Modelling of the Potential Distribution of Angelim Species in the Amazon for Conservation of Genetic ResourcesAbstract
Dinizia excelsa Ducke and Hymenolobium excelsum Ducke are economically important species in the Legal Amazon, widely exploited for construction, shipbuilding, and rustic furniture, making them vulnerable to illegal extraction and extinction. This study aimed to map and predict the distribution of these species, assess their future environmental adaptations through ecological niche modeling, and identify priority areas for genetic conservation and sustainable use. Modeling was based on 19 bioclimatic variables, 14 soil properties, and four algorithms. Current distribution considered the reference period (2009-2019), while projections were evaluated under two climate scenarios (SSP2-4.5 and SSP5-8.5) for 2041-2060, 2061-2080, and 2081-2100. Results showed similar dispersion patterns, with reductions in suitable areas across most states, except Acre, where H. excelsum suitability increased. Overall, D. excelsa proved more sensitive to climatic variations. The modeling highlighted vulnerable areas and provided essential guidance for targeted conservation strategies under climate change scenarios.
Keywords:
Climate change; environmental suitability; ecological niche
1. INTRODUCTION AND OBJECTIVES
Climate change has increasingly affected tropical forests, particularly the Amazon rainforest, threatening biodiversity and key ecosystem services (Tan et al., 2023; Jha & Dev, 2024; Dao et al., 2024). Rising temperatures and altered precipitation regimes have intensified tree mortality, reduced forest biomass, and altered ecosystem functioning (Arruda et al., 2024).
Deforestation and forest fires have made Brazil a major contributor to greenhouse gas emissions associated with land-use change, causing substantial losses of forest carbon stocks (Sampaio et al., 2023). In 2023, deforestation in the Brazilian Legal Amazon reached 8,022.79 km² (INPE, 2024). Illegal logging, selective timber exploitation, and agricultural expansion have intensified biodiversity loss and threatened timber specie (Leite et al., 2023).
Dinizia excelsa Ducke, commonly known as “angelim”, is native to South America, and widely distributed across the Brazilian Amazon. Its durable wood is extensively used in civil and naval construction (Sousa et al., 2023). Due to its economic importance, the species has been intensely exploited, raising concerns about its long-term conservation.
Hymenolobium excelsum Ducke, known as “angelim-da-mata”, occurs mainly in dry land forests of Amazonas and Pará (Ribeiro et al., 2016). The species is widely used in construction and remains subject to illegal exploitation in vulnerable areas (CNCFlora, 2024). Despite its economic relevance, information regarding the effects of climate change on its environmental suitability remains limited.
Climate change has shifted plant distributions toward higher altitudes and latitudes, altering the natural ranges of many tree species and reducing environmentally suitable areas in some cases (Zhao et al., 2021). These changes may affect ecosystem functioning and challenge forest management and conservation (Fyllas et al., 2022). Although these distributional shifts were not directly observed in this study, they were inferred from modeled patterns of environmental suitability.
Ecological niche modeling has been widely used to estimate relative environmental suitability based on associations between species occurrence records and environmental variables (Liu et al., 2022; Cordeiro et al., 2023). Although such models do not directly represent biological adaptation or demographic resilience, they provide valuable information on potential climatic favorability under different scenarios. Studies involving D. excelsa and H. excelsum can support conservation planning and sustainable forest management by identifying areas of relative suitability under future climate conditions (Zhao et al., 2021; Souza et al., 2024).
This study aimed to identify patterns of environmental suitability for Dinizia excelsa and Hymenolobium excelsum in the Legal Amazon under current and future climate scenarios. Ecological niche modeling was used to assess spatial trends in environmental suitability and identify priority areas for conserving genetic resources and supporting sustainable forest management.
Aimed at researchers, environmental managers, policymakers, and professionals involved in sustainable forest management and biodiversity conservation, this study supports the development of conservation and management strategies by identifying areas of relative environmental suitability under climate change scenarios (Baidya & Saha, 2024). By providing spatially explicit information for public policy and in situ conservation planning, it contributes to understanding the potential responses of ecologically and socioeconomically important forest species to future climatic conditions (Cordeiro et al., 2023; Marques et al., 2024; Souza et al., 2024).
Despite the use of multiple algorithms and evaluation metrics, ecological niche models remain approximations influenced by data quality, spatial bias, and modeling assumptions. Therefore, the results should be interpreted as relative patterns of environmental suitability and potential spatial trends rather than definitive predictions of species distributions.
2. MATERIALS AND METHODS
Occurrence records for Dinizia excelsa and Hymenolobium excelsum were obtained from SpeciesLink platform (CRIA), the Global Biodiversity Information Facility (GBIF), the Reflora Virtual Herbarium, and published georeferenced studies (Moscoso et al., 2013; Carim et al., 2015; Lewis et al., 2017; Canetti et al., 2021; Rocha et al., 2023; Martorano et al., 2025; Monteiro-Oliveira et al., 2025). Only records with explicit geographic coordinates were retained. Occurrence records were restricted to 2009-2019 to represent contemporary distributions and reduce temporal mismatches with recent land-use dynamics and sampling effort. This temporal decoupling between occurrence records and climatic baselines is a standard in ecological niche modeling and reflects the use of long-term climate averages to characterize environmental suitability.
All occurrence data are publicly available through the cited databases or publications, and the cleaned dataset is available from the corresponding author upon request. These data were used to model and generate distribution layers for the species across Brazilian phytogeographic domains.
To ensure spatial consistency, only records with precise geographic coordinates were retained, while duplicated, imprecise, or erroneous records were excluded. Point verification was performed using the Tidyverse package in the R environment (Wickham & Wickham, 2017; R Studio Team, 2024). To reduce spatial sampling bias and autocorrelation, occurrence records were spatially filtered using a minimum distance of 5 km between points with the spThin package in R. This threshold approximates the spatial resolution of the bioclimatic variables (~4-5 km) and minimizes the influence of clustered sampling on model calibration, following common practice in ENM/SDM studies (Aiello-Lammens et al., 2015; Baker et al., 2024; Moudrý et al., 2024; Bührs et al., 2026).
