A Resistant‐Kernel Model of Connectivity for Amphibians that Breed in Vernal Pools

Abstract

Pool-breeding amphibian populations operate at multiple scales, from the individual pool to surrounding upland habitat to clusters of pools. When metapopulation dynamics play a role in long-term viability, conservation efforts limited to the protection of individual pools or even pools with associated upland habitat may be ineffective over the long term if connectivity among pools is not maintained. Connectivity becomes especially important and difficult to assess in regions where suburban sprawl is rapidly increasing land development, road density, and traffic rates. We developed a model of connectivity among vernal pools for the four ambystomatid salamanders that occur in Massachusetts and applied it to the nearly 30,000 potential ephemeral wetlands across the state. The model was based on a modification of the kernel estimator (a density estimator commonly used in home range studies) that takes landscape resistance into account. The model was parameterized with empirical migration distances for spotted salamanders (Ambystoma maculatum), dispersal distances for marbled salamanders (A. opacum), and expert-derived estimates of landscape resistance. The model ranked vernal pools in Massachusetts by local, neighborhood, and regional connectivity and by an integrated measure of connectivity, both statewide and within ecoregions. The most functionally connected pool complexes occurred in southeastern and northeastern Massachusetts, areas with rapidly increasing suburban development. In a sensitivity analysis estimates of pool connectivity were relatively insensitive to uncertainty in parameter estimates, especially at the local and neighborhood scales. Our connectivity model could be used to prioritize conservation efforts for vernal-pool amphibian populations at broader scales than traditional pool-based approaches.

Potential range of impact of an ecological trap network: the case of timber stacks and the Rosalia longicorn

“…However, the insufficient amount of field data on Rosalia longicorn dispersal behaviour, mentioned earlier, is the real problem in reliable modelling of this process. Theoretically, most appropriate diffusion models of dispersal demand precise information on dispersal distances as well as the isolation power of different habitat types (Compton et al 2007;Ovaskainen 2004Ovaskainen , 2008. Such data with sufficient accuracy are available only for a few flying beetle species studied with radio-tracking techniques (Dubois and Vignon 2008;Tini et al 2018).…”

“…In the least-cost patch approach, the increase in migratory cost is treated as an increase in isolation between subpopulations or habitat patches (Cushman et al 2009). We recommend, therefore, that when the spatial structure of Rosalia longicorn habitat is studied, the distance from the nearest ecological trap should be included in analyses based on the resistant kernel approach (Compton et al 2007). However, the insufficient amount of field data on Rosalia longicorn dispersal behaviour, mentioned earlier, is the real problem in reliable modelling of this process.…”

A landscape index of ecological integrity to inform landscape conservation

“…For example, our watershed-based metrics which evaluate the integrity of aquatic systems use a watershed kernel that honors how terrain and land cover affect the movement of water and water-born pollutants to a site, which is clearly more appropriate than treating all locations in the watershed the same. Similarly, our connectedness metric uses a resistant kernel (Compton et al 2007) to represent how organisms 515 and ecological processes move across the landscape in response to environmental resistance (Zeller et al 2012). We are unaware of other approaches that adopt these specific kinds of kernel estimators to evaluate ecological integrity, although our traversability metric (which is a version of connectedness), is used as a component of The Nature Conservancy's (TNC) terrestrial resilience (Anderson and Ferree 2010).…”

“…Introduced by Compton et al (2007), the resistant kernel is a hybrid between two existing approaches: the standard kernel estimator as described above and least-cost paths 175 based on resistant surfaces. Resistant surfaces (also referred to as cost surfaces) are being increasingly used in landscape ecology to model ecological flows in heterogeneous landscapes (Zeller et al 2012).…”

Climate‐smart management of biodiversity

“…Locations with high local landscape resistance will constrain the ability of species (especially those that are encumbered by dispersal barriers) to track suitable climates. We estimated landscape resistance using an existing map with a 90-m resolution (Anderson et al 2012) that was produced using a resistance-kernel estimator (Compton et al 2007) with a 3-km kernel (Fig. 1d).…”