Transboundary conservation hotspots in China and potential impacts of the belt and road initiative

Biodiversity hotspots often span international borders, thus conservation efforts must as well. China is one of the most biodiverse countries and the length of its international land borders is the longest in the world; thus, there is a strong need for transboundary conservation. We identify China's transboundary conservation hotspots and analyse the potential effects of the Belt and Road Initiative (BRI) on them to provide recommendations for conservation actions.


| INTRODUC TI ON
International border areas often harbour rich and endemic biodiversity (Fowler et al., 2018;Huang et al., 2012;Liu et al., 2020). For example, as a politically sensitive area located on the border of North and South Korea, the Korean Demilitarized Zone harbours more than 4000 species (Lee & Miller-Rushing, 2014). Globally, the distribution of 53.8% vertebrates cross international borders (Mason et al., 2020). However, transboundary species face great conservation challenges because international borders often separate wildlife populations artificially, and the creation of physical barriers to separate nations prevents transboundary wildlife movement (Linnell et al., 2016). For example, 1506 native species whose geographical range traverse the US-Mexico border are threatened by the Border Wall (Peters et al., 2018). Furthermore, protected areas (PAs) often stop at political boundaries, thereby failing to safeguard the full distribution of species (Kark et al., 2015). In addition, differences in governance and unequal status of legal protection of nations on either side of the border, makes it hard to control poaching and smuggling (Nijman et al., 2016). For example, 43,399 kg of pangolin material and 518 whole individuals were seized on the border between Myanmar and China between 2010 and 2014 (Nijman et al., 2016).
Consequently, conservation actions that span borders are needed to avoid biodiversity loss. These actions can prevent the isolation of small populations, permit effective combating of poaching (Scholte et al., 2013), coordinate conservation priorities, identify and control extensive and large-scale threats like climate change, and lead to sharing of data and experience to improve protection effectiveness (Kark et al., 2015;Ma et al., 2020). A high-profile example of conservation action along borders is the European Green Belt. This project snakes along the line of the former Iron Curtain and spans 23 countries and six biogeographical regions to improve landscape connectivity and biodiversity conservation effectiveness (Vasilijević & Pezold, 2011).
The need for increased conservation efforts along China's borders will become more acute because of the Belt and Road Initiative (BRI), which was formally proposed in 2013. This initiative involves over 70 countries and is the most extensive international transportation infrastructure construction project ever developed (https:// eng.yidai yilu.gov.cn/). Construction of infrastructure like road and rail networks is often considered one of the most influential human interventions to the earth's ecosystems (Laurance et al., 2004(Laurance et al., , 2014Popp & Boyle, 2017). Historically, large transportation infrastructure projects have had significant negative impacts (Laurance et al., 2001(Laurance et al., , 2009. Once a road is built, it is almost permanent in the environment, bringing long-term risks to the surrounding ecosystems (van der Ree et al., 2015). This is particularly critical in roadless areas including border regions, because the first cut can bring a rapid increase in human pressure (Laurance, 2015;Selva et al., 2015). The BRI will promote development in border areas so that will also bring conservation challenges to transboundary species. Linear infrastructure of the BRI may hinder the dispersal of species and pose threats such as roadkill, noise and pollutions (Hughes, 2019). The BRI could be accompanied by logging, urbanization or agriculturalization, which can lead the loss of habitat for surrounding species (Hughes, 2019). Furthermore, the growth of human population along with the construction of infrastructure will increase the risk of biological invasion (Liu et al., 2019). Conservation planning in advance is therefore needed to reduce the possible loss of biodiversity caused by the BRI.
To advance the understanding of the importance of international borders to conservation globally and to guide regional planning, we compiled a list of transboundary terrestrial vertebrates and identified the hotspots of transboundary conservation in China. We then evaluated whether existing PAs are sufficient to protect these hotspots. Finally, we analysed the potential risks and opportunities brought by the BRI.

| China's transboundary terrestrial vertebrates
We compiled a list of transboundary terrestrial vertebrates in China from the International Union for Conservation of Nature (IUCN) Red List database (https://www.iucnr edlist.org/). We downloaded data of all species of mammals, birds, amphibians and reptiles from the database and filtered those living in terrestrial ecosystems. We then filtered these species based on their geographical ranges, to retain species live both in China and other neighbouring countries.
Furthermore, we filtered the species by their distribution codes and retained those with codes of 'Extant', 'Possibly Extant', 'Native', and for birds we excluded 'Passage'. The retained species were classified as the transboundary terrestrial vertebrates in China.