A total of 33 climatic and edpahic variables, were used to model the potential distribution of D. excelsa and H. excelsum. Nineteen bioclimatic variables were obtained from the WorldClim Global Climate Data database (version 2.1) at a spatial resolution of 2.5 arc-minutes (~4-5 km), summarizing long-term temperature and precipitation patterns derived from monthly climate data (Table 1) (Molloy et al., 2014; Fick & Hijmans, 2017).
Bioclimatic, edaphic, and topographic predictors used in the ecological niche modeling of Dinizia excelsa and Hymenolobium excelsum, including variable description and units. Topographic predictors represent terrain-derived descriptors of landscape heterogeneity rather than direct ecological drivers.
Considering the influence of soil attributes on plant development (Alvarez et al., 2022), 14 edaphic variables from the 15-30 cm soil layer were obtained from the SoilGrids database of the International Soil Reference and Information Centre (ISRIC - World Soil Information) (ISRIC, 2020) and incorporated into the modeling process (Table 1). To complement the characterization of the species’ environmental niche, topographic variables representing landscape heterogeneity were incorporated. Elevation (m) and slope were obtained from the EarthEnv database (Amatulli et al., 2018). Elevation was considered a proxy for broad-scale climatic and edaphic gradients rather than a direct ecological driver of species distributions. During multicollinearity assessment, elevation was evaluated jointly with climatic variables and interpreted cautiously because of its strong spatial structure and potential collinearity. Slope was retained because of its ecological relevance to drainage patterns, soil stability, and microhabitat variation.
Prior to model calibration, all environmental raster layers were standardized to ensure spatial compatibility. Climatic variables from WorldClim (2.5 arc-minutes) were used as the reference grid, while edaphic (250 m) and topographic (1 km) layers were resampled to the same spatial resolution, geographic extent, and coordinate reference system. Resampling was performed using bilinear interpolation for continuous variables, and all layers were clipped to the Legal Amazon extent. Consequently, the final dataset used for model calibration and projection had a uniform spatial resolution of approximately 2.5 arc-minutes (~4-5 km).
The datasets and modeling framework used in this study are summarized in Table 2, including data sources, spatial resolution, temporal coverage, and key modeling components.
Description of environmental and occurrence data used in ecological niche modeling of Dinizia excelsa and Hymenolobium excelsum. Original spatial resolution and temporal coverage of each dataset are indicated; all environmental layers were spatially standardized and resampled to a common grid of 2.5 arc-minutes (~4-5 km) prior to model calibration.
Principal Component Analysis (PCA) was used as an exploratory tool to identify environmental gradients and reduce predictor redundancy. Components explaining at least 95% of the cumulative variance were retained (PC1-PC33). Although the first components accounted for most of the total variance, variables with high absolute loadings (|loading| ≥ 0.50) in the retained components were considered ecologically relevant and selected for further evaluation, after which multicollinearity among selected predictors was assessed to avoid redundancy (Evangelista-Vale et al., 2021).
Future environmental suitability projections were generated using an ensemble of three CMIP6 Global Climate Models (GCMs): CNRM-CM6-1, MIROC6, and IPSL-CM6A-LR, selected for their widespread use in ecological and biogeographical studies and their performance in representing climatic patterns in South America (Ortega et al., 2021). For each GCM, bioclimatic variables were obtained from the WorldClim database at a spatial resolution of 2.5 arc-minutes (~4-5 km), corresponding to future climate scenarios. Environmental suitability projections were generated independently for each model and combined using an ensemble approach based on the arithmetic mean of suitability values.
This strategy was adopted to reduce individual model biases and better represent inter-model variability in future climate projections. Thus, the resulting maps represent consensus patterns across multiple climate models rather than outputs from a single GCM, following current best practices in ecological niche modeling.
Model calibration and validation were performed considering the spatial extent of the Brazilian Legal Amazon, which was defined as the accessible area (M) (Barve et al., 2011) for both species, representing the region potentially accessible to the species over relevant time periods. All background, pseudo-absence generation, model training, and internal validation procedures were restricted to this same spatial extent. Model validation was conducted using internal partitioning procedures implemented in the ENMTML framework, in which occurrence data were randomly split into training and testing subsets across model replicates, and model performance was evaluated using multiple metrics (AUC, Kappa, TSS, Jaccard, and Sørensen indices).
Future climate projections were based on two Shared Socioeconomic Pathways (SSPs): SSP2-4.5, an intermediate-emission scenario with partial mitigation, and SSP5-8.5, a high-emission scenario. Projections were generated for three future periods (2041-2060, 2061-2080, and 2081-2100) using WorldClim data (Fick & Hijmans, 2017). These scenarios were selected to represent a broad range of plausible future climatic conditions commonly adopted in climate-impact studies.
Species distribution models were calibrated using four algorithms implemented in the ENMTML framework (Andrade et al., 2020): Support Vector Machine (SVM; Prasad et al., 2006), Maximum Entropy with default settings (MaxEnt; Anderson & Gonzalez, 2011), Random Forest (RDF; Prasad et al., 2006), and Bayesian Gaussian models (GAU; Williams & Barber, 1998). These algorithms represent distinct statistical and machine-learning approaches commonly applied in ecological niche modeling.
Algorithm-specific parameter tuning was not performed; instead, default settings implemented in the ENMTML R packages were adopted for all modeling techniques. This approach ensured methodological consistency across algorithms and reduced the risk of overfitting, particularly given the limited number of occurrence records. For Maximum Entropy models, default feature classes (linear, quadratic, and hinge) and a regularization multiplier of 1.0 were used.
Background and pseudo-absence data were generated automatically in ENMTML using an environmentally constrained background approach. A common set of 1,000 pseudo-absence points was generated for each species and applied across all algorithms and model replicates, following recommendations that this number provides reliable model performance while minimizing class imbalance and environmental overrepresentation (Barbet-Massin et al., 2012; Descombes et al., 2022).
Model performance was evaluated using multiple complementary metrics, including Area Under the Receiver Operating Characteristic Curve (AUC), with values above 0.8 indicating good predictive performance (Fielding & Bell, 1997); the Kappa statistic; and the True Skill Statistic (TSS), which jointly considers sensitivity and specificity. Threshold-dependent metrics included the Jaccard and Sørensen indices, calculated following Leroy et al. (2018). In these formulations, correctly predicted presences (CPP), incorrectly predicted presences (IPP), unpredicted actual presences (UAP), and correctly predicted absences (CPA) were used to quantify model agreement.
