Monsoon Climate and Anthropogenic Influences Shape Primate Distributions Across the Southeastern Edge of the Qinghai-Tibet Plateau
Chen Li, Yi-hao Fang, Guo-peng Ren, and Yin Yang are joint first authors.
Wen Xiao, Rui-dong Wu, and Yin Yang are joint senior authors.
Funding: This work was supported by the National Natural Science Foundation of China (32471737, 32260251, 32371567, and 32360137), the Natural Science Foundation of Yunnan Province (202401BF070001-008 and 202301AT070247), and the Talent and Platform Project within the Yunnan Provincial Science and Technology Department (202205AM070007, 202105AM070008, YNWR-CYJS-2018-052, and YNWR-QNBJ-2019-262).
ABSTRACT
The southeastern edge of the Qinghai-Tibet Plateau (Yunnan, China) exhibits high biodiversity but stark differences in species richness between its western Longitudinal Range Gorge (LRG) and eastern Yunnan Plateau (YP). We collected distribution data for 16 primate species in Yunnan and analyzed palynological records over the past 20 ka from 21 localities to identify the biogeographic, climatic, and anthropogenic factors that have driven the present-day distribution of primates in this region. By integrating local ecological knowledge, field surveys, species distribution models, niche utilization rates, and historical vegetation and land use changes, we found that spatial–temporal shifts in the monsoon climate have been a critical factor in shaping primate species richness on the southeastern edge of the Qinghai-Tibet Plateau. Compared to the YP, the LRG receives more precipitation, has more limited seasonal temperature variation, and has higher minimum temperatures during the coldest month. These conditions have facilitated the development of moist evergreen broadleaf forests, which represent a more suitable habitat for the 14 primate species that inhabit this area. In contrast, the drought-adapted forests of the YP support only one primate species. Palynological records indicate that the differentiation of the LRG and YP predates human influence. However, over the past 2000 years, anthropogenic habitat loss and hunting have significantly affected the distribution of primates. The ranges of gibbons, langurs, and snub-nosed monkeys are now restricted to the central and northern regions of the LRG and have disappeared from lower elevations. Lorises have disappeared from their northernmost range. In contrast, the distribution of macaques has remained relatively stable. The Yangtze-Red River-24° N line marks the biogeographic boundary of high primate species richness and biodiversity in the LRG and southeastern Yunnan. Our research suggests that changes in monsoon climate have fundamentally shaped contemporary species richness, while recent anthropogenic pressures have caused ‘range contraction’ for many taxa.
1 Introduction
Large-scale patterns of animal biogeography and species richness have been shaped by species evolutionary history, ecological barriers to dispersal, the presence of refugia (Anthony et al. 2007; Wiens and Donoghue 2004; Beaudrot and Marshall 2011), interspecific competition, environmental heterogeneity (Stein et al. 2014), and climate change (Davies et al. 2011; Svenning et al. 2015; McCain et al. 2018; Skeels et al. 2023). In the case of climate, temperature and precipitation are among the most important factors that affect the ability of species to expand their distribution and exploit newly emerging habitats (Hawkins et al. 2003; Evans et al. 2005; Reed and Fleagle 1995; Torres-Romero and Olalla-Tárraga 2015). For example, regions with the highest biodiversity globally are primarily concentrated in warm and rainfall-rich tropical biomes (Rahbek et al. 2019). In recent decades, the ever-increasing human impact on natural landscapes, including deforestation, urbanization, industrial agriculture, and the burning of fossil fuels, has altered climate stability, resulting in what has been described as the world's 6th major extinction event (Sebastián-González et al. 2019; Stobo-Wilson et al. 2020; Storch et al. 2022). In the case of South China, intensified human disturbance combined with temperature changes have led to the local extinction of 11 species of large mammals, including elephants, rhinos, pandas, and water deer over the past 300 years (Wan et al. 2019). For primates, 80% of China's 26 existing species face extinction threats, including 15 to 18 species with wild populations of fewer than 3000 individuals. In addition, two species of gibbons have been extirpated from China in the past few recent decades (Grueter et al. 2009; Fan et al. 2014). Projections indicate that in the absence of strict conservation measures, China could see an additional 51 to 87% decline in primate range distributions by 2100 due to agricultural expansion (Li, Li, et al. 2018).
Located on the southeastern margin of the Qinghai-Tibet Plateau (hereafter referred to as QTP), Yunnan Province in China represents the convergence of three biodiversity hotspots: the Eastern Himalayas, the Southwest Mountains, and the Indo-Burma. Together, this region is the most biodiverse among all world regions at this same latitude (Myers et al. 2000; Yang et al. 2004). Occupying only 4.1% of China's total land area, Yunnan harbors more than 19,000 species of higher plants and over 2300 species of vertebrates, accounting for more than 50% of China's plant and vertebrate species. The areas of greatest biodiversity are located in Yunnan's western and southeastern regions (Qian et al. 2020; Wang et al. 2022; Yuan et al. 2022; Li and Pimm 2016; Yang et al. 2025), with the Yangtze River–Red River–24° N as a possible biogeographic boundary to faunal exchange. This boundary may have originated (i) in response to the formation of different monsoon climates along the Jinsha River and the Red River during the Pleistocene (~2–1 Ma), resulting from geological uplifting of the Yunnan Plateau (hereafter referred to as YP) in the late Pliocene (~3 Ma) and large-scale differential uplifting of mountains in the longitudinal range gorge (hereafter referred to as LRG) approximately 4–1.6 Ma (Fan et al. 2013; Huang et al. 2016; Zhu 2023) or (ii) in response to species population decline and extinctions due to human activities in the central YP over the past several thousand years (Maimaitiming et al. 2013; Zhang et al. 2021; Zhang et al. 2022; He et al. 2022). However, given a limited primate fossil record across Yunnan in response to its acidic soils, there is a limited fossil record from which to test these two alternatives (Hou et al. 2019; Table S1). Moreover, historical documents, such as chorographies and local gazetteers, often failed to distinguish between different but closely related animal species (e.g., apes, macaques, and langurs), resulting in taxonomic ambiguities that persisted until the early 20th century (Niu et al. 2022; Wen 2019; Yang 2021). Therefore, we examined geological and climatological changes that have occurred in Yunnan over the past 20,000 years in order to better understand the set of factors that have promoted and constrained biodiversity, in particular primate species diversity and distribution across the region.
