Slow natal dispersal across a homogeneous landscape suggests the use of mixed movement behaviours during dispersal in the Darwin's frog

Dispersal is a key process in ecology and evolution. Both theoretical and empirical evidence in actively dispersing organisms support the general notion that the use of nearly straight dispersal paths is a highly efficient way to maximize dispersal success in heterogenous landscapes. In homogeneous landscapes, in contrast, the benefits of a straighter dispersal path could be outweighed by an increase in risk costs, favouring the evolution of tortuous dispersal paths resulting in a relatively slow dispersal. Empirical support for this theoretical prediction, however, has remained elusive. To explore this theoretical prediction, we studied the movement behaviour of the southern Darwin's frog, Rhinoderma darwinii, a fully terrestrial amphibian inhabiting a highly homogeneous environment (i.e. South American temperate forest). Using spatial capture–recapture data collected over a 4-year period in wild populations, in combination with statistical and simulation modelling, we found evidence of a slow natal dispersal lasting one year or more. In contrast, adults exhibited high site fidelity, having a median annual displacement of 3.64 m. A correlated random walk model produced synthetic distributions of juvenile annual displacement that were nearly identical to the empirical data, suggesting that a plausible explanation of juvenile dispersal is the use of routine movements (with high path tortuosity) over short temporal scales (<3 months) integrated over the year along a relatively straight dispersal path. We predict that for species living in homogenous landscapes, this behaviour likely reduces many of the costs associated with the transient stage of dispersal. Specifically, periods of routine movements might reduce risk costs (e.g. dying due to starvation or predation), while the integration of these periods along a straight line maximizes dispersal distance while minimizing energetic costs.