Only models with consistent performance across evaluation metrics were retained for ensemble construction. Detailed information on algorithm configuration, parameter settings, and pseudo-absence generation is provided in Supplementary Table S1.
The resulting maps of relative environmental suitability were generated from ENMTML outputs in R (R Core Team, 2024) and further processed and visualized using QGIS (QGIS Development Team, 2024). Continuous suitability outputs were converted into binary maps using a threshold based on the maximum sensitivity plus specificity criterion. The total area of suitable habitat was calculated for each state and scenario, and percentage changes were computed relative to baseline conditions (2009-2019). These outputs were subsequently interpreted as spatial indicators to support discussions on conservation priorities, genetic resource protection, and sustainable management under climate change scenarios, rather than as definitive predictions of species distributions.
3. RESULTS
3.1. Analysis of the distribution of environmental suitability of species
A total of 126 occurrence records for Dinizia excelsa and 83 for Hymenolobium excelsum were compiled from online databases and literature sources. After data cleaning procedures, including the removal of duplicate, spatial errors, and records with insufficient geographic precision, 121 records for D. excelsa and 50 for H. excelsum were retained. Following spatial filtering and model preparation, 101 occurrence points for D. excelsa and 42 for H. excelsum were used for modeling (Figure 1).
Occurrence points of (A) Dinizia excelsa and (B) Hymenolobium excelsum in the Legal Amazon.
3.2. Selected variables
Principal Component Analysis (PCA) was used to describe the main climatic and edaphic gradients. Components explaining at least 95% of the cumulative variance were retained (PC1-PC33). Variables with high absolute loadings (|loading| ≥ 0.50) within the retained components were considered ecologically relevant and subsequently evaluated for multicollinearity before modeling.
The most influential climatic variables were: annual average temperature (ºC) (Bio1) (PC30: 0.73), temperature seasonality (%) (Bio4) (PC23: 0.63), minimum temperature in the coldest month (ºC) (Bio6) (PC33: 0.73), temperature in the driest quarter (ºC) (Bio9) (PC25: 0.56), mean temperature in the coldest quarter (ºC) (Bio11) (PC31: 0.83), accumulated rainfall in the year (mm) (Bio12) (PC24: 0.72), rainfall accumulated in the wettest quarter (mm) (Bio16) (PC28: 0.56) and rainfall accumulated in the driest quarter (mm) (Bio17) (PC26: 0.50) (Table 1). The most important soil variables related to soil characteristics were identified for elevation (m) (PC22: 0.62), organic carbon density (COD) (hg/m3) (PC21: 0.70), and sand (g/kg) (PC32: 0.74) (Table 3).
The models analyzed for D. excelsa presented values close to 1.0 in the evaluation index AUC, with standard deviation, respectively: RDF (0.97±0.02), GAU (0.97±0.03), SVM (0.97±0.03) and MXD (0.96±0.02). For H. excelsum, the GAU algorithm obtained the best performance in the AUC index (0.93±0.01), accompanied by low standard deviation values in all models (Table 4).
Although high AUC values were obtained, these metrics should be interpreted cautiously, as AUC may be inflated by spatial structure, background extent, and prevalence, particularly in presence-pseudoabsence models. Therefore, model evaluation emphasized consistency across multiple complementary metrics rather than AUC alone (Fielding & Bell, 1997; Hijmans, 2012; Grimmett et al., 2020; Bracho-Estévanez et al., 2024).
Under current climatic conditions, the relative environmental suitability of Dinizia excelsa and Hymenolobium excelsum was estimated using an ensemble of four ecological niche modeling algorithms (MXD - Maximum Entropy Default, SVM - Support Vector Machine, GAU - Bayesian Gaussian Process, and RDF - Random Forest) (Figure 2). Between 2009 and 2019, both species exhibited broad areas of higher suitability across Amazonas, Roraima, Pará, Amapá, Maranhão, Tocantins, Rondônia, and Mato Grosso, with lower suitability in Acre.
Spatial patterns of relative environmental suitability for Dinizia excelsa and Hymenolobium excelsum under current climatic conditions, derived from ensemble ecological niche models. Suitability values represent continuous indices of relative environmental favorability (scaled between 0 and 1) and do not correspond to binary presence-absence or absolute probability of occurrence.
Low-suitability areas (Figure 2A-B), were concentrated mainly in the southern, southwestern, and northern Amazon. In these regions, H. excelsum showed relatively higher suitability values than D. excelsa, although suitability remained low for both species. These current suitability patterns provide the baseline for interpreting projected changes under future climate scenario.
Based on this baseline, future projections indicate spatial shifts in environmental suitability for both species under climate change scenarios. For D. excelsa, the SSP2-4.5 scenario in 2041-2060, projected low environmental suitability across all states of the legal Amazon (Acre, Amazonas, Roraima, Pará, Amapá, Maranhão, Tocantins, Rondônia and Mato Grosso) (Figure 3A). Areas with lower climatic suitability were mainly concentrated along the phytogeographic domain Amazon margins, except in central Amazonas (Figures 3A-C). Isolated patches of higher suitability persisted in Amazonas, Rondônia, and at the Amazonas-Pará border (Figures 3A-C).
Future spatial patterns of relative environmental suitability for Dinizia excelsa and Hymenolobium excelsum under projected climate change scenarios, derived from ensemble ecological niche models. Suitability values represent continuous indices of relative environmental suitability, scaled between 0 and 1, and do not correspond to binary presence-absence outputs. Projections are based on the ensemble mean of three CMIP6 Global Climate Models (CNRM-CM6-1, MIROC6, and IPSL-CM6A-LR).
AUC values were interpreted comparatively among models and alongside other evaluation metrics rather than as absolute measures of predictive accuracy. In the remaining scenarios, low-suitability areas expanded, reducing isolated patches of higher suitability for D. excelsa (Figures 3D-F). However, a climatically suitable area between Amazonas and Acre remained stable (Figure 3F).
