Priority areas for the conservation of perennial plants in China

1 downloads 0 Views 2MB Size Report
Priority areas for the conservation of perennial plants in China. Ming-Gang ... 6000 higher plants are threatened or near threatened (Qin and Zhao,. 2014), meaning that .... analyses (Table S1): (1) Bio01: Annual Mean Temperature; (2) Bio03: ..... tween the abiotic environment and evolutionary processes is consid- ered an ...
BIOC-06840; No of Pages 8 Biological Conservation xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Biological Conservation journal homepage: www.elsevier.com/locate/bioc

Priority areas for the conservation of perennial plants in China Ming-Gang Zhang a,b, J.W. Ferry Slik c, Ke-Ping Ma b,⁎ a b c

Institute of Loess Plateau, Shanxi University, Taiyuan 030006, China State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China Faculty of Science, Universiti Brunei Darussalam, Jln Tungku Link, Gadong BE1410, Brunei Darussalam

a r t i c l e

i n f o

Article history: Received 29 November 2015 Received in revised form 8 June 2016 Accepted 12 June 2016 Available online xxxx Keywords: Plant diversity Species distribution model Collection bias Systematic conservation planning Priority areas for conservation Land use planning

a b s t r a c t With over 35,000 higher plants recorded, China is among the countries with the highest plant diversity. However, due to increasing human population, land-use intensification, and economic development, the habitat of most species is under considerable threat. Here we develop conservation priority maps covering all of China based on plant species distribution models in combination with spatially explicit decision making tools for systematic conservation planning. Our aim was to find spatial scenarios that maximize the success of conservation goals while minimizing the required land area to achieve these goals, so that economic development can proceed with minimal damage to existing biodiversity resources. We built species distribution models for 7427 vascular plant species at a 10 × 10′ resolution covering whole China, using geo-referenced herbarium collections and detailed environmental data, corrected for spatial bias using a null model. Based on these models we mapped: (1) species richness centers for common species (3535), endemic species (1965) and Chinese red list species (1927); (2) priority areas for conservation, distinguishing between conservation targets for common species (15% of the predicted suitable habitat), endemic species (25% the predicted suitable habitat) and red list species (35% the predicted suitable habitat); and (3) downscaled land-use pattern in each priority area for conservation. Clear priorities for the development of a sustainable and feasible biodiversity conservation strategy can now be provided based on our maps at national and regional levels. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction Strong environmental gradients, complex topography and a long geological history have resulted in a diverse Chinese flora (Wang et al., 2012). Over 35,000 higher plants have been recorded in this region (Editorial Committee of Flora Reipublicae Popularis Sinicae, 1959– 2004), and over half of them are endemic (Huang et al., 2012; Ying and Zhang, 1994). As a developing country famous for its extremely high biodiversity, solving the conflict between plant conservation and economic development is a major challenge for both scientists and the government (Corlett, 2015; Sang et al., 2011). Several efforts have been made to conserve the biodiversity of China over the past decades. According to the recent list provided by Ministry of Environmental Protection of China (MEP), over 2600 protected areas (national parks, natural parks, nature reserves, protected landscapes, etc.) have been established, the total area of which reaches ~15% of China's land surface (MEP, 2007). From 2008 onwards, using the red list categories and criteria developed by International Union for Conservation of Nature (IUCN), the Chinese Ministry of Environmental Protection and Chinese Academy of Sciences evaluated the threat status of over 35,000 higher plants of China (MEP and CAS, 2013). The results showed that over ⁎ Corresponding author. E-mail address: [email protected] (K.-P. Ma).

