Potential impact of land‐use change on habitat quality in the distribution range of crocodile lizards in China

Abstract The over‐exploitation of land resources poses a serious threat to biodiversity on a global scale. Changes in land‐use and human exploitation have had a major impact on wild populations and their habitat in China. We assessed how habitat quality has changed over time (1995–2020). Specifically, we analyzed how the habitat quality of crocodile lizard has changed over time based on multi‐temporal land‐use data (1995, 2000, 2010, 2015 and 2020) using a land‐use transfer matrix and habitat quality model. The results showed that the main landscape types in the study area were arable land (21.21% of the area) and woodland (69.59% of the area) during the period. Construction land (land used for development) had decreased by 991 km2, a decrease rate of 59.84% from 1995 to 2000, and increased to 2349 km2, an increase rate of 71.69% from 2000 to 2020. The proportion of grasslands and areas with water were negligible and overall, did not vary significantly in size over the study period. The main feature of land use change in the study area was the loss of grasslands and woodlands through development. The habitat quality model indicated that habitat quality was highest and degradation was lowest in Dayao mountain, Guxiu town, Qichong village and Beituo town. Habitat quality improved in Daguishan and Luokeng areas. Habitat quality was good in Daping mountain and Linzhouding, but they were highly fragmented with patches of low‐quality habitat of varying sizes. Habitats were severely degraded in the Dateng Gorge area. The rate of habitat degradation has slowed over time in the study area, but gradually increased in degradation intensity, and low‐quality habitats were widely distributed and overlapped with the crocodile lizards distribution area. We recommend that protected areas for the crocodile lizard be more closely monitored and managed to halt further decline in habitat quality.


