Habitat Distribution Pattern of François’ Langur in a Human-Dominated Karst Landscape: Implications for Its Conservation

Abstract

The Mayanghe National Nature Reserve, a key habitat for the endangered François’ langur (Trachypithecus francoisi), faces significant anthropogenic disturbances, including extensive distribution of croplands, roads, and settlements. These human-modified features are predominantly concentrated at elevations between 500 and 800 m and on slopes of 10–20°, which notably overlap with the core elevation range utilized by François’ langur. Spatial analysis revealed that langurs primarily occupy areas within the 500–800 m elevation band, which comprises only 33% of the reserve but hosts a high density of human infrastructure—including approximately 4468 residential buildings and the majority of cropland and road networks. Despite slopes >60° representing just 18.52% of the area, langur habitat utilization peaked in these steep regions (exceeding 85.71%), indicating a strong preference for rugged karst terrain, likely due to reduced human interference. Habitat type analysis showed a clear preference for evergreen broadleaf forests (covering 37.19% of utilized areas), followed by shrublands. Landscape pattern metrics revealed high habitat fragmentation, with 457 discrete habitat patches and broadleaf forests displaying the highest edge density and total edge length. Connectivity analyses indicated that distribution areas exhibit a more continuous and aggregated habitat configuration than control areas. These results underscore François’ langur’s reliance on steep, forested karst habitats and highlight the urgent need to mitigate human-induced fragmentation in key elevation and slope zones to ensure the species’ long-term survival.

1. Introduction

Spatial utilization, referring to how wildlife access and use resources across heterogeneous habitats, is a key determinant of gene flow, population viability, and long-term species survival [1]. As a core concept in wildlife ecology, it provides essential theoretical grounding for biodiversity conservation efforts [1,2]. Traditionally, studies have emphasized home range estimation and movement patterns to infer habitat suitability [3,4], using field surveys [5,6], radio telemetry and GPS tracking [7,8], and camera trapping [9]. However, for highly arboreal and cryptic primates such as François’ langur (Trachypithecus francoisi), an endangered colobine species endemic to the karst forests of Southwest China and northern Vietnam, traditional ground-based tracking methods remain challenging due to their preference for steep, forested limestone terrains and their elusive behavior. Consequently, research on this species has primarily relied on direct observations focused on spatial use, sleeping site selection, and seasonal movement patterns [10,11,12,13]. Foundational work in Fusui, Mayanghe, has laid the groundwork for understanding their spatial ecology [13,14].

Landscapes are composed of mosaics of land cover patches, defined by both composition (types and proportions of habitat) and configuration (spatial arrangement of components) [15,16]. Studies typically adopt patch-based or landscape-scale approaches to quantify the responses of wildlife to landscape variables such as forest area, matrix quality, or edge effects [17,18,19,20], often supported by remote sensing techniques [4,8,21]. Primates are especially vulnerable to dynamic anthropogenic disturbances, including habitat loss and fragmentation [22,23,24], and their persistence depends on effective landscape-scale management strategies [25,26]. However, several challenges persist: narrow spatial extents of analyses [25,26], difficulties in separating interacting landscape effects [27], and varying species-specific responses to land-use change [27,28,29]. Moreover, scale sensitivity remains a key concern—different primate species exhibit habitat associations at vastly different spatial scales [30,31], necessitating scale-appropriate landscape assessments [15,32].