6000 higher plants are threatened or near threatened (Qin and Zhao, 2014), meaning that over 18% of all higher plants in China need urgent conservation action. A major problem with the current protected area system in China is that it lacks a sound spatial design at the macro-scale (Zhang et al., 2012, Wan et al., 2014). Furthermore, in the past decades the effectiveness of those reserves in protecting threatened species has been questioned (Zhang et al., 2015). Most protected areas were selected based on their inaccessibility or unsuitable nature for other purposes rather than their biodiversity value per se (Sang et al., 2011). This bias is especially problematic for species that specialize on productive or potentially productive landscapes because they may not be well represented in the current protection system. Therefore, there is a real and urgent need for more systematic spatial planning to identify priority areas for protection. Meanwhile, those areas should conserve most of China's biodiversity but leave enough room for economic development (Ardron et al., 2010; Huggins, 2005). Once conservation priority areas have been identified, it is important to analyze current and planned future land use. China's forests are facing serious risk of being converted into economically more productive landuse types, such as tree plantations and agriculture (Zhang et al., 2014). Recent reviews give some sensible recommendations on land-use planning within protected and agricultural areas, which include: restoration towards natural forest, designating corridors that facilitate migration of

http://dx.doi.org/10.1016/j.biocon.2016.06.007 0006-3207/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

2

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

plants and animals, and maintenance of diverse landscape mosaics (Tambosi et al., 2014). Due to the large area of China and the wide distribution of priority areas, multiple land use planning strategies will be needed. In this paper, we used species distribution modeling in combination with systematic conservation planning and decision making tools (Possingham et al., 2000) to: (1) map the botanical richness patterns of common species, endemic species and red list species across China; (2) identify priority areas for conservation based on the current distribution of species; and (3) explore current land use in the proposed priority areas to determine the threats to conservation faced by each priority conservation area. 2. Methods 2.1. Species data We obtained collection information for c. 4.5 million specimens of Chinese vascular plants present in 42 major herbaria from the Chinese Virtual Herbarium (bhttp://www.cvh.org.cn/cms/en/N, accessed September 2012). As a first step, we selected the 1.1 million records for woody species (trees, shrubs and lianas) for further analysis. Most of these records had no latitude and longitude information, so they were georeferenced by mapping the label locations on a high resolution map of China. This resulted in 464,045 specimens with latitude and longitude information. Based on these georeferenced collections, species presences were scored in 10′ grid cells, avoiding duplicate species records in each grid cell. We used the 10′ spatial resolutions because it provided a good balance between geo-referencing accuracy and the spatial resolution of available species occurrence data. Species that were present in fewer than 5 grid cells were removed from the analysis because no statistically sound habitat association analyses could be performed with these. This meant that the final species distribution analyses made use of 371,712 records belonging to 157 plant families representing 6828 species. Furthermore, following the three steps above, 867 herbaceous plant species were also included in the analysis. These herb species were chosen because they were evaluated as threatened or near threatened in the China biodiversity red list (MEP and CAS, 2013). Finally, all of these species were re-classified into three categories according to the China biodiversity red list: common species, endemic species and red list species. 2.2. Environmental predictors Initially, 35 environmental predictors were selected to model the species distribution patterns. These included 19 bioclimatic predictors (1950–2000) plus altitude of the WORLDCLIM dataset (bwww. worldclim.orgN) for China at 10′ resolution, and 15 soil variables selected from the FAO database for poverty and insecurity mapping (FAO, 2002). The FAO soil properties had a spatial resolution of 5′, so we resampled all soil layers into 10′ grid cells using ArcGIS 9.3. The whole mainland of China was thus covered by 34,230 grid cells. Because multi-colinearity of variables can result in over-fitting in species distribution modeling (Graham, 2003; Pearson et al., 2006), we removed highly correlated environmental predictors. For both bioclimatic and soil predictors, we used spearman's rank correlation tests to select the least correlated variables (Spearman's rho b 0.75). From correlated variables with Spearman rho higher than 0.75 only the ecologically most meaningful factors were kept. This procedure eventually resulted in the following climatic variables being included for further analyses (Table S1): (1) Bio01: Annual Mean Temperature; (2) Bio03: Isothermality (P2/P7) × 100 (P2: Mean Diurnal Rang; P7: Temperature Annual Range); (3) Bio07: Temperature Annual Range; (4) Bio12: Annual Precipitation; (5) Bio15: Precipitation Seasonality; and (6) Elevation. Of the soil predictors the following variables were included in the analysis (Table S2): (1) BS-T: base saturation% topsoil; (2) CE-S: CEC

