The challenge of assaying landscape connectivity in a changing world: A 27-year case study in the southern Great Plains (USA) playa network.

Many habitat resources fluctuate in availability due to natural environmental variability and anthropogenic influences. These fluctuations pose challenges to organisms attempting to move from one habitat patch to another, and also pose challenges to detecting and managing factors impacting landscape connectivity. Our understanding of these relationships is further hampered by lack of precedence on how to quantify dynamic connectivity. The ephemeral freshwater wetlands of the southern Great Plains of the USA (playas) form a dynamic habitat network that serves as a case study of these challenges and allows us to propose a suite of connectivity metrics to monitor changes in network topology and evaluate the management importance of individual wetlands. We used satellite imagery to examine inundation patterns of >7000 playas in a 29,083 km 2 portion of Texas on 80 dates from 1984 to 2011. Based on historic locations of playa basins, approximately 85% of playas (particularly those ≤10 ha) have lost the capacity to hold water even during regionally wet times, resulting in a ∼69% reduction of surface water area. These losses were associated with proximity to cropland, with total cropland acreage increasing by 0.07–17.34% of county land area during our focal time span in 10 of the 14 counties in our study area. We examined connectivity at wetland and whole-network scales to determine effects of playa losses on network topology and thus on connectivity. We evaluated 11 metrics for this purpose, which quantified the number of wet playas present on each date, their degree of connectedness, their clustering, path redundancy within the network, overall network topology, the importance of individual playas in various roles, and the size of a single playa that would provide equivalent connectivity (amount of reachable habitat) as in the actual network. Topology has thinned over the past three decades with playa losses, reflected in increasing graph density, average path length, degree of connectivity for highly linked hubs, and average number of cutpoints. Similarly, graph diameter is currently less than half of the historic potential maximum, and the equivalent connected area has declined by over 23% since 1984 (and by over 82% relative to historic values). These patterns suggest that path redundancy through the network has declined such that dispersers currently have fewer connectivity options compared to a few decades ago. Relatively high transitivity scores indicate that the playa network is still populated with a large (but diminishing) number of wetlands, and the dwindling surface water present in the remaining playas is not compensating for playa losses over time. Average coalescence distances are currently higher than the dispersal capacity of many organisms, meaning that the playa network is fragmented such that only an extremely vagile disperser (capable of moving at least 18–45 km) would be able to traverse the landscape via the remaining wetlands, even if all were wet simultaneously. These findings illustrate the importance of using multiple indicators in assaying dynamic connectivity and provides a framework of possible metrics to use for monitoring and assessment of any dynamic habitat network. [ABSTRACT FROM AUTHOR]