Assessment of the Impact of Climate Change on the Potential Distributions of Melliferous Plant Species on the Yucatan Peninsula, Mexico: Implications for Conservation Planning

Abstract

Climate change is altering environmental conditions, which can, in turn, change the geographic distribution and flowering patterns of plant species, affecting both the plants themselves and their pollinators. The responses of melliferous plant species to climate change in southeastern Mexico are poorly understood, which hinders an accurate assessment of their vulnerability and the resulting ecological impacts. As understanding the mechanisms that influence the distribution and susceptibility of these species is essential, the present study examined how climate change affects their potential distribution areas and spatial distribution patterns. This information will serve as reference data for the implementation of conservation strategies and inform the selection of species for reforestation. Ecological niche models were used to estimate the potential distributions of 92 melliferous species under both current environmental conditions and two climate change scenarios projected for the 2041–2060 period (SSP245 and SSP585). Changes in distribution patterns were then assessed by evaluating the range size of each species and analyzing the spatio–temporal trends in species richness. The results revealed that suitable habitats shifted for approximately 80% of melliferous species, with more significant habitat loss observed under the SSP585 scenario than under SSP245. Although a significant decrease in melliferous plant species richness was expected in future scenarios, richness was slightly higher (by 10% for SSP245 and 5% for SSP585) than that observed under current environmental conditions. Under SSP245 conditions, species richness areas expanded to encompass almost the entire region, although this contrasted drastically with the SSP585 scenario, where areas with the highest concentration of species richness contracted significantly and areas with low species richness expanded. These projections are of potential use for conservationists and environmental management authorities, providing crucial insights into the future distributions of several melliferous plant species in the region, the potential impacts of climate change on their habitats, and the vulnerability of threatened species to changing climatic conditions.

1. Introduction

Climate change represents one of the most pressing challenges confronting humanity in the 21st century, and its impacts are already evident in various aspects of global ecosystems [1]. Pollinators are a vital component of these ecosystems, especially bees, which depend on melliferous plants for food and survival. Melliferous plants produce nectar and pollen, which are collected by bees to produce honey and other bee products [2,3]. However, climate change is altering environmental conditions, which, in turn, alters the geographic distribution and flowering patterns of plant species, affecting both the plants and their pollinators [4,5,6]. In this context, melliferous plants take on even greater importance, as they are not only essential for the survival of pollinators but also key for mitigating the effects of climate change, restoring damaged ecosystems, ensuring global food security, and providing useful health products [7,8]. Pollinators, particularly bees, are essential for both food production and biodiversity maintenance [9]. Globally, approximately 75% of flowering plant species depend on pollinators for reproductive success. In the Neotropics, where biodiversity is particularly high, the mutualistic relationships among melliferous plants and pollinators are especially critical for maintaining ecosystem function. Climate change disrupts these interactions by altering the phenology of melliferous species and reducing the availability of nectar and pollen [5,6]. Variations in temperature and precipitation regimes cause temporal mismatches between floral resource availability and pollinator activity [10,11,12]. For instance, temperature-induced shifts in flowering periods can reduce synchrony with foraging behavior, negatively impacting pollinator nutrition and population viability. Additionally, extreme climatic events—such as prolonged droughts and heatwaves—increase physiological stress in plants, diminishing nectar secretion and exacerbating food shortages for pollinators [13,14]. Elevated temperatures also heighten the vulnerability of bee colonies to pathogens and parasites, further compromising pollinator survival. In addition, honey production, a key ecosystem service in tropical and subtropical rural economies, is also affected by climate variability, as honey yield and its physicochemical properties (e.g., flavor, aroma, and composition) are strongly influenced by floral sources, climate, and geography [15].

Despite the threats described above, certain neotropical melliferous plants exhibit adaptive traits that confer resilience in response to drought, heat, and nutrient-poor soils [10,16,17]. For example, some melliferous plants in arid ecosystems may store water in root tissues or minimize transpiration [18]. These traits not only ensure the persistence of floral resources for pollinators but also support ecosystem restoration by stabilizing soils, preventing erosion, and enhancing water retention in degraded areas [16,17,18]. In the context of pollinator habitat loss driven by land use change and climate stressors [19,20], the ecological plasticity of melliferous plants is particularly valuable. Their incorporation into restoration strategies in deforested, desertified, or urbanized areas can facilitate the recovery of pollinator communities and promote native plant diversity [21,22].

Pollinator decline poses a significant risk to food security, particularly in the Neotropics, where many high-value crops—such as fruits, vegetables, legumes, and oilseeds—are pollinator-dependent [9,12,23,24]. Reductions in pollinator populations can lead to decreased crop yields, affecting food availability and increasing economic vulnerability. The strategic conservation and use of climate-resilient melliferous species can enhance pollination services, support agricultural productivity, and promote agroecosystem biodiversity. Diversifying nectar and pollen sources within agricultural landscapes, particularly through the integration of native melliferous flora, can help mitigate climate impacts and foster more sustainable and resilient production systems [25,26,27]. In this sense, the diversification of honey crops on agricultural land can contribute to mitigating the effects of climate change, thus promoting more balanced and sustainable ecosystems [27].

Species distribution modeling correlates georeferenced occurrence data with bioclimatic predictors to delineate potential habitat suitability [28]. Widely used across taxonomic groups [29,30,31], species distribution models (SDMs) rank among the most powerful methods for projecting climate–driven shifts in species’ ranges [32]. Contemporary implementations employ diverse algorithms and standardized workflows to quantify ecological niches, map habitat distribution, and evaluate species’ responses to environmental change [33]. By operating at multiple spatial scales, SDMs enhance our understanding of population–level distribution patterns and yield actionable insights for conservation planning [34], including the identification of priority areas for threatened taxa, the design of reserve networks, and the assessment of landscape connectivity [35,36,37]. In the neotropical context, SDMs effectively predict suitable habitats for honey-producing tree species under current and future climate scenarios. Nevertheless, model reliability hinges on the representativeness and accuracy of occurrence records, the selection of ecologically relevant variables, and rigorous calibration. In regions with sparse or spatially biased data, outputs should be validated in the field and then evaluated by someone with expert knowledge to ensure ecological realism.