| INTRODUC TI ON
Habitat quality refers to the ability of an ecosystem in a specific time and space to provide suitable and sustainable environments for organisms (Regolin et al., 2021). Habitat quality and availability can be used as proxies for biodiversity (Sharp et al., 2018). Understanding the spatiotemporal variability of habitat quality is important for expanding ecological conservation of wildlife (e.g., protect genetic diversity, predict population dynamics) (Crawford & Nusha, 2018; Thornton et al., 2013). In general, habitat quality varies with the intensity of nearby land use (Liu et al., 2022). Land use types, intensities and patterns alter the condition of natural resources and thus affect the survival and reproduction of wildlife (Dai et al., 2019;Whittington et al., 2019). Biogeochemical cycles and habitat quality for animals and plants are changed because of increased human disturbance (Abbott et al., 2019;Kiskaddon et al., 2019;Lin et al., 2020;Powers & Jetz, 2019). And changes in these cycling processes may have adverse effects on the structure and function of ecosystems.
With urban expansion and development of land in developing countries, habitat quality is increasingly influenced at the landscape level, which has made habitat conservation to be an urgent issue (Liu et al., 2022).
Urbanization and industrialization have accelerated since the 20th century, and the over-exploitation of land resources poses a severe threat to biodiversity (Deng et al., 2021). Because overexploitation of land can result in habitat degradation, fragmentation and loss (Brudvig et al., 2015). Several studies have concluded that land-use and land-cover changes (LULCC) activities are intensifying, and that wildlife habitat is increasingly being developed for agriculture and infrastructure (Jha & Bawa, 2006;Karki et al., 2018;Khan et al., 2021;Newbold et al., 2015). Evidence from different taxa and geographical regions suggested that land-use was not equally affected all organisms in terrestrial ecological communities and that different functional groups of species may respond differently (Felipe-Lucia et al., 2020;Newbold et al., 2020). The Researchers expected large carnivore populations to decline more in disturbed land than other animal groups (Newbold et al., 2020). However, amphibians and reptiles are the two most vulnerable groups of terrestrial vertebrates, being at a significantly higher risk than mammals and birds for threats such as habitat loss and fragmentation (Mayani-Parás et al., 2019). Amphibians and reptiles generally have low dispersal abilities and are more habitat specialists than other vertebrates, making them particularly sensitive to landscape changes (Audrey et al., 2016;Joly et al., 2003;Wang et al., 2020). Therefore, habitat degradation and destruction are the focus of amphibian conservation. Despite many related studies have been conducted in mammals, birds, amphibians, it is amazing that little attention has been paid on reptiles (Gibbons et al., 2000) and are likely to be at a high risk of extinction (IUCN, 2006). The destruction and fragmentation of habitats reduce the structural complexity and functional integrity of habitats occupied by reptiles (Liu et al., 2016).
Several factors, such as habitat loss, water pollution, climate change and mining, have been identified as negatively affecting breeding activities, reproduction and survival for reptiles (Becker et al., 2007;Gardner et al., 2007). These processes cause significant interference to the survival, reproduction and spread of reptiles, affecting species composition and community structure (Hung et al., 2017).
Lately, these processes have also caused population declines due to the obstruction of population genetic exchange, reducing the range size of the species and resulting in local population extirpation (Mayani-Parás et al., 2019).
Thus, effectively assessing and monitoring biodiversity and habitat quality changes and identifying the mechanisms causing these changes are essential for ecological management in fastchanging and human-dominated regions (Sun et al., 2019). There are three primary methods commonly used to evaluate changes in biodiversity and habitat quality: traditional field and habitat surveys (Do Nascimento et al., 2020), assessments of ecological indicators (Coates et al., 2016;Riedler & Lang, 2018) and simulations using ecological models (Akbari et al., 2021;Sallustio et al., 2017). Traditional terrestrial habitat monitoring methods are often time consuming, and their accuracy is difficult to assess due to differences between subjects (Lengyel et al., 2008). The Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) when used to evaluate biodiversity indicators or proxies of biodiversity, is a powerful tool to monitor biodiversity dynamics and habitat quality, especially in areas with limited available data on biodiversity (Sharp et al., 2016).
Among the InVEST models, the habitat quality assessment model relies on the proximity of habitats to human land-use and the intensity of land-use (Sharp et al., 2018). Habitat quality is affected by habitat suitability, threats due to habitat quality reduction factors, habitat sensitivity to reduction factors and access to the habitat (Lee & Jeon, 2020). InVEST models introduced habitat quality as a proxy for biodiversity assessment (Gong et al., 2019). This approach allows for a rapid assessment of the status and changes in biodiversity status as a proxy for a more detailed biodiversity status (Sun et al., 2019).
The crocodile lizard (Shinisaurus crocodilurus Ahl, 1930) is a monotypic species in the monotypic family Shinisauridae. It is an ancient lineage from the Pleistocene, with ~100 million years of evolutionary history (Xie et al., 2021). Individuals of this species are diurnal, semiaquatic, viviparous and occur in rocky streams in cool mountain forests in southern China and northern Vietnam (Huang et al., 2008;van Schingen, Schepp, et al., 2015). The species is threatened with extinction due to continued deforestation, habitat destruction and poaching. As such, it is listed as endangered by the International Union for Conservation of Nature (IUCN) . Here, in order to fully analyze changes in habitat quality across the crocodile lizards distribution range, in conjunction with a habitat quality model, we set two main objectives: (1) analyzing land-use change in the study area from 1995 to 2020 and (2) assessing habitat quality in the crocodile lizards distribution areas of Guangdong and Guangxi.

