Predicting suitable habitat of the Chinese monal (Lophophorus lhuysii) using ecological niche modeling in the Qionglai Mountains, China

Introduction

Understanding the distribution of suitable habitats and its influencing factors are crucial for wildlife conservation and management (Austin, 2002). Ecological niche models have developed as excellent tools for predicting habitat distribution of species that are difficult to investigate (Peterson, Ball & Cohoon, 2002; Mota-Vargas et al., 2013), because they can predict the distribution of species’ habitats at a large spatial scale based on species presence data and environmental variables, without the need for extensive surveys and detailed descriptions of physiological and behavioral characteristics (Morrison, Marcot & Mannan, 2012). In particular, MaxEnt, a presence-only modeling approach based on the maximum entropy principle (Phillips, Anderson & Schapire, 2006), outperforms other models in prediction accuracy (Elith et al., 2006; Phillips, Anderson & Schapire, 2006), transferability (Tuanmu et al., 2011), and performance with small sample sizes (Pearson et al., 2007; Costa et al., 2010). It offers great potential for addressing endangered and poorly known bird species with scarce occurrence data (e.g., Botero-Delgadillo, Páez & Bayly, 2012; Marcondes et al., 2014; Tobón-Sampedro & Rojas-Soto, 2015).

The Chinese monal (Lophophorus lhuysii) is the largest Galliform species (with a mean length of 76 cm and a mean weight of 3.18 kg) distributed in high-montane regions, mainly inhabiting subalpine scrubs, as well as subalpine and alpine meadows at an elevation of 3,000–4,900 m (MacKinnon, Phillipps & He, 2000; Madge, McGowan & Kirwan, 2002). This species is endemic to southwest China, and is found primarily in southeast Gansu, southeast Qinghai, western Sichuan, and northwest Yunnan (Lei & Lu, 2006; Lu, 2015). It has been listed on appendix I of CITES since 1975 (CITES, 2016) and classified as a vulnerable species on the IUCN red list since 1994 (IUCN, 2015). In 1989, the Chinese government started to legally protect the species as a first-class, nationally-protected wildlife species (Ministry of Forestry of People’s Republic of China, Ministry of Agriculture of People’s Republic of China, 1989). Recently, the Chinese monal was identified as one of the endemic bird species in China with the highest conservation values in terms of phylogenetic diversity (Chen, 2013). Its population size, however, remains small (10,000–25,000 individuals in total, BirdLife International, 2015) and is inferred to be in a state of continuous decline because of illegal hunting and on-going habitat degradation and fragmentation (BirdLife International, 2015). It is therefore necessary to establish effective conservation programs targeting the Chinese monal.

However, knowledge of the natural habitats of the Chinese monal is limited, as cryptic behavior and inaccessible habitats of the species make it difficult to survey the natural populations. Most of the previous studies were conducted in the late twentieth century and only provided simple descriptions of habitat use (Ma, 1989; Zhang, 1995), population density and structure (He & Lu, 1985; Long et al., 1998), breeding ecology (He et al., 1986), activity pattern (Ma, 1988), and feeding habits (Lu et al., 1986). There has not been a quantitative analysis of the natural habitats of the Chinese monal at a large spatial scale in any part of its geographical range and this limited knowledge has constrained conservation development for the Chinese monal.

The Qionglai Mountains are a part of the biodiversity hotspot of the Mountains of Southwest China (Mittermeier et al., 2011) that supports many endangered and endemic wildlife species. As the geographical center of the distribution of the Chinese monal (Lei & Lu, 2006; Lu, 2015; Fig. 1), this region is critical for the conservation of the species. In this study, we used the ecological niche modeling approach, with species records and predictors selected from a large set of environmental variables, to predict the current potential suitable habitats for the Chinese monal in the Qionglai Mountains. Our objectives were to (1) delineate the distribution of suitable habitats of the Chinese monal, (2) identifycritical environmental factors influencing the species’ habitat suitability, (3) compare the prediction with existing nature reserve network and provide conservation suggestions based on the results.

Materials & Methods

Results

Optimized variable selection showed that the model with the lowest AICc was built with: maximum temperature of the warmest month (bioclim5), precipitation of the driest quarter (bioclim17), annual maximum of EVI (evi maximum), base level values of EVI (evi base level), slope, and distances to residential locations (d_resident) and to roads (d_road) (Tables S3 and S4). The seven predictors were therefore selected for the final MaxEnt modeling. We obtained average values of model evaluation indices after 20 replicates: training AUC = 0.966, test AUC = 0.953, maximum Kappa = 0.813, and maximum TSS = 0.882. High scores of model evaluation indices, both the threshold-independent and threshold-dependent, indicated that the habitat suitability model produced by MaxEnt performed excellently (Manel, Williams & Ormerod, 2001; Araújo et al., 2005).

