2.1 Species Data

We obtained occurrence records from the literature, the Global Biodiversity Information Facility (https://www.gbif.org/), the National Specimen Information Infrastructure (http://www.nsii.org.cn/), the Chinese Virtual Herbarium (http://www.cvh.ac.cn/), Sichuan University Herbarium (http://mnh.scu.edu.cn/list/introduction/), New York Botanical Garden (https://sciweb.nybg.org/science2/hcol/vasc/index.asp.html/), and Kew Royal Botanic Garden (http://apps.kew.org/herbcat/gotoHomePage.do), as well as our field surveys conducted in 2017 and 2021. All occurrence records were reviewed and cross-checked using Flora of China (http://www.efloras.org/) and The Plant List (http://www.theplantlist.org/) to ensure accuracy and eliminate errors or problematic entries. To reduce the potential influences of overfitting biases, occurrence records for each species were thinned (10 km) using SDMtools in ArcGIS 10.7 (Zhou et al. 2021). This resulted in a total of 44 occurrence records selected for our study (20 for T. doichangensis and 24 for T. verticillata; Supporting Information S1: Table A1; Figure A1).

2.2 Environmental Data

Environmental data were obtained from the WorldClim database (https://www.worldclim.org), which includes 19 bioclimatic variables for the present period (1970–2000) and projections for the 2050s (2040–2060) and 2070s (2060–2080), as well as elevation variables (Supporting Information S1: Table A2). Detailed descriptions of these environmental variables are provided in Supporting Information S1: Table A2. Future climate data were based on the Community Climate System Model (CCSM4), with a total of four representative concentration pathways (RCPs) considered: RCP2.6, RCP4.5, RCP6.0, and RCP8.5, which represent low, medium, high, and highest concentration greenhouse gas emissions scenarios, respectively (van Vuuren et al. 2011). Map data were sourced from the Resource and Environment Science and Data Center (https://www.resdc.cn/Default.aspx).

To avoid the influence of highly correlated variables, Pearson's correlation analysis was performed to select the final bioclimatic variables. Using the Band Collection Statistics tool in ArcGIS, any pair of variables with a correlation coefficient |r | ≥ 0.8 was assessed, and the variable with the lesser contribution to the model was omitted. Variables with minimal effects on species distribution and growth were removed, while those with potential ecological relevance to the species were retained (Supporting Information S1: Figure A2). The elevation factor was incorporated directly into the analysis and assumed to remain constant over the coming decades (Zhang et al. 2023). As a result, nine environmental variables were retained to simulate the potential distribution of T. doichangensis, and seven environmental variables were retained for T. verticillata (Tables 1 and 2).

2.3 Species Distribution Modeling

Ecological niche models (ENMs) are valuable tools for investigating species distribution patterns (Peterson et al. 2011; Warren and Seifert 2011). Among various modeling methods, the MaxEnt model, which is based on maximum entropy theory, is particularly noted for its exceptional performance in predicting species geographic distribution (Ab Lah et al. 2021; Zeng et al. 2023; Elith et al. 2011). Numerous studies have demonstrated the reliability of the MaxEnt model in simulating potential species distributions (He et al. 2023; Li et al. 2023; Shi et al. 2023; Gao et al. 2022).

Species distributions were predicted using MaxEnt v3.4.1 (Elith et al. 2011). Twenty percent of occurrence data were randomly selected as the test data set, while the remaining 80% were used as the training data set (Orefice and Innocenti 2024). Background points were set to 10,000, with the maximum number of iterations of the model set to 1000 and the number of repetitions set to 10. Other parameters were set to default values (i.e., regularization multiplier = 1, convergence threshold = 10⁻⁵; Syfert et al. 2013). The jackknife test was used to identify the most dominant environmental factors, and the threshold value for each environmental factor was determined from the response curve. The area under the receiver operating characteristic curve (AUC) was used to evaluate the reliability of the model's predictions. The AUC value ranges from 0 to 1, with higher values indicating greater reliability in model predictions (Elith et al. 2006; Fielding and Bell 1997).

The outputs from the Maxent model were converted into raster format and reclassified to identify potentially suitable areas. The cloglog output from MaxEnt represents climate suitability on a scale from 0 to 1, with higher values indicating more favorable conditions for species survival (Lee et al. 2021). A 10% training presence cloglog threshold was used to classify suitability levels: unsuitable area (< 10% training presence cloglog threshold), low suitability (10% training presence cloglog threshold–0.6), medium suitability (0.6–0.8), and high suitability (0.8–1). In addition, response curves for environmental variables were used to assess the suitability range of environmental variables for species growth. A presence probability greater than 0.5 indicates that the corresponding environmental variable range is suitable for species survival (Zhao et al. 2021).