1.1 Topography and Monsoon Climate Differentiation
The Yangtze River and the Red River divide Yunnan into two major regions: the YP in the east and the western pan three parallel rivers area, known as the Longitudinal Range Gorge (LRG) (Figure S2). The LRG features north–south trending mountains and rivers from northwestern Yunnan to Xishuangbanna. Yunnan supports a range of climate types including tropical, subtropical, temperate, and alpine, and a highly diverse flora. It is composed principally of mountainous terrain (94%) that ranges in elevation from 76 m in the southeast to 6740 m in the northwest (Yang et al. 2004). The geological history of the QTP is not well understood (Wang et al. 2014; Spicer et al. 2021). However, there is a consensus that uplifting of the Hengduan Mountains on the southeastern margin (in northwestern Yunnan, with an average elevation of 3000–4000 m) was a rapid and recent event, occurring primarily between the late Miocene and the late Pliocene (6–2.5 Ma) (Clark et al. 2004; Sun et al. 2011; Wang et al. 2012). Additionally, due to the uplifting of the QTP and its extrusion towards the southeast margin, the Sichuan-Yunnan block at the southeastern edge of the QTP was carried along with the Indochina block and rotated clockwise (Xu et al. 2003; Zhang and Li 2016; Li, Jiang, et al. 2018; Yu et al. 2022) (Figure S2). Consequently, the Sichuan-Yunnan block underwent uplifting during the late Pliocene to middle Pleistocene (3.6–0.12 Ma), contributing to the formation of the YGP, which is located alongside the Yangtze Block (with an average elevation of 1000-2000 m) (Wu et al. 2020; Zhang and Li 2016; Li, Jiang, et al. 2018). Moreover, since the early Miocene, right-lateral strike-slip shearing between the Yangtze Block and the Indochinese Block led to large-scale differential uplifting and erosion processes on the west side of the shear zone (the Honghe-Ailao Mountain fault zone). This formed a series of north–south mountains, such as the Cangshan Mountains (the Yunling Mountains' southern section) and the Ailao Mountains (with an average elevation of 2500–3500 m), some 4–1.6 Ma (Clark et al. 2004; Wang et al. 2006; Li et al. 2014).
Following the intensification of the monsoon climate due to the uplifting of the northwestern YP in the Miocene, the topographical differences between eastern and western Yunnan had a considerable impact on its modern climate (Yao et al. 2012; Huang et al. 2016). Specifically, since the Late Miocene, the temperature in northern Yunnan has dropped, whereas the temperature in southern Yunnan has remained largely stable. In both the south and north, summer precipitation has increased, while winter precipitation has decreased, leading to greater seasonal differences. The southwest monsoon, which originated from the equatorial Indian Ocean, and the East Asian monsoon, which originated from the Pacific Ocean, act to converge on the YGP, creating a distinct transition zone (Wang and Ho 2002). During the summer, the north–south mountains (ridges averaging 3000 m a.s.l.) in the LRG function as a barrier blocking the southwest monsoon from moving west to east. At the same time, the deep valleys allow warm and humid airflow from the Bay of Bengal and the South China Sea to pass through in a south–north direction and reach higher elevations in northwestern Yunnan (He et al. 2007; Pan et al. 2012) (Figure S2). This has resulted in the dispersal of southern, tropical plants and animals into the valleys of central and northwestern Yunnan along the LRG (corridor-barrier effect) (Yang et al. 2004). In eastern Yunnan, the YP (averaging 2000 m a.s.l.) impedes the East Asian monsoon from carrying humid air from the Pacific into central Yunnan (Yu et al. 2023). During the winter, Yunnan is primarily influenced by westerly winds, leading to sunny and warmer weather. When the southern branch of the west winds passes through the Bay of Bengal, it brings moisture to Yunnan from the southwest. Consequently, during the winter season, rainfall in western and southeastern Yunnan is greater (350 mm vs. 50 mm) than that in the central and northeastern regions (Zhou and Yan 2007; Duan et al. 2016; Yu et al. 2023). In the south, the valleys are wide, gently sloping, and low in elevation, facilitating the movement of water vapor during periods of rain. Therefore, annual precipitation tends to be higher in the west than in the east and in the south than in the north (Duan et al. 2016).
In the eastern region of Yunnan, when the Siberian high anomaly intensifies, strong cold air or cold waves (there is an average of 9.3 cold waves per year) can invade the YP. This results in significant oscillations in wind direction, humidity, and temperature between summer and winter in Yunnan's central and northeastern regions (Yan et al. 2023; Yang, Yang, et al. 2023). Historically, intensification and regional differences in the monsoon climate in Yunnan have led to the emergence of arid-adapted sclerophyllic evergreen broadleaf forests in northwestern and central Yunnan (Huang et al. 2016) and to the extinction of more humid-adapted plant species in the YP (Su et al. 2013; Jia et al. 2015; Wang et al. 2015; Zhang et al. 2015).
1.2 Primate Model and Research Hypotheses
Non-human primates (hereafter referred to as primates) are the third most speciose mammalian radiation (currently the IUCN recognizes 527 primate species) and are considered flagship species in many forest ecosystems (Estrada et al. 2017). Yunnan is home to 16 primate species from 9 genera and 4 families, which exploit all forest types ranging from an altitude of 76 m to 4600 m (Yang et al. 2004; Li, Li, et al. 2018). Yunnan has the highest primate diversity in China (Li et al. 2023; Table S2), with 14 species distributed to the east of the Yangtze and Red Rivers and only two primate species, the rhesus macaque (Macaca mulatta) and the Tibetan macaque (Macaca thibetana) present in the YP. The distribution of primate species in Yunnan strongly correlates with the distribution of other terrestrial vertebrate species diversity across the region (Yang et al. 2025). As a result, the primate assemblage can serve as a proxy for analyzing factors that underlie patterns of vertebrate species diversity across the region. This occurs in part because individual primate species exploit particular forest types, and thus collectively, primates exploit a broad set of ecological conditions. For example, the stump-tailed macaque (Macaca arctoides) lives only in tropical and subtropical humid evergreen broadleaf forests (Fooden et al. 1985; Gupta 2001), while gibbons (Hylobatidae spp.) do not occur in relatively dry, subtropical evergreen needleleaf forest or sclerophyll evergreen broadleaf forests (Li et al. 2023). Factors that affect primate distributions and species richness in Yunnan also affect the species distributions and richness of many other vertebrate taxa.
