Connectivity

Habitat Connectivity

TerrAdapt’s habitat connectivity models quantify the proximity to the nearest patch of core habitat in ecological distance (also known as cost-weighted distance) rather than the straight-line Euclidean distance between locations. Ecological distance reflects the cumulative cost of routes that seek to avoid areas that are the most costly to move through. Cost is dependent on the movement ecology of the target species, but generally reflects the behavioral avoidance of certain habitat types (e.g., many species avoid areas of high human population density) as well as the mortality risk of crossing certain landscape features (e.g., risk of vehicle collision when crossing highways).

TerrAdapt’s connectivity models begin by identifying the core areas that are to be connected. Cores are mapped based on methods developed by the Washington Habitat Connectivity Working Group and implemented in the Gnarly Landscape Utilities. This approach uses a local neighborhood algorithm to identify concentrations of high quality habitat that are highly connected and meet minimum size thresholds.

The cost to move across grid cells between core areas is specified by a ‘resistance’ model tailored to each species or ecosystem. Environmental factors that may influence the cost of movement may include landcover, vegetation characteristics, climate, topography, and human pressures (e.g., human population density, transportation infrastructure, energy infrastructure, etc.).

To map connectivity to the nearest core or between cores, we use methods developed by the Washington Habitat Connectivity Working Group and implemented in the software program LinkageMapper. These methods use least-cost-corridor and least-cost kernel algorithms to identify routes between cores that minimize cost-distance based on the resistance model.

As noted above, it is often difficult to validate connectivity models due to lack of genetic or GPS data for many species. Our Greater sage-grouse habitat model was validated by landscape genetic analysis of microsatellite DNA collected from birds across the Columbia Basin and found to capture the key movement barriers in the region, such as major highways, transmission lines, and irrigated agricultural lands. The grizzly bear, fisher, lynx, and wolverine models were evaluated qualitatively by a small number of GPS collared animals making long-distance movements that aligned with predicted movement corridors. Vehicle-wildlife collision data supports the key highway crossings predicted by our ecosystem-based models.

---

Climate Connectivity

TerrAdapt’s climate connectivity models start with identifying core habitat areas for each species or ecosystem for the current year and calculating the cost-weighted distance to the nearest core given the landscape’s resistance to movement as specified in the species/ecosystem resistance surface (as described above in the methods of the connectivity section). Core habitat (based on projected future habitat for the species/ecosystem) and cost-weighted distance to the nearest core is also calculated at decadal intervals in future years out to 2100 under a range of Global Climate Models and GHG scenarios. Climate connectivity is then assessed as the average of the cost-weighted distances across all decades and scenarios. This average captures the areas commonly predicted to be used by species and ecosystems as their ranges shift over time. Uncertainty in climate connectivity can be calculated as the standard deviation of the cost-weighted distances across all decades and scenarios.

Validating climate connectivity models is not possible because they project future conditions that are unknowable. However, the same resistance models used in the habitat connectivity models described above are also used to map climate connectivity. Validating these models based on present day observations can provide a measure of confidence when projecting them into the future. Still, future projections are inherently uncertain and care must be taken to interpret them as potential futures, rather than likely outcomes.