| Study area
The distribution range of crocodile lizards is restricted to southern China and northern Vietnam, where suitable habitat consists of small, isolated, fragmented and steadily shrinking habitat patches (Huang et al., 2008;Le & Ziegler, 2003;van Schingen, Ihlow, et al., 2014). Within Guangdong and Guangxi, populations are relatively scattered and far apart ( Figure 1). Therefore, we selected part of the Pearl River Basin as the primary research area, including all crocodile lizards distribution areas (102°14′ to 115°53′ E, 21°31′ to 26°49′ N). This river spans the Yunnan-Guizhou Plateau, the hills of Guangdong and Guangxi and the Pearl River Delta Plain from west to east (He et al., 2018;Wang et al., 2021). The climate in the study region has subtropical monsoon features, where the annual average temperature is approximately 14-21°C. And annual precipitation ranges between 1200 and 2200 mm , decreasing from southeast to northwest and primarily falling during April-September. The dominant vegetation is composed of evergreen forests (~65.3%), followed by cropland (~18.1%) .
They are mainly distributed in the middle of the basin, which happens to be in the transitional areas of high-to-low elevations in Guangxi province ). Land-use and land-cover maps from 1995Land-use and land-cover maps from , 2000Land-use and land-cover maps from , 2010Land-use and land-cover maps from , 2015Land-use and land-cover maps from and 2020 (1 × 1 km) were used in this research. Data from the crocodile lizards' distribution area mainly include the Dayao mountain, the Guxiu area, the Mengshan area, the Qichong area, the Beituo area, the Daguishan area, the Luokeng area and the Maoming. Crocodile lizards have been reported from all of these areas (Huang et al., 2008;Zhang, 1991). County-level administrative zoning map and protected area boundary data were analyzed. The county-level administrative zoning map was obtained from the Ministry of Natural Resources of China (http://bzdt.ch.mnr.gov.cn). Land-use and land-cover maps came from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn).

| Land use transfer matrix
Land use data were classified in three levels, according to the "China Land Use/Land Cover Remote Sensing Monitoring Data Classification System" (https://www.resdc.cn/). We reclassified landscape types into 14 different types (Table 1). Then, we overlaid land use data from 1995, 2000, 2005, 2010, 2015 and 2020 to construct the land-use transfer matrix, input/output direction and the area of each type of land-use within the study area using the spatial analysis tools in ArcGIS10.6 (ESRI, America).

| InVEST-Habitat quality model
The InVEST model allows for the calculation of habitat quality by combining the sensitivity of landscape type and the intensity of external threats by assessing the service function of biodiversity based on habitat quality (Peng et al., 2018). In ecology, the InVEST model has been successfully used to assess land-use change and regional habitat quality. Plant ecology, animal ecology or bird ecology studies tend to target specific species and populations in target regions, assessing the habitat quality of biodiversity service functions F I G U R E 1 Geographic location of the study area, regional hydro-topographic configuration and occurrence data for the target species. (Bhagabati et al., 2014). Habitat quality was determined by a function using four factors: (1) the relative impact of each threat, (2) the relative sensitivity of each habitat type to each threat, (3) the distance between habitats and (4) sources of threats (Chen et al., 2016).
At the pixel scale, the threat level of each pixel cell was translated into habitat quality using the total threat level and a half-saturation function. The formula we used follows (Sharp et al., 2014): where Q xj is ecological habitat quality value of land use type j, H j is a habitat quality score ranging from 0 to 1, where non-habitat land-use types are given by a score of 0, and perfect habitat classes score 1. In our study, H j is the habitat suitability in Table 3. k is the half-saturation constant (Liang & Liu, 2017; and z is a constant.
where D xj represents the total threat level of the grid x in LULC or habitat type j, y indexes all grid cells on r's raster map and Y r indicates the set of grid cells on the raster map of r. Note that each threat map can have a unique number of grid cells due to variation in raster resolution. r is the weight; r y is the number of stress factors on the grid unit; x is the accessibility level of grid x; S jr is the sensitivity of landscape j to stress factors, ranging from 0 to 1; i rxy is the stress factor influence distance. If S jr = 0 then D xj is not a function of threat r. In our study, S jr is the sensitivity of different land use types to different ecological threat factors in Table 3. Also, note that threat weights are normalized so that the sum across all threat weights equals 1. The impact of threat r that originates in a grid cell y, r y on habitat in a grid cell x is given by i rxy . It is represented by the following equations, mainly including the linear or exponential distance-decay function: where d xy is the linear distance between grid cells x, y and d rmax is the maximum effective distance of the reach across the threat space.
Generally, the impact of a threat on a habitat decreases as the distance from the degradation source increases, so that grid cells that are more proximate to threats will experience higher impacts (Sharp et al., 2014).
We referred to InVEST model manual and related research, combined with the actual situation of the study area and crocodile lizards distribution areas (Huang et al., 2008), to determine the relevant parameter values (Sharp et al., 2014). We considered arable land, reservoirs, urban land, rural settlements and construction land as the main ecological threats to crocodile lizard habitat quality ( Table 2).
The ecological threats are weighted, reflecting the intensity of interference with the habitat types. We set the maximum range of action of each stressor, which means that the interference intensity of the stressor to the habitat types decreases with increasing distance. At the same time, we chose the attenuation function to describe the mode of threat mitigation in space. We assigned a value to the sensitivity of these threat factors (Table 3)-the higher the value, the more sensitive it is to ecological threats.