Southeastern Mexico exhibits a significant conservation deficit, characterized by limited protected area coverage, high rates of land use change, and insufficient management of biodiversity-rich ecosystems. This region harbors a high diversity of melliferous plant species, many of which are threatened by deforestation, agricultural expansion, and habitat fragmentation. In particular, the Yucatan Peninsula is a large peninsular region in southeastern Mexico, bordered by the Gulf of Mexico and the Caribbean Sea. Covering an area of 450,000 km² and containing the states of Quintana Roo, Yucatan, and Campeche, it is considered a biotic province in the Neotropical region and presents high diversity [38], including approximately 2300 plant species across 956 genera and 161 families [39,40]. Melliferous plants are a key component of the flora of the Yucatan Peninsula, with an estimated 900 plant species that could act as meliponiculture resources to produce honey, resin, and wax [3,41]. Despite the great importance of melliferous flora, the potential impact of climate change on the future distribution and composition patterns of these species has been the subject of limited research. The lack of targeted conservation strategies for these species compromises pollination networks, affects local livelihoods dependent on apiculture, and undermines ecosystem resilience. Addressing this deficit requires the integration of melliferous plant conservation into regional land use planning and ecological restoration initiatives. Therefore, understanding the impacts of climate change on these species is essential for designing effective conservation, propagation, and management strategies in the face of future climate scenarios [19]. Species distribution models are a valuable tool for assessing and predicting the impacts of climate change on the potential distribution of plant species and vegetation [42,43,44].

Climate change poses complex challenges for pollinator survival and ecosystem stability. In this context, melliferous plants are a fundamental tool for preserving biodiversity, ensuring crop pollination, and mitigating the effects of climate change while generating various products for human use and consumption. Promoting the conservation and cultivation of melliferous species that have adapted to new climatic conditions will not only support the survival of pollinators but will also contribute to ecosystem restoration and global food security. Therefore, melliferous plants are a key element in the fight against climate change, and their conservation and promotion must be made a priority in global environmental and agricultural policies. The aim of the present research was, therefore, to use species distribution modeling to project the current geographic distributions of the melliferous plant species of the Yucatan Peninsula and model the potential changes to these distributions under future climate scenarios. This data will provide an improved understanding of distribution trends and possible distribution regions, which can be used to inform the implementation of conservation strategies, help locate known suitable habitats for ex situ conservation, and facilitate reforestation in the future.

2. Materials and Methods

2.1. Study Area

The Yucatan Peninsula is located in the southeastern region of Mexico, bounded by the coordinates 19°33′ N and 89°17′ W (Figure 1) and bordered to the north and east by the Caribbean Sea, to the south by Guatemala and Belize, and to the west by the Gulf of Mexico and the state of Tabasco. With an area of approximately 151,515 km², this region comprises the states of Quintana Roo, Yucatan, and Campeche, representing 7.7% of the area of Mexico. According to the Köppen climate classification modified by García [45], the Yucatán Peninsula exhibits a tropical climate with warm, humid, and semi-arid subtypes, characterized by predominant summer rainfall. The region experiences three distinct seasons: a hot and dry period from March to May, a rainy season from June to October, and a cooler season from November to February, characterized by occasional short rains and winter storms known as nortes [46]. Average annual temperatures range between 24 °C and 28 °C, with two thermal zones—western and eastern—divided by a north–south isothermal boundary at 26 °C. Annual precipitation generally falls between 1000 and 1200 mm, following a gradient from drier conditions in the northwest (<600 mm) to more humid conditions in the south and southeast (>1500 mm) [47,48]. Vegetation varies across the region, with low deciduous forest dominating the northern areas, medium subdeciduous forest in central and northern sections, and medium semi-evergreen forest in the south [49,50].

2.2. Melliferous Plant Species Selection

The present study selected 92 melliferous plant species (Table S1), based on their relative importance to beekeeping, ecology, and agroforestry, as established by a review of the following studies: Sousa-Novelo et al. [52], Villanueva-Gutiérrez [53,54], Carnevali et al. [55,56], Porter-Bolland [57,58], Porter-Bolland et al. [59], Alfaro-Bates et al. [60], Villanueva-Gutiérrez et al. [61,62], Flores-Guido and Vermont-Ricalde [63], Ramos-Díaz et al. [64], Castillo-Cazares et al. [65], Cetzal-Ix et al. [66], Coh-Martínez et al. [67,68], Baena-Díaz et al. [2], Briceño-Santiago et al. [69], Villalpando-Aguilar et al. [70], Pérez-Morfi et al. [71], Ríos-Oviedo et al. [72], and Zúñiga-Díaz et al. [3]. Occurrence records were compiled for all melliferous plant species sampled in the Yucatan Peninsula, Mexico, by the Global Biodiversity Information Facility (GBIF, https://www.gbif.org/; accessed on 17 November 2022) (see Table S1). To minimize potential sampling biases and prevent model overfitting, several data-cleaning procedures were applied. First, duplicate records and those lacking precise geographic coordinates were removed, with records located within 3 km of another record of the same species then filtered out to ensure spatial independence and improve distance-based accuracy in the species occurrence data. This thinning process was conducted using the spThin package [73] in R [74]. Additionally, to detect and exclude species occurrence associated with erroneous climate values [75], an environmental heterogeneity filter was applied, as implemented in the ArcMap 10.2 SDMtoolbox [76]. Finally, after all the above-described steps were performed, any species (~14 species) that were left with fewer than ten unique occurrence records in the GBIF database were also eliminated (see Table S1).