Since the precise distribution of individual primate species across Yunnan has not been well documented, we gathered data on the distribution of each of Yunnan's 16 primate species through literature reviews, local ecological knowledge, field investigations, camera trapping, and online interview surveys. In addition, we analyzed fossil pollen records for the past 20,000 years at 21 sites that varied in elevation and climatic conditions to reconstruct how historical changes in climate affected primate habitat types across Yunnan (e.g., Xiao et al. 2015; Cheng et al. 2024). Our goal was to use species stacking models and ArcGIS tools to determine (1) how primate biodiversity across the southeastern edge of the Qinghai-Tibet Plateau was historically shaped by changes in the monsoon climate (i.e., increased diversity in wetter and warmer regions) and (2) how human activities (i.e., infrastructure, habitat conversion, and hunting) over the past thousand years have further impacted primate species distributions and the threat of local extinctions.
2 Methods
2.1 Study Area
Yunnan is located from 21°8′N to 29°15′ N and from 97°31′ E to 106°11′E (Figure S2). The average annual precipitation is approximately 1100 mm (Duan et al. 2016). Annual average temperatures range between 15°C and 18°C (Yan et al. 2021).
2.2 Modeling Data Acquisition
2.2.1 Distribution Database of Non-Human Primates
We collected occurrence data for primates in Yunnan from 248 scientific publications, including peer-reviewed articles, technical reports, scholarly books, the GIBF database, and specimens at the Kunming Zoological Museum. Additionally, we gathered local ecological knowledge on primate distributions over the past 60 years by conducting 3D map-based interviews with local villagers and online surveys (e.g., Cui et al. 2016; Ren et al. 2017). Our team also recorded field data using camera trapping and transect sampling methods during the period from 2010 to 2023. All data acquisition procedures adhered to the methodologies outlined in the Supporting Information (Text S1). Ultimately, we collected 6065 data points on the distribution of the 16 primate species during the 1960s (n = 6065) and again in the present (n = 5410) (Figure S3). When populations disappeared, we documented the reasons for these disappearances through interviews and field investigations. We also extracted historical data on primate species distributions across Yunnan from the literature and noted explicitly stated reasons for population extirpation. In cases of unclear reasons, we used Google Earth Pro images to identify whether the habitats had been converted to farmland, residential areas, plantations, urban centers, or road and rail networks (Table S3).
2.2.2 Environmental Variables and Habitat Data
We collected environmental data using 48 monthly climate datasets, as well as elevational data sourced from the WorldClim global climate data platform (http://worldclim.org/). These datasets included average temperature (labeled as tavg1-12), precipitation (labeled as prec1-12), maximum temperature (labeled as tmax1-12), and minimum temperature (labeled as tmin1-12) for each month of the year. These datasets have a resolution of 30 s, equivalent to a spatial resolution of 1 km2.
Based on our field observations indicating that primates avoid pure eucalyptus, rubber, Chinese fir, and tea plantations, we excluded such artificial forests from our analysis (see Table S4). In addition, we leveraged land use data sourced from the Environmental Systems Research Institute (ESRI) via https://livingatlas.arcgis.com/andcoverexplorer, drawing upon the vegetation datasets developed by Tuanmu and Jetz (2014). These datasets are composed of 12 vegetation categories tailored for biodiversity and ecosystem modeling. We focused on data related to coniferous, evergreen broadleaf, deciduous broadleaf, and mixed forests, as these are the primary vegetation types utilized by primates (Table S4). Furthermore, recognizing the detrimental impact of major roads on adjacent forest cover, we incorporated OpenStreetMap (OSM) road data into our habitat suitability criteria, thereby enhancing the rigor and relevance of our habitat filtering process.
2.3 Modeling Techniques
2.3.1 Species Distribution Modeling
Properly generated pseudo-absence data can enhance the performance of species distribution models (SDMs) (Text S2). In order to address the lack of absent site data in Yunnan's primate distribution database, we utilized the bioclimatic envelope model to create species-specific climate zones, incorporating 48 months of climate data and altitude as environmental variables. Based on expert knowledge, closely related primate species with similar diets and ecologies were grouped together in the bioclimatic envelope models (Text S3). After implementing the bioclimatic envelope model for each group, we randomly selected pseudo-absences outside the optimal climate zone for each primate species within that group. These points were then incorporated into the selected models (Text S2), Random Forest down sampled (RFdown sampled), Boosted Regression Tree (BRT), and MaxEnt, for simulations. This drastically reduced the use of pseudo-absences as background data and thereby enhanced model efficiency.
To mitigate sampling bias and overfitting the RFdown sampled, BRT, and MaxEnt models, the R package spThin (Aiello-Lammens et al. 2015) was implemented to spatially spread out the distribution points, ensuring a minimum of 1 km of adjacent distance between each species' occurrence point (e.g., Yang, Chen, et al. 2023). For RFdown sampled, an equivalent number of pseudo-absences were selected matching the spatially thinned points (Valavi et al. 2022). Both the BRT and MaxEnt models used 2000 pseudo-absences. Eighty percent of the species' sampled points were randomly assigned to the training set, while the remaining 20% served as the validation set. The pseudo-absences, spatially thinned distribution points, climate data, and elevation were then integrated into the best-performing models of RFdown sampled, BRT, and MaxEnt for species distribution simulations (Valavi et al. 2022). The AUC (area under the curve) was used to evaluate the accuracy of the models, with values over 0.7 considered acceptable (Vale et al. 2014). The species' distribution predictions from all three models were averaged, and the potential habitat map was expressed as a binary variable (presence/absence) using the no-omission method. In SDMs, we refrained from excluding highly collinear climate factors, as our primate species analysis revealed that models incorporating these factors better reflected their actual distributions than those excluding them (Figure S4; Alves-Ferreira et al. 2024). An expert workshop was convened to select the simulation outcomes that best aligned with actual species distributions (Table S5).
2.3.2 Expert Knowledge Filters
As SDMs usually tend to overestimate the true biogeographic ranges of species, we employed expert knowledge-based filters to delineate habitats that are unsuitable for each species. These filters consider critical variables such as land use patterns, vegetation types, plantations, roads, and natural barriers (Table S6; Text S4). By applying these filters and exclusions, we created a habitat distribution map that closely reflected the actual habitats of each species. The clipping process was executed using R 4.1.1 (R Core Team 2021), and the specific expert knowledge filters for each species are detailed in Table S6. After generating the distribution maps of 16 primate species, the species richness pattern map was overlaid and created in ArcGIS 10.8.