clay subsoil (CEC = cation exchange capacity); (3) CN-T: C:N ratio class topsoil; (4) CP-T: organic carbon pool topsoil; (5) Depth: effective soil depth; (6) Drain: soil drainage class; (7) NN-T: nitrogen% topsoil; (8) Prod: soil production index; (9) Text.: textural class subsoil. In total 15 of the original 35 predictors were kept to model species distributions. 2.3. Species distribution model building and significance testing In order to model species distributions we used the modeling application Maxent (ver. 3.3.3k; b www.cs.princeton.edu/~schapire/maxent/ N) (Phillips et al., 2006). Maxent was specifically developed to model species distributions with presence-only data. Of available species distribution modeling algorithms, Maxent has been shown to perform best when few presence records are available (Wisz et al., 2008), while it is also the least affected by location errors in occurrences (Graham et al., 2007). Maxent was run with the following modeling rules: (1) for species with 5–10 collection records linear features were applied, (2) for species with 10–14 records quadratic features were applied, while (3) for species with N15 records hinge features were applied (Raes and ter Steege, 2007). As a measure of the accuracy of the SDMs, we used the threshold independent area under the curve (AUC) of the receiver operating characteristic (ROC) plot produced by Maxent. All measures of SDM accuracy require absences (Liu et al., 2011). When these are lacking, as is the case here, they are replaced by pseudo-absences or sites randomly selected at localities where no species presence was recorded (Phillips et al., 2006, Merow et al., 2013). However, when SDM accuracy measures are based on presence-only data and the background data, the standard measures of accuracy (e.g. the often used measure AUC N 0.7) do not apply (Raes et al., 2009; Raes and ter Steege, 2007). When building the presence-only models for these species with low prevalence, the AUC values tend to inflate (van Proosdij et al., 2015). Therefore, we applied the null-model developed by Raes and ter Steege (2007) to test the AUC value of an SDM developed with all presence records against the AUC values expected by chance. However, this assumes that collection localities represent a random subset of the study areas environmental space. In many cases this is not a valid assumption due to collecting biases (Kleidon and Mooney, 2000; Tsoar et al., 2007). To check for collecting bias in our dataset we tested whether our 3068 collection localities formed a random subsample of China's environmental predictor space. To do this we divided each of the 15 environmental predictors into 10 equal-interval bins based on the ranges observed for whole China (34,230 grid cells) (Loiselle et al., 2008). We then tested whether the observed frequency distributions represented by the 3068 collection localities differed from those observed for whole China using a Chi-square test. This showed that for 14 of the 15 environmental predictors, our collection locations represented nonrandom subsamples of China's environmental predictor space. To correct for this we developed a bias corrected null model by testing each species model AUC value against 1000 AUC values that were generated by randomly sub-sampling from all the available collection localities. When the observed AUC value fell in the top 95% of randomly generated AUC values, it was considered to have a significant non-random distribution and was used in our further analyses. For all the 6828 available woody species of China, 6560 species showed a significantly non-random distribution (AUC value ≥ 95% C.I.), while all of the 867 herbs passed the null model test. Therefore, 7427 species were included in the final analyses. 2.4. Species richness pattern and priority areas for conservation of China To define whether a species was present or absent in a grid cell, the following thresholds in Maxent were applied: ‘sensitivity specificity equality’ or the ‘sum maximization’ (for SDMs represented by 5–9

