Spatio-temporal connectivity: assessing the amount of reachable habitat in dynamic landscapes

1. Landscape heterogeneity and habitat connectivity affect species movements, playing an important role in determining the likelihood of species persistence. However, landscape connectivity is usually evaluated using static snap-shots, which do not account for the sequential interactions among habitat patches through time. 2. We developed a network-based model of landscape dynamics, and corresponding connectivity metrics, to account for the reachable habitat across space and time. We illustrate the behaviour of these metrics, using fragmented forested landscapes in the Atlantic Forest of Brazil. We parametrized the models using the dispersal capacities of selected bird and small mammal species. 3. We found that when considering spatio-temporal links, connectivity is estimated to be on average 30% higher (with a maximum of 150% higher) than what is estimated from purely spatial models. This higher degree of spatio-temporal connectivity arises due to connections through temporal stepping-stone patches that appear (habitat gain) and disappear (habitat loss) over time. Species with short dispersal distances (<1000 m) particularly benefited from the spatio-temporal connections. The contribution of spatio-temporal connectivity to habitat reachability increased with higher habitat loss rates. Moreover, it depended on the amount of habitat in the landscape, being higher at intermediate habitat amounts (similar to 30%). 4. We showed that accounting for spatio-temporal connectivity is critical for understanding ecological patterns and processes in dynamic landscapes, and that a series of purely spatial connectivity metrics underestimates the actual connectivity patterns across time. The proposed spatio-temporal connectivity approach and metrics can be applied to evaluate the effective connectivity patterns and trends in a variety of dynamic landscapes, avoiding the potential overestimates of population isolation and extinction probabilities that may result from widely used purely spatial connectivity models.