2.4 Biogeographic Distribution Pattern Analysis
2.4.1 Testing Relationships Between Climate and Primate Species Richness
NUR is a measure of the proportion of area actually utilized by species across its climatically suitable regions and can be used to compare differences in ecological climate niches in areas with high and low primate species richness. A high value reflects the importance of the relevant variable in determining species richness. We also calculated the range in precipitation, temperature, and seasonality across 19 bioclimatic variables for known primate habitats in the central YP, where only the rhesus macaque (Macaca mulatta) resides. Subsequently, we compared and analyzed differences in major climate factors between regions that host at least two primate taxa other than macaques. This included regions with up to 6 primate species (≥ 3 taxa) and regions in central Yunnan that contained only M. mulatta. These data were analyzed using ArcGIS 10.8 and R 4.1.1 (R Core Team 2021).
2.4.2 Quaternary Habitat Identification
By searching Google Scholar and CNKI (a Chinese literature database), we found 26 paleopalynological studies from the past 30 years that documented a long-term sequence of fossil pollen assemblages. These assemblages occurred at elevations ranging from 600 to 3900 m a.s.l. across 21 different locations in Yunnan covering the past 20,000 years. These paleopalynological studies contain data on past plant species composition, vegetation types, and human activities resulting in habitat conversion and include interpretations of the paleoclimate and paleoenvironments during specific time periods (Xiao et al. 2015; Li, Liao, et al. 2024). From these publications, we extracted the vegetation types and their shifts at different locations every 1000 years as well as the timing of significant pollen changes indicating climate or human-driven impacts on vegetation. In this way, we were able to distinguish between climate- and human-induced changes in vegetation composition and track the location of primate habitats over the past 20,000 years. The accuracy of the extracted vegetation data was reviewed by phytogeographer Prof. Zhu Hua.
2.4.3 Species Distribution Changes Analysis
To explore changes in the distribution of primate species in Yunnan between the 1960s and 2020s, we utilized the weighted mean center and spatial statistics (standard deviation ellipse) in the Arcmap Spatial Statistics Toolbox (Scott and Janikas 2009). In this way, we generated the distribution centers and standard deviation ellipses for each of the five primate taxa (Lorisidae, Hylobatidae, Macaca, Trachypithecus, and Rhinopithecus) in the 1960s and in the 2020s. We recorded the coordinates, migration distance, and migration direction of the distribution centers (mean gravity center) at the beginning and the end of this 60-year period. If the standard deviation ellipse contracted, that indicated that the distributional range of the taxa had shrunk. The contraction or expansion of the standard deviation ellipse reflects the change in the spatial distribution and degree of clustering of primate populations. The migration direction and the distance from the gravity center of the distribution indicate the amount (km) and direction of distributional contraction due to the disappearance of local populations. We calculated the maximum, minimum, and median elevation (elevational center of gravity) by extracting the elevation data of distribution points and generating elevation box plots for the elevational distributional ranges of the five primate taxa and 16 primate species. Differences in elevational distribution between the 1960s and 2020s for the five primate taxa and the 16 primate species were assessed via Mann–Whitney U tests (with p < 0.05 being significant).
3 Results
3.1 Distribution of Nonhuman Primates in Yunnan
The distributional range for each of Yunnan's 16 primate species is presented in Figure 1 (Table S8). Macaca mulatta has the largest current range and is distributed across the entire region (183,360 km2). In contrast, Trachypithecus shortridgei has the smallest range (479 km2) and is restricted to the northwest tip of the Durong Valley (Table 1; Figure 1). The AUC values of our SDMs were robust (≥ 0.98 for all primate species), indicating reliable results (Table 1). Our findings also show that primate species richness is greatest in the LRG west of the Yangtze and Red Rivers, with a maximum of 8 species per km2. In addition to M. mulatta, the Tibetan macaque (Macaca thibetana) is the only other primate species occurring in northeastern Yunnan. This appears to be the result of the dispersal of individuals from north of the Yangtze River during the dry season, as reported by local villagers. The lowest species richness was observed in the central YP, with only M. mulatta present (Figure 1).
| Serial number | Species | Habitat distribution (km2) | AUC |
|---|---|---|---|
| 01 | Nycticebus bengalensis | 79,445 | 0.99 |
| 02 | Xanthonycticebus intermedius | 18,583 | > 0.99 |
| 03 | Macaca leonina | 64,036 | 0.98 |
| 04 | Macaca arctoides | 92,855 | > 0.99 |
| 05 | Macaca mulatta | 183,360 | > 0.99 |
| 06 | Macaca assamensis | 86,669 | > 0.99 |
| 07 | Macaca thibetana | 12,146 | 1 |
| 08 | Rhinopithecus bieti | 8802 | > 0.99 |
| 09 | Rhinopithecus strykeri | 786 | 1 |
| 10 | Trachypithecus shortridgei | 479 | > 0.99 |
| 11 | Trachypithecus melamera | 15,384 | > 0.99 |
| 12 | Trachypithecus crepusculus | 61,576 | > 0.99 |
| 13 | Nomascus concolor | 38,316 | > 0.99 |
| 14 | Nomascus leucogenys | 6778 | 1 |
| 15 | Hylobates lar | 14,513 | 1 |
| 16 | Hoolock tianxing | 14,630 | > 0.99 |
3.2 Geographic Distribution and Changes in Primate Species Richness
3.2.1 Climatic Differentiation Effect on Primate Species Richness
Among the 19 bioclimate factors examined, Bio13 (precipitation of the wettest month) is the most important in affecting primate species richness. This is followed by Bio12 (annual precipitation), Bio4 (temperature seasonality), Bio16 (precipitation of the wettest quarter), and Bio18 (precipitation of the warmest quarter) (Figure 2). The results of the partial correlation analysis indicated that Bio16, Bio13, Bio18, Bio12, and Bio3 (isothermality) all have a positive impact on primate species richness, while Bio4 (temperature seasonality) has a negative impact (Figure S5). When primate species richness was ≥ 6, Bio16 (precipitation of the wettest quarter) (551–994 mm) had the greatest influence on species richness, followed by Bio13 (precipitation of the wettest month) (193–380 mm), Bio18 (precipitation of the warmest quarter) (533–984 mm), Bio4 (temperature seasonality: standard deviation×100) (305–481), and Bio12 (annual precipitation) (933–1882 mm). This is best explained by the fact that precipitation of ≥ 550 mm during the summer season likely results in an increase in flower, fruit, and leaf production of tree species that are part of the primate diet. In contrast, in the central YP, where only M. mulatta is present, the corresponding values for precipitation variables were lower (Bio16: 368–671 mm, Bio13: 133–244 mm, Bio18: 284–671 mm, and Bio12: 628–1213 mm) but temperature seasonality was greater (Bio4: 392–623), indicating that reduced precipitation and extreme seasonal fluctuations in temperature favor habitats that are unsuitable for most primate species (Table 2).