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

records); fixed ‘10 percentile presence’ (for SDMs represented by ≥10 records). Once the threshold was set, a presence/absence matrix was created by stacking the presence/absence layers of SDMs, the rows representing the grid cells covering China and the columns representing the presence/absence of the 7427 modeled species. Species diversity of all species and the three sub-categories were mapped using ARCGIS 9.3. Marxan is decision making software that can be used to solve the minimum set reserve design problem (Possingham et al., 2000). This tool aims to minimize the space for conservation, while meeting userdefined conservation targets (Margules and Pressey, 2000). The importance of each grid cell for conservation was determined using the ‘selection frequency’ (Game and Grantham, 2008). Selection frequency represents the importance of each grid cell to achieve the target for a given conservation goal (Ardron et al., 2010). The decision making tool is based on a random search algorithm; whereby each run will produce a set of specific optimization result. The ‘selection frequency’ is determined by the summed number of times each grid cell was chosen across all runs. Grid cells with high ‘selection frequency’ reflect a high compositional dissimilarity with other grid cells, while a low ‘selection frequency’ indicates a high compositional overlap with other grid cells. For computational capabilities reasons, we up-scaled the grid cells to the resolution of 20 × 20′. This step results in 8260 planning units covering the whole continental China. Considering the low ‘opportunity cost’ in these areas with high biodiversity persistence (Cameron et al., 2008), we set the ‘cost’ of each planning unit as 10,000/(species diversity), i.e. it is considered more efficient to conserve the areas with high richness. The conservation feature represents the presence/absence of the 7427 significant species in each planning unit. Our conservation targets were set as flowing rules: the 15% of suitable habitat included in the conservation network for common species; 25% for endemic species, and 35% for red list species. These targets were set to control the conservation priority for each category while the total areas fall within a reasonable range (15%–20% of the land surface of whole China). Another important parameter is the boundary length modifier (BLM), which was used to control the level of fragmentation allowed to occur in the reserve system by minimizing the reserve system boundary length relative to its size (Stewart and Possingham, 2005). First, the BLM value was set at 10 levels (0.0001, 0.001, …, 1000, 10,000). Secondary, we set the Marxan ran 100 times, for each of the 10 levels the average boundary length and total surface area was plotted to find the optimum level of BLM. When the BLM value higher than 1000 the boundary length started to decrease while the total area started to increase fast (Fig. S1). Finally, we set the algorithm to run 100 times (BLM = 1000) to calculate the ‘selection frequency’ for each planning units. For those planning units with ‘selection frequency’ higher than 80 were then chosen as the priority areas for conservation. 2.5. Land use data and downscaled land use pattern analysis A recent land use map for all of China (2009) was obtained from the ESA (European Space Agency) bhttp://ionia1.esrin.esa.int/N. The original data set was built at a resolution of 10″ (1/360°) and contained 23 land use types. We resampled this original map to produce a map with a resolution of 30″ (1 km × 1 km) to facilitate landscape pattern analysis. Then, the 23 land use types were reclassified into six broad