3.1 Maxent Performance and the Importance of Environmental Variables

The results indicated that the AUC values exceeded 0.9 (T. doichangensis: 0.984, T. verticillata: 0.974), implying high reliability in the simulation outcomes, which effectively reflected the responses of both island and continental species to climate change.

For the continental species T. doichangensis, the analysis revealed that the mean diurnal range (Bio2), isothermality (Bio3), and precipitation of the warmest quarter (Bio18) were the most influential factors affecting its predicted distribution. The cumulative contribution of these three variables was 87.3% (Table 1). Other factors, such as precipitation of the wettest month (Bio13), elevation, and precipitation of the coldest quarter (Bio19), had a moderate influence, contributing 10.5% to the model (Table 1). In contrast, precipitation seasonality (Bio15), precipitation of the driest month (Bio14), and mean temperature of the wettest quarter (Bio8) had the least influence on the predicted distribution of T. doichangensis (Table 1). Overall, temperature was found to have the greatest impact on the potential distribution of the continental species, followed by precipitation and elevation (Table 1).

For the island species T. verticillata, the dominant environmental factors influencing species distribution were precipitation seasonality (Bio15), mean temperature of the wettest quarter (Bio8), temperature seasonality (Bio4), and elevation. These four factors collectively contributed 88.7%, with the next most influential variables being mean diurnal range (Bio2, 6.6%), precipitation in the warmest quarter (Bio18, 2.9%), and annual precipitation (Bio12, 1.9%). The three highest permutation importance values were precipitation seasonality (Bio15), mean diurnal range (Bio2), and elevation. In total, precipitation factors accounted for the greatest cumulative contribution (37.7%), followed by temperature factors (30.1%), and elevation (32.3%). Precipitation had the greatest influence on the potential distribution range of island species, followed by elevation, while temperature factors had the least influence (Table 2).

3.2 Present Potential Habitat

The current potential distribution of the continental species T. doichangnsis is mainly in southwest Yunnan, China, eastern Myanmar, and some scattered areas in northern Laos and northern Thailand. Low-, medium-, and high-suitability areas account for 60.17%, 16.34%, and 23.49%, respectively. Current high-suitability areas are primarily located in Yunnan, China, and eastern Myanmar (Figure 1a). The current potential distribution of the island species T. verticillata includes Hainan (China), south-central and central Vietnam, Malaysia, Sulawesi, Sumatra, Java, northeastern Kalimantan, and the Philippines. The low-, medium-, and high-suitability areas account for 76.89%, 12.15%, and 10.96%, respectively. Current high-suitability areas are mainly in northern Sumatra, Sulawesi, scattered areas in northern Java, northeast Kalimantan, and the southern Philippines (Figure 1b).

3.3 Potential Future Distribution Pattern of the Continental Species T. doichangensis

For the narrowly distributed continental species T. doichangensis, the potential suitable areas across different periods were primarily centered in southwest Yunnan, eastern Myanmar, northern Laos, and northern Thailand, with southwest Yunnan and northern Laos being the most suitable (Figure 2a–h).

Under different climate scenarios, the potential high- and medium-suitability areas exhibited significant changes. Under low-concentration (RCP2.6) and medium-concentration (RCP4.5) greenhouse gas emissions scenarios, the potential high-suitability areas initially decreased and then increased. In contrast, under high-concentration (RCP6.0) and highest-concentration (RCP8.5) emissions scenarios, the potential high-, medium-, and low-suitability areas decreased significantly (Figure 2a–h).

Overall, the potential distribution areas of the narrowly distributed continental species T. doichangensis are expected to expand significantly, particularly in regions bordering southwest China, eastern Myanmar, northern Thailand, and northern Laos (Figures 3 and A4).

3.4 Future Potential Distribution Pattern of the Island Species T. verticillata

For the widely distributed island species T. verticillata, areas of high suitability decreased under the four climate scenarios, especially in Kalimantan. Additionally, some unsuitable habitats in Sulawesi were transformed to low-suitability areas. From the present to the 2050s, potential high-suitability habitats were transformed into medium- and low-suitability habitats under all four climate scenarios (Figures 1b and 4a–d).

Overall, the island species T. verticillata is expected to experience range contraction in its potential distribution areas (Supporting Information S1: Figure A4). The primary areas of range contraction are Hainan, China, south-central and central Vietnam, and the coastal areas of western Sumatra, Java, northeast Kalimantan, and Malaysia (Figure 5).