2.3. Environmental Variables

For each species, a subset of potential predictor variables was selected from interpolated climate data (~1 km² resolution) obtained by WorldClim 2.1 [77] (available at https://www.worldclim.org/data/; accessed on 17 November 2022). This bioclimatic dataset comprises 19 climate variables that represent essential aspects of temperature and precipitation, derived from interpolated data obtained from global meteorological stations during the 1970–2000 period [77]. To minimize collinearity among these variables and to model overfitting, a set of non-redundant environmental variables (see Table S2) was selected for each species by eliminating variables using Pearson’s correlation coefficient (r) > 0.80 and variance inflation factor >10 [78,79,80], with an average of eight variables selected per species. The environmental predictors selected are recognized as key factors influencing the ecophysiology and distribution of the melliferous plant species under study [81]. The overlap of occurrence records with the Yucatan Peninsula ecoregion [51] and the Neotropical biogeographical province [82] was used to define the model calibration area for each species. This approach was selected because both regions reflect the species’ historical and ecological boundaries.

Future climate projection variables for the year 2050 were sourced from the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) [83]. These scenarios serve as a key tool for understanding potential climate outcomes, impacts, risks, and mitigation strategies [84]. For climate model simulations, the present study selected MIROC6 and IPSL-CM6A-LR projections from general circulation models (GCMs), using two Shared Socioeconomic Pathway (SSP) scenarios—SSP245 and SSP585—derived from the Coupled Model Intercomparison Project Phase 6 (CMIP6). These scenarios model potential future climate conditions under different greenhouse gas emission trajectories [83,85], with SSP245 representing a moderate emissions-reduction scenario, predicting a temperature increase of 2.7–3.4 °C by 2100, and SSP585 assuming high emissions and limited climate mitigation, corresponding to a rise of 4.3 °C by the end of the century [83,86,87].

2.4. Ecological Niche Models

Ecological niche models for each species were developed using MaxEnt v.3.4.1, which applies the maximum entropy method to model ecological niches based on presence data and environmental variables [88,89,90]. Although widely used for species distribution modeling due to its effectiveness with presence-only data, MaxEnt has several limitations [88,89,90,91], including the assumption that species are in equilibrium with their environment and that occurrence records are representative and unbiased, which may not hold true, particularly in data-deficient regions. The model’s performance is sensitive to sampling bias, spatial autocorrelation, and the selection of environmental variables. Additionally, MaxEnt does not account for biotic interactions, dispersal limitations, or anthropogenic barriers, potentially leading to overestimations of suitable habitat. These limitations underscore the need for careful model calibration, validation, and ecological interpretation. Model generation was carried out using the kuenm package [80] in R [74], enabling detailed calibration and the selection of optimal model parameters based on statistical significance, performance, and simplicity. Fifty-three candidate solutions (n = 29 per species) were evaluated using four different sets of environmental predictors and seventeen regularization multiplier values (RM: 0.5–8.0). Parameter optimization was achieved by testing various regularization multipliers (0.1 to 1, at intervals of 0.1, and 1 to 4, at intervals of 1) alongside multiple feature types (linear, quadratic, product, and threshold) [91]. The best solutions were selected based on low omission error (<0.05) and low Akaike information criterion (AICc) values [92,93,94,95]. Model performance was assessed using omission error [96], the area under the curve (AUC) [97,98], a partial ROC test [99], and true skill statistics (TSS) [100] (see Table S3). The models were trained using 70% of the marine occurrence data and validated with the remaining 30%, thus representing interpolation performance, as both datasets originate from the same spatial extent. Despite being non-overlapping, the validation subset shares the spatial domain with the training data. To reduce the uncertainty associated with random data partitioning, 100 replicates were generated for each species, region, and modeling method. Model transferability was evaluated using independent datasets. All models were fitted with the full set of explanatory variables without applying variable selection procedures.

The final models were then transferred to future environmental conditions [95,101,102]. For MaxEnt projections, the present study allowed for both “unconstrained extrapolation” and “extrapolation by clamping”, which manage predictions outside the training data range in different ways. “Unconstrained extrapolation” enables model predictions to be made beyond observed environmental conditions, while “extrapolation by clamping” restricts predictions to conditions similar to the training data, thus minimizing unrealistic future projections [88,92]. Finally, potential habitat suitability maps were generated using the raster package [103] in R [74]. Continuous probability maps (0–1) were transformed into binary presence–absence maps using a threshold that facilitated a 10% omission of training occurrence records, corresponding to a 5% false-negative rate. This approach minimized commission errors (false positives), thus yielding more conservative estimates of species distributions [104].

3. Results

3.1. Model Evaluation and Variable Contributions

The present study obtained a total of 504,939 occurrence records for the 92 melliferous plant species selected, a number which, after data cleaning and thinning, was reduced to 6733 unique occurrence records used for model building (Table S1). The average number of occurrence records obtained per species was 77, with a range of 10 to 442. The mean AUC of the 92 melliferous plant species was 0.968 (±0.035 SD), indicating that all models were valid under the given parameter setting conditions (Table S3). Under future climate scenarios, the AUC values ranged from 0.927 to 0.992 (SSP245) and 0.919 to 0.991 (SSP585) (Table S3). The models exhibited good performance, with significant values in both the pROC test (1.62 to 1.99, p < 0.05) and TSS (0.46 ≤ TSS ≤ 0.51). Among the bioclimatic variables selected, the minimum temperature of the coldest month, the annual temperature range, and the precipitation of the driest month were the most important in terms of their percentage contribution (Table S2). The variables with the highest permutation importance were temperature seasonality, the maximum temperature of the warmest month, and the minimum temperature of the coldest month (Table S2).