3

categories: cropland, forest, shrubland, grassland; sparse vegetation and open land (Table S3). We mainly focused on habitat loss and fragmentation within each conservation priority area. Habitat loss provides information on how severely the forest is encroached by other land use types while habitat fragmentation provides information on how land use breaks up larger forest patches into smaller ones (Weng, 2007). In our study, forest was set as the focal class. We then specifically looked at percentage of landscape (PLAND), patch density (PD), mean patch size (MPS) and mean proximity index (MPI) of forests patches, those indexes can be used to evaluate the degree of habitat loss together with forest patch isolation and fragmentation (details explained in Table 1). Furthermore, we calculated the Shannon's diversity index (SHDI) at the landscape level. Shannon's diversity index is a popular measure of diversity in community ecology, here applied to landscapes. SHDI increases as the number of different patch types increases or the proportional distribution of area among patch types becomes more equitable. Additionally, we obtained the distribution of current national nature reserves from the State Environmental Protection Administration of China (MEP) and resampled the map to get a resolution of 20′. The current national nature reserves were handled at a coarse resolution and it mainly serves the purpose of showing the overlap between the current protected areas and our suggested conservation priority areas. The land use pattern analysis was performed using FRAGSTATS (version 4.2) (McGarigal et al., 2012). 3. Results China was divided into 34,230 grid cells at a spatial solution of 10′, while the plant collections used in our study covered 3068 of these grid cells (9%). A total of 7427 species passed the collection bias corrected null model test, indicating that they exhibit significant habitat preferences. These species were classified into three categories: common species (3535), endemic species (1965) and red list species (1927). We put 717 species that are both endemic and red list species in the red list species category due to the higher conservation priority of red list species. The species diversity patterns all show strong latitudinal gradients. The high species richness regions identified for all species could be subdivided into 11 centers: (a) Southern Tibet, (b) Gaoligong Mountains, (c) Southern Yunnan province, (d) Western Guangxi province, (e) Guizhou Plateau, (f) Daxue Mountains, Qionglai Mountains, (g) Wuling Mountains, Dalou Mountains, (h) Qinling Mountains, Daba Mountains, (i) South of the Nanling Mountains, (j) Wuyi Mountains, Daiyun Mountains, and (k) Tianmu Mountains (Fig. 1, Table 2). The high species richness centers for endemic species overlapped with the 11 identified centers for all species, except for Southern Tibet and South of the Nanling Mountains. For red list species southwest China showed great importance, especially in Gaoligong Mountains, Southern Yunnan province and Western Guangxi province. The ‘selection frequency’ of each grid cell was calculated (Fig. 2). The assembled grid cells with ‘selection frequency’ over 80 were set as high conservation priority patches. Based on this threshold, 20 high conservation priority patches were identified (Fig. 3). These proposed priority areas for conservation include a variety of vegetation types and cover

Table 1 Landscape metrics utilized for landscape pattern characterization. Landscape-level: describe the spatial patterns of all land-use types as a whole. Class-level: describe the spatial patterns of different land-use types individually. Metrics

Abbreviation

Description

Range

Percentage of landscape Patch density Mean patch size Large patch index Mean proximity index Shannon's diversity index

PLAND PD MPS LPI MPI SHDI

The proportion of total area occupied by a particular land-use type The number of patches per 10,000 ha The average area of all patches in a landscape The area of the largest patch of the corresponding patch type divided by total landscape area Measuring the degree of isolation and fragmentation of a patch A measure of patch diversity, which represents the amount of “information” per patch

0 b PLAND ≤ 100 PD N 1 MPS N 0 0 b LPI ≤ 100 PROX ≥ 0 SHDI ≥ 0

Class-level Class-level Class-level Class-level Class-level Landscape-level

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

4

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

Fig. 1. A. Species richness pattern for all species. B. Species richness pattern for common species. C. Species richness pattern for endemic species. D. Species richness pattern for red list species. a–k indicate the 11 major species richness centers (deep red). All maps are in Albers projection. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

17% of the surface area China. The proposed protected areas showed very limited overlap with current protected system (Table 3; Fig. 2). The land use pattern analysis showed that in the middle of China, most forests have been replaced by other land use types (PLAND b 30%). For the conservation priority areas located in southwest China vegetation restoration is not urgent currently due to the high forest coverage (30% b PLAND b 60%). However, challenge from habitat fragmentation still exists. Currently, three proposed conservation Table 2 High richness centers for all species, common species, endemic species and China biodiversity red list species. ++: major centre; +: minor centre; PPAs: proposed protected areas; y/n means the proposed protected areas overlap with the high richness centers or not. Main ranges a. b. c. d. e. f. g. h. i. j. k.

Southern Tibet Gaoligong Mountains Southern Yunnan province Western Guangxi province Guizhou Plateau Daxue Mountains, Qionglai Mountains Wuling Mountains, Dalou Mountains Qinling Mountains, Daba Mountains South of the Nanling Mountains Wuyi Mountains, Daiyun Mountains Tianmu Mountains