For H. excelsum, the SSP2-4.5 scenario, in the period 2041-2060, projected higher environmental suitability in Acre, Amazonas, Roraima Amapá and in the West and North of Pará (Figure 3G). Conversely, reductions in climatic suitability were observed in eastern Legal Amazon (Maranhão and Pará) and southern areas (Rondônia and Mato Grosso) (Figures 3G-I). Compared with D. excelsa, it was verified that H. excelsum retained broader climatically suitable areas throughout the Legal Amazon (Figure 3).
Under SSP5-8.5 scenario, in the periods 2041−2060, 2061−2080, and 2081−2100, suitability recovery of areas with climate adequacy was observed the occurrence of H. excelsum showed recovery of climatically suitable areas mainly in northern Amazonas and Roraima (Figures 3J-L). Greater losses of suitability were projected for southern Amazonia (Rondônia and Mato Grosso) and Maranhão. Nevertheless, the overall trend indicated recovery and climatic adaptation of the species across the Legal Amazon (Figures 3J-L).
In percentage terms, under the SSP2-4.5 scenario, projected climatically suitable areas for D. excelsa increased progressively during 2041-2060, 2061-2080, and 2081-2100, with the largest increases by 2081-2100 occurring in Acre (+508.09%), Mato Grosso (+10.35%), and Amazonas (+6.55%) (Supplementary Table S2). These high percentage increases reflect changes from very low baseline suitability values, where small absolute gains result in large relative differences. In contrast, significant reductions in suitable areas were projected for Maranhão, Amapá, Roraima, Pará, Tocantins, and Rondônia by 2081-2100 (Supplementary Table S2).
Under SSP585 scenario, in the period 2081−2100, D. excelsa showed an expressive increase in climatically suitable areas in Acre (+659.22%) (Supplementary Table S2). However, these high percentage values reflect relative changes from near-zero baseline suitability rather than ecologically meaningful expansions. In the same scenario and period, the greatest losses of suitable areas were projected for Maranhão (−91.69%), Amapá (−73.83%), and Roraima (−68.85%), whereas the smallest reduction occurred in Tocantins (−0.79%) (Supplementary Table S2).
Priority areas for in situ conservation correspond to regions consistently showing high environmental suitability across current and future scenarios (Figures 2 and 3), combined with relative gains or persistence in suitable area at the state level (Supplementary Table S2). These areas are mainly concentrated in Amazonas, Pará, and Mato Grosso, which showed stable or increasing suitability patterns over time. High percentage increases observed in Acre reflect very low baseline suitability and should be interpreted with caution.
4. DISCUSSION
The results indicate spatial patterns of relative environmental suitability under current and future climate scenarios rather than definitive species distributions. Despite the limitations of correlative ecological niche models, the methodological refinements adopted here identified consistent suitability patterns relevant to comparative and conservation analyses. Model calibration was restricted to the Brazilian Legal Amazon, the geographic focus of this study; therefore, projections should be interpreted only within this region. Although Hymenolobium excelsum also occurs outside Brazil, its broader environmental niche may not have been fully represented because model calibration was restricted to the Brazilian Legal Amazon. Therefore, extrapolation beyond the study area should be interpreted with caution.
Climate change was associated with negative shifts in the environmental suitability of both species across most states of the Legal Amazon, except Acre, where increases were projected, particularly for H. excelsum. These increases should be interpreted cautiously because they mainly reflect near-zero baseline suitability rather than ecological recovery. D. excelsa showed greater sensitivity to climate change than H. excelsum, which retained comparatively higher suitability
The identification of priority areas for in situ conservation, particularly in Amazonas, Pará, and Mato Grosso, highlights regions with high and persistent environmental suitability across current and future scenarios. These areas likely represent core habitats for maintaining viable populations and genetic diversity, reinforcing their importance for long-term conservation. Although Acre showed high percentage increases in suitability, these changes likely reflect low baseline values rather than substantial ecological expansion of suitable habitats. The persistence of suitable conditions in these regions suggests greater climatic stability, an important factor for conservation planning under climate change. Thus, this study contributes to strategies aimed at mitigating climate change effects on these ecologically and socioeconomically important Amazonian timber species.
Climate change has been associated with reductions in environmental suitability for plant species, potentially increasing extinction risk and genetic diversity loss, particularly among vulnerable or threatened species (Exposito-Alonso et al., 2022). In the Amazon biome, changes in temperature regimes, forest fires, and deforestation are major drivers of environmental change (Fátima et al., 2021).
These processes should not be interpreted as deterministic outcomes, but as part of the broader climatic and environmental context influencing the relative environmental suitability of D. excelsa and H. excelsum. Deforestation is one of the main drivers of temperature increase, promoting forest fragmentation, population isolation, and reduced genetic diversity (Lawrence et al., 2022). Plant species may respond by migrating to climatically suitable regions or adapting through phenotypic plasticity, gene flow, and in situ changes in allelic frequencies, potentially resulting in locally adapted ecotypes (Gougherty et al., 2021). Exposito-Alonso et al. (2022) further demonstrated that genomic diversity is strongly associated with geographic range, with losses exceeding 10% already reported for both endangered and non-endangered species.
Studies on timber forest species have shown that climate change may increase or reduce distribution areas, particularly under rising temperatures and altered precipitation patterns. These changes may also promote species migration from the Amazon to other regions, an important consideration for future management and conservation strategies (Morais et al., 2024).
The Eastern Amazon is considered one of the regions most affected by projected species losses and deforestation until 2050 (Gomes et al., 2019). Similarly, our results showed significant reductions in suitable areas for D. excelsa and H. excelsum across most states of the Legal Amazon, corroborating previous studies that reported reductions in environmentally suitable areas for Amazonian forest species under climate change scenarios (Morais et al., 2024).
The geographical distribution of plant species results from long-term interactions with environmental conditions and is not affected uniformly by climatic factors (Pérez-Suárez et al., 2024). However, climate change is occurring faster than the adaptive or migratory capacity of many species (Morais et al., 2024). Species unable to adapt or migrate before climate change affects their vegetative and reproductive capacity may experience local genetic diversity loss, extinction, reduced ecosystem services, and lower carbon stocks (Pérez-Suárez et al., 2024).