| Primate distribution | Climatic factor | Description | Min value | Max value | Niche utilization rate |
|---|---|---|---|---|---|
| ≥ 3 taxa (6 species) | bio16 | Precipitation of wettest quarter/mm | 551 | 994 | 0.50 |
| bio13 | Precipitation of wettest month/mm | 193 | 380 | 0.49 | |
| bio18 | Precipitation of warmest quarter/mm | 533 | 984 | 0.48 | |
| bio12 | Annual Precipitation/mm | 933 | 1882 | 0.40 | |
| bio4 | Temperature seasonality | 305 | 481 | 0.40 | |
| bio6 | Min temperature of coldest month/°C | 2 | 12 | 0.39 | |
| bio11 | Mean temperature of coldest quarter/°C | 7 | 18 | 0.37 | |
| bio15 | Precipitation seasonality/mm | 76 | 95 | 0.36 | |
| bio1 | Annual mean temperature/°C | 13 | 23 | 0.35 | |
| bio8 | Mean temperature of wettest quarter/°C | 18 | 26 | 0.34 | |
| Macaca mulatta | bio16 | Precipitation of wettest quarter/mm | 368 | 671 | / |
| bio13 | Precipitation of wettest month/mm | 133 | 244 | / | |
| bio18 | Precipitation of warmest quarter/mm | 284 | 671 | / | |
| bio12 | Annual Precipitation/mm | 628 | 1213 | / | |
| bio4 | Temperature seasonality | 392 | 623 | / | |
| bio6 | Min temperature of coldest month/°C | −11 | 10 | / | |
| bio11 | Mean temperature of coldest quarter/°C | −5 | 17 | / | |
| bio15 | Precipitation seasonality/mm | 65 | 97 | / | |
| bio1 | Annual mean temperature/°C | 1 | 23 | / | |
| bio8 | Mean temperature of wettest quarter/°C | 8 | 27 | / |
The top five bioclimate factors ranked by NUR were the same as the top five climate factors ranked by the RFdown sampled model, confirming the robustness of the latter (Figure 2). Among these five bioclimatic factors, Bio12, Bio13, Bio16, and Bio18 are highly correlated (Pearson correlation coefficient r = 0.94–0.99). Each of these factors is associated with warmer and wetter conditions. Moreover, temperature during the coldest month in the primate habitat in central YP can reach −11°C, whereas in LRG, temperatures during the coldest month do not fall below 2°C. In central Yunnan, extended periods of freezing weather in winter are likely to impose high energetic costs of thermoregulation on primate foragers, limiting their ability to survive.
3.2.2 Environmental Changes During the Quaternary
Based on quaternary paleopalynology data from 21 different locations over the past 20,000 years, the main vegetation types in the central YP, between 1500 m to 2000 m a.s.l., fluctuated between sclerophyllous evergreen broadleaf forest and subtropical evergreen needleleaf forest (Figure 3). Both types of vegetation are unsuitable for gibbons, leaf monkeys, and lorises, which inhabit humid tropical and subtropical evergreen broadleaf forests. Notably, the vegetation at Yilong Lake in Shiping County, at 1414 m a.s.l., near southwestern Yunnan, gradually transitioned from tropical monsoon forests to sclerophyllous evergreen broadleaf forest between 7000 and 3000 years ago. This occurred as the monsoon climate strengthened and rainfall decreased. With the onset of increasing human modification of the natural environment beginning 2000 years ago, the vegetation further transitioned into subtropical evergreen needleleaf forest.
Similarly, during the period from 7000 to 3000 years ago in the central YP of northwestern Yunnan, the vegetation fluctuated between deciduous broadleaf forests, cool-temperate conifer forests, and cold-temperate conifer forests at an elevation of between 2700–3900 m a.s.l. In contrast, the forests in the lower elevational areas of the LRG tended to fluctuate along a tropical and subtropical evergreen broadleaf forest vegetation continuum that adapted to moist environments. Notably, primate habitats at Ganglanba in Jinghong City, at 600 m a.s.l., shifted from tropical rainforest to subtropical evergreen needleleaf forest over the past 2000 years. This transition was driven by human activities, including deforestation, expanded agricultural production, and urbanization. Pollen composition indicates that large-scale human activities, mainly deforestation and agricultural cultivation, occurred at 15 localities in Yunnan between 260 and 3000 years ago, with a median date of 2000 ± 745 years ago. Additionally, the anthropogenic impact on habitat conversion was later observed at high elevations of 3200 to 3400 m.
3.2.3 Recent Biogeographical Changes in Primate Species Distribution
The distribution gravity centers and standard deviation ellipses for the five major primate taxa (gibbons/hylobatids, lorises, langurs/Trachypithecus, macaques/Macaca, snub-nosed monkeys/Rhinopithecus) in Yunnan in the 1960s and in the 2020s are depicted in Figure 4. The Hylobatidae showed the most significant changes in distributional ranges, with their gravity center moving 111.51 km northwest from Jinggu County to Yunxian County (Table 3). Their standard deviation ellipse sharply declined toward the north, suggesting a major loss in their southern distributional range. For Lorisidae, the gravity center shifted from the Mojiang Hani Autonomous County in the 1960s to Ning’er County in the 2020s, moving 22.38 km southeast. The reduction in their standard deviation ellipse indicates that these two loris species have been extirpated from their northernmost distribution. The gravity centers of the three Trachypithecus species remained in Fengqing County during both periods. However, compared to the 1960s, the 2020s gravity center has moved 22.49 km northwest, and the standard deviation ellipse has decreased slightly. Lastly, both Macaca and Rhinopithecus species exhibited northward trends in the distribution of their gravity centers between the 1960s and the 2020s. Both groups had a migration distance of less than 5 km. Their standard deviation ellipses remained largely unchanged, indicating stability in their horizontal distribution over the past 60 years.
