Disclaimer: Supplementary materials have been peer-reviewed but not copyedited.

nyas12172-sup-0001-FigS1.tif8814KFigure S1. A schematic of the process for identifying a least-cost path (LCP). The shaded patches represent suitable habitat, a wetland type. (A) Other wetlands exist to the northeast (cattail icon), and mountains representing the most unsuitable habitat are present to the southwest. (B) The landscape between habitat patches is converted to a continuous surface of triangles (the TIN) with nodes attributed by the level of resistance (here, mountain, in green, equals highest resistance; wetland, in blue, is lowest; and the remainder, in brown, is in between). (C) A path (in red) that minimizes the sum of all costs was derived from a function that balances distance and resistance (Equation 2), to find the LCP. The same two patches depicted again in (D), and then as part of a cluster of patches in (E), with every patch-to-patch combination mapped (28 paths).
nyas12172-sup-0002-FigS2.tif8838KFigure S2. Examples of results for a small subset of the range of the northern copperhead. The left panel depicts the resistance grid, which is transformed to a triangular irregular network (TIN) with inset habitat patches in the middle panel. The right panel shows the modeled least-cost paths among patches.
nyas12172-sup-0003-FigS3.tif10862KFigure S3. Consistency in parcel importance, as measured by the number of species for which a parcel was important (increasing from blue to red; note that the same range of colors is used in each figure, although the absolute numbers differ) in both current day and the 2050s (left panel) and the number of species for which a parcel was important for all three time periods (current day, 2050s, 2080s), right panel.
nyas12172-sup-0004-FigS4.tif8815KFigure S4. Model of management actions to take based on known relationships of a species with climate and other variables and four potential types of distribution responses.
nyas12172-sup-0005-FigS5.tif8814KFigure S5. Model of management actions for maintaining patch to patch connectivity. This model begins with the final actions from Figure S4.
nyas12172-sup-0006-TableS1.pdf61KTable S1. Environmental layers used for suitability modeling with a short description and derivation for each.

Table S2. Evaluation metrics (mean ± standard error) for random forest models of 26 species in New York's Hudson Valley.

Table S3. Descriptive statistics for predicted patches and paths for 21 species in three time periods in New York's Hudson Valley.


Appendix S1. Collecting and smoothing climate data.

Appendix S2. Modeled patches and connections for each species within each time period.

Appendix S3. Species distribution model validation metadata.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.