All species

Common species

Endemic species

Red list species

PPAs

++ ++ ++ ++ ++ ++

+ + ++ ++ + +

+ ++ ++ ++ ++ ++

+ ++ ++ ++ + +

n y y y n y

++

+

++

+

y

++

+

++

+

y

++

++

+

+

n

++

++

++

++

y

++

+

+

+

y

patches (Fig. 3; patch No.: 2, 3 and 19) are still well covered by forests (PLAND N 60%). At the arid region of northwest China, four proposed conservation priority areas (Fig. 3; patch No.: 8, 9, 10 and 11) with low vegetation coverage were found. 4. Discussion 4.1. Botanical richness patterns of all species, common species, endemic species and red list species Eleven high richness centers were identified when using all species, and most of these overlap with mountain ranges in southern China. This is good news because those mountain ranges of southern China are still well forested and will remain so in the near future because little development is planned for these areas (Tang et al., 2006). The importance of mountain areas from southern China in conserving China's biodiversity have been shown by several studies (Fang et al., 2012; Lopez-Pujol et al., 2011b; Tang et al., 2006). Multiple reasons may explain the high diversity in these mountainous regions. On the one hand, the complex topography and strong environmental gradients in mountain areas promotes the isolation and speciation of species (Stein et al., 2014; Svenning et al., 2010). Gaoligong Mountains, Daxue Mountains, and Qionglai Mountains are part of the Hengduan Mountain ranges, and previous studies have highlighted that the complex topography may have promoted the biodiversity of this region (Sherman et al., 2008). Molecular studies also support these findings (Liu et al., 2013; Wang et al., 2008). Western Guangxi province, Guizhou Plateau, Wuling Mountains, Dalou Mountains, Qinling Mountains, Daba Mountains form the geographical barrier between southern and northern floras, because their east to west

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

5

Fig. 2. Current national nature reserves of China (the green circular) and the ‘selection frequency’ of planning units of China (rescaled to a 20 × 20 km resolution, Albers projection). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

orientation blocks the movement of the southeast monsoon from the Pacific Ocean. On the other hand, in tropical and subtropical regions, scientists have mainly invoked historical and ecological processes as drivers of the high biodiversity (Brown, 2014). The high richness centers of Southern Tibet, Southern Yunnan province, South of the Nanling Mountains, Wuyi Mountains, and Daiyun Mountains have a strong

tropical affiliation. Due to the long-term stable geological and climatic conditions, the plants there have mainly dispersed and adapted in situ (Hortal et al., 2011; Qian et al., 2015). The steady-state relationship between the abiotic environment and evolutionary processes is considered an important driving force for the high species diversity in areas with strong tropical affiliation (Brown, 2014).

Fig. 3. 20 identified priority areas for conservation and reclassified landscape patterns. Albers projection.

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

6

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

Table 3 Landscape pattern character for each identified priority area for conservation. PLAND: Percentage of landscape; PA: percentage of the total area that have already been covered by protected areas; PD: Patch density; MPS: Mean patch size; LPI: Large patch index; MPI: Mean proximity index; SHDI: Shannon's diversity index.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Geographic area

Area (ha)

PLAND

PA

PD

MPS

LPI

MPI

SHDI

Changbai Mountains, Zhangguangcailing Mountains, Longgang Mountains Xiaoxing'anling Mountains Middle Daxing'anling Mountains Yanshan Mountains Southern Taihang Mountains Northern Shanxi province Jiaodong Peninsula Altai Mountains Poluokenu Mountains Tianshan Mountains Western Kunlun Mountains Middle Himalayas Liupan Mountains, Qinling Mountains, Daba Mountains Daxue Mountains, Qionglai Mountains Boshula Mountains, Taniantaweng Mountains, Shaluli Mountains Gaoligong Mountains Southern Yunnan province Western Guangxi province Wuyi Mountains Xianxialing Mountains, Tianmu Mountains

36,953,700 5,649,700 3,227,100 21,328,000 5,641,200 2,068,000 2,561,000 23,096,900 4,802,600 7,233,900 16,034,400 4,025,600 42,165,100 21,588,300 36,387,300 4,008,000 8,664,100 12,445,000 2,238,100 7,422,100

43.83 64.66 71.51 21.32 19.41 28.05 8.67 10.6 17.92 7.07 1.04 26.78 31.86 52.41 31.55 57.89 43.53 49.73 80.23 58.91