| Data processing
According to the guidance of the user manual (Sharp et al., 2014), rasterise land use data. All threats should be measured in the same scale and units (i.e. all measured in density terms or all measured in presence/absence terms) and not some combination of metrics (Sharp et al., 2014). Areas classified as "No Data" in the threat maps were reclassified. When the pixel did not contain a threat, we set the threat level for that pixel to zero (Sharp et al., 2014). According to the natural breakpoint method in ArcGIS software, the grid habitat quality of each study period was divided into four categories: poor (0-0.2), medium (0.2-0.5), good (0.5-0.7) and high (0.7-0.1) (Deng et al., 2021). from 2000 to 2020. The proportion of grasslands area and water areas was negligible and overall they did not vary a lot in size over the study period ( Figure 2).
The land-use transfer matrix showed that arable land and grassland gained land converted from woodland, as well as conversely woodland gained land converted from arable land and grassland.
At the same time, construction land was growing in a faster way, with construction land gaining land converted from grassland and woodland during the study period (

| Habitat quality in the Pearl River Basin
Based on the habitat quality calculations (Table 5), the habitat quality in the study area showed a "decrease-increase" trend from 1995 to 2020, consistent with the results of the land-use transfer matrix.
The standard deviation of the habitat quality index increased from 0.3218 to 0.3250 between 1995 and 2015 ( Table 5). The maximum habitat degradation degree decreased from 0.1301 to 0.1285 from 1995 to 2015. Nevertheless, the maximum of the habitat degradation degree increased to 0.1332 after 2015 (Table 5). The habitat quality model showed that habitat quality within the study area did not vary significantly over time scales. Low habitat quality areas were widely distributed, mainly concentrated in counties and districts around the crocodile lizards' range ( Figure 3).

| Habitat quality of Shinisaurus crocodilurus distribution area
We calculated the habitat quality index of the crocodile lizards distribution area separately (  (Figures 4 and 5).

| The impact of land-use change on the crocodile lizards' habitat
In our study, the land-use transition matrix was used to explore the temporal and spatial changes in land-use types in the lizards' distribution range. We found that the main landscape types in the study area were arable land and woodland during the period 1995 to 2020.
Over time, the construction land shows a "decrease-increase", es-