François’ langur, a Class I protected species under China’s Wildlife Protection Law and listed as Endangered (EN) on the IUCN Red List, persists in small, isolated populations across northern Vietnam and China’s Guangxi, Guizhou, and Chongqing provinces, facing critically declining numbers due to poaching, deforestation, infrastructure development, and agricultural expansion [33,34,35,36]. Habitat degradation and fragmentation have severely compromised population connectivity, making the maintenance of functional landscapes and inter-patch corridors essential for genetic exchange and long-term viability [37]. In this context, landscape genetics has emerged as a powerful tool to assess ecological connectivity under complex landscape conditions [37]. The Mayanghe National Nature Reserve (MNNR), which hosts the largest remaining population of François’ langurs, represents a crucial research platform. Quantitative analysis of spatial utilization and anthropogenic disturbances in this reserve will not only deepen our understanding of the species’ ecological needs, but also provide evidence-based guidance for conservation planning. This study proposes an integrative and replicable framework that combines disturbance ecology with habitat suitability modeling, aiming to support conservation of François’ langur and inform landscape-scale strategies for other endangered primates.

To better understand the impacts of habitat fragmentation on François’ langur, this study integrates field observations with landscape metrics and spatial modeling in the MNNR. We aim to quantify how landscape composition, spatial configuration, and human disturbances influence habitat use, identify key habitat features, and assess the extent of anthropogenic impacts. The following sections introduce the study area, data collection, analytical methods, and modeling framework used to address these objectives.

2. Materials and Methods

2.1. Study Area

The study was conducted in the MNNR, located in the transitional zone between Yanhe Tujia Autonomous County and Wuchuan Gelao and Miao Autonomous County in Guizhou Province, China. The reserve lies between latitudes 28°37′30″ N and 28°54′20″ N, and longitudes 108°03′58″ E and 108°19′45″ E, covering a total area of 311.13 km², including a core protection zone of 105.43 km² (Figure 1). The terrain within the reserve is dominated by typical karst landforms, with elevations ranging from 280 m to 1441 m [13]. The MNNR experiences a warm and humid mid-subtropical monsoon climate, characterized by abundant rainfall, moderate humidity, sufficient thermal resources, distinct seasonal changes, a short frost period, and an extended growing season [35]. The MNNR represents a complex socio-ecological landscape, spanning 7 townships and 40 administrative villages [35]. Conservation and management efforts, particularly for François’ langur populations, are overseen by the Mayanghe National Nature Reserve Administration. This work is carried out at three field management and monitoring substations—Liangqiao, Gongxikou, and Wuchuan—which are coordinated under a unified administrative station.

2.2. Presence Data Collection

From October 2017 to August 2018, presence data of François’ langur were collected within the MNNR. The complex and rugged limestone karst topography limited the applicability of random or uniformly spaced transect designs for field surveys [12,13]. Instead, presence locations were identified based on GPS coordinates obtained from multiple sources, including direct sightings, sleeping sites, foraging traces, fecal remains, and reports from experienced reserve personnel who have extensive field knowledge of François’ langur behavior and habitat use. To reduce the risk of model overfitting, a 700 m buffer was applied around each presence point. When buffer zones overlapped, one point was randomly selected within the overlapping area. This buffer distance reflects the average daily travel range of François’ langur [13]. After this filtering process, the initial 229 presence records were consolidated into 98 spatially independent presence sites.

2.3. Anthropogenic Disturbance Distribution Data

Anthropogenic disturbance data, including settlement, road, and cropland distribution, were extracted using a combination of remote sensing and field-based methods. Building footprints and road networks within the study area were manually digitized using Google Earth Pro—with settlements as point features (one point per structure) and roads as line features. The digitized data were exported in Keyhole Markup Language (KML) format, imported into ArcGIS, and converted into Shapefile (SHP) format using standard conversion tools. Cropland distribution was mapped through field verification integrated with existing spatial datasets. The researchers utilized the sub-compartment (Xiaoban) polygon layer from the MNNR’s 2015 Second-Class Forest Resource Inventory, preloaded onto mobile field devices. During ground-truthing surveys, cropland areas were identified, and their boundaries were verified and annotated directly on the Xiaoban layer using a mobile GIS application. The field-verified vector data was then topologically corrected and integrated into the spatial analysis through overlay operations (e.g., intersection and union) with other environmental layers, ensuring consistency between ground observations and analytical outputs. Ensuring accurate and up-to-date land-use information.