Studies on tree and timber species distributions under climate change suggest a tendency for migration toward higher-altitude regions (Zhao et al., 2021; Pérez-Suárez et al., 2024). Similarly, both species analyzed here showed increased suitable areas in Acre across all scenarios and periods, possibly associated with its relatively higher altitudes compared with most Amazonian regions.
The angelim species D. excelsa and H. excelsum are ecologically and socioeconomically important in the Brazilian Amazon and are increasingly threatened by deforestation and population decline. This study highlights the potential effects of climate change on their occurrence and identifies areas with higher relative environmental suitability that may support in situ conservation planning for both species.
5. CONCLUSIONS
This study identified spatial patterns of environmental suitability for Dinizia excelsa and Hymenolobium excelsum under future climate scenarios in the Legal Amazon. Suitability changes varied among states and scenarios, with a general reduction in suitable areas and localized increases, particularly for H. excelsum. However, increases in Acre should be interpreted cautiously, as they reflect near-zero baseline suitability rather than ecological recovery. Thus, interspecific differences represent patterns of suitability rather than deterministic outcomes or evidence of resilience.
Areas consistently identified as suitable across scenarios may support in situ conservation and genetic resource management, although results should be interpreted within the limitations of correlative ecological niche models. The findings highlight the usefulness of environmental suitability modeling as a decision-support tool under climate change while emphasizing the need for interpretation and ecological data.
DATA AVAILABILITY
The datasets generated and/or analyzed during the current study are available at public repositories: Species occurrence data from SpeciesLink (CRIA) and GBIF (with DOIs and access data listed below). Data are openly available without restriction and properly cited in the manuscript.
Species occurrence: SpeciesLink (CRIA): Dinizia excelsa Ducke (https://specieslink.net/search/download/20260723145859-0012722-1); Hymenolobium excelsum Ducke (https://specieslink.net/search/download/20260723150232-0020517-1).
GBIF: Dinizia excelsa Ducke (GBIF.org (15 November 2024) GBIF Occurrence Download https://doi.org/10.15468/dl.bs76fm); Hymenolobium excelsum Ducke (GBIF.org (15 November 2024) GBIF Occurrence Download https://doi.org/10.15468/dl.aapnqs)
R scripts used in this study are available at Github: https://github.com/carolinebezerra7-netizen/R/blob/e74df127cc9bede1c436b88c43306a13abd14068/R%20scripts%20for%20ecological%20niche%20modelling%20of%20Dinizia%20excelsa%20and%20Hymenolobium%20excelsum
SUPPLEMENTARY MATERIAL
The following online material is available for this article:
Table 1
Table S2
REFERENCES
-
Aiello-Lammens ME, Boria RA, Radosavljevic A, Vilela B, Anderson RP. SpThin: An R package for spatial thinning of species occurrence records for use in ecological niche models. Ecography 2015; 38(5):541-545. https://doi.org/10.1111/ecog.01132
» https://doi.org/10.1111/ecog.01132 -
Alvarez F, Morandi PS, Marimon-Junior BH, Exavier R, Araújo I, Mariano LH, Muller AO, Felpausch TR, Marimon BS. Climate defined but not soil-restricted: the distribution of a Neotropical tree through space and time. Plant Soil 2022; 471:175-191. https://doi.org/10.1007/s11104-021-05202-6
» https://doi.org/10.1007/s11104-021-05202-6 -
Amatulli G, Domisch S, Tuanmu MN, Parmentier B, Ranipeta A, Malczyk J, Jetz W. A suite of global, cross-scale topographic variables for environmental and biodiversity modeling. Scientific data 2018; 5(1):1-15. https://doi.org/10.1038/sdata.2018.40
» https://doi.org/10.1038/sdata.2018.40 -
Anderson RP, Gonzalez Júnior I. Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with Maxent. Ecological Modelling 2011; 222(15):2796-2811. https://doi.org/10.1016/j.ecolmodel.2011.04.011
» https://doi.org/10.1016/j.ecolmodel.2011.04.011 -
Andrade AFAD, Velazco SJE, De Marco Júnior P. ENMTML: An R package for a straightforward construction of complex ecological niche models. Environmental Modelling & Software 2020; 125:104615. https://doi.org/10.1016/j.envsoft.2019.104615
» https://doi.org/10.1016/j.envsoft.2019.104615 -
Arruda DM, Schaefer CEGR, Fonseca RS, Fernandes-Filho EI, Veloso GV, Gomes LC et al. Vegetações amazônicas e terras indígenas ameaçadas pelas próximas mudanças climáticas: Previsão de impacto nos biomas brasileiros. Austral Ecology 2024; 49(1): e13394. https://doi.org/10.1111/aec.13394
» https://doi.org/10.1111/aec.13394 -
Baidya A, Saha AK. Exploring the research trends in climate change and sustainable development: A bibliometric study. Cleaner Engineering and Technology 2024; 18:100720. https://doi.org/10.1016/j.clet.2023.100720
» https://doi.org/10.1016/j.clet.2023.100720 -
Baker DJ, Maclean IMD, Gaston KJ. Effective strategies for correcting spatial sampling bias in species distribution models without independent test data. Diversity and Distributions 2024; 30(3): e13802. https://doi.org/10.1111/ddi.13802
» https://doi.org/10.1111/ddi.13802 -
Barbet-Massin M, Jiguet F, Albert CH, Thuiller W. Selecting pseudo-absences for species distribution models: how, where and how many? Methods in Ecology and Evolution 2012; 3(2): 327-338. https://doi.org/10.1111/j.2041-210X.2011.00172.x
» https://doi.org/10.1111/j.2041-210X.2011.00172.x -
Barve N, Barve V, Jiménez-Valverde A, Lira-Noriega A, Maher SP, Peterson AT, Soberón J, Villalobos F. The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecological Modelling 2011; 222:1810-1819. https://doi.org/10.1016/j.ecolmodel.2011.02.011
» https://doi.org/10.1016/j.ecolmodel.2011.02.011 -
Bracho-Estévanez CA, Arenas-Castro S, González-Varo JP, González-Moreno P. Spatially explicit metrics improve the evaluation of species distribution models facing sampling biases. Ecological Informatics 2024; 84: 102916. https://doi.org/10.1016/j.ecoinf.2024.102916
» https://doi.org/10.1016/j.ecoinf.2024.102916 -
Bührs M, Zepp H, Rienow A. Integrating species distribution models (SDM) in spatially explicit GIS-tools to support nature-sensitive urban planning. Basic and Applied Ecology 2026; 91: 61-73. https://doi.org/10.1016/j.baae.2026.01.005
» https://doi.org/10.1016/j.baae.2026.01.005 -
Canetti A, Braz EM, Mattos PP, Basso RO, Filho AF. A new approach to maximize the wood production in the sustainable management of Amazon forest. Annals of Forest Science 2021; 78(3): 67. https://doi.org/10.1007/s13595-021-01079-8
» https://doi.org/10.1007/s13595-021-01079-8 -
Carim MDJV, Guimarães JRS, Tostes LDCL, Takiyama LR, Wittmann F. Composition, structure and floristic diversity in dense rain forest in the Eastern Amazon, Amapá, Brazil. Acta Scientiarum Biological Sciences 2015; 37(4): 419-426. https://doi.org/10.4025/actascibiolsci.v37i4.27536
» https://doi.org/10.4025/actascibiolsci.v37i4.27536 -
Centro de Referência e Informação Ambiental (CRIA) (2024). SinBiota: sistema de informação ambiental do Programa Biota/FAPESP. Available at: Available at: http://www.biotasp.org.br/sia/ Access in: nov. 15, 2024.