Relevant variables

Evaluation of percent contribution of each variable to the model illustrated that three predictors had over 10% relative contribution to the habitat suitability model of the Chinese monal, contributing 78% collectively (Table 2). Precipitation of the driest quarter (bioclim17) made the largest contribution, followed by maximum temperature of the warmest month (bioclim5) and distance to the nearest residential locations. Similarly, these three predictors received the highest AUC values when used in isolation in the jackknife test (Table 2), indicating they are better at discriminating suitable from non-suitable habitat as compared with the other variables. The response of logistic probability of monal occurrence on the two critical climatic predictors were both unimodal, that the probabilities were at the peak at 31 mm for precipitation of the driest quarter and at 15 °C for maximum temperature of the warmest month (Figs. 2A and 2B). The response curve also showed that monals were more likely to occur at sites further away (>5,000 m) from the nearest human residential locations (Fig. 2C).

Table 2:

Relative importance of environmental variables in the habitat suitability model of the Chinese monal.

Variables	Percent contributiona	Jackknife of AUCb
bioclim17	36.2	0.856
bioclim5	31.0	0.840
d_resident	10.5	0.771
evi base level	8.8	0.689
evi maximum	6.0	0.683
slope	4.8	0.631
d_road	2.7	0.642

DOI: 10.7717/peerj.3477/table-2

Notes:

aThe relative contribution of each variable to predictive model, shown as mean value of 20 replicates.

bJackknife test of variable importance, expressed as AUC (area under the receiver operating characteristic curve) for models using each variable alone. A higher gain indicates a variable with more information for modelling when used in isolation, shown as mean value of 20 replicates.

Note

bioclim17

precipitation of the driest quarter

bioclim5

maximum temperature of the warmest month

d_resident

distance to residential location

evi base value

annual base level value of EVI

evi maximum

annual maximum EVI

d_road

distance to roads

Habitat status

We created a binary suitable/unsuitable habitat map after applying the maximum TSS threshold (where habitat suitable index = 0.208). The suitable habitats for the Chinese monal covered an area of 2,490 km², corresponding to 9.48% of the entire Qionglai Mountains region. The majority of suitable habitats for the Chinese monal distributed in the central mountain regions, mainly stretching along the boundaries of Lixian-Wenchuan, Wenchuan-Baoxing, Xiaojin-Baoxing, and Baoxing-Tianquan (Fig. 3, Table 3). Ten existing nature reserves are located in the Qionglai Mountains, covering over a quarter of the entire region. Over 50% of the suitable monal habitats were situated within nature reserves. Wolong Nature Reserve provided the largest area of suitable habitats for the Chinese monal, while Heishuihe, Labahe, and Fengtongzhai Nature Reserve also had high proportions of suitable habitat (Table 4). Two general regions with large areas of potential suitable habitat were outside of any nature reserves: one in the northern region of Lixian County, and the other in northwest Baoxing County (Fig. 3).

Discussion

Our study represents the first attempt at predicting the suitable habitat of the Chinese monal. Variable analysis revealed that bioclimatic variables were the most influential predictors on the habitat suitability of the Chinese monal. Higher habitat suitability was constrained in narrow ranges of both precipitation of the driest quarter and maximum temperature of the warmest month, suggesting that Chinese monal was highly sensitive to climate under extreme periods. The monals appeared to prefer habitats far away from residential locations, suggesting human disturbance as a crucial pressure for the species. Although roads represent remarkable sources of disturbance to wildlife that could lead to road-kill (Mumme et al., 2000) and barrier effect (Shepard et al., 2008), the impact of roads was relatively weak in our niche model (Table 2). It is likely that a greater part of occurrence records were sampled in more easily accessible sites in proximity to roads (Fig. 1), and such a sampling bias had decreased the importance of roads disturbance in the model. Despite the sampling bias towards roads, we found an relation that habitat suitability increased with increasing distance from roads in the niche model predictions (Fig. 2D). As an explanation, we suggest that our predictive habitat suitability was more sensitive to bioclimatic conditions than to human disturbance; sites with bioclimatic conditions preferred by the Chinese monal might be infrequently crossed by roads.

The predicted suitable habitats for the Chinese monal were highly fragmented. Specialized niche requirements (i.e., narrow climate preferences) constrained their suitable habitats to subalpine and alpine regions. High mountain ridges and deep valleys throughout the Qionglai Mountains separated the suitable haibtats into small and isolated patches, especially for those distributed in the north of Lixian County, south of Baoxing County, and the boundary area of Lixian-Xiaojin-Wenchuan County (Fig. 3). As a strong disturbance, human residential areas further restricted and separated the suitable habitats. The fragmentation might have negative effects on the Chinese monal, which has low dispersal ability (With & Crist, 1995), reducing the probability of genetic exchange (Höglund et al., 2011) and colonizing suitable empty patches (Stamps, Buechner & Krishnan, 1987) by increasing the resistance to individual movement between isolated habitat patches.