| Taxa | Gravity center 1960s | Gravity center 2020s | Moving direction | Moving distance (km) | ||
|---|---|---|---|---|---|---|
| Longitude (°) | Latitude (°) | Longitude (°) | Latitude (°) | |||
| Lorisidae | 101.41617 | 23.207159 | 101.442523 | 23.007378 | Southeast | 22.38 |
| Macaca | 100.093335 | 24.795621 | 100.068983 | 24.827453 | Northwest | 4.31 |
| Rhinopithecus | 98.946992 | 26.830088 | 98.946068 | 26.858062 | Northeast | 3.11 |
| Trachypithecus | 99.84428 | 24.366807 | 99.72189 | 24.535571 | Northwest | 22.49 |
| Hylobatidae | 100.610847 | 23.561741 | 100.353811 | 24.536686 | Northwest | 111.51 |
Elevational range changes of primate species from the 1960s to the 2020s are shown in Figure 5 and Table 4. The maximum, minimum, and median values of the box plots for the 1960s and 2020s reflect upper and lower ranges as well as the elevational distribution center (Table S9). Compared to the elevational ranges in the 1960s, the patterns of the 2020s indicated that the upper and lower elevational limits of Macaca and Rhinopithecus did not change. In the case of Trachypithecus and the Hylobatidae, the upper limits remained unchanged; however, their lower elevation limits increased by 84 m and 110 m, respectively. The elevational gravity center (median value) of the Hylobatidae increased from the 1960s to the 2020s by up to 413 m (W = 58,980, p < 0.00001). Although these changes in the elevational gravity center were not significant for the Lorisidae, by the 2020s, there was a decrease of 222 m in their elevational gravity center compared to that of the 1960s. The 2020s elevational gravity centers of the three other primate groups did not change significantly compared with the 1960s. They increased by 14 m for Macaca, 2 m for Rhinopithecus, and 54 m for Trachypithecus.
| Altitude | Lorisidae | Macaca | Rhinopithecus | Trachypithecus | Hylobatidae | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1960s | 2020s | 1960s | 2020s | 1960s | 2020s | 1960s | 2020s | 1960s | 2020s | |
| Max (m) | 2307 | 2307 | 3856 | 3856 | 4233 | 4233 | 2851 | 2851 | 2870 | 2870 |
| Min (m) | 295 | 295 | 328 | 328 | 2215 | 2215 | 375 | 459 | 549 | 659 |
| Mid (m) | 1217 | 995 | 2022 | 2008 | 2981 | 2983 | 1932 | 1986 | 2039 | 2452 |
| W | 2655 | 1,388,672 | 27,862 | 96,422 | 58,980 | |||||
| p | 0.1558 | 0.9454 | 0.7454 | 0.1747 | < 2.2e-16 | |||||
Among the Hylobatidae, the elevational distribution ranges of black crested gibbons (Nomascus concolor) were found to have changed significantly compared to the 1960s (W = 34,815, p < 0.0001) (Figure 5). Their elevational gravity center (median value) increased by 33.6 m, and their lower altitude limit increased by 551 m. For the skywalker gibbon (Hoolock tianxing), the elevational gravity center increased by 27.5 m from the 1960s to the 2020s, but this was not significant. The elevational gravity centers of Macaca arctoides, Macaca assamensis, Rhinopithecus bieti, Rhinopithecus strykeri, and Trachypithecus crepusculus in the 2020s increased between 6 and 245 m from the 1960s. This increase was not statistically significant for any of these five primate species. The elevational gravity centers for Nycticebus bengalensis (Lorisidae) and Macaca mulatta decreased by 222 m and 88 m, respectively, by the 2020s compared to the 1960s. This decrease was not significant for either species. Due to a lack of comparative data, we could not examine changes in the elevational gravity centers of Xanthonycticebus intermedius (Lorisidae), Macaca leonina, Macaca thibetana, Trachypithecus melamera, and Trachypithecus shortridgei (Table S9).
3.2.4 Reasons for Recent Loss of Primate Ranges
Between the 1960s and the 2020s, primates disappeared from 655 of 6065 distribution sites (10.8%). The primary causes of primate population loss were hunting (47.56%), followed by habitat loss [32.56%, including agricultural reclamation (13.17%), plantations (13.05%), lumber extraction (3.78%), and infrastructure (2.56%)], habitat degradation (10.61%), and habitat fragmentation (9.27%) (Figure 6).
4 Discussion
In the southeastern edge of the QTP, 14 primate species are found west of the Yangtze and Red Rivers and south of 24° N. In addition, rhesus macaques have expanded into central Yunnan and Tibetan macaques range into northeastern Yunnan. Fossil pollen records from 21 sites confirm that the arid climate of the YP has persisted for at least 20,000 years, fostering the growth of sclerophyllous evergreen broadleaf and subtropical evergreen needleleaf forests across an elevational band of 1500–2700 m. These forests do not provide suitable habitats for gibbons, langurs, snub-nosed monkeys, lorises, and certain species of macaques, which are adapted to relatively humid tropical/subtropical evergreen broadleaf forests. Our findings indicate that the topography-induced onset and shifting of the Asian monsoon climate resulted in a significant reduction in primate species richness in the YP compared to the LRG by at least 20,000 years ago. Across this region of China, primate species richness is positively driven by monsoon precipitation and negatively affected by extreme seasonality that produces low precipitation and low temperatures. Habitats with annual precipitation between 933 and 1882 mm and temperatures above 2°C during the coldest month of the year correlate with high primate species diversity in Yunnan and support at least 6 co-existing primate species. In contrast, the central YP, with an annual precipitation between 628 and 1213 mm and winter temperatures that frequently drop below zero, hosts only one primate species, the rhesus macaque.
Moreover, the fossil pollen records and our field surveys show that over the past 2000 years, significant anthropogenic habitat transformation, habitat loss, and hunting for meat, fur, and traditional medicine have horizontally and vertically reshaped primate distributions. This resulted in range contraction and population extirpation of primate populations inhabiting edge and lower elevational distributions, as occurred in snub-nosed monkeys over the past few hundred years in the LRG (Li et al. 2002; Wang et al. 2019). In this regard, human activities over the past 2000 years have not significantly modified patterns of primate species richness in Yunnan. However, those same activities have significantly reshaped primate species distributional ranges both horizontally and vertically, leading to severe range contractions in at least 12 species across all five major primate families. Based on a recent study (Yang et al. 2025, Figure S6) linking high primate species richness and high vertebrate species richness across Yunnan, our findings confirm a natural biogeographical boundary for terrestrial vertebrate biodiversity at the Yangtze-Red River-24° N intersection, marking a gradient of increasing species richness towards the west and the south and decreasing species richness towards the east and north.
