find that interspecific competition strongly influences the evolution of traits involved in resource use, with a weaker effect on plumage signals, and very little effect on song. Our results provide compelling evidence that interspecific exploitative competition contributes to ecological trait diversification among coexisting species, even in a large continental radiation. In comparison, signal traits mediating mate choice and social competition seem to diversify under different evolutionary models, including rapid diversification in the allopatric stage of speciation.

Fig 1. Biogeography of tanagers (Thraupidae).

(A), Phylogeny of study species, with colors at tips denoting geographical distributions corresponding to the areas shown in (B). (B), Map of the regions used for ancestral range estimation. Key to regions as defined by ref. [37]: a, Gulf-Caribbean slope, Pacific arid slope, and Chocó lowlands; b, Chiriquí-Darién highlands; c, subtropical Pacific; d, northern Andes; e, central and southern Andes; f, central South America and Pampas; g, Atlantic forest; h, Patagonia; i, Amazonia (north and south) and northern South America; j, Tepuis.

Fig 2. Plotted is the relative support for the best model that incorporates interspecific competition (MC, DD_lin, or DD_exp) when compared to the best model that does not incorporate interspecific competition (BM or OU) ± the standard deviation across fits to the sample of stochastic maps for each trait (see S4 Data).

Filled points represent model fit cases in which the mean relative support for a model with competition is higher than for a model excluding competition (i.e., relative support > 0.5). Species were grouped by diet (A-D), habitat (E-G), and year-round territoriality (H). Trait numbers (1–27) are defined in S1 Table. BM, Brownian motion; DD_exp, exponential diversity-dependent model; DD_lin, linear diversity-dependent model; MC, matching competition; OU, Ornstein-Uhlenbeck.

Fig 3

Trait disparity through time between allopatric and sympatric lineages for the following four representative scenarios: (A) Beak pPC3 for year-round territorial taxa (trait number 4, S1 Table), (B) whole song frequency pPC2 for year-round territorial taxa (trait number 22, S1 Table), (C) ln(mass) for frugivores (trait number 1), and (D) male coloration pPC2 for frugivores (trait number 17, S1 Table). The best-supported model is indicated. Each panel depicts the mean disparity, calculated as the average squared Euclidean distance, between allopatric lineages and between sympatric lineages through time in 100 datasets simulated with the MLE parameter values for the best-fit model (left) and the distribution of mean sympatric and allopatric disparity values at the tips (right). Red lines depict the mean disparity values in the empirical datasets (see S5 Data). BM, Brownian motion; DD_exp, exponential diversity-dependent model; ln(mass), log-transformed mass; MC, matching competition; MLE, maximum likelihood estimate; OU, Ornstein-Uhlenbeck; pPC2, phylogenetic principal component 2; pPC3, phylogenetic principal component 3.

Fig 4. The tempo of ecological and social trait evolution.

(A) Mean MLE of the σ² (evolutionary rate) parameter from the BM model fit to standardized trait values show that song traits evolve more rapidly than resource-use or plumage traits in either sex. Plotted are mean ± SD of mean values for each class of trait from fits across 100 posterior trees. (B) Mean MDI estimates similarly indicate that song traits accumulate within-clade disparity more rapidly than resource-use or plumage traits in either sex. Plotted are mean ± SD of mean values for each class of trait calculated across 100 posterior trees (see S6 Data). BM, Brownian motion; MDI, morphological disparity index; ML, maximum likelihood.