4 DISCUSSION
4.1 Influence of environmental variables
In the field of Francois’ langur ecology, habitat selection and use have been an important research topic for researchers. The factors of habitat selection have been divided into biological, geographical, and anthropogenic factors (Han et al., 2023). Our results suggested that bioclimatic variables made the largest contribution to the distribution of Francois’ langur, vegetation, water source and anthropogenic variables were also important factors influencing the distribution of Francois’ langur. As demonstrated by numerous previous studies,, the distribution of Francois’ langur was largely restricted to areas characterized by karst topography, limestone cliffs, and caves with mixed conifer-broadleaf forests as they provided quality food resources and cover for predator avoidance (Wang et al., 2011; Wen et al., 2010; Zeng et al., 2013). Additionally, its distribution was concurrently influenced by various factors, including temperature, precipitation and vegetation (Enstam & Isbell, 2004; Han et al., 2011).
We used the MaxEnt model to construct patterns of habitat suitability for the Francois’ langur in China. According to our model prediction, temperature-associated bioclimatic variables were of great importance in predicting habitat suitability for Francois’ langur, while precipitation-associated bioclimatic variables were the most important habitat predictor for Francois’ langur. Numerous studies have shown that in recent years the frequency of precipitation in southwest China has decreased, while the intensity, total precipitation, and occurrence of extreme rainstorm have increased (Yan et al., 2023; Lu et al., 2022; Ding, 2014; Feng et al., 2012). As a result, river water level have risen, and the original valley area with food and habitat caves for Francois’ langur have been flooded, posing a direct threat to the living habitat of Francois’ langur. In addition, temperature and precipitation have a great influence on the growth rates of plant species, and fruits, seeds, and especially young leaves are the main food of Francois’ langur. Therefore, temperature and precipitation indirectly determined the distribution of Francois’ langur by affecting the availability of potential food resources (Huang et al., 2008; Zhou et al., 2011; Zhou et al., 2006). Based on the response curves in our study, the most suitable distribution of Francois’ langur exhibited the following temperature characteristics: temperature annual range of 21-23℃, mean of monthly temperature range (max temp-min temp) of < 6℃, and minimum temperature of coldest month of 9-11℃. The precipitation seasonality (coefficient of variation) and precipitation of coldest quarter of Francois’ langur’s most suitable distribution were < 65 mm and 70-80 mm, respectively. This result is consistent with the climatic conditions of the extant range of Francois’ langur (Han & Hu, 2010; Li et al., 2019; Niu et al., 2016).
Vegetation-related variable, the normalized difference vegetation index, was another important variable. Francois’ langur prefers areas with relatively high NDVI, which is consistent with Han et al. (2023), Niu et al. (2016), Zeng et al. (2013) and Zeng (2011), who showed that Francois’ langur habitat was mainly in the valley areas on both sides of the rivers, covered by evergreen broad-leaf forests with higher tree canopy density and vegetation cover, which provided availability of better food resources and suitable cover for predator avoidance. Meanwhile, distance to rivers and altitude were the less important variables. Previous studies have indicated that Francois’ langur was primarily distributed in valleys characterized by dominant evergreen broadleaf forests. These areas offer abundant food and water sources, and culverts and karst caves along rivers provide enough suitable sleeping sites for Francois’ langur (Han et al., 2011; Hu, 2011; Wang et al., 2011). In addition, the elevation of Francois’ langur distribution ranges from 300 to 1800m, with seasonal changes in climate and food (Li et al., 2019; Li et al., 2015).
For the anthropogenic variables, distance to roads (Including railways) was found to be the more important variable compared to distance to housing. Most previous studies have assessed the impact of linear infrastructure on wild animals and have typically shown that distribution decreases as road density increases (Céline Clauzel et al., 2013; Clauzel et al., 2015). Roads are likely to act as barriers to wildlife movement (Coffin, 2007; Zhu et al., 2011). Therefore, roads may have negative impact on the activity and distribution of Francois’ langur. In this study, distance to roads was positively correlated with the occurrence probability of Francois’ langur. However, distance to housing had a limited influence on the distribution prediction of Francois’ langur. This finding aligns with Han et al. (2023), Zeng et al. (2013), Wang et al. (2011), who reported that Francois’ langur has become accustomed to human disturbances, especially, crops grown on agricultural land may supply the food resources for Francois’ langurs (Han et al., 2023).
4.2 Characteristics of the current potential distribution
Our model estimated potential suitable habitat of 144207.44 km2 for Francois’ langur across the range, based on the Max TSS and average TPT thresholds in the MaxEnt output. Hu et al. (2011) showed that Francois’ langur was restricted to only five isolated sites (Dashahe, Baiqing, Mayanghe, Kuankuoshui and Yezhong) with a total area of about 912 km2 in Guizhou.Chen (2006) estimated potential Francois’ langur habitat in Guangxi at approximately 3216.23 km2. Han and Hu (2010) confirmed that the distribution area of Francois’ langur in Chongqing Jinfo Mountain was about 50 km2. Our model predicted the largest proportion of Francois’ langur habitat in China than the total habitat in three provinces (Guizhou, Guangxi, and Chongqing) in previous studies, which were based on survey records or estimates at a small scale, such as one or more nature reserves. Importantly, our model analysis was based on a machine-learning algorithm, species records from the entire range in China, and a set of environmental variables including climatic factors. Furthermore, our model prediction for the Francois’ langur was based on a larger and more comprehensive scale than prior methods (Chen, 2006; Han et al., 2011, 2023; Hu et al., 2011).