6.52 2.86 0 0.75 5.71 0 0 0 0 0 0 12.9 6.08 15.79 15.42 0 5.56 0 0 0

1.6 0.27 0.29 2.7 3.4 3.2 3.0 1.29 2.06 1.4 0.64 2.0 2.3 1.3 2.3 0.73 1.5 1.6 0.17 0.87

2817.73 23,876.47 24,290.53 781.34 570.91 877.46 286.21 818.96 871.26 518.88 160.95 1323.83 1392.87 4029.34 1360.17 7918.43 3000.16 3164.31 46,041.03 6800.16

24.44 63.93 70.59 3.58 0.89 7.11 0.46 1.65 2.77 0.81 0.018 3.68 11.29 44.42 11.97 45.96 25.13 28.14 79.97 45.69

2384.14 6965.33 3477.97 73.97 6.61 26.63 1.02 44.54 20.7 6.47 0.17 36.32 585.1 4534.59 479.94 1756.97 1416.51 1775.84 3212.79 1198.35

1.45 0.95 0.88 1.19 0.76 1.05 0.91 1.19 1.56 1.06 0.79 1.29 1.41 1.17 1.47 1.09 1.1 1.09 0.7 1.09

Endemic species is one of the most effective surrogates for identifying conservation priorities or hotspots (Myers et al., 2000; Lopez-Pujol et al., 2011a; Huang et al., 2012). The high richness centers for endemic species in our study strongly matched with the identified high richness centers for all species (Fig. 1, Table 2). Protecting the high richness regions was suggested to be the most direct way for conserving endemic species. Red list species are more vulnerable to extinction due to their limited geographic ranges. Thus, detecting the richness centers for red list species is crucial for conservation. Most red list high species centers are located in the southwest of China, especially in Yunnan province. However, due to statistical limitations in species distribution modeling of species with few collection sites (species with records b 5) no statistically analyses could be performed to detect the potential habitat. This problem may lead to a distorted identification of richness patterns, especially for endemic and red list species (Ma et al., 2013). We suggest that further efforts are necessary to map the spatial distribution pattern for species with narrow geographic ranges; this could be especially useful for improving the quality of the priority areas for conservation.

4.2. Priority areas for conservation and current national nature reserves of China On the basis of Marxan's solution we delimited 20 areas of high conservation priority in China. The twenty patches covered the whole longitudinal and latitudinal gradient of the country and included a wide variety of vegetation types. Similar to high species richness centers, most conservation priority areas are concentrated around mountain ranges. Only a limited amount of priority areas was located in the tropics, probably because the species composition in the tropics is highly similar to that of subtropical regions in China. Thus, the subtropical mountain regions of China are critical for optimal conservation. Recently, Zhang et al. (2016) defined the 11 biogeographical regions of China (Fig. S2) and suggested that conserving species richness centers in each biogeographical region is essential for biodiversity conservation. In this study, we found that the boundaries between those biogeographical regions show especially high conservation value; similar to an earlier study in Yunnan province, (Zhang et al., 2012). On the one hand, these boundaries encompass many species within relatively small surface areas due to the steep floristic gradients (Zhang et al., 2012). On the other hand, along with the floristic boundaries the topography and climatic conditions also show dramatic changes (Yang and Xu, 2003; Zhu, 2013).