4.1 The Impact of Dominant Environment Variables on Continental and Island Species

Temperature, precipitation, and elevation are crucial climatic factors that influence species distribution and ecosystems (Punyasena et al. 2007). Our study revealed that precipitation and temperature conditions significantly shape species distribution patterns. Specifically, the cumulative contribution of temperature factors affecting the potential distribution of the continental species T. doichangensis was 74.2%, while precipitation factors contributed 37.7% to the potential distribution of the island species T. verticillata.

For mountain species inhabiting high-elevation areas, low temperatures limit plant growth and development, thereby influencing species distribution. Hence, temperature plays a crucial role in constraining the distribution and development of mountain species (Dong et al. 2023; Yu et al. 2019). The continental species T. doichangensis is distributed in lower montane evergreen broad-leaved forests (Mao et al. 2010; Zhu and Ashton 2021). The high suitability of T. doichangensis clustered in high-elevation areas also indicates that species at high elevations are sensitive to temperature changes. With climate change, lower temperatures at high elevations are likely to drive continental species to migrate upward in search of suitable habitats. Although precipitation and elevation had less impact on the distribution of continental species in our study, these factors remain important for species growth and distribution.

The island species in our study predominantly thrives in Southeast Asia, a region characterized by the prevalence of monsoons. The Asian monsoon has a direct impact on rainfall and temperature patterns in these regions, creating a regular seasonal cycle between dry and wet periods (Jiang et al. 2017; Tan, Malabrigo, et al. 2020). These climatic fluctuations affect plant growth, development, and reproduction, facilitating better adaption to seasonal changes (Sun et al. 2020). With climate change, an increase in extreme weather events, such as droughts and heavy rainfall can be expected (Reichstein et al. 2013; Wang et al. 2024), which will likely have significant consequences for the distribution of island species.

Furthermore, our study highlights the significant impact of topographic factors on both continental and island species. Interestingly, when excluding other factors (e.g., land-use change, human activities), the influence of topography on island species is considerably greater than that on continental species. We speculate that changes in the upper and lower elevation ranges may have contributed to this result. Affected by climate change, both the upper and lower limits of the species' elevation distribution is expected to shift (Zu and Wang 2022). The elevation range of island species is lower than that of continental species, meaning that island species have a greater upward migration potential than continental species (Tables 1 and 2).

4.2 Differences in Climate Response Between Continental and Island Plant Species

Species survival is closely linked to climate change, with some researchers suggesting that species with narrower thermal niche breadths, smaller geographic ranges, or smaller body sizes are at a greater risk of extinction (Malanoski et al. 2024). However, our study demonstrated an opposite viewpoint: under scenarios of low and moderate greenhouse gas emissions, the highly suitable distribution area for the continental species in this study is expected to first decrease and then increase in the future. In contrast, under scenarios of high emissions, highly suitable areas are expected to continue shrinking. This suggests that continental species with narrow distribution ranges may not be able to adapt to rapid climate change and migrate to the most suitable areas quickly (Figure 2).

Additionally, with global warming, temperature restrictions on vegetation distribution in certain regions may be lifted, facilitating the further expansion of species (Zhong et al. 2023). Temperatures increases are driven by rising greenhouse gas concentrations (Wang et al. 2024). Hence, even if continental species can adapt to escalating temperature conditions to a certain extent, the decline in highly suitable habitats remains unavoidable.

Unlike continental species, island species are generally more sensitive to climate change and often lack the ability to adapt quickly or migrate to more suitable areas. This is evident in the study by Tanaka et al. (2012), which predicted that climate change would reduce the potential habitats of Abies species in Japan. Our findings support this view, with the suitable area for the island species T. verticillata predicted to decrease (Figures 4 and A4).