» http://www.biotasp.org.br/sia/ -
Centro Nacional de Conservação da Flora - CNCFlora. (2024). Hymenolobium excelsum in Lista Vermelha da flora brasileira versão 2012.2. Available at: Available at: http://cncflora.jbrj.gov.br/portal/pt-br/profile/Hymenolobiumexcelsum Access in: nov. 15, 2024.
» http://cncflora.jbrj.gov.br/portal/pt-br/profile/Hymenolobiumexcelsum -
Cordeiro AL, Tomaz JS, Bezerra CS, Meneses CHSG, Aguiar AV, Wrege MS et al. Prediction of the geographic distribution and conservation of Amazonian Palm trees Astrocaryum acaule MART. and Astrocaryum aculeatum MART. Revista Árvore 2023; 47: e4719. https://doi.org/10.1590/1806-908820230000019
» https://doi.org/10.1590/1806-908820230000019 -
Dao PU, Heuzard AG, Le TXH, Zhao J, Yin R, Shang C, Fan C. The impacts of climate change on groundwater quality: A review. Science of The Total Environment 2024; 912: 169241. https://doi.org/10.1016/j.scitotenv.2023.169241
» https://doi.org/10.1016/j.scitotenv.2023.169241 -
Descombes P, Chauvier Y, Brun P, Righetti D, Wüest RO, Karger DN, Zurell D, Zimmermann NE. Strategies for sampling pseudo-absences for species distribution models in complex mountainous terrain. bioRxiv 2022. 2022(03): 24.48569. https://doi.org/10.1101/2022.03.24.485693
» https://doi.org/10.1101/2022.03.24.485693 -
Evangelista-Vale JC, Weihs M, José-Silva L, Arruda R, Sander NL, Gomides SC, et al. Climate change may affect the future of extractivism in the Brazilian Amazon. Biological Conservation 2021; 257: 109093. https://doi.org/10.1016/j.biocon.2021.109093
» https://doi.org/10.1016/j.biocon.2021.109093 -
Exposito-Alonso M, Booker TR, Czech L, Gillespie L, Hateley S, Kyriazis CCM. Genetic diversity loss in the Anthropocene. Science 2022; 377(6613): 1431-1435. https://doi.org/10.1126/science.abn5642
» https://doi.org/10.1126/science.abn5642 -
Fátima B, Bottino MJ, Nobre P, Nobre CA. Deforestation and climate change are projected to increase heat stress risk in the Brazilian Amazon. Communications Earth & Environment 2021; 2(1): 1-8. https://doi.org/10.1038/s43247-021-00275-8
» https://doi.org/10.1038/s43247-021-00275-8 -
Fick SE, Hijmans RJ. WorldClim 2: new 1 km spatial resolution climate surfaces for global land areas. International Journal of Climatology 2017; 37(12): 4302-4315. https://doi.org/10.1002/joc.5086
» https://doi.org/10.1002/joc.5086 -
Fielding AH, Bell J. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation. Cambridge University Press 1997; 24(1): 38-49. https://doi.org/10.1017/S0376892997000088
» https://doi.org/10.1017/S0376892997000088 -
Fyllas NM, Koufaki T, Sazeides CI, Spyroglou G, Theodorou K. Potential Impacts of Climate Change on the Habitat Suitability of the Dominant Tree Species in Greece. Plants 2022; 11(12): 1616. https://doi.org/10.3390/plants11121616
» https://doi.org/10.3390/plants11121616 -
Global Biodiversity Information Facility (GBIF.org) (2024) Occurrence download. 2024. Available at: Available at: https://doi.org/10.15468/dl.bs76fm ; Available at: https://doi.org/10.15468/dl.bs76fm; https://doi.org/10.15468/dl.aapnqs . Access in: nov. 15, 2024.
