Mapping climate adaptation corridors for biodiversity—A regional-scale case study in Central America

percentage of protection, number of overlapping protected areas (excluding marine) and Key Biodiversity Areas (KBAs), and mean forest biomass. We used a 1000 m buffer around corri- dor polylines to calculate all these variables (except corridor length) using the ‘extract’ function in the ‘terra’ R package [57] and vector operations (Buffer, Overlap analysis, and Intersection) in QGIS Desktop v3.24.1. We chose a 1000 m buffer (2 km corridor width) because this is con- sidered a “rule-of-thumb” minimum corridor width for facilitating long-term gene flow and habitat recolonization [58]. Because variable distributions were often not normal, we used Spearman’s correlation tests to compare different ecological and conservation variables. Our climate adaptation corridor priority index was based on the premise that shorter, pro- tected pathways with intact vegetation that link large lowland protected areas to large, high- elevation patches of contiguous forest offer the greatest potential as climate adaptation corri- dors, particularly if they overlap with areas of high biodiversity (e.g., KBAs). We therefore applied a combination of corridor, start node, and end node variables that reflected this prem- ise (S4 Table in S1 Appendix) into a principal component analysis (PCA). We eliminated over- lapping number of protected areas for each corridor and end node area because these were highly correlated with corridor length and end node elevational range, respectively (Spear- man’s rho = 0.72 and 0.76, respectively). We inverted corridor length such that shorter corri-

dors would be prioritized and we normalized all input variables (mean of 0 and standard deviation of 1) prior to PCA calculations. Based on visual inspection of the screeplot, we decided to use the first 3 principal components, which collectively explained 63% of variation in the data, to calculate priority index values for each potential corridor. Index values were cal- culated as the distance from the origin within the 3-dimensional space defined by the first 3 principal components. We designated high-, medium-, and low-priority index values based on visual inspection of their distribution (see results) and the need to identify a relatively small number of high-priority potential climate adaptation corridors for future, finer-scale investiga- tion. Finally, we also compared the ecological characteristics and conservation status of corri- dors and associated start and end nodes across high-, medium-, and low-priority corridor groups to aid interpretation. We applied nonparametric Kruskal-Wallis tests to compare dif- ferences among these groups and if there were significant differences, we then applied Dunn’s test with Holm’s p-value adjustment to perform pairwise comparisons using the ‘dunn_test’ function in the ‘rstatix’ R package [59]. Results Basic characteristics of start and end nodes and potential climate adaptation corridors Overall, end nodes were generally smaller, spanned wider elevational gradients, and had more intact forest biomass than start nodes, but were also poorly protected. Median start and end

node sizes were 50 and 11 km2, respectively (Fig 2A). Despite these notable size differences, end nodes spanned a median of 695 m in elevation, whereas start nodes only spanned a median of 142 m (Fig 2B). Because we selected start nodes as protected areas, these were all fully protected, but median end node protection was only 4% (Fig 2C), even though end nodes were selected as contiguous, intact forest patches (median = 73 Mt C) (Fig 2D). This result indicates that intact highland forest habitat is often unprotected in Central America. However, start node forest biomass was lower (median = 50 Mt C) despite greater protection. Finally, potential climate corridors originating from all 475 start nodes only terminated in 192 of the possible 529 end nodes (36%). This was because some pathways converged on the same end nodes and because we only modeled least cost paths between each start node and the 5 geo- graphically closest end nodes. PLOS ONE Mapping climate adaptation corridors for biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0304756 May 31, 2024 7 / 19