| Habitat quality change of crocodile lizards
The InVEST model showed that the low-quality habitats were widely distributed, mainly in the periphery of the crocodile lizard's distribution areas. High-quality habitats were concentrated TA B L E 4 The land-use transfer matrix from 1995 to 2020s (hm 2 ).  Ziegler et al., 2019). Crocodile lizards are "living fossils", and the only surviving member of their family (Xie et al., 2021). The ecological niche of crocodile lizards are in valleys below 800 m.a.s.l. and appears to be restricted to tiny sections of clean and remote streams Wu et al., 2007;Zhao et al., 2006;Ziegler et al., 2019). Guangxi and Luokeng might have been the source of an initial population expansion (Huang et al., 2014). Initial field surveys showed that between 1977 and 1991, the main distribution sites of crocodile lizards in Guangxi were within DYS, BT, DGS, QC, GX and MS (Zhang, 1991;Zhang et al., 2005).  Long et al., 2007;van Schingen, Pham, et al., 2014). And crocodile lizards prefer habitats with sandy water substrates because the abundance of sand in the water body provides a buffering effect and also enables crocodile lizards to climb from out of the water to land .
Thus, the high-quality habitats located in mountainous forest areas have not been subject to significant anthropogenic disturbance for the time being, which provides conducive areas for the continued reproduction of the crocodile lizards.
Our results showed that the habitat quality in the Dateng Gorge has been poor in the past 25 years, with high levels of habitat degradation. This might be connected with the construction of water conservancy and hydropower projects. The Dateng Gorge is a canyon in the lower reaches of the Qianjiang River in the West River system of the Pearl River Basin, formed by the Qianjiang waterway between the Dayao Mountains and the Lotus Mountains (Yang et al., 2017).
The connection of the mountains may provide a migration channel for the crocodile lizards, which may be a fundamental reason for its presence (Yang et al., 2017). Upon completion of the Dateng Gorge Water Conservancy Project, the downstream area of the ditch in the Dawandu sub-field where crocodile lizards had been recorded, especially the creeks where crocodile lizards are widely distributed, will be submerged to the middle reaches (Yang et al., 2017). Hydropower facilities fragment streams into several channel segments and can alter the flow and sediment regimes (Csiki & Rhoads, 2014;Fantin-Cruz et al., 2015;Takahashi & Nakamura, 2011) and inhibit the dispersal of riparian plants and the migration of aquatic organisms (Andrea et al., 2012;Chen et al., 2015;Fencl et al., 2015;Perkin et al., 2015;Zhang et al., 2021). The situation has resulted in the loss of better quality habitat for crocodile lizards or even a break in flow, which had a significant negative impact on the growth and development of crocodile lizards (Yang et al., 2017). Moreover, during the construction of mining roads, large amounts of blasting and excavation debris were dumped into the stream, causing the pollution of inhabited streams (van Schingen, Pham, et al., 2014;van Schingen, Schepp, et al., 2015;Yu et al., 2005). And local villagers often use electro-fishing and poisonous chemicals to fish in the stream and this can kill all of the crocodile lizards in the water (Huang et al., 2008),

| Conservation suggestions
In the face of a massive crisis of deteriorating habitat quality for the lizards, while coping with local habitat destruction due to agricultural purposes, agreements with respective local farms helped to keep at least core zones of important habitats intact in the crocodile lizards' nature reserve in China (van Schingen, Schepp, et al., 2015). Second, the Chinese government should encourage the development of the local economy and educate local people about the laws relating to wildlife conservation and prohibit the capture or trade of crocodile lizards. Third, the nature reserves should be expanded to restore forest conditions within the reserves to create more suitable habitats for the crocodile lizard.
Further, Chinese crocodile lizards could be bred in captivity in nature reserves and released back into nature to restore the wild populations (Huang et al., 2008). Therefore, a breeding station was constructed in 2003, and the first round of crocodile lizards released back into the wild (Long et al., 2007;Zollweg, 2011Zollweg, , 2012. In 2009, 30 crocodile lizards were released into the Guangdong Luokeng Crocodile Lizard Provincial Nature Reserve (Zhong, 2009).  (Hu, 2020). The efforts have already led to a stable and even slightly increasing subpopulation within the Daguishan Nature Reserve in 2011 (Zollweg, 2012).

ACK N OWLED G M ENTS
We thank the National Natural Science Foundation of China

CO N FLI C T O F I NTE R E S T
The authors declare that this research was conducted out with commercial and/or financial concerns and is free of potential conflicts of interest.

O PEN R E S E A RCH BA D G E S
This article has earned Open Data and Open Materials badges. Data and materials are available at https://doi.org/10.5061/dryad.s1rn8 pk9n.

DATA AVA I L A B I L I T Y S TAT E M E N T
https://doi.org/10.5061/dryad.s1rn8 pk9n

E TH I C S S TATEM ENT
The animal study was reviewed and approved by Institutional Animal