2.4. Distribution Area Estimation

Kernel Density Estimation (KDE), a non-parametric method for estimating probability density functions and associated home ranges based on presence data [38,39,40] (Kernohan et al., 2001; Hemson et al., 2005), was employed to estimate the spatial use patterns of the species. KDE has been widely applied in wildlife studies, including home range estimation for giant pandas [4]. The KDE probability density function is defined as [4,38]:

$${\widehat{f}}_{\left(x\right)}={\displaystyle \frac{1}{n{h}^2}}{\displaystyle \sum }_{i=1}^n K\left[{\displaystyle \frac{x-X_i}{h}}\right]$$

where ${\widehat{f}}_{\left(x\right)}$ is the estimated density at a location, n is the total number of presence points, h is the smoothing parameter (bandwidth), X_i represents the i-th presence location, and K [.] is a two-dimensional symmetric kernel function [4,38]. The optimal bandwidth h was determined using Likelihood Cross-Validation (CVh), as implemented in Animal Space Use 1.3 Beta [4,38]. The resulting home range surfaces were generated and visualized using ArcGIS 10.5.

Spatial Analysis of Habitat Characteristics and Anthropogenic Disturbance

Using ArcGIS, we conducted integrated spatial analyses to quantify habitat characteristics and anthropogenic disturbance patterns across the karst landscape of the MNNR. To assess habitat factors, the study area was stratified into standardized topographic and land cover classes as follows:

Elevation (E_i): ≤500 m, 501–800 m, 801–1100 m, >1100 m.

Slope (S_i): <10°, 10–20°, 20–30°, 30–40°, 40–50°, 50–60°, >60°.

Aspect (A_i): Eight cardinal directions (N, NE, E, SE, S, SW, W, NW).

Land cover (V_i): Coniferous forest, mixed forest, broadleaf forest, shrubland, bamboo forest, cropland, and other types (based on data provided by the Administration).

Zonal statistics were used to quantify the areal distribution of each habitat class, while anthropogenic disturbances (settlements, roads, cropland) were analyzed across the same elevation categories and a simplified set of slope classes. The François’ langur distribution (from KDE analysis) was then overlaid with all habitat and disturbance layers to assess (i) habitat selection patterns across elevation (E_i), slope (S_i), aspect (A_i), and (ii) disturbance exposure gradients along the elevational and slope dimensions.

2.5. Landscape Pattern Analysis

Landscape pattern analysis was employed to quantify the spatial configuration and composition of the species’ distribution area, adopting a hierarchical framework comprising landscape-level and class-level metrics. Although patch-level indices characterize individual patch features, they were excluded from this study due to their limited relevance to the broader landscape dynamics being examined. The analysis was conducted using FRAGSTATS 4.2 software, with a focus on metrics that capture spatial heterogeneity and fragmentation at both the class and landscape levels (

Table 1. Description of class-level and landscape-level metrics used in the landscape pattern analysis.

Metric	Full Name	Unit	Description
CA	Class Area	m2	Total area of a specific land cover or landscape class.
NP	Number of Patches	Count	Number of discrete patches of a particular class, indicating fragmentation.
MPS	Mean Patch Size	m2	Average area of all patches in a given class (MPS = CA/NP).
PLAND	Percentage of Landscape	%	Proportion of the landscape occupied by a specific class.
LPI	Largest Patch Index	%	Proportion of the total landscape area comprising the largest patch.
ED	Edge Density	m/ha or m/km2	Total edge length per unit area, indicating boundary complexity.
TE	Total Edge	m	Sum of all edge lengths in the landscape.
CONTAG	Contagion Index	Dimensionless (0–100)	Measures the extent of clumping or interspersion among patch types.
SHDI	Shannon’s Diversity Index	Dimensionless	Quantifies landscape diversity based on patch type richness and evenness.
SHEI	Shannon’s Evenness Index	Dimensionless (0–1)	Measures the evenness of the distribution among patch types.
AI	Aggregation Index	%	Indicates the degree to which patches of the same class are aggregated.