3.2. Species Richness Patterns

The species richness map for current environmental conditions predicted the highest level of melliferous plant species richness on the north coast, west, and south of the Yucatan Peninsula, with a richness of up to 75 species (Figure 2A). The lowest species richness (≤17 species) was observed in small portions of the northern and southwestern region of the peninsula (Figure 2A). Comparing the species richness obtained for future scenarios to that under current conditions revealed a potential range increase in species richness (~82 species), expanding to cover most of the Yucatan Peninsula and concentrating, under the SSP245 scenario, mainly on the northern coast and the central and eastern regions (Figure 2B). Under the SSP585 scenario, an increase in species richness (~80 species) was observed compared to the current scenario, with the highest level of richness concentrated along the northern and southeastern coasts and in the south-central region of the peninsula (Figure 2C).

3.3. Changes to Suitable Areas

Under present conditions, the habitat with a suitable climate for the 92 melliferous plant species varies in area from 8817 to 151,515 km², resulting in an average of 95,319 km² (

Table A1. Summary of changes in the range size of climatic suitability of melliferous plant species in the Yucatan Peninsula, Mexico. See the Section 2 for abbreviation definitions.

Species	Current (km2)	SSP 245			SSP 585
		Gain (km2)	Stable (km2)	Loss (km2)	Gain (km2)	Stable (km2)	Loss (km2)
Acacia angustissima (Mill.) Kuntze	66,481.3	58,901.5	51,519	1428.6	7024	730.9	65,750.3
Aeschynomene americana L.	142,631	9019	142,631		149	783	141,848
Albizia lebbeck (L.) Benth.	117,920	28,572	115,481	2439	30159	45,282	72,638
Albizia tomentosa (Micheli) Standl.	131,742	18,508	130,741	1001	6797	4109	127,633
Allophylus cominia (L.) Sw.	123,438	16,014	118,952	4486	18,064.2	9629.1	113,808.8
Ambrosia hispida Pursh	21,294.3	11,137	20,611.3	683	9144	18,617	2677.3
Azadirachta indica A. Juss.	8817	16,823.4	6952.3	1911.6	35,049.6	8746.4	70.5
Bauhinia divaricata L.	124,134.3	7965	85,318	38,816.3	17,862.5	90,898.5	33,235.7
Bauhinia herrerae (Britton & Rose) Standl. & Stayer.	119,738	31,122.7	119,501.5	236.5	10,174.9	99,149	20,589
Bignonia diversifolia Kunth	116,774.6	27,065	114,531.7	2242.9	57,733.3	24,263.6	42,653
Brosimum alicastrum Sw.	105,793.7	27,939	92,726.8	13,066.9		23,963.8	81,829.9
Bursera simaruba (L.) Sarg.	148,847.8		136,326.5	12,521.2		25,759	123,088.8
Byrsonima bucidifolia Standl.	114,188	25,182.5	113,554	634	6985.6	71,293.9	42,894.1
Cassia fistula L.	62,293.7	73,429	35,118.6	27,175.1		18,881.2	43,412.5
Cedrela odorata L.	112,162	30,984	110,325	1837	30,729	103,998.7	8163.2
Ceiba pentandra (L.) Gaertn.	66,527.3	55,242	63,564.5	2962.8	72,951	57,261	9266.3
Chrysophyllum mexicanum Brandegee ex Standl.	145,319	5061	144,767	552	2212	33,359.3	111,959.6
Cissus verticillata (L.) Nicolson & C.E. Jarvis	139,576.5	10,936.5	139,457.5	119	11,271	36,321	103,255.5
Cnidoscolus souzae McVaugh	98,524	32,829	63242.8	7373.1	11,560	2495	96,029
Coccoloba uvifera (L.) L.	16,195.9	8615	9269.6	6926.3	15,016.2	8296	7899.9
Cordia gerascanthus L.	62,971.6	34,792.4	62,971.6		25,466.7	3367.3	59,604.3
Cordia sebestena L.	13,834.6	22,198	13,463.3	371.3	58,765	234.7	135,99.9
Cornutia pyramidata L.	75,611.7	38,466.3	75,611.7		15,953	5447.3	17,601.3
Crescentia cujete L.	78,058.6	73,848.4	77,645.6	413	29,260.3	1666	75,392.6
Croton glabellus L.	123,951	10,799	116,587	7364	16,021	102,239	21,712
Croton icche Lundell	82,226	52,095	82,226		35,647	44,833	37,393
Croton reflexifolius Kunth	140,378	10,919	139,138	1240	731	5502	134,876
Dalbergia glabra (Mill.) Standl.	123,296	23,115.9	120,489	2837	19,178	77,365.4	45,960.5
Diospyros cuneata Standl.	113,233	38,282	113,233		5203	343	112,890
Distimake dissectus (Jacq.) A.R. Simoes & Staples	32,529.33	113,064.6	32,529.3		7474		32,529.3
Ehretia tinifolia L.	82,306	20,026	77,745.7	4560.3	64,447	32,086	50,220
Enterolobium cyclocarpum (Jacq.) Griseb.	86,486.8	29,745.2	85,210.6	1276.2	52,567.4	39,012.8	47,474
Erythrostemon yucatanensis (Greenm.) Gagnon & G.P. Lewis	92,816	7042	66,018	26,798	4377	15,112	77,704
Eugenia axillaris (Sw.) Willd.	98,700.5	21,322	90,414	8286.5	33,362	81,963.2	16,737.3
Eugenia winzerlingii Standl.	81,167	33,564	79,111	2056	26,099	55,015	261,52
Exothea diphylla (Standl.) Lundell	128,711	6083	124,393	4318	22,468	63,964	64,747
Gliricidia sepium (Jacq.) Kunth ex Walp.	138,147.2	12,063	135,886.9	2260.3	15,816	102,432.2	35,715