| Mapping transboundary conservation hotspots
We downloaded distribution maps of transboundary species from the IUCN Red List (IUCN, 2021) and BirdLife International and the Handbook of the Birds of the World (BirdLife International, 2018).
We then refined the distribution range (R 4.1.0, terra package) (Hijmans, 2022b) for each species according to its suitable habitat types (i.e. land cover types) and elevation range, which were obtained from the IUCN Red List. Land cover data were obtained from Jung et al. (2020), which is consistent with the IUCN habitat classification, and elevation data were obtained from WorldClim (https://world clim.org/) (Fick & Hijmans, 2017). All raster layers were rescaled to a spatial resolution of 1 km and were under spatial reference coordinate system of WGS1984.
We created 10 km, 50 km and 100 km buffer zones on both sides of China's border as border region (made in ArcGIS 10.2.2) for subsequent analysis. We visually checked the results, and found the geographical locations of transboundary conservation hotspot were similar when using different buffer zones (see Figure S2). Finally, we chose to present the 100 km results in this article to cover a larger area for conservation planning. This was also based on the consideration that the dispersal range of most terrestrial vertebrates (Minor & Lookingbill, 2010;Paradis et al., 1998;Saura et al., 2017) is within 100 km. If an individual animal disperses across international borders, most do not extend beyond 100 km.
We used this border region to crop the distribution maps of transboundary terrestrial species in China. Within the border region, each specie has a distribution layer with a value of 0 or 1 in each 1-km 2 cell, where 1 represents presence and 0 represents absence.
All species were then weighted by their Red List category, assuming Least Concern (LC) as 1, Near Threatened (NT) as 2, Vulnerable (VU) as 3, Endangered (EN) as 4 and Critically Endangered (CR) as 5 (Balaguru et al., 2006). We valued DD as 3 because DD species are often considered potentially at risk of extinction (Jaric et al., 2016).
However, excluding the 65 DD species did not affect the main results. The weighted distribution layers were stacked to obtain a weighted-richness map. Finally, we extracted the top 30% of cells with highest values in the weighted-richness map as conservation hotspots. The 30% was chosen as the threshold because according to the 2030 action target 3 of the 15th meeting of the Conference of the Parties to the Convention on Biological Diversity (COP15) (Convention on Biological Diversity, 2020), it is necessary to protect 30% of land and sea globally by 2030. We aggregated neighbouring cells in the hotspot map into a patch, and patch less than 100 km apart (from their centre points) were further aggregated into a hotspot (R 4.1.0, grainscape package, see Figure S1) (Chubaty et al., 2020). Four hotspots were finally identified ( Figure 1a).

| Coverage and connectivity of PAs within the border region
We obtained map layers of PAs in China's neighbouring countries from the World Database on Protected Areas (UNEP-WCMC, 2017) and supplemented China's PAs from Yang, Chen, et al. (2018). For some PAs which are point data in the WDPA dataset, we constructed circles around the points with areas equal to the sizes listed in the attribute table. In each hotspot, we calculated percentage of area in each country and their PA coverages ( Table 2). We also calculated PA coverage for each species and compared the differences between Classes using Kruskal-Wallis Test with a post-hoc Conover's all-pairs comparison test (R 4.1.0, PMCMRplus, multcompView package) (Graves et al., 2019;Pohlert, 2022).
Following Saura et al. (2017), we calculated the connectivity of PAs in each hotspot, using the Confer 2.6 software (Saura & Torné, 2009). The connectivity index of PAs, ProtConn index, represents the proportion of connected PAs to the study area. First, we calculated the area of each PA in the hotspot (a). Then we made buffer zones of 500 km for each of the above-mentioned PAs to capture possible 'springboard' PAs (T) following Saura et al. (2017). The area of springboard PAs was set at 0. Next, we calculated distances (x) between pairs of these PAs, and these distances were then converted by negative exponents (formula 1) as the probability of direct movement between the two PAs (Pij).
In this formula, D is a predefined dispersal kernel, which is set at 100 km, also known as the dispersal range of most terrestrial vertebrates.
Finally, the connectivity index of the PA-ProtConn was calculated according to the following formula (2): In this formula, a i and a j are the area of PA i and j in the hotspot (L). p * ij is the maximum probability of movement between two PAs including the probability of movement connected by 'springboard' or other PAs. A L is the area of the research area, which is the area of the hotspot in this analysis (Saura et al., 2017). All raster layers were rescaled to a spatial resolution of 500 m under coordinate system of EPSG:32648-WGS 84 / UTM zone 48 N (R 4.1.0, raster package) (Hijmans, 2022a).