» https://doi.org/10.15468/dl.bs76fm» https://doi.org/10.15468/dl.aapnqs -
Gomes VH, Vieira IC, Salomão RP, Ter Steege H. Amazonian tree species threatened by deforestation and climate change. Nature Climate Change 2019; 9(7): 547-553. https://doi.org/10.1038/s41558-019-0500-2
» https://doi.org/10.1038/s41558-019-0500-2 -
Gougherty AV, Keller SR, Fitzpatrick MC. Maladaptation, migration and extirpation fuel climate change risk in a forest tree species. Nature Climate Change 2021; 11(2):166-171. https://doi.org/10.1038/s41558-020-00968-6
» https://doi.org/10.1038/s41558-020-00968-6 -
Grimmett L, Whittingham MJ, Wilson RJ. Presence-only species distribution models are sensitive to sample prevalence. Ecological Modelling 2020; 431: 109194. https://doi.org/10.1016/j.ecolmodel.2020.109194
» https://doi.org/10.1016/j.ecolmodel.2020.109194 -
Hijmans RJ. Cross-validation of species distribution models: removing spatial sorting bias and calibration with a null model. Ecology 2012; 93(3): 679-688. https://doi.org/10.1890/11-0826.1
» https://doi.org/10.1890/11-0826.1 -
Instituto Nacional de Pesquisas Espaciais - INPE. (2024). Coordenação geral de observação da terra. Programa de Monitoramento da Amazônia e demais biomas. Desmatamento - Amazônia Legal. Available at: Available at: https://terrabrasilis.dpi.inpe.br/app/dashboard/deforestation/biomes/legal_amazon/increments Access in: dez. 21, 2024
» https://terrabrasilis.dpi.inpe.br/app/dashboard/deforestation/biomes/legal_amazon/increments -
Jha MK, Dev M. Impacts of Climate Change. In: Azrour M, Mabrouki J, Alabdulatif A, Guezzaz A, Amounas F. (eds) Smart Internet of Things for Environment and Healthcare. Studies in Computational Intelligence, vol 1165. Springer, Cham: 2024. https://doi.org/10.1007/978-3-031-70102-3_10
» https://doi.org/10.1007/978-3-031-70102-3_10 -
Lawrence D, Coe M, Walker W, Verchot L, Vandecar K. The Unseen Effects of Deforestation: Biophysical Effects on Climate. Frontiers in Forests and Global Change 2022; 5: 756115. https://doi.org/10.3389/ffgc.2022.756115
» https://doi.org/10.3389/ffgc.2022.756115 -
Leite ED, Costa CLA, Oliveira RS, Cavalcante HCC. Impactos ambientais causados pelo desmatamento no Brasil. Revista Livre de Sustentabilidade e Empreendedorismo 2023; 8(1): 19-38. https://doi.org/10.5281/zenodo.7538544
» https://doi.org/10.5281/zenodo.7538544 -
Leroy B, Delsol R, Hugueny B, Meynard CN, Barhoumi C, Barbet-Massin M, Bellard C. Without quality presence-absence data, discrimination metrics such as TSS can be misleading measures of model performance. Journal of Biogeography 2018; 45(9): 1994-2002. https://doi.org/10.1111/jbi.13402
» https://doi.org/10.1111/jbi.13402 -
Lewis GP, Siqueira GS, Banks H, Bruneau A. The majestic canopy-emergent genus Dinizia (Leguminosae: Caesalpinioideae), including a new species endemic to the Brazilian state of Espírito Santo. Kew Bulletin 2017; 72(3): 48. https://doi.org/10.1007/s12225-017-9720-7
» https://doi.org/10.1007/s12225-017-9720-7 -
Liu D, Lei X, Gao W, Guo H, Xie Y, Fu L et al. Mapping the potential distribution suitability of 16 tree species under climate change in northeastern China using Maxent modelling. Journal of Forest Research 2022; 33: 1739-1750. https://doi.org/10.1007/s11676-022-01459-4
» https://doi.org/10.1007/s11676-022-01459-4 -
Marques, M. J., Bezerra, C. S., Tomaz, J. S., Lopes, R., Wrege, M. S., Aguiar, A. V et al. Predição de distribuição geográfica e modelagem de nicho ecológico de açaizeiros na Amazônia. Pesquisa Agropecuária Tropical 2024; 54: e78108. https://doi.org/10.1590/1983-40632024v5478108
» https://doi.org/10.1590/1983-40632024v5478108 -
Martorano LG, Brienza Junior S, Moraes JRSC, Nascimento W, Lisboa LSS, Correa DL, Santos TM, Lima RF, Magalhães KRS, Dias CTS. Topoclimatic zoning in the Brazilian Amazon: enhancing sustainability and resilience of native forests in the face of climate change. Forests 2025; 16(6): 1015. https://doi.org/10.3390/f16061015
» https://doi.org/10.3390/f16061015 -
Molloy SW, Davis RA, Van Etten EJB. Species distribution modelling using bioclimatic variables to determine the impacts of a changing climate on the western ringtail possum (Pseudocheirus occidentals; Pseudocheiridae). Environmental Conservation 2014; 41(2): 176-186. https://doi.org/10.1017/S0376892913000337
» https://doi.org/10.1017/S0376892913000337 -
Monteiro-Oliveira D, Scudeller VV, Ferreira MJ, Lopes MTG, Ramos SLF, Capucho HLV, Silva LS, Oliveira JH, Freitas JGC, Tomazello-Filho M, Assis-Pereira G. Radial variation of carbon and nitrogen isotopes in Dinizia excelsa Ducke (Fabaceae) wood. Acta Botanica Brasilica 2025; 39: e20240238. https://doi.org/10.1590/1677-941X-ABB-2024-0238
» https://doi.org/10.1590/1677-941X-ABB-2024-0238 -
Morais ILL, Lima AA, Santos INLD, Meneses C, Silva RF, Lopes R, Lopes MTG. Climate Change Impact on the Distribution of Forest Species in the Brazilian Amazon. Sustainability 2024; 16(8): 3458. https://doi.org/10.3390/su16083458
» https://doi.org/10.3390/su16083458 -
Moscoso V, Albernaz AL, Salomão RP. Niche modelling for twelve plant species (six timber species and six palm trees) in the Amazon region, using collection and field survey data. Forest Ecology and Management 2013; 310: 652-662. https://doi.org/10.1016/j.foreco.2013.08.064
» https://doi.org/10.1016/j.foreco.2013.08.064 -
Moudrý V, Bazzichetto M, Remelgado R, Devillers R, Lenoir J, Mateo RG, Lembrechts JJ, Sillero N, Lecours V, Cord AF, Barták V, Balej P, Rocchini D, Torresani M, Arenas-Castro S, Man M, Prajzlerová D, Gdulová K, Prošek J, Marchetto E, Zarzo-Arias A, Gábor L, Leroy F, Martini M, Malavasi M, Cazzolla Gatti R, Wild J, Šímová P. Optimising occurrence data in species distribution models: sample size, positional uncertainty, and sampling bias matter. Ecography 2024; 2024(12): e07294. https://doi.org/10.1111/ecog.07294
» https://doi.org/10.1111/ecog.07294 -
Ortega G, Arias PA, Villegas JC, Marquet PA, Nobre P. Present-day and future climate over Central and South America according to CMIP5/CMIP6 models. International Journal of Climatology 2021;41(15):6713-6735. https://doi.org/10.1002/joc.7221
» https://doi.org/10.1002/joc.7221 -
Pérez-Suárez M, Ramírez-Albores JE, Martínez-Campos ÁR. Predicting the impacts of climate change on potential suitability habitats of three Juniperus trees in Mexico. Plant Ecology 2024; 225:37-51. https://doi.org/10.1007/s11258-023-01374-6
» https://doi.org/10.1007/s11258-023-01374-6 -
Prasad AM, Iverson LR, Liaw A. Newer classification and regression tree techniques: bagging and random forests for ecological prediction. Ecosystems 2006; 9(2):181-199. https://doi.org/10.1007/s10021-005-0054-1
» https://doi.org/10.1007/s10021-005-0054-1 -
QGIS Development Team. QGIS Geographic Information System. Open source geographic information system software for spatial data analysis and visualization. QGIS version 3.x, Open Source Geospatial Foundation, 2024. [Access in: 8 jan. 2025]. Available at: Available at: https://www.qgis.org
» https://www.qgis.org -
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; 2024. [Access in: jan. 26 2025]. Available at: Available at: https://www.R-project.org/
» https://www.R-project.org/ -
Reflora Herbário Virtual. (2024). Available from: Available from: https://reflora.jbrj.gov.br/reflora/herbario Virtual/ Access in: nov. 15, 2024.