Gouania lupuloides (L.) Urb.	72,115	68,242.1	72,115		25,012.4	604	71,511
Guazuma ulmifolia Lam.	100,702	43,996	99,131	1571	44,109.2	68,540	32,162
Guettarda combsii Urb.	68,918	12,264	67,543.5	1374.5	47,996	35,977	32,941
Gymnanthes lucida Sw.	37,411	63,537	35,721.3	1687.7	66,937.7	3500.9	33,908.1
Gymnopodium floribundum Rolfe	150,491		89,456.8	61,034.2		91,218	59,273
Haematoxylum campechianum L.	151,515		66,816.3	846,98.6		28,762.3	122,752.6
Hampea trilobata Standl.	117,743	27,474	113,766	3977	22,019.7	103,206.8	14,536.2
Havardia platyloba (Berteru ex DC.) Britton & Rose	77,717	37,349	76,753	964	14,540.2	73,019	4698
Ipomoea carnea Jacq.	100,238	86,325	57,409	1765	22,384	16,682	83,556
Ipomoea corymbosa (L.) Roth ex Roem.	148,119.8	2553.1	49,132.3	98,987.5	3074	42,815	105,304.8
Ipomoea crinicalyx S. Moore	81,602	60,226.4	81,504.9	97.1	305,66	11,654.8	69,947.2
Jacquemontia pentantha G. Don	133,249	16,042.5	56,030	77,219	3001.9	5022.8	128,226.1
Lonchocarpus hondurensis Benth.	129,623.2	21,891.8	129,063.2	560	20,596.8	94,139.8	35,483.4
Lonchocarpus punctatus Kunth	94,271.9	52,070.4	92,762.1	1509.8	5424	3465.7	90,806.2
Lonchocarpus rugosus Benth.	119,069	14,516	105,070	13,999	18,352	17,533	101,536
Luehea speciosa Willd.	76,503.6	63,586.3	76,218.6	285	29,343	4793	71,710.6
Lysiloma latisiliquum (L.) Benth.	115,431	26,561	111,248.5	4182.5	28,908	59,795	55,636
Machaonia lindeniana Baill.	61,104	57,093	60,938.3	165.7	14,383.6	27,786	33,318
Metopium brownei (Jacq.) Urb.	93,537	52,532	93,537		567,03	81,602	11,935
Milleria quinqueflora L.	85,735	19,948	81,221	4514	26,683	31,279	54,456
Mimosa bahamensis Benth.	118,513	23,356	116,207	2306	23,020	20,635.3	97,877.7
Mimosa pudica L.	53,602.2	69,192.1	53,533.8	68.4	9544.4	36,483	17,122.2
Murraya paniculata (L.) Jack.	95,700	50,273	80,177.8	15,522.2	48,979	20,640.4	75,059.5
Neomillspaughia emarginata (H. Gross) S.F. Blake	113,599	35,597	113,599		7946	1732	111,867
Operculina pinnatifida (Kunth) O’Donell	48,485	88,769	48,185		33,399		48,185
Phithecellobium albicans (Kunth) Benth.	75,761	36,277	75,761		38,658	18,194	57,567
Phyllanthus brasiliensis (Aubl.) Poir.	64,497	62,601	63,008	1489	45,412	5163.4	59,333.6
Piscidia piscipula (L.) Sarg.	148,189	94.8	96,604.9	51584.1		88,600	59,589
Pisonia aculeata L.	103,334	44,300	103,334		40,688	57,174	46,160
Platymiscium yucatanum Standl.	68,792	81,759	68,792		11,686		68,792
Pluchea carolinensis (Jacq.) G. Don	103,321.3	46,579.7	103,115.3	206	33,222	39,947.3	63,373.9
Polanisia viscosa (L.) DC.	23,098.8	3869.3	21,901.8	1197	27,041	23,098.8
Poutenia campechiana (Kunth) Baehni	83,002	17,656	77,404	5598	13,337	66,009	16,993
Pseudobombax ellipticum (Kunth) Dugan	113,210.7	36,320	112,188.7	1022	37,301.1	82,102.7	31,108
Psidium sartorianum (O. Berg.) Nied.	110,481.8	41,033.2	110,481.8		19,056	7508	102,973.8
Ruellia ciliatiflora Hook.	120,124	27,347	118,760	1364	11,850	14,607	116,989
Ruellia inundata Kunth	81,718	14,743	72,183	9555	42,245	7391	74,327
Sabal mexicana Mart.	117,060	32,087	114,747	2313	29,115	12,424.6	104,635.4
Sabal yapa C. Wright ex Becc.	80,422	35,947.3	75,432	4990	51,420.3	75,143.4	5278.5
Scaevola plumieri (L.) Vahl	129,971	19,963	119,785	10,186	6073	79,003.6	50,967.3
Schoepfia schreberi J.F. Gmel.	128,529.6	22,985.4	128,529.6		6962.9	8015	120,514.6
Senegalia gaumeri (S.F. Blake)	74,436	69,979	74,436		65,240	22,259	52,177
Serjania yucatanensis Standl.	56,542.7	30,704	37,999.7	18,543	46,489	33,879.7	22,663
Sideroxylon salicifolium (L.) Lam	77,452.9	31,132.8	76,416.9	1036	37,146.4	56,890.7	20,562.2
Stachytarpheta jamaicensis (L.) Vahl	93,882.7	42,750.6	85,475.8	26.7	44,744	25,607.7	68,275
Swartzia cubensii (Britton & P. Wilson) Standl.	20,381	48,943.3	15,490.6	4890.4	85,690.2	17,374.8	34.8
Tabebuia rosea (Bertol.) DC.	76,924	72,664	75,968.8	955.2	54,278.7	64,167.3	12,756.7
Talisia floresii Standl.	107,574	34,574	107,064	510	33,941	20,439	87,135
Tecoma stans (L.) Juss. ex Kunth	61,207.5	31,654.2	52,989.5	8218	58,342	19,471.5	41,736
Terminalia buceras (L.) C. Wright	144,004.9		61,340.9	82,664		74,539	69,465.9
Thouinia paucidentata Radlk.	101,792.3	27,489	94,047.3	7745	26,717	73,975	27,817.3
Trema micranthum (L.) Blume	62,962.7	40,998.9	61,478	1484.7	58,796.3	61,907.7	1055
Verbesina gigantea Jacq.	62,962.7	40,998.9	61,478	1484.7	58,796.3	61,907.7	1055
Viguiera dentata (Cav.) Spreng.	149,974		114,684.7	35,289.2		79,225	40,934.6
Vitex gaumeri Greenm.	145,920.7		23,377.1	122,543	51.3	77,294.8	68,625.9

