Central American tropical forests are biodiversity hotspots providing crucial ecosystem services for human livelihoods. However, climate change poses a significant threat to these forests, potentially altering species composition, driving forest type transitions, and causing irreversible habitat and biodiversity loss. Understanding the specific impacts of climate change on these ecosystems is vital for effective conservation strategies. This study focuses on the potential changes in the environmental suitability of major plant functional types (PFTs) across Central America under various climate change scenarios. The research aims to identify regions experiencing significant transitions, predict potential connectivity losses, and assess the risk of mountaintop extinctions for montane species. This information is crucial for developing targeted conservation efforts, including the establishment of biological corridors and the expansion of protected areas in vulnerable regions. The implications of these changes extend beyond biodiversity, affecting ecosystem services like ecotourism and timber production, highlighting the urgency of the research.

Existing research indicates a general trend towards dry vegetation types in Central American forests under climate change. However, there's a need for more detailed assessments at regional and PFT-specific levels, considering small-scale heterogeneity in climate, topography, and soil conditions. Studies have shown rapid range shifts of species in response to warming, along with potential local extinctions. The concept of mountaintop extinctions, where species lack alternative habitats at higher altitudes, also requires more regional-specific investigation. Previous research on Central American forests indicates a general trend towards drier vegetation types; however, this study aims to provide more nuanced insights by considering regional disparities and the specific responses of different PFTs. The importance of habitat connectivity in mitigating the negative effects of climate change on biodiversity has also been highlighted in past research, but this study seeks to quantify this connectivity loss for different PFTs in Central America.

The study area encompasses Central America (68°-100°W, 4°-24°N), including parts of Mexico and northern South America to improve model calibration. Tree species were classified into PFTs based on abiotic tolerances and structural/physiological attributes, distinguishing between wet and dry lowland forests (<1000 m) and coniferous and montane forests at higher altitudes. Further sub-classification considered resource acquisition strategies (acquisitive and conservative). A “generalist” PFT was included for species showing traits of both categories. Four representative species were selected per PFT based on literature review, trait data analysis, broad occurrence within ecoregions, and data availability (minimum 200 records). Species presence records were compiled from GBIF, BIEN, and other sources, thinned to one record per grid cell to reduce sampling bias. Predictor variables (bioclimatic, topographic, and edaphic) were obtained from CHELSA, GMTED2010, and SoilGrids at 30-arc-second resolution. Present-day data (1979-2013) and climate projections from three GCMs (CCSM4, HadGEM2-AO, MPI-ESM-LR) and three RCPs (2.6, 4.5, 8.5) were used. A stepwise procedure addressed multicollinearity. Species distribution models (SDMs) were built using the R package SSDM, employing ANN, CTA, GAM, GLM, MARS, MAXENT, SVM, and RF. Model performance was evaluated using AUC, CBI, and a novel calibration statistic. Models meeting a 0.7 threshold were used. To minimize algorithm bias, models were ensembled using unweighted arithmetic means across algorithms and replicates. Potential range shifts were analyzed by plotting presence density across latitude and altitude. Fragmentation analysis used a 3x3 moving window approach to classify grid cells into interior, patch, transitional, perforated, and edge categories based on the proportion of cells covered by each PFT and their connectivity.

Model projections (2061-2080) forecast transitions from wet to generalist or dry forest PFTs across large areas. Wet-adapted PFTs are projected to shift latitudinally and lose connectivity, with significant fragmentation along the Caribbean coast (Honduras, Nicaragua, Panama). Montane and coniferous species show upslope shifts in their lower distribution boundary (approximately 100 m for RCP2.6, 200 m for RCP4.5, and 500 m for RCP8.5), indicating a high risk of mountaintop extinction. The wet conservative PFT experienced the most substantial projected losses (up to 78%), with increasing fragmentation. Generalist PFTs expanded their range considerably under all scenarios. Dry forest PFTs showed only slight area increases, while coniferous areas decreased slightly, but without significant changes in fragmentation. Despite overall small increases in the area of wet acquisitive PFTs, several connectivity bottlenecks appeared along the Caribbean coast. The analysis highlights increasing proportions of fragmented forest patches, particularly impacting wet forest species which support significant amphibian, bird, mammal, and reptile biodiversity—many already threatened by extinction or fragmentation. Connectivity bottlenecks were observed in the Mesoamerican Biological Corridor, a crucial migration route.

The findings support existing studies on Central American forest transitions toward drier vegetation types but reveal regional disparities and PFT-specific responses due to small-scale heterogeneity in environmental factors. The southward shift in most PFTs may lead to increased competition and distribution limits due to low connectivity between Central and South America. Diverging latitudinal and altitudinal trends in wet forest species compromise habitat connectivity and exacerbate fragmentation from land use. Connectivity bottlenecks in the Mesoamerican Biological Corridor pose a severe risk to species relying on this migration route. Increased fragmentation reduces available forest interior habitat, crucial for many species. Upslope shifts in montane species, limited by the highest elevations they already occupy, raise serious concerns about mountaintop extinction. These changes will dramatically affect biodiversity, ecosystem services (ecotourism), and economic activities (timber production). Transitions from wet to dry forests will impact carbon sequestration capacity, with reduced assimilation and growth rates in drier forests. Increased competition between pine species and drought-resistant species is also expected, potentially exacerbated by pests and El Niño events. Future research should consider other plant groups (lianas, palms, grasses) and their interaction with climate change and forest disturbances.

This study reveals both general trends and hotspots of forest type transitions in Central America under various climate change scenarios. Decreasing climatically suitable areas for wet forest species, impaired habitat connectivity, and mountaintop extinctions represent major threats. Urgent policy interventions are needed, including the delineation of biological corridors and expansion of protected areas. Future research should focus on growth responses to climate change in different biomes and vegetation types to inform management decisions. Trait-based approaches, as employed here, offer valuable insights for combining ecological understanding with practical application.

The study relies on species distribution models (SDMs), which have inherent limitations, such as assumptions of species-environment equilibrium and potential biases. The projections do not provide a timeline for species range shifts or extinction, as other factors (disturbances, resource availability, dispersal strategies) influence species survival. Anthropogenic factors like urbanization and land use further constrain the available area for range shifts. Extrapolation beyond the range of the training data might underestimate climate change effects in some areas. The study focuses primarily on tree species and does not account for the full complexity of the ecosystem, including the impact on other plant groups and the potential for species interactions to alter the projections.