| Assessing the threats brought by the BRI
We downloaded a map layer of the BRI routes from the World Bank database (https://datab ank.world bank.org/) (Reed & Trubetskoy, 2019). Six economic corridors that cross China's borders are planned along the BRI (https://eng.yidai yilu.gov.cn/), where infrastructure construction will likely to be the largest and cause the highest human pressures. We retained the routes belonging to each economic corridor according to the attributes of the routes. We also downloaded a published map of the BRI (Geographic Data Sharing Infrastructure, College of Urban and Environmental Science, Peking University, http://geoda ta.pku.edu.cn) to complement routes belonging to the economic corridors.
We overlaid the BRI routes (spatial lines) with refined distribution maps of all transboundary species, to identify the species whose distribution is traversed by the BRI (terra package, R 4.1.0).
Because the BRI may have different impacts on different taxa, we further obtained threat information of transboundary species from the IUCN Red List database. We grouped these threats into three categories to match the environmental impacts that To further assess the potential impacts of the BRI, we selected routes that are identified as 'under construction' and 'operational' (Reed & Trubetskoy, 2019), because they already have an impact on their surrounding environment. We made a 2 km (on both sides) buffer zone along selected routes, which was termed as 'roadeffect-zone' in road ecology and often span from several hundred meters to 3.5 km (Forman & Alexander, 1998;Husby, 2017). We

| Transboundary conservation hotspots, PA coverage and connectivity
We identified four transboundary hotspots, that is, the south-   Figure S3 and Figure S4).

| Potential impacts of the BRI
Routes in the BRI corridors traversed all four transboundary hotspots ( Figure 1a) and intersected distribution ranges of 82.4% (1619/1964) of the transboundary species (Appendix S1). More than half (918) of these species are sensitive to ecological risks posed by the BRI. Compared to direct risks, accompanying and long-term risks will affect more species ( Figure 2). Although distributions of more species of birds are traversed by the BRI (984), they are the least sensitive to the BRI risks (215).
Amphibians have the highest proportion of species exposed to at least one risk (0.87, 87/100), followed by reptiles ( Table S2). a Since the grids on which the border line located were not assigned to either country in calculation, the sum would be slightly <100%.  Ma et al., 2020) are endemic to this area and each has a tiny population. Without effective transboundary conservation, poaching and habitat loss and degradation will be difficult to regulate (Trinh-Dinh et al., 2022;Wang et al., 2021). Also, the SW hotspot has insufficient PA coverage (12.2%) and low connectivity (ProtConn 1.68%), especially in China, Myanmar and India ( Table 2). Both China and Myanmar should make efforts to increase PA coverage and curb rampant wildlife smuggling across the border (Nijman et al., 2016;Tan et al., 2022). In the N and NE hotspots, termed the Amur basin, China, Mongolia and Russia have reached a series of transboundary conservation cooperation agreements since 1956 (Simonov & Egidarev, 2018). The N hotspot is small but harbours 220 transboundary species. Most of the N hotspot (63.1%) were covered by China's Hulun Lake National Nature Reserve, which is currently threatened by human activities and climate change (Zheng et al., 2016). The Hulun Lake basin is a Ramsar PA, and it is an important breeding ground for rare transboundary wading birds, such as red-crowned crane (Grus japonensis), white-naped crane (Antigone vipio) and Baer's pochard (Aythya baeri). Lake levels and water quality are declining, and species living on wetland are further threatened by climate change, overgrazing and overuse of water resources (Zheng et al., 2016).