Our model predicted a total of 34265.96 km2 of moderate- and high-suitability habitat for Francois’ langur, which only accounted for 23.76% of predicted suitable habitat. The moderate- and high-suitability habitats, primarily located in southwest Guangxi, east of Chongqing and the Guizhou-Chongqing border, was nearly 10 times the area of previous surveys (Chen, 2006; Han & Hu, 2010; Hu et al., 2011). However, the low suitable habitat was distributed over a larger area as fragmentation, which is consistent with Guan et al. (2022) and Niu et al. (2016), who suggested that the Francois’ langur habitats presented severe fragmentation, complex patch shape, weak patch aggregation, and decentralization, and typically showed that habitat area decreased and further fragmented with increasing human activities. The current potential distribution in our study basically covered and was larger than the historical distribution areas of Francois’ langur recorded in “A guide to the mammals of China” (Smith et al., 2010) and “China’ s mammal diversity and geographic distribution” (Jiang et al., 2015). Our results can provide a scientific reference for further field survey on the distribution of Francois’ langur in China.
4.3 Distribution change in the future
Our results indicate that, compared to the current potential habitat, the suitable habitats of Francois’ langur will decrease in all future periods under climate change. In the future climate prediction, the temperature in southwest China are expected to continue increasing, and show a more obvious difference in the increase between the northern and southern regions (Yang et al., 2022; Zhang, 2020).Climate change will lead to changes in species habitat, such as increased precipitation and more extreme precipitation events (Feng et al., 2023; Yue, 2022). From 2021 to 2100, the area of suitable habitat will greatly contracted, especially in the medium- and high-suitability areas along the border between Guizhou and Chongqing. Moreover, most of the patchy areas of low-suitability will cease to exist. The centroid of habitat shifted to the southeast at a rate of approximately 2.84km/year from current to 2100. Previous studies have indicated that changes in annual precipitation, annual mean temperature, precipitation and temperature seasonality, as well as land use/ land cover changes, could reduce suitable habitat for the large mammals such as Asiatic black bear (Ursus thibetanus ), Asian elephant (Elephas maximus ) and western hoolock gibbon (Hoolock hoolock ), and thus increase their extinction risk (Deb, 2016). In addition, our results are consistent with the description that primates and other large mammals are highly dependent on a complete forest ecosystem for their food and security requirements, and habitat fragmentation will accelerate local population extinctions (Guan et al., 2022). For Francois’ langur, habitat fragmentation caused by climate change and anthropogenic disturbance would be the most important factor in changing the distribution of this species.
4.4 Conservation recommendations
Our results showed that the current predicted habitats of Francois’ langur were over more than the history areas, but the distribution habitats of Francois’ langur would be threatened to contract and fragment dramatically in the future. We suggest to expand the survey range of their habitat, reduce human disturbance and enhance habitat connectivity by establishing ecological corridor especially the border of Chongqing and Guizhou and further protection on these areas. Moreover, conservation of the Francois’ langur should not be based only on the existing distribution areas, but should also focus on conservation beyond existing habitats, especially the expansion areas in the study, which should be considered to join the national programme as protected areas.
AUTHOR CONTRIBUTIONS
Yaqiong Wan: Conceptualization (lead); data curation (equal); formal analysis (equal); writing - original draft (lead); writing - review and editing (equal). Luanxin Li: Data curation (equal); writing - original draft (equal). Jiang Zhou: Investigation (equal); writing - review and editing (equal). Yue Ma: Data curation (equal); writing - original draft (equal). Yanjing Zhang: data curation (equal); methodology (equal). Yan Liu:writing - review and editing (equal). Jiaqi Li: Supervision (equal); writing - review and editing (equal). Wei Liu:Supervision (equal); writing - review and editing (equal).
ACKNOWLEDGMENTS
We thank the nature reserves for their support to collect data: Guangxi Xialei Nature Reserve, Guangxi Encheng Nature Reserve, Guangxi Chongzuo White-Headed Langur Nature Reserve, Guangxi Nonggang National Nature Reserve, Guizhou Dashahe Nature Reserve, Guizhou Yezhong Nature Reserve, Guizhou Mayanghe National Nature Reserve, Guizhou Kuankuoshui National Nature Reserve, and Chongqing Jinfoshan National Nature Reserve.
APPENDIXES
APPENDIX 1: Pearson’ s correlation coefficients between 19 climate variables, and intensity of color represents greater positive (blue) or negative correlation (orange).