).

2.6. Habitat Classification and Metric Calculation

Habitat patches were classified by overlaying the François’ langur habitat distribution (derived from KDE) with land cover data in ArcGIS 10.3. Using the Clip tool, land cover types within langur-occupied areas were extracted and reclassified into the following habitat quality categories: optimal habitat—broadleaf forest; suitable habitat—shrubland and mixed forest; marginal habitat—coniferous forest and cropland. The landscape metrics were then computed separately for each habitat class using FRAGSTATS, enabling us to evaluate the spatial structure and fragmentation patterns associated with varying habitat suitability.

3. Result

3.1. Topographic and Land Cover Differentiation in the MNNR

Statistical analysis of the DEM and land-cover layers revealed elevations between 500–800 m encompassing 101.27 km² (32.55%) of the reserve (

Table 2. Summary of topographic features, land cover types, and human disturbance distribution in Mayanghe National Nature Reserve.

(A) Topographic Conditions
Category	Code	Class (Unit)	Area (km2)	Proportion (%)
Elevation	E1	<500 m	34.10	10.96
	E2	500–800 m	101.27	32.55
	E3	800–1100 m	122.06	39.23
	E4	>1100 m	53.70	17.26
Slope	S1	<10°	43.90	14.11
	S2	10–20°	117.12	37.64
	S3	20–30°	92.49	29.73
	S4	30–40°	38.92	12.51
	S5	40–50°	14.63	4.70
	S6	50–60°	3.52	1.13
	S7	>60°	0.56	0.18
Aspect	A1–A8	All directions	–	Evenly distributed across directions; no dominant aspect detected
(B) Land Cover Composition
Code	Land Cover Type		Area (km2)	Proportion (%)
V1	Cultivated land		81.47	26.18
V2	Shrubland		52.29	16.81
V3	Others (bare land, construction)		13.09	4.21
V4	Broadleaf forest		73.71	23.69
V5	Mixed broadleaf–coniferous forest		50.99	16.39
V6	Coniferous forest		38.87	12.49
V7	Bamboo		0.72	0.23
(C)Human Disturbance Distribution by Elevation and Slope
Type	Category	Code	Class	Value	Proportion (%)
Cultivated land	Slope	S1	<10°	21.19 km2	20.47
	Slope	S2	10–20°	49.28 km2	47.60
	Slope	S3	20–30°	25.66 km2	24.79
	Elevation	E2	500–800 m	39.82 km2	38.46
	Elevation	E3	800–1100 m	35.51 km2	34.30
Roads	Slope	S2	10–20°	176.90 km	43.60
	Elevation	E2	500–800 m	155.42 km	38.31
Settlements	Slope	S2	10–20°	1966 sites	44.00
	Elevation	E2	500–800 m	1966 sites	44.00

A), while the 800–1100 m band covers an even larger 122.06 km² (39.23%), jointly accounting for more than 70% of the total area. Slope classes are similarly concentrated (Table 2A): 10–20° slopes occupy 117.12 km² (37.64%), and 20–30° slopes another 92.49 km² (29.73%). Extremely steep terrain (>60°) is rare, representing only 0.18% of the reserve. Slope aspect differentiation is minimal, with no single orientation exceeding 20% of total area, indicating a broadly even azimuthal distribution. Land cover patterns reveal pronounced human influence. Cropland is the most extensive class, covering 81.46 km² (26.18%), followed closely by broadleaf forest at 73.71 km² (23.69%) (Table 2B). Overall forest vegetation (broadleaf + coniferous + mixed types) amounts to 142.87 km² (45.92%), providing nearly half of the reserve’s surface with woody cover.