, Figure 3 and Figure 4). Under the SSP245 scenario, the mean gain in suitable habitat was 35,251.3 km², with the suitable habitat for 32 species increasing by more than 50% (Table A1, Figure 3 and Figure 4). The SSP245 scenario indicated a mean area of stable suitable habitat of 84,116.1 km², with more than 80% of the current suitable habitat cover retained for 74 species (Table A1, Figure 3 and Figure 4). This scenario revealed a 28,130.2 km² mean decrease in suitable habitat, with the suitable habitat for only 20 species decreasing by more than 10% and decreasing by less than 10% for 72 species (Table A1, Figure 3 and Figure 4). For the SSP585 scenario, the mean gain in suitable habitat was 28,130.2 km², with the suitable habitat for 23 species increasing by more than 50% (Table A1, Figure 3 and Figure 4). The SSP585 scenario indicated a mean area of stable suitable habitat that significantly decreased to 38,973.3 km², and more than 80% of the current suitable habitat was maintained for only 16 species (Table A1, Figure 3 and Figure 4). Contrary to the SSP245 scenario, a mean suitable habitat area of 56,875.9 km² was lost for SSP585, corresponding to a significant increase, in contrast to that seen for SSP245, while the suitable habitat for 84 species decreased by more than 10% and by less than 10% for only 8 species (Table A1, Figure 3 and Figure 4).

3.4. Habitat Suitability of Vulnerable Melliferous Plant Species

Of the 92 melliferous species selected, 3 are currently on the IUCN Red List [105], with Macawood (Platymiscium yucatanum) and Coloc (Talisia floresii) listed as Near Threatened and the Black olive tree (Terminalia buceras) as Endangered. The current suitable habitat for P. yucatanum measured 68,792 km² and was mainly distributed along the northern coast and central and southeastern regions of the Yucatan Peninsula (Figure 5). Under the SSP245 scenario, the entire suitable habitat area was projected to remain stable up to the year 2060, while an area of more than 80,000 km² was gained, covering most of the Yucatan Peninsula, especially in the northern, southwestern, and southeastern regions (Figure 5). However, under SSP585 conditions, the entire projected suitable habitat was lost, with gains of only 11,000 km² in the southwestern part of the peninsula, corresponding to a 92% reduction in area relative to the current conditions and an 83% reduction relative to the SSP245 scenario conditions (Figure 5). For T. floresii, the current distribution area measured 107,574 km² and was mainly distributed along the northern coast and the central-eastern region of the Yucatan Peninsula (Figure 5). Under the SSP245 scenario, 99% of the suitable habitat area remained stable, with gains of more than 34,000 km² that covered most of the Yucatan Peninsula, mainly in the northern, western, and southern regions (Figure 5). In contrast, under SSP585 conditions, 81% of the current suitable habitat was lost, while over 33,000 km² was gained, with suitable habitat remaining solely in the northern and southwestern regions of the peninsula. This corresponded to a 49% reduction in total suitable habitat area relative to the current conditions and was 61% lower than that obtained in the SSP245 scenario (Figure 5). The current distribution of T. buceras measured 144,004.9 km², which was mainly found along the western coast and in the southern region of the Yucatan Peninsula (Figure 5). Under both climate change scenarios, close to half of the current distribution was lost (42.5% under SSP245; 51% under SSP585) while no gains in area were obtained, with suitable habitat areas remaining solely in the southern region of the peninsula (Figure 5).

4. Discussion

Climate change affects organisms in diverse ways [106,107]. Some species move to areas with similar climate conditions, while others remain in their current locations and either adapt or face the risk of extinction. Even among species that change their distribution ranges, responses to these conditions can vary—some ranges may expand, contract, or remain unchanged in size [108]. The results obtained by the present study reveal that both the number of species facing habitat loss and the extent of that loss in the future are low under the SSP245 scenario, corresponding to 37 species losing habitat and a lost area of x̄ = 12,362.8 km². Dramatically higher results were obtained for these metrics under SSP585 conditions, corresponding to 75 species losing habitat and a lost area of x̄ = 56,875.9 km². Contrary to expectations of a marked decline, the overall species richness increased slightly—by 10% under SSP245 and 5% under SSP585—compared to current conditions. However, changes in the spatial patterns of species distribution were observed. Under the current scenario, the areas of highest richness were concentrated on the northern coast and the western and central-southern regions of the peninsula, while under the SSP245 scenario, these areas of species richness expanded to encompass almost the entire peninsula, covering more areas of the northern coast and the eastern and central regions. This pattern changed drastically under the SSP585 scenario, wherein the areas of highest species richness concentration were significantly smaller, remaining only along the northern and southeastern coasts, while areas with low species richness increased.