The association between increased precipitation and primate richness in Yunnan is consistent with previous studies linking Asian primate diversity to areas with increased rainfall (Reed and Fleagle 1995; Wang et al. 2013). Similarly, Gouveia et al. (2014) found a strong correlation between forest canopy height and primate richness, but not precipitation, globally except in Asia. Although the relationship between canopy height, resource productivity, and annual rainfall remains complex and unclear, these factors are critical in understanding patterns of niche expansion, niche complexity, and species diversity (Tao et al. 2016; Kay et al. 1997; Hanya et al. 2011). Zhao et al. (2025) found that the establishment of humid evergreen broadleaf forests in Asia requires precipitation levels that exceed 600 mm in the wettest quarter of the year. This supports our finding that habitats in the LRG characterized by precipitation exceeding 551 mm in the wettest month support the coexistence of gibbons, langurs, snub-nosed monkeys, and tropical/subtropical macaques and also promote areas of humid evergreen broadleaf forests with high primate biodiversity.
Evergreen tree species exposed to cold and dry conditions in winter and spring have evolved resilient leaves to withstand these conditions (Spicer 2017). These adaptations create “monsoon fingerprints” on the leaf structure, including smaller leaves, fibrosis, leatherization, and increased trichome cover and secondary metabolites (Yang et al. 2015; Spicer 2017; Li et al. 2020; Zhu 2023). These leaves present digestive challenges to many primate species (Simmen et al. 2012; Windley et al. 2022). The mountains west of the Red River Fault Zone and the YP act as filters of moisture carried by the Southeast monsoon and East Asian monsoon entering into central and eastern Yunnan, respectively. This has resulted in vegetation and forest type differences between central YP, the LRG, and southeastern Yunnan. In the YP, sclerophyllous evergreen broadleaf forests emerged during the Late Pliocene, approximately 3.6–2.6 mya (Huang et al. 2016). Increased precipitation seasonality and ongoing aridification could further transform sclerophyllous evergreen broadleaf forests into subtropical evergreen needleleaf forests or savannas (Zhu 2023; Huang et al. 2016; Ratnam et al. 2016). This environmental transition likely led to the extinction of primate taxa from central Yunnan and is supported by the fossil record, which confirms that macaques, leaf-eating monkeys, and gibbons were present in northeastern and central Yunnan during the late Miocene (8–6 mya) (Ji et al. 2020; Ji et al. 2022; Jablonski 1993; Table S2). A similar climatic-ecological process likely led to the extinction of a large-bodied ape (Gigantopithecus blacki) in the adjacent Chongzuo region of Guangxi Province during the late Miocene (Zhang et al. 2024). In Yunnan, the monsoon filter resulting from the Himalayan orogeny, and the reinforcement of the monsoon climate impacted the biogeographical distribution and species richness of several animal taxa, including primates and other species of terrestrial vertebrates (Huang et al. 2016; Yang et al. 2025).
Presently, Yunnan's northwest, west, southeast, and a small northeastern section of the Wumeng Mountains preserve tropical rainforests (900–1800 m a.s.l.), tropical seasonal rainforests (100–1200 m a.s.l.), monsoon evergreen broadleaf forests (900–2100 m a.s.l.), or semi-humid and humid subtropical evergreen broadleaf forests (1300–2900 m a.s.l.) (Zhu and Ashton 2021; Zhu 2022, 2023). These habitats support lorises, gibbons, langurs, and tropical/subtropical macaques (e.g., Macaca leonina and M. arctoides). Higher altitudes host cool-temperate conifer (mixed conifer-broadleaf forests) (2600–3500 m a.s.l.) and cold-temperate conifer forests (3000–4300 m a.s.l.) along with cold winter temperatures. These habitats are exploited by snub-nosed monkeys (Rhinopithecus spp.), which represent a radiation of cold- and low-oxygen-adapted primates (ibid, Yu et al. 2016). In contrast, central and eastern Yunnan, along with parts of the northeast and southeast, are dominated by subtropical evergreen needleleaf forests (1500–2500 m a.s.l.), sclerophyllous evergreen broadleaf forests (1600–4000 m a.s.l.), and temperate conifer forests. Only the highly adaptable and widely distributed rhesus macaques (Macaca mulatta) manage to thrive at elevations that vary from 481 m to 3856 m and habitat types ranging from tropical rainforests to alpine conifers or even alpine shrubs and grasslands. In Yunnan, the humid tropical and subtropical evergreen broadleaf forests of the LGR appear to have served as refuges for tropical and subtropical primates, such as the lorises, gibbons, langurs, and tropical/subtropical macaques for at least the past 20,000 years.
Over the past 2000 years, and especially during the last 100 years, large-scale anthropogenic habitat transformation and hunting for meat and traditional medicine have significantly reduced primate populations, resulting in isolated subpopulations, low genetic diversity, and highly fragmented distributions. In Yunnan, three taxa—snub-nosed monkeys, langurs, and gibbons—have moved to higher latitudes and elevations compared to their 1960s distributions. Notably, gibbons have undergone the most significant range shift, in part due to the extirpation of two tropical species, Nomascus leucogenys and Hylobates lar, from China in the past few decades (Li et al. 2023). Our study found that compared to the 1960s, the lower limits and centers of gibbon elevational distribution had significantly increased by the 2020s. These findings align with those by Fan and Jiang (2010) and Yang et al. (2021) highlighting the dramatically reduced range of the black crested gibbon (Nomascus concolor) due to hunting, habitat fragmentation, and habitat loss. In the Huanglian Mountain, Luchun County, Yunnan, black crested gibbons are functionally extinct, with only 1–3 remaining groups (Ni and Ma 2006). Our investigations show that hunting has resulted in the extirpation of Trachypithecus crepusculus from the same region.