» https://reflora.jbrj.gov.br/reflora/herbario Virtual/ -
Ribeiro ES, Souza RATM, Paula MH, Mesquita RRS, Moreira EL, Fazion H. Espécies florestais comercializadas pelo estado de Mato Grosso. Biodiversidade 2016; 15(2): 2-20. https://periodicoscientificos.ufmt.br/ojs/index.php/biodiversidade/article/view/3957
» https://periodicoscientificos.ufmt.br/ojs/index.php/biodiversidade/article/view/3957 -
Rocha LM, Rosa C, Secco H, Lopes EV. Hotspots and hotmoments of wildlife roadkills along a main highway in a high biodiversity area in Brazilian Amazonia. Acta Amazonica 2023; 53(1): 42-52. https://doi.org/10.1590/1809-4392202201871
» https://doi.org/10.1590/1809-4392202201871 -
RSTUDIO. Undelete and data recovery software. Integrated development environment free software for R for statistical analysis. R version 3.4.1, Boston, 2024. [Access in: 8 jan. 2025]. Available at: Available at: https://www.rstudio.com/
» https://www.rstudio.com/ -
Sampaio NN, Chaves CCS, Ferreira AF. Redução de emissões de gases do efeito estufa por desmatamento evitado-REDD+ na lei do Amazonas (4.266/2015) e a convenção quadro sobre mudanças do clima. Revista Ibero-Americana de Humanidades. Ciências e Educação 2023; 9(4): 1241-1256. https://doi.org/10.51891/rease.v9i4.9315
» https://doi.org/10.51891/rease.v9i4.9315 -
Sousa ACM, Costa ALA, Castro PHC, Lima AC, Silva GP, Araújo DG et al. Production of Dinizia excelsa Ducke (Fabaceae), seedlings subjected to increasing doses of controlled release fertilizer. Colloquium Agrariae 2023; 19: 182-191. https://doi.org/10.5747/ca.2023.v19.h522
» https://doi.org/10.5747/ca.2023.v19.h522 -
Souza DAT, Lima AA, Wrege MS, Aguiar AV, Bezerra CS, Meneses CHSG et al. Impactos das mudanças climáticas na distribuição natural de espécies de várzea alta e baixa na Amazônia. Revista Árvore 2024; 48(1): e4808. https://doi.org/10.53661/1806-908820244826375
» https://doi.org/10.53661/1806-908820244826375 -
Tan X, Wu X, Huang Z, Fu J, Tan X, Deng S et al. Increasing global precipitation whiplash due to anthropogenic greenhouse gas emissions. Nature Communications 2023; 14(1): 1-15. https://doi.org/10.1038/s41467-023-38510-9
» https://doi.org/10.1038/s41467-023-38510-9 -
Wickham H, Wickham MH (2017) Package tidyverse. Easily install and load the “Tidyverse”. Available at: Available at: https://www.tidyverse.org Accessed on: Jan 2025.
» https://www.tidyverse.org -
Williams CK, Barber D. Bayesian classification with Gaussian processes. IEEE Transactions on pattern analysis and machine intelligence 1998; 20(12): 1342-1351. https://doi.org/10.1109/34.735807
» https://doi.org/10.1109/34.735807 -
World Soil Information (ISRIC) (2024) SoilGrids - global gridded soil information. Available at: Available at: https://www.isric.org/explore/soilgrids Accessed on 21 dez 2024.
» https://www.isric.org/explore/soilgrids -
Zhao Z, Guo Y, Zhu F, Jiang Y. Prediction of the impact of climate change on fast-growing timber trees in China. Forest Ecology and Management 2021; 501: 119653. https://doi.org/10.1016/j.foreco.2021.119653
» https://doi.org/10.1016/j.foreco.2021.119653
-
FINANCIAL SUPPORT
This study was supported by the Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES), Brazil, under Notice No. 16/2020 - PROCAD-SPCF (Process No. 88881.516217/2020-01). Additional support was provided by National Council for Scientific and Technological Development (CNPq) through Research Productivity Grants awarded to Carlos Henrique Salvino Gadelha Meneses (Process No. 306661/2025-0), Santiago Linorio Ferreyra Ramos (Process No. 305280/2022-8), Ricardo Lopes (Process No. 308815/2023-8), Maria Teresa Gomes Lopes (Process No. 306943/2025-5), and Ananda Virginia de Aguiar (Process No. 307064/2023-9).
Edited by
-
Associate editor:
Bruno Araujo Furtado de Mendonça http://orcid.org/0000-0003-0288-0024