Furthermore, the findings obtained suggest that the average change in habitat shifts for most of the species of interest will, in the future, be negative. Under SSP245, richness expanded broadly across the region; however, the SSP585 projections revealed a contraction of high-richness zones and an expansion of areas with low richness. These results partially contrast with previous SDM-based studies. For instance, Cayuela et al. [109] and Crimmins et al. [110] reported consistent declines in both plant richness and range contractions in response to warming conditions, particularly for species with narrow climatic niches. Similarly, Tennakoon et al. [111] and Rahimi and Jung [112] identified reduced habitat suitability and altered flowering phenology as major constraints on pollinator–plant dynamics under high-emissions scenarios. Hulme [113] emphasized that species with phenological sensitivity to temperature shifts may experience spatial–temporal mismatches, leading to functional biodiversity loss despite minor changes in numerical richness. The slight richness increase observed in the present study may reflect model artifacts or transient redistribution effects rather than true gains in ecological function. This underscores the importance of incorporating phenological data and species–specific functional traits in SDMs to more accurately assess biodiversity responses to climate change, particularly for species with key ecological roles, such as melliferous plants. According to Escobar-Lujan et al. [114], species richness decreases in areas with potentially higher increases in temperature and sea level, with similar results obtained by other studies [115,116]. For instance, predicting a species’ future distribution requires consideration of both its capacity to colonize new areas and its ability to persist in existing locations or avoid localized extinction. The area of suitable habitat for many melliferous plant species decreased significantly under the SSP585 scenario. This was exemplified by the results obtained under SSP585 conditions for the three species listed in IUCN risk categories [105], which showed an overall reduction in suitable habitat area by ~70% of the current conditions and the complete disappearance of some projected areas with discontinuous suitable habitat. Furthermore, tree species with limited distributions or small, fragmented populations are more vulnerable to the impacts of climate change [117,118].

The findings of the present study, which indicate an expansion of suitable habitat under an optimistic climate change scenario but a significant decrease under a pessimistic scenario, are consistent with those found by previous studies [119,120], including for other taxa in the same region [114]. The varied responses of plant species to climate stress can be attributed to differences in functional traits, dispersal capacity, edaphic specialization, and species–specific vulnerability. Traits such as drought tolerance, phenological plasticity, and reproductive strategy influence resilience to environmental change. Dispersal limitations may restrict range shifts, while dependence on specific soil types constrains habitat suitability under shifting climates. Additionally, sensitivity to biotic interactions, pathogens, or habitat fragmentation can exacerbate vulnerability, resulting in heterogeneous responses across taxa. In general, while the ecological responses and behaviors of plant species will vary based on their environmental needs, future climate conditions are expected to drastically alter their geographic distributions and vegetation structures. Some species may disperse to new habitats, others may adapt to changing conditions, while some could face extinction [120,121,122].

Among the 19 bioclimatic variables used in the species distribution models, the minimum temperature of the coldest month, the annual temperature range, and the precipitation of the driest month made the strongest contributions for the melliferous plant species of interest. The variables associated with extreme environmental conditions were also found to be crucial in explaining the distribution of the melliferous plant species. Precipitation, closely linked to both the rainy season and temperature, plays a key role in shaping species distribution. These climatic factors help assess the influence of environmental variables across a region, offering insights into the suitable habitats for each species [123]. Similarly, temperature–related bioclimatic variables are useful in defining species distributions, especially in tropical areas [124,125]. Beyond climate variables, environmental factors, such as distance from water sources, pollinator numbers, and human activity, can also influence melliferous plant species distribution [126,127]. While the results obtained in the present study based on these variables are both interesting and informative, they should be interpreted with caution. However, research on biodiversity and climate modeling in the Neotropics reveals critical limitations in the representation and predictive accuracy of tropical SDMs. Bellard et al. [19,128] emphasized that tropical regions, despite hosting the highest levels of global biodiversity, remain underrepresented in climate impact assessments, leading to spatial biases in global projections. Hortal et al. [129,130] further identified significant knowledge shortfalls—including taxonomic, distributional, and ecological data gaps—that compromise the development of reliable SDMs for tropical taxa, particularly plants with specialized mutualisms, such as those reliant on honeybee pollination. These studies highlight the urgency of improving data completeness and model resolution in the Neotropics, where species exhibit narrow climatic tolerances and complex biotic interactions.

Integrating high–quality occurrence data, functional traits, and biotic dependencies into SDMs is essential for enhancing the robustness of biodiversity forecasts. Such improvements are pivotal for informing conservation strategies under climate change conditions, particularly in biologically rich yet data-deficient tropical systems. Species distributions are influenced by factors other than the climatic variables that the models used here accounted for, which should still be considered [33,131]. These factors include ecological processes, such as density–dependent interactions and population dynamics [132,133], as well as historical processes [134,135]. Additionally, when climate change surpasses natural variability, species may employ various compensatory mechanisms, such as adjusting their physiological patterns [136]. Plant species can alter their growing seasons and shift their geographic ranges to access wetter conditions, particularly those with stringent climatic and habitat requirements [137,138]. Coupled with the impact of restricted species migration capacity and habitat fragmentation, this limits the accuracy of SDM predictions. These results align with projections that some species will shift toward higher altitudes and latitudes in response to future climate warming [136]. However, a species’ distribution is strongly constrained by dispersal capacity, and SDM outputs are highly dependent on assumptions regarding dispersal limitations and ecological interactions [33,139].

