The allometry of movement predicts the connectivity of communities

Abstract

Connectivity has long played a central role in ecological and evolutionary theory and is increasingly emphasized for conserving biodiversity. Nonetheless, connectivity assessments often focus on individual species even though understanding and preserving connectivity for entire communities is urgently needed. Here we derive and test a framework that harnesses the well-known allometric scaling of animal movement to predict community-level connectivity across protected area networks. We used a field translocation experiment involving 39 species of southern African birds to quantify movement capacity, scaled this relationship to realized dispersal distances determined from ring-and-recovery banding data, and used allometric scaling equations to quantify community-level connectivity based on multilayer network theory. The translocation experiment explained observed dispersal distances from ring-recovery data and emphasized allometric scaling of dispersal based on morphology. Our community-level networks predicted that larger-bodied species had a relatively high potential for connectivity, while small-bodied species had lower connectivity. These community networks explained substantial variation in observed bird diversity across protected areas. Our results highlight that harnessing allometric scaling can be an effective way of determining large-scale community connectivity. We argue that this trait-based framework founded on allometric scaling provides a means to predict connectivity for entire communities, which can foster empirical tests of community theory and contribute to biodiversity conservation strategies aimed at mitigating the effects of environmental change.

Keywords: birds; dispersal; landscape; network; translocation experiment.

Fig. 1.

A framework for the connectivity of communities. (A) Species-specific dispersal kernels can be summarized based on (B) the mean distances moved (or other summary statistics) and allometric scaling equations based on morphology and/or related traits. With allometric scaling, (C) predictions for movement and dispersal can be made for all species in a regional community pool, creating a multilayer network. (D) This multilayer network can then be leveraged in a variety of ways to interpret connectivity based on network theory. In this example, each species is a different color. For D, shown are potential linkages across a hypothetical network for all species in the community.

Fig. 2.

The allometry of movement capacity and its scaling with dispersal. (A) Movement distance during translocation experiments is predicted by wing length across a savanna bird community. Shown are the predicted relationships and associated 95% CI for movement parameters as a function of wing length. Dot colors represent species (n = 39). (B) Flight distances from the experiment explain mean dispersal distances from ringing data. Shown is the fit of a nonlinear model to 13 species for which at least five records of dispersal distances were available from the SAFRING database.

Fig. 3.

Scaling movement to patch and landscape connectivity for bird communities across protected areas in the lowveld of Eswatini. (A) Predicted mean dispersal distances (1/α_k) for 39 species used in translocation experiments. (B) The relationship of wing length to patch connectivity for each protected area i and species k. (C) Variation in the probability of connectivity and metapopulation capacity, two landscape metrics of connectivity that reflect reachable habitat and metapopulation viability, across species. (D) Community-level protected area connectivity, C, where darker shaded protected areas (polygons) are more connected at the community level. Transparent polygons are protected areas not considered. Lines represent potential linkages based on the mean dispersal distances across species. Gray shading represents elevation (light gray indicates low elevation, lowveld). Map projected in Universal Transverse Mercator (UTM) Zone 36S.

Fig. 4.

Network connectivity of communities derived from allometric scaling predicts bird communities across protected areas in Eswatini. (A) The predicted probability of occurrence as a function of species-specific patch connectivity, C_ik. Solid lines represent species-specific predictions, with color gradient reflecting wing length. Dotted line is the average prediction, with dark and light shaded regions representing uncertainty (80 and 95% CIs, respectively) in average predictions. (B) Estimated species richness (dot indicates mean; error bars 80%, 95% CI) of birds across eight protected areas as a function of community-level connectivity, C_i. r and β are based on posterior distributions of the log-log relationship between species richness and community connectivity.