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

protected areas and KBAs were positively correlated with corridor length across all countries (respective Spearman’s rho = 0.76 and 0.75; p < 2.2e-16), but percentage corridor protection was not (Spearman’s rho = -0.02, p = 0.32). Mirroring corridor protection, corridors in Pan- ama and El Salvador had the greatest and least forest biomass, respectively (medians = 74 and 39 Mt C) (Table 1, Fig 3F). Across other countries, median corridor forest biomass ranged from 52 to 67 Mt C and percentage corridor protection was positively correlated with corridor forest biomass (Spearman’s rho = 0.50, p < 2.2e-16). Corridors in Colombia and Costa Rica were among the most heavily forested (medians = 67 and 65 Mt C, respectively), behind only Panama, despite poor and moderate levels of corridor protection (Colombia: median = 19% protection; Costa Rica: median = 36% protection). In summary, corridors located in Panama, Belize, and Honduras were generally the most protected and those located in Panama, Colom- bia, Costa Rica, and Belize were generally the most forested. Longer corridors generally over- lapped with the most protected areas and KBAs but were not necessarily more protected overall. Climate adaptation corridor priority index Climate adaptation corridor priority index values ranged from 0.21–7.11 and followed a right- skewed distribution (S2 Fig in S1 Appendix). Considering this and that only a small percentage of the 2375 potential corridors we identified might be pursued for conservation efforts, we

assigned priority index values < 2 as low (n = 1290; 54%), 2–4 as medium (n = 925; 39%), and > 4 as high (n = 160; 7%). We encountered several clusters of high-priority corridors, including the Yucatan Peninsula (mostly Mexico and Belize), the coastal lowlands to Sierra Madre in Guatemala, the coastal lowlands to the Cordillera Occidental of Colombia, eastern Honduras/northern Nicaragua, southeastern Nicaragua/northern Costa Rica, and converging on La Amistad International Park of Costa Rica and Panama from both the Pacific and Carib- bean coasts (Fig 4). All countries except El Salvador contained at least one high-priority corri- dor, but El Salvador had a relatively high density of medium-priority corridors. Much of Nicaragua was covered by low-priority corridors, particularly the western portion of the coun- try where there are few large protected areas. Analysis of the ecological characteristics and conservation status of potential climate adap- tation corridors according to priority index values demonstrated that high-priority corridors were significantly longer, spanned wider elevational gradients, had more forest biomass, were more protected, and overlapped with more protected areas and KBAs compared to medium- and low-priority corridors (Dunn’s p � 0.003) (Table 2 and Fig 5A–5F). For example, high- priority corridors spanned median elevational ranges of 2185 m (compared to 1868 m and 1829 m for medium- and low-priority corridors, respectively) (Table 2 and Fig 5B) and had median protection of 46% (compared to 37% and 38% for medium- and low-priority corri-

dors, respectively) (Table 2 and Fig 5D). These findings were largely consistent with our priori- tization design, except for corridor length, which was positively correlated with the number of overlapping protected areas and KBAs (Spearman’s rho = 0.72 and 0.66, respectively; p < 2.2e- 16). Additionally, high-priority corridors consistently led to end nodes that were larger, spanned wider elevational ranges, and were more protected compared to medium- and low- priority corridors (Table 2 and Fig 5G–5I) (Dunn’s p � 3.70e-15). Similarly, high-priority cor- ridors consistently originated from larger start nodes compared to medium- and low-priority corridors (Dunn’s p � 1.23e-6), but start nodes of high-priority corridors also spanned nar- rower elevational ranges compared to those of medium- and low-priority corridors (Dunn’s p � 0.02) (Table 2, Fig 5J and 5K). However, all lowland protected areas have considerably shorter elevational gradients than highland target areas (Table 1 and Fig 2B). In summary, PLOS ONE Mapping climate adaptation corridors for biodiversity PLOS ONE | https://doi.org/10.1371/journal.pone.0304756 May 31, 2024 10 / 19