Wade carefully: water parametrization in structural landscape connectivity modeling

Structural connectivity models provide critical insights on how movement potential varies across landscapes and can support decisions on where to prioritize land protection and restoration. Water is a common component of landscape models used to inform conservation strategies, but the impacts of water parameterization on model behavior are poorly understood. We assessed how various parameterization approaches for water bodies, such as rivers, lakes, and shorelines, in structural, omnidirectional circuit-theory connectivity models affect patterns in the primary output, cumulative current flow values, in terrestrial grid cells. We ran two structural modeling approaches with ten simulated and six real-world landscapes using three methods for assigning the resistance of water relative to terrestrial resistance values: low, high, or neutral values drawn randomly from a distribution informed by the terrestrial resistance grid cells. We then compared the cumulative current outputs from each combination. The influence of how water was parameterized was consistent across all landscapes. When resistance values for water grid cells were low, the mean values of cumulative current on land were lower, indicating a dampening effect. When water resistances were high, the means and standard deviations of current flow on land were higher, with very high values (a pattern often used to indicate key barriers related to habitat fragmentation and loss) often observed near water. Assigning neutral resistance values to water had the least impact on the current flow across land, relative to a null model. We demonstrate water parameterization has important impacts on the outputs of terrestrial structural connectivity models that can complicate model interpretation. When the modeler’s intent is to inform land protection or restoration actions in response to human modification, the choice to treat water as neutral minimizes the interaction between water and land grid cells, clarifying output interpretation and use.