Fig 1.
Social systems include non-family-living species (55% in our data set, e.g., the blue tit Parus caeruleus [a]), in which parent–offspring associations do not extend beyond nutritional independence and individuals that do not engage in cooperative breeding; family-living species (31% in our data set, e.g., the Siberian jay Perisoreus infaustus [b]; see also S1 Fig), in which offspring remain with their parents beyond nutritional independence but do not aid in the rearing of future broods; and cooperative breeding species (13% in our data set, e.g., the apostlebird Struthidea cinerea [c]), in which offspring remain with their parents beyond nutritional independence and help them in subsequent breeding attempts or engage in redirected helping at nests of relatives. In a small number of species (1% in our data set), e.g., in the guira cuckoo Guira (d), cooperative breeding primarily involves nonrelatives (“non-kin cooperatively breeding species”). (a) Image credit: Per Harald Olsen/NTNU. (b) Image credit: Michael Griesser. (c) Image credit: Michael Griesser. (d) Image credit: Beatrice Murch.
Fig 2.
Ancestral state reconstruction (based on maximum likelihood) and estimated evolutionary transitions of bird social system (N = 2,968 species).
Pie charts plotted at each node represent the estimated posterior proportion of the 3 social systems: non-family living (green), family living (orange), and cooperative breeding families (blue).
Fig 3.
Estimated transition rates of the best-fitting model (a) and statistical evaluation of the different transition models of the evolution of avian social systems (b). In the best-fitting transition model, arrow thickness is proportional to the estimated transition rates, and the size of the circles is proportional to the relative abundance of the 3 social systems among the species in the sample. No Fam = non-family living; Fam = family living; Coop = cooperative breeding families. Directions of the arrows indicate modelled transitions: a single arrow between 2 states pointing in both directions reflects transition rates constrained to be equal, a single arrow pointing in 1 direction reflects transitions only in 1 direction, and 2 arrows between states reflects unconstrained transition rates. AIC = Akaika information criterion.
Fig 4.
Ecoclimatic and life-history correlates of non-family-living (green dotted line), family-living (orange solid line), and cooperative breeding species (blue dashed line); N = 2,968 bird species (excluding cooperative breeding species with non-kin helpers only).
Lines reflect the predicted probabilities of occurrence of respective social systems estimated from phylogenetically informed multinomial models (see Table 1). Family-living and cooperative breeding species are associated with locations that have abundant but variable precipitation (PC1), a longer mean growing season (PC2), and a higher among-year variance in net primary productivity (NPP) during the growing season (PC5). Moreover, these species live in denser habitats (PC7) and have a larger body size (PC8). Cooperative breeding species are associated with higher within-year variance in NPP (PC3). MGS, mean growing season.
Fig 5.
Global abundance of non-family-living, family-living, and cooperative breeding species in birds (number of species per 0.5° x 0.5°) and global patterns of the 3 most influential ecoclimatic parameters (duration of the mean growing season [MGS]; included in PC2), annual variance in precipitation (square-root transformed; included in PC1), and within-year variance in net primary productivity (NPP; included in PC3).
Figures were plotted using the letsR package [34]. Abbreviations: sqrt, square-root transformed.
Table 1.
Multinomial phylogenetically controlled generalized linear mixed models comparing the effect of ecoclimatic and life-history variables on the evolution of non-family-living, family-living, and cooperative breeding species; N = 2,968 bird species (excluding cooperative breeding species with non-kin helpers only).
Coefficients reflect the results of multinomial phylogenetic regression models with “cooperative families” as the reference category in the analyses and thus not shown per se. Significant factors are highlighted in bold. The principal component analyses (PCAs) resulting in PC1–8 are shown in S2 Table. The factor social system assessment specified whether it was assessed based on the time offspring remained with their parents beyond independence (using 50 days as a threshold to differentiate between non-family-living and family-living species; see [17]), breeding behavior, or social information. MCMC = Markov chain Monte Carlo.