Although transboundary conservation cooperation between
The NE hotspot provides habitat for important large predators, including Siberian tiger and Amur leopard (Panthera pardus).
Conservation of these large predators requires large and wellconnected habitats. More than 95% of the Siberian tiger population lives in the Sikhote-Alin Mountains in Russia (Miquelle et al., 2006 (Xu et al., 2021). This successful conservation experience should be replicated to protect more transboundary threatened species.
We found that the six main economic corridors of the BRI traversed all four hotspots and traversed the distribution range of 82.4% transboundary species within the border region. Mammals and reptiles are sensitive to direct risks (Figure 2), so we need to design speciesspecific conservation actions during the planning of the BRI routes.
It is better to circumvent their active areas. Alternatively, the combination of well-designed ecological corridors and road fences can minimized the risk of road kill (Weller, 2015). Birds and amphibians are sensitive to accompanying risks and long-term risks, which means as roads are built and operated, more species in these two Classes may be at risk. Therefore, long-term adaptive management is more important for them, such as continuous monitoring of forest and water resources, and adaptation of PAs to climate change (Li et al., 2015;Li & Gao, 2020).
Human pressure (indexed by night light) increased significantly after the proposal of the BRI in the whole border area, especially in the four hotspots. Increasing human pressure could bring more conflicts between humans and wildlife in the future. Increase movement of people will also likely increase the risk of biological invasion (Laurance & Burgues Arrea, 2017;Liu et al., 2019). To prevent these effects, it will be important to establish sufficient transboundary PAs. Fortunately, the BRI showed some positive impacts on landscape in the NE hotspot (Figure 3), which may be attributed to successful habitat restoration projects in these areas (http://www.brigc. net/zcyj/yjkt/20201 1/t2020 1125_102825.html). In fact, conservation has been put at the core of the BRI since 'The Green Belt and Road' construction was proposed in 2015. Four Chinese ministries and commissions jointly released specific guidelines on The Green Belt and Road in 2017 (https://eng.yidai yilu.gov.cn/). However, as the BRI expands and begins to operate, habitat restoration will become an enormous challenge. Conservation planning and design in advance could be more cost-effective.
It is likely that the full extent of the impacts of the BRI will take a long time to measure due to the time lag between large-scale infrastructure construction and its environmental impact (Ascensão et al., 2022;Ng et al., 2020). Besides, negative effects on the individual or population scale of one specie need to be measured at a fine spatial scale (Fahrig & Rytwinski, 2009), which requires detailed longterm study. Many other groups of taxa like insects and plants were not included in this research, but they may face similar impacts with vertebrates from the BRI and also need transboundary conservation.
For example, Magnolia grandis, a critically endangered tree species that is distributed in the Sino-Vietnamese border region, was predicted to shift their distribution across the border, which called for transboundary cooperation (Blair, Galante, et al., 2022;Blair, Le, & Xu, 2022).
Hopefully, these less charismatic species can also benefit from conservation suggestions of this study. In any case, ecological infrastructure (Li & Shvarts, 2017) and strategic environmental impact assessment and ecological protection planning (Blair, Galante, et al., 2022;Blair, Le, & Xu, 2022;Laurance & Burgues Arrea, 2017) will be necessary for transboundary species conservation in the hotspots.
In addition to reducing the potential negative impacts of the BRI itself, we suggest conservation efforts take advantage of opportunities brought by the BRI. By connecting more than 71 countries (https://eng.yidai yilu.gov.cn/) and about 2/3 of the world's population (World Bank, https://www.world bank.org/), the BRI can be an extremely wide platform for dialogue and consultation, for knowledge and data sharing, and for joint planning. Such dialogue is particularly important today given the political instability in Myanmar and Afghanistan, the cooling of China-India relations, and the conflict between Russia and Ukraine, all of which may affect the transboundary cooperation on biodiversity conservation. The BRI also brings economic prosperity and cultural exchange (Liu & Dunford, 2016), which can reduce ideological gaps and build common conservation goals. Countries that earn dividends from the BRI are obliged by multilateral environmental agreements, like the Convention on Biological Diversity, to maintain ecological services along the routes.
It is important that development brought by BRI be part of an explicit strategy to promote transboundary conservation cooperation.
Policymakers and conservationists in the Belt and Road member countries have to reach a consensus by putting biodiversity conservation at the core (Lechner et al., 2018).

ACK N O WLE D G E M ENTS
We thank T. M. Lee and Y. Liu for their comments on an earlier version of this manuscript.

FU N D I N G I N FO R M ATI O N
This work was funded by National Natural Science Foundation of China (#31822049) and during the writing of the manuscript Colin Chapman was supported by the Wilson Center.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
Additional supporting information may be found in the online version of this article. Some data used for the analyses are freely downloadable online; two datasets, including the heatmap and hotspots of transboundary species, the rasterized protected area layers in research area, are available on the Dryad repository at: https://doi. org/10.5061/dryad.573n5 tb9x.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/ddi.13670.