In contrast, snub-nosed monkeys (Rhinopithecus spp.), which inhabit elevations above 2300 m, face fewer human threats and have experienced only minor range changes. The black-and-white snub-nosed monkey (Rhinopithecus bieti) has been protected by law since the 1980s, and this has resulted in an increase from 13 groups and ~1300 individuals to 23 groups of over 3600 individuals over the course of 40 years (Xiao 2021). This occurred without any significant changes in range distribution. In contrast, human activities in China have significantly impacted the elevational distribution of black snub-nosed monkeys (R. strykeri), as almost all habitats below 2300 m have been converted into human settlements. The species is also present in Myanmar, where it continues to inhabit lower elevation forests (1800 m a.s.l.) that have not been severely disturbed or highly fragmented by human activities (Yang et al. 2019, 2022).
The rhesus macaque (M. mulatta), with a downward-shifting center of gravity in elevational distribution, is commensal with humans. Despite habitat loss, it can survive in highly disturbed environments such as urban areas, temples, and tourist sites. These primates raid crops, scavenge garbage, and are often provisioned by local inhabitants. However, increased human-macaque conflict has resulted in retaliatory killings and captures by farmers (Fan et al. 2018; Song et al. 2020; Li, Fang, et al. 2024). Many provisioned populations have exceeded the normal troop size, further increasing human-macaque conflict and increasing the potential for zoonotic disease transmission.
The two species of lorises are the only primates in Yunnan that have shifted their distributional range to lower elevations and latitudes, with the Bengal slow loris (Nycticebus bengalensis) extirpated from its northernmost range in the upper Salween River valley. The last record of N. bengalensis in Gongshan County dates to the 1990s (Huang 1999). Villagers who witnessed the species reported that slow lorises historically inhabited semi-moist evergreen broadleaf forests at 1200–2000 m. However, there have been no sightings in the last 20–30 years. Field investigations found that their habitat has been transformed into farmland, residential areas, and cemeteries. Both loris species inhabit low elevation forests. However, they are found outside of protected areas and are characterized by very small populations and low population density. Therefore, formulating and implementing emergency conservation measures for lorises through China's Animal Species with Extremely Small Population Program (Yang et al. 2020) and establishing protected areas are vital to prevent their extinction in Yunnan.
Recent analyses of long-term precipitation records in Asia reveal that climate change has altered the direction of the South Asian summer monsoon. This has caused drought, while the southeast monsoon is intensifying, leading to extreme weather events and high rainfall and landslides during the monsoon season and more arid conditions in the dry season (Bollasina et al. 2011; Sun and Ding 2010; Singh et al. 2014; Edirisinghe et al. 2023). These changes are likely to significantly impact the distribution and population persistence of primates and other animals across the region over the next several decades and centuries. In the face of human-induced climate change, habitat conversion, expanding infrastructure, and human population growth, paleovegetation data and local ecological knowledge offer valuable insights into how future climate and habitat conditions are expected to adversely affect biodiversity across the region. Studies exploring how primates respond to changes in precipitation, temperature, habitat conversion, extreme weather events, and human activities provide significant information for designing and implementing effective conservation and management plans (Zhang et al. 2019; Ameca et al. 2023).
In conclusion, we examined the manner in which changes in the monsoon climate and human activities over the past 20,000 years have shaped patterns of primate distribution, species richness, and population extirpation across the southeastern edge of the Qinghai-Tibet Plateau, a region of high terrestrial vertebrate and plant biodiversity. We provide new insights into the set of climate and habitat requirements that serve to constrain and promote primate persistence in the 21st century. Given considerable overlap in the habitat requirements of primates and other taxa of terrestrial vertebrates, our study has important implications for understanding how climate change over the coming decades is expected to affect current and future changes in species distributions and biodiversity across Southeast Asia. The methods used in this study can be applied to investigate the causes of historical changes in species richness and distribution across a diverse set of animal taxa, particularly in regions where recent fossil and historical records are limited.
Author Contributions
Chen Li: formal analysis, investigation, methodology, visualization, writing – original draft. Yi-hao Fang: data curation, formal analysis, investigation, writing – original draft. Guo-peng Ren: conceptualization, formal analysis, funding acquisition, methodology, software, writing – original draft. Yan-peng Li: funding acquisition, investigation, project administration, resources. Zhi-pang Huang: data curation, funding acquisition, investigation, project administration, resources. Liang-wei Cui: data curation, funding acquisition, investigation, resources. Dionisios Youlatos: methodology, validation, writing – review and editing. Paul A. Garber: visualization, writing – original draft, writing – review and editing. Xi-jun Ni: conceptualization, validation, writing – review and editing. Hua Zhu: methodology, validation, writing – review and editing. De-wen Luo: formal analysis, visualization. Xin Liu: investigation, visualization. Meng-ran Chu-yuan: investigation. Ying-ping Tian: investigation. Ying-chun Li: investigation. Xiang-le Zeng: investigation. Dong Yan: investigation. Gen-hui Li: investigation. Wen Xiao: conceptualization, data curation, funding acquisition, project administration, resources, supervision, validation, writing – review and editing. Rui-dong Wu: conceptualization, data curation, funding acquisition, project administration, resources, supervision, validation, writing – original draft, writing – review and editing. Yin Yang: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, supervision, validation, visualization, writing – original draft, writing – review and editing.
Acknowledgments
We are grateful to the Editor and two anonymous reviewers for their useful and constructive comments on the manuscript. We would like to thank the Administration Bureau of Yunnan Forestry and Glasslands and bureaus of multi-national and provincial nature reserves in Yunnan Province for their altruistic support. This work was supported by the National Natural Science Foundation of China (32471737, 32260251, 32371567, and 32360137), the Natural Science Foundation of Yunnan Province (202401BF070001-008 and 202301AT070247), and the Talent and Platform Project within the Yunnan Provincial Science and Technology Department (202205AM070007, 202105AM070008, YNWR-CYJS-2018-052, and YNWR-QNBJ-2019-262). P.A.G. wishes to thank Chrissie, Sara, Jenni, Dax, and Saffron for their love and support during the writing of this manuscript.
Conflicts of Interest
The authors declare no conflicts of interest.
Open Research
Data Availability Statement
The data and code that support the findings of this study are openly available in Figshare at https://doi.org/10.6084/m9.figshare.28691960. Climate datasets were obtained from WorldClim at http://worldclim.org/. Land use data were obtained from the Environmental Systems Research Institute at https://livingatlas.arcgis.com/landcoverexplorer and EarthEnv at https://www.earthenv.org/landcover (version1). Plantation data was obtained from Zenodo at https://zenodo.org/record/5879022#.ZBh_e3bMJdg. Road location data were obtained from OpenStreetMap at http://www.openstreetmap.org.