Many subalpine meadows within the Qionglai Mountains are summer pasture for the grazing yaks. Grazing could lead to direct disturbance, accelerating habitat degradation and fragmentation, and increasing the probability of poaching (Lu et al., 1986; Ma, 1988; Long et al., 1998). Additionally, gathering of herbs for use in traditional Chinese medicine in spring and summer, especially the collection of Fritillaria spp., an important food source of the Chinese monal, could result in a reduction of food abundance and thus a indirect disturbance to the Chinese monal (Fuller & Garson, 2000; Lu, 2015). Our study, however, failed to incorporate these remarkable local threats into modeling due to the lack of data. If grazing and herb gathering are considered, the actual suitable monal habitat might be even smaller, more fragmented, and more variable than our prediction.

Our results raise concerns for the status of the Chinese monal in the Qionglai Mountains, and the species may be at greater risk than has previously been considered. As a typical high-montane species with large home range, the Chinese monal has the potential to serve as an umbrella or flagship species for high-montane ecosystems. Conservation of these ecosystems will likely contribute to the maintenance of regional biodiversity, at least for the montane Galliformes (Roberge & Angelstam, 2004; Rowland et al., 2006; McGowan, Zhang & Zhang, 2009). Based on our results, we have several conservation suggestions targeting the Chinese monal. Our prediction could guide future field surveys for locating new populations in the areas that were predicted to be suitable habitat but lack investigation (Raxworthy et al., 2003; Menon et al., 2010). The large and well-connected habitat patches distributed in the boundary area of Wenchuan, Baoxing, Lushan, and Dayi County appear to provide ideal and important habitats for the Chinese monal and should be treated as a priority area for conservation. However, a large portion stretching along the northern boundary of Baoxing County is unprotected by any nature reserves and lacks systematic survey (Fig. 3). Similarly, large ribbon-like patches of suitable habitats in northeast of Lixian County and north of Wenchuan County was also lacking protection and survey (Fig. 3). Our first suggestion, therefore, is to conduct ad hoc surveys for determining the actual occurrence of Chinese monal in these two areas. If there are indeed some populations in these two regions, we suggest that new nature reserves should be designated in order to fill these two obvious conservation gaps. The second suggestion is to study the dispersal capacity of the Chinese monal for further assessing habitat quality and fragmentation. Such ecological knowledge could be helpful for determining the actual distribution of animals in relation to the distribution of suitable environmental habitats (Pulliam, 2000; Lu et al., 2012). For instance, it will be useful to determine whether the Zagunao river valley is a geographic barrier that Chinese monals cannot cross through, and whether gene flow can occur between the monal populations isolated in the south and north of Baoxing County (Fig. 3).

Although human disturbance appeared to be a crucial pressure for Chinese monals (He et al., 1986), we found that Chinese monals did not respond avoiding roads. For instance, during our survey at Wenchuan in 2015, we observed that two Chinese monal pairs, almost at a daily bases, crossed provincial road 303 through the top of the Huayan tunnel (E102°58′, N30°51′; Fig. 1), a well-known location for shooting Chinese monal among wildlife photographers. Huayan tunnel has a length of 570 m, and the large area of natural shrub and meadow preserved on its top serves as a corridor for the monals. However, this does not mean that roads are not a significant threat for the species, as road-kill of Galliformes are not uncommon (Clevenger, Chruszcz & Gunson, 2003; D’Amico et al., 2015). Our third suggestion, therefore, is to build more corridors and green bridges across roads to facilitate road-crossing or dispersal by Chinese monals.

For the purpose of ‘game food’, Chinese monals had been often poached by local people before 1980s, which was considered as the main cause of a substantial decline of the species (He et al., 1986; Lu et al., 1986; Long et al., 1998; Lei & Lu, 2006). Unfortunately, we found self-made traps for capturing monals and other Galliformes during our surveys, both inside and outside nature reserves, suggesting that poaching is still continuing nowadays. Even though we found little evidence of human impact on monal habitat distribution, our field observation suggested that illegal hunting potentially threaten survival of the Chinese monals and their sympatric Galliform species. Moreover, other short-term and seasonal local disturbance such as yak grazing and herb collection could increase the probability of poaching (Lu et al., 1986; Ma, 1988; Long et al., 1998). Therefore, our fourth suggestion is to limit local disturbances, such as poaching, yak grazing, and herb collection, for maintaining existing populations and habitats of the Chinese monal as well as other montane Galliformes.

Conclusion

This study used the ecological niche modeling approach to predict current suitable habitat of the Chinese monal in Qionglai Mountains. Their suitable habitat was associated with about 31 mm precipitation of the driest quarter, about 15 °C of maximum temperature of the warmest month, and far from the nearest human residential locations (>5,000 m). The predicted suitable habitats of the Chinese monal was highly fragmented covering an area of 2,490 km². A total of 54.78% of suitable habitat was under the protection of existing nature reserve network, but there were obvious conservation gaps as two regions with large area of well-connected suitable habitats were out of any nature reserve. Finally, this study provide conservation management suggestions in terms of ad hoc surveys targeting potential suitable habitats to determine occurrence of Chinese monals, more ecological studies regarding its dispersal capacity, establishment of more corridors and green bridges across roads in suitable habitats and limitation of local disturbances such as poaching, yak grazing, and herb collection.

Supplemental Information