The impact of local extinction can affect the entire range of species, where those with life histories more prone to population extinction typically occupy a smaller proportion of their potential suitable habitats [121,140]. In melliferous plant species, their limited habitat extent is associated with increased extinction risk, which could lead to the loss of their pollinators [141,142]. Recent reports of declining wild bee populations are concerning because of their potential implications, where reductions in bee diversity—whether through increased rarity or extinction—pose significant risks to the pollination of wild flora and agricultural crops, with broad ecological and economic ramifications [142]. Consequently, developing conservation strategies that mitigate the effects of climate change and other human-driven pressures on melliferous plant species is of the utmost importance [143]. One possible course of action is to intentionally plant these species in areas that have recently become or are projected to soon become climatically suitable [144,145]. Our results suggest that the northern and southeastern coasts and the south-central region of the Yucatan Peninsula are critical points for conservation; however, it is also essential to assess their suitability for the species. In the face of changing conditions, species will experience selection pressure to either adapt to a new environment or migrate to different habitats. However, without the mitigation of barriers and habitat degradation, these species are likely to face extinction or local extirpation [146].

While climate change is a major threat to biodiversity because it increases extinction risk, restoring degraded habitats, improving landscape connectivity, protecting critical areas, and facilitating species introduction into suitable new habitats can help mitigate some of the adverse impacts of climate change on melliferous plants. In southeastern Mexico, shifts in the distribution and abundance of native honey-producing plants—driven by land use change and climate variability—pose significant ecological and socioeconomic challenges. These species are essential for sustaining native pollinator networks and local apiculture–based economies; however, they are increasingly threatened by deforestation, agricultural expansion, and altered precipitation regimes [147,148,149]. The decline in melliferous flora directly affects bee populations and traditional livelihoods, exacerbating rural vulnerability [147,149]. Despite their wide use in predicting potential shifts in species ranges under climate change conditions, recent meta-analyses have highlighted both their strengths and limitations. Thuiller et al. [150] emphasized that although SDMs effectively delineate climatic suitability at a broad spatial scale, they are constrained by key assumptions, including species–climate equilibrium, niche conservatism, and the exclusion of biotic interactions, dispersal barriers, and evolutionary responses. Briscoe et al. [151] further demonstrated that SDM performance is context–dependent, with significant variation across taxa, modeling algorithms, and geographic regions. Notably, projections frequently overestimate potential range expansion, particularly under high–emissions scenarios, due to the limited integration of ecological and demographic processes, thus explaining the need to enhance SDMs through hybrid approaches incorporating mechanistic data, trait–based ecology, and land use dynamics [150,151]. SDMs offer a robust framework for identifying climatically suitable habitats under current and future scenarios, guiding conservation strategies aimed at preserving plant–pollinator dynamics. Through the application of predictive spatial analyses, SDMs can inform the design of ecological corridors and the prioritization of areas for habitat restoration or the assisted migration of key species. Incorporating socioeconomic data into niche–based conservation planning enhances the alignment of ecological priorities with community-based management practices. This integrative approach is critical for maintaining biodiversity and the cultural and economic resilience of apiculture in the region. These projections are valuable for the conservation and sustainable management of species in the face of climate change [152]. Additionally, it is crucial to note that using SDMs enhances our understanding of the threats facing plant species, their potential future distributions, and patterns of species richness and community composition. These models are highly valuable in shaping conservation and management strategies and priorities [153], and their insights are essential for refining conservation planning and improving the reliability of biodiversity forecasts in the face of accelerating climate change.

5. Conclusions

Climate change is altering weather patterns on a global scale, leading to changes in resource availability, temperature, and precipitation. These alterations lead to the displacement, contraction, or loss of habitat for numerous species, affecting their geographic distribution and population viability. As a result, critical conservation hotspots are being identified in locations where high biodiversity, endemism, and increasing ecological vulnerability converge. This situation requires the design and implementation of comprehensive policy actions, including adaptation and mitigation plans based on scientific evidence, the incorporation of ecological connectivity into territorial planning, and the strengthening of both in situ and ex situ conservation strategies under future climate scenarios.

The results obtained in the present study show that bioclimatic variables related to temperature and precipitation significantly influence the distribution of suitable habitat for melliferous plant species. The projections performed in the present study indicate that climate change will alter habitat suitability for most of the 92 species of interest, resulting in a slight decline in species richness across the Yucatan Peninsula by the 2050s. These findings underscore the potential impact of climate change on the distribution and conservation of melliferous flora in southeastern Mexico. However, SDMs have several limitations inherent to their assumptions and input data. One of the main limitations is the assumption of equilibrium between the current distribution of species and environmental conditions, which may not be met in dynamic scenarios or for species with limited dispersal capacity. Furthermore, many SDMs assume unlimited dispersal, ignoring geographic, biological, or anthropogenic barriers that restrict potential expansion. Another key limitation is the omission of relevant non–climatic variables, such as land use, edaphic factors, or biotic interactions, which can significantly influence the actual distribution of species. These simplifications can lead to overestimating or underestimating suitable habitat areas, compromising the accuracy of predictions under environmental change scenarios.

The present study provides essential information for conservation planning, offering projections of future distributions of melliferous plant species under climate change scenarios and identifying habitat shifts and species vulnerabilities in the region. Restoration ecology should focus on identifying and implementing realistic strategies that help plant species adapt to new climatic conditions. The findings obtained in the present study are crucial for developing adaptation strategies and management plans for endangered trees and the reforestation of areas near the core zones designated for specific species. While these results offer meaningful insights for developing more effective evidence-based conservation strategies, other factors, such as climate change adaptability and diffusivity, should also be considered.

To mitigate the effects of climate warming in regions such as the Yucatan Peninsula, it is essential to reduce human pressures derived from land use. Various strategies should be implemented, such as educating stakeholders, designating protected rivers, removing barriers, and enhancing riparian shading. Current and projected species richness, derived from species distribution models (SDMs), can inform the strategic selection of protected areas in the Yucatan Peninsula. This approach is critical for conserving tropical plant diversity and enhancing ecosystem resilience to climate change while sustaining essential ecosystem services. When assessing community dynamics and trophic interactions, it is important to consider both regional spatial extents and local conditions. While climate change poses a significant threat to biodiversity, other anthropogenic pressures must also be addressed to ensure successful biodiversity conservation.