Human disturbances, including cultivated land, roads, and settlements, were predominantly distributed in areas with moderate slopes and mid-elevation zones. Cultivated land was most extensive on slopes of 10–20° (47.60%), followed by those of 20–30° (24.79%) and <10° (20.47%). In terms of elevation, 38.46% of cultivated land was located at 500–800 m, and 34.30% at 800–1100 m. Similarly, road networks were concentrated on slopes of 10–20° (43.60%) and elevations of 500–800 m (38.31%). Settlements showed a highly clustered pattern, with 44.00% located in the 10–20° slope range and the same proportion within an elevation range of 500–800 m.

3.2. Habitat and Anthropogenic Disturbance Distribution Patterns

3.2.1. Altitudinal Patterns

The main distribution zone was concentrated at 500–800 m, accounting for 33.42 km² (48.53%) of the total distribution range (Figure 2A). Secondary distribution occurred below 500 m (25.84%) and between 800 and 1100 m (25.20%), while areas above 1100 m showed minimal use (0.45%). Distribution intensity—defined as the proportion of available area actually used—was highest at elevations below 500 m (52.17%) and declined with increasing elevation, reaching only 0.45% above 1100 m (Figure 3A). Anthropogenic disturbances such as cropland, roads, and settlements were also concentrated in mid-elevation zones. Cropland was most abundant between 500 and 800 m (39.82 km², 38.46%) and between 800 and 1100 m (35.51 km², 34.30%) (Figure 4A), while roads were most densely distributed in the 500–800 m range (155.42 km, 38.31%) (Figure 4A). Similarly, 44% of all settlements (n = 1966) were situated within the 500–800 m elevation range, with an additional 32.8% (n = 1466) found between 800 and 1100 m (Figure 4A).

3.2.2. Slope Gradient Distribution

François’ langur used all slope classes but exhibited non-linear preferences. The largest occupied area occurred on slopes of 20–30° (20.99 km², 30.56%) (Figure 2B), with utilization intensity increasing with slope steepness. While slopes < 10° showed only 10.57% usage, extremely steep slopes (>60°)—despite their limited availability—showed the highest relative intensity (85.71%) (Figure 3B), indicating behavioral adaptation to avoid heavily disturbed, flatter areas. Cropland was primarily distributed on gentle to moderate slopes (Figure 4B), with 47.6% located on 10–20° slopes (49.28 km²), and smaller proportions on <10° (20.47%) and 20–30° (24.79%). Roads followed a similar pattern (Figure 4B), with 43.6% of their total length on 10–20° slopes and only a minimal distribution in steeper areas. Settlements were also concentrated on slopes of 10–20° (54.03%) and <10° (36.93%) (Figure 4B).

3.2.3. Slope Aspect Neutrality

François’ langur showed no significant preference across aspect classes (Figure 2C). Their habitat was evenly distributed across all eight cardinal directions, with each sector accounting for approximately 10% of total use, and utilization intensity hovering around 20% (Figure 5A). This suggests a general behavioral indifference to solar orientation, likely due to the complex terrain and forest cover buffering microclimatic variation. No obvious variation in slope selection was observed under different levels of human disturbance (Figure 5).

3.2.4. Vegetation Type and Landscape Pattern Characteristics of François’ Langur Habitat Distribution

The analysis of vegetation use revealed a clear preference by François’ langurs for specific habitat types, indicating strong vegetation stratification in habitat selection (Figure 2D). Broadleaf forest was the most heavily selected vegetation type, encompassing 25.57 km², which accounted for 37.19% of the total langur distribution area (Figure 3C). This habitat also showed the highest relative distribution intensity, with François’ langurs occupying 34.69% of all available broadleaf forest in the reserve. Shrubland functioned as a secondary habitat, supporting 13.89 km² of langur activity (20.20% of total range), with a moderate utilization rate of 26.20%.

The landscape metrics revealed distinct spatial patterns between the used and control areas across the three habitat types (H1, H2, H3) (

Table 3. Basic landscape pattern indices of François’ langur between used and control areas in Mayanghe Nature Reserve.