Identifying the macro-scale priority areas for conservation is the most important foundation for developing conservation strategies for the whole country (Myers et al., 2000). Our results are especially useful to indicate the locations that can maximize the coverage of species while minimizing the amount of land necessary to achieve this. Current protected areas show a scattered spatial distribution pattern and generally did not match the conservation priority areas identified in this study (Table 3); this means the poor biodiversity representation of the current protected areas. Furthermore, we point out two special challenges for future studies. One challenge is the sustainability of proposed priority areas for conservation. The intersection of biogeographical regions can encompass many species within relatively small surface areas, but because these areas form boundaries between floristic regions, they are also sensitive to climatic change. Changes in climate could lead to a shift of floristic boundaries (Camarero et al., 2006), which may reduce the conservation importance of the currently proposed conservation priority areas. More importantly, because of the high diversity and complexity of the species communities in the identified priority conservation areas, the vegetation will be difficult to recover once damaged. This will mean that monitoring and management programs will have to be implemented to maintain the conservation value of the identified regions. 4.3. Planning land use in proposed priority areas for conservation As the world's largest developing country, China is facing the conflict between economic development and nature conservation (Liu et al., 2003). Human activities have severely affected current biodiversity patterns and this situation is ongoing. We found that the land use patterns across China show great regional differences. In the middle of China, most forests in our four proposed conservation patches (Fig. 3; patch No.: 4, 5, 6 and 7) have been replaced by other land use types (PLAND b 30%). This will highly increase the risk of local extinction together with alien species expansion (Hansen and Clevenger, 2005), which will increase the instability of plant communities. Although those regions are identified as conservation priority areas, their conservation will be hampered by the extremely high cost of restoration projects. In this case, it may be more effective to focus on enhancing ex-situ conservation for the key species (Li and Pritchard, 2009) rather than building new protected areas. For the conservation priority areas located in southwest China (Fig. 3; patch No.: 13, 14, 15, 16, 17 and 18), vegetation restoration is not urgent currently because of their high forest coverage (30% b PLAND b 60%).

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

However, two proposed conservation areas face high levels of habitat fragmentation (low MPI value): Qinling and Daba Mountains (Fig. 3; patch No. 13) and the middle Hengduan Mountains (patch No. 15). This fragmentation results in isolation of forests leading to dispersal limitation and increased the risk of local extinctions. In this case efforts should be focused on the construction of ecological corridors to promote dispersal and increase the stability of these forest ecosystems (Cotter et al., 2014). Currently, most parts of the three proposed conservation patches (Fig. 3; patch No.: 2, 3 and 19) are still well covered by forests (PLAND N 60%). Compared with other patches, the total area of these three proposed conservation priority areas is relatively small and low risk of habitat loss and habitat fragmentation is expected. The problem caused by high coverage of forests is the low landscape diversity (SHDI). However, the intermediate landscape diversity could promote the β diversity of local communities and improve the effectiveness of local conservation management (Tscharntke et al., 2012). Additionally, four proposed conservation priority areas (Fig. 3; patch NO.: 8, 9, 10 and 11) are located at the arid region of northwest China. The maintenance of the remaining forests there depends highly on the moisture conditions in mountain valleys (Sang, 2009). Considering the representativeness of the local ecosystem, more attentions should be given to special ecosystems rather than species diversity per se. 5. Conclusion Selecting priority areas for conservation which can conserve most of China's biodiversity in the most space-conserving way is the most effective conservation strategy available. The current national nature reserve system is conserving a very limited set of China's biodiversity and does not overlap much with the conservation priority areas identified in our study. Our identified conservation priority areas include most vegetation types and plant species of China, while leaving enough room for economic activities. Currently, most of our proposed priority areas are still well covered by forests. This means that it is still possible to adapt China's conservation strategy by setting up a forest protection and nature reserve system that will actually protect most of China's plant diversity. Acknowledgements We thank Guo-Ke Chen, Jun.-Jie Wang and Wu-Bing Xu geo-referenced the specimens' data of China biodiversity red list species, Yin-Bo Zhang provided the distribution map of the national nature reserves of China. Three anonymous reviewers provided insightful comments on the early version of this manuscript. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.biocon.2016.06.007. References Ardron, J.A., Possingham, H.P., Klein, C.J., 2010. Marxan Good Practices Handbook Pacific Marine Analysis and Research Association, Victoria, BC, Canada. Brown, J.H., 2014. Why are there so many species in the tropics? J. Biogeogr. 41, 8–22. Camarero, J.J., Gutierrez, E., Fortin, M.J., 2006. Spatial patterns of plant richness across treeline ecotones in the Pyrenees reveal different locations for richness and