Many species begin migrating to higher latitudes and elevations in response to climate change, leading to changes in distribution areas (Meng, Zhou, Jiang, et al. 2019; Zhou et al. 2021). Our results show that both continental and island species are forecast to shift towards higher elevations (Figures 2 and 4). Surveys of natural forest habitats in Southeast Asia have suggested that biodiversity in this region is highly vulnerable to climate change, especially in the Indo-Burma region and central Malaysia, where many species are increasingly moving to higher elevations in response to climate warming, which increases the risk of habitat loss (Namkhan et al. 2022). Rising greenhouse gas emissions will accelerate climate change (Tan, Wu, et al. 2020). Our study indicates that, as greenhouse gas emissions increase, both island and continental species are expected to experience varying degrees of contraction in their potentially highly suitable areas (Figures 1, 2 and 4). Because habitat availability decreases with elevation, the shift of island species to higher elevations under climate change poses a significant extinction risk (Figure 4). Interestingly, our results also suggest that continental species may migrate to lower latitudes (Figure 2). Although there is ample evidence of species migrating to higher latitudes or elevations in response to climate change, evidence of their migration to lower latitudes is limited (Franco et al. 2006). We propose that the observed movement towards lower latitudes is likely attributable to niche conservation, prompting plants to migrate towards habitats that are more conducive to their survival. It is important to note that in subtropical and tropical regions, there are no high-elevation areas acting as buffer zones, such as the high mountain ranges in the Himalaya–Hengduan Mountains, which can provide microclimatic refuges and reduce extinction risks (Meng, Zhou, Jiang, et al. 2019). Therefore, mountaintops in low-latitude tropical regions are likely to serve as refugia. However, we did not observe significant low-latitude migration in island species, leading us to speculate that island species may lack similar buffer zones.

Based on our analysis of the potential distribution patterns of continental and island species, our findings align with our initial hypothesis that widely distributed island species are likely to be more significantly impacted by future climate change, while narrowly distributed continental species may experience fewer effects. Specifically, our study of the widespread island species T. verticillata in Southeast Asia indicated a less favorable future distribution (Figures 1, 4 and A4). In contrast, the continental species T. doichangensis, which currently has a limited distribution in Yunnan, China, and Thailand, shows promising future distribution trends (Figures 1, 2 and A4).

One possible explanation for our findings is the role of genetic diversity, which is critical to the evolutionary history and future adaptability of species (Jump et al. 2009). Low genetic diversity can impede a species' ability to adapt to environmental changes, and we speculate that the observed distribution patterns in our study may be influenced by the genetic diversity of the species involved. In earlier studies, we found that the genetic diversity and gene flow of the continental species T. doichangensis were higher than those of the island species T. verticillata (Hu et al. 2022; unpublished data). Moreover, studies on the genus Geodorum have shown that narrowly distributed species tend to have higher genetic diversity (Zhu et al. 2023). Therefore, we suggest that, in addition to ecological factors, intrinsic characteristics of species, such as genetic diversity, are crucial for their adaptation to climate change.

Furthermore, island plant species face greater challenges in dispersing across geographical barriers than continental species. There is no evidence to suggest that Trigonobalanus species possesses the ability to disperse over long distances across oceans. As a result, island species generally exhibit weaker resilience to environmental risks than continental species. Therefore, we propose that continental species are likely to be more resilient to climate change than island species.

4.3 Conservation Implications

In Asia, the genus Trigonobalanus plays a crucial role in maintaining ecological balance and preserving regional biodiversity, thereby supporting the rich biodiversity of Southeast Asia. To our knowledge, there are currently no studies on the future distribution of Trigonobalanus. Therefore, our study helps fill this gap. Our predictions indicate that southwest Yunnan, China, and eastern Myanmar are high-suitability areas for T. doichangensis (Figure 1a). Meanwhile, northern Sumatra, Sulawesi, scattered areas in northern Java, northeast Kalimantan, and the southern Philippines are high-suitability areas for island species of T. verticillata (Figure 1b). The natural habitats in these areas are optimal environments for species growth. We found that, except for Myanmar, Java, and the Philippines, where there was no population distribution, the remaining areas aligned with the real distribution areas of T. doichangensis and T. verticillata. These highly suitable habitats are predicted to experience varying degrees of fragmentation in the future, particularly for the island species (Figures 2 and 4). Therefore, we propose the establishment of protected areas (PAs) to maintain habitat stability in regions most affected by climate change, such as Sumatra. Additionally, our results suggest that some areas where island species are present, such as Hainan, China, and Vietnam, are at risk of disappearing in the future because of climate change (Figure 4). These areas, highly affected by climate change, should be prioritized for conservation, supplemented by ex situ conservation efforts to prevent potential extinction of island species.

High-elevation regions, particularly mountaintops in tropical areas, are predicted to serve as crucial refuges under climate change, offering microclimatic conditions that buffer extinction risks. Furthermore, our study suggests that the continental species T. doichangensis will likely exhibit greater adaptability to climate change than the island species T. verticillata. This implies that conservation efforts for T. doichangensis may yield more significant results, suggesting that this species should be given priority. Based on our fieldwork, there is an urgent need for a more comprehensive assessment of the endangered status of both T. doichangensis and T. verticillata, especially since the current threat status of T. verticillata may be more severe than that indicated by the IUCN (Near Threatened, https://www.iucnredlist.org).