Type	CA (m2)		NP		MPS		PLAND		LPI		ED		TE (m)
	Used Area	Control Area	Used Area	Control Area	Used Area	Control Area	Used Area	Control Area	Used Area	Control Area	Used Area	Control Area	Used Area	Control Area
Optimal	2560.23	5860.71	80	315	32.00	18.61	37.23	20.43	20.56	8.98	28.53	23.00	196,260	659,880
Suitable	2579.49	9313.74	110	474	23.45	19.65	37.51	32.47	10.23	4.96	44.21	47.86	304,080	1,372,860
Marginal	1503.36	12,251.43	267	803	5.63	15.26	21.86	42.71	1.74	6.97	43.93	76.01	302,130	2,180,460

). In distribution areas, NP was consistently lower (H1: 80; H2: 110; H3: 267) and the MPS notably larger (H1: 32.00 ha; H2: 23.45 ha; H3: 5.63 ha) than in control areas. PLAND was higher in used areas (H1: 37.23%; H2: 37.51%; H3: 21.86%), suggesting selection for areas with higher habitat availability. Interestingly, LPI was lower in used areas. ED showed mixed trends—lower in H1, but higher in H2 and H3—indicating variable edge complexity among habitats. The total edge (TE) and class area (CA) were consistently lower in distribution areas. At the landscape level, CONTAG was slightly higher in the distribution area (49.01) than in the control area (47.46) (

Table 4. Aggregaton varation characteristics of François’ langur space use patterns in Mayanghe Nature Reserve.

Type	CONTAG	SHDI	SHEI	AI
Control area	47.46	1.20	0.75	86.10
Distribution area	49.01	1.21	0.74	88.06

), indicating a more clumped and cohesive landscape where habitat patches are more aggregated. SHDI and SHEI showed minimal differences between areas (SHDI: 1.21 vs. 1.20; SHEI: 0.74 vs. 0.75), suggesting comparable levels of landscape diversity and compositional evenness. AI was also higher in the distribution area (88.06) than in the control area (86.10), reflecting a higher degree of spatial connectivity and lower fragmentation. These results suggest that the distribution area exhibits a more continuous and aggregated habitat configuration, which may support improved habitat quality and facilitate species movement.

4. Discussion

Understanding the spatial utilization patterns of François’ langur requires an integrated analysis of biotic, topographic, and anthropogenic factors. In the MNNR, these factors interact in complex ways to shape habitat use, with topography playing a foundational role in determining the spatial distribution of the species. Given that slope and elevation are largely constant over ecological timescales, they serve as reliable predictors of habitat suitability. Our findings support previous research indicating that François’ langurs predominantly inhabit steep slopes (≥30°), particularly in riverine areas characterized by cliffs and caves, which offer critical shelter and protection [12,35,41,42].

The preference for these steep, rugged habitats is likely influenced not only by topographic stability but also by vegetation characteristics. The inaccessibility of these areas for agricultural expansion and logging has enabled the persistence of evergreen broadleaf forests with high plant diversity, offering year-round foraging opportunities. In contrast, low-slope areas further from rivers have been extensively modified by human activities, including farmland expansion and settlement development [41]. While langurs occasionally exploit crops such as sweet potatoes, corn, and fruit during specific seasons, these resources serve as supplementary rather than staple dietary components.

Vegetation structure also plays a critical role in both nutrition and predator avoidance. Consistent with earlier studies [11,43,44], our results confirm that François’ langurs are primarily folivorous, relying heavily on the abundant foliage in broadleaf forests [13]. These forests also provide vital cover, as observed in alarm responses and infant-carrying behavior during raptor encounters [45]. Shrublands, which are typically found on steep slopes where tall tree growth is limited, serve as important seasonal foraging grounds due to the presence of fruit-bearing species [14,43]. Conversely, coniferous forests are rarely utilized, likely due to their low food value. The anthropogenic disturbances in the MNNR are increasingly pronounced, particularly in buffer and experimental zones. The extensive road network, high settlement density, and recent infrastructure developments driven by poverty alleviation policies have altered the landscape significantly. However, there are indications of positive change. In historically disturbed areas, shifts in local energy sources from firewood to electricity have led to notable reductions in vegetation harvesting.

Elevation was another important factor influencing langur distribution, with core activity concentrated between 500 and 800 m. Limited use of areas above 1100 m likely reflects both climatic constraints and reduced vegetation quality. Interestingly, slope aspect—a factor known to influence habitat use in many large terrestrial mammals—did not significantly affect langur distribution in this study, possibly due to the species’ relatively small home ranges and more specialized habitat preferences [46]. Overall, the spatial ecology of François’ langurs in the MNNR underscores the importance of steep, forested habitats that offer both food and refuge, and highlights the dual influence of persistent topographic conditions and dynamic human activities [35,47]. These findings provide a valuable basis for future conservation planning, emphasizing the need to protect steep-slope broadleaf forests and mitigate the expansion of anthropogenic infrastructure in ecologically sensitive areas.

Landscape pattern metrics reveal that fragmentation is particularly pronounced in suboptimal habitats, such as shrublands and mixed coniferous–broadleaf forests. These areas commonly occur in ecotonal zones—transitions between farmlands and optimal habitats or between core and marginal habitat patches. Fragmentation within these transitional zones is further compounded by complex terrain, which generates highly heterogeneous habitat mosaics. Such discontinuity limits the suitability of these patches for sustained occupancy by François’ langur, thereby reducing the overall ecological resilience of the landscape. Based on spatial pattern and composition, habitat quality across the reserve can generally be categorized as medium to low in terms of conservation viability. To mitigate ongoing degradation and promote long-term species persistence, targeted conservation measures are urgently needed. First, stricter control of anthropogenic disturbances is essential, particularly by limiting new infrastructure development in ecologically sensitive zones. Second, ecological restoration should prioritize degraded corridors and fragmented patches to re-establish landscape connectivity. Third, regional development policies must balance socioeconomic development with biodiversity conservation, especially in areas adjacent to key habitat zones. Moreover, interspecific competition further exacerbates habitat stress, spatial overlap, and resource competition between François’ langurs and domestic goats in certain parts of the reserve [36]. Intensive grazing not only accelerates vegetation degradation in shrublands, but also increases habitat fragmentation and edge effects in transitional zones [12,43].

This study provides valuable insights into the spatial distribution and habitat preferences of François’ langurs in a karst environment. However, several limitations should be noted. First, due to the challenging terrain and limited accessibility, data collection relied partly on indirect evidence and opportunistic observations, which may underrepresent some habitat use patterns. Second, the study did not incorporate detailed behavioral data or seasonal variations in habitat use, which could offer a more comprehensive understanding of ecological needs. Future research should prioritize long-term monitoring using a combination of GPS collar tracking and camera traps, along with systematic behavioral studies. Additionally, integrating habitat quality metrics and anthropogenic disturbance assessments would further enhance conservation planning for this endangered species.

Implications for Conservation

To enhance habitat suitability and support population viability, conservation efforts should prioritize the protection of core forest areas, restrict further infrastructure development, and restore degraded landscapes. Specific strategies include regulating grazing to promote vegetation regeneration and managing ecotonal zones to reduce edge effects and maintain habitat integrity.

5. Conclusions

This study reveals that François’ langurs in the MNNR preferentially utilize elevations of 500–800 m and slopes of 20–30°, with a strong association with evergreen broadleaf forests that provide essential food and shelter. Despite the dominance of broadleaf forest, the habitat is highly fragmented, with 457 discrete patches and limited cohesion, particularly in suboptimal areas characterized by high edge density and poor connectivity. These patterns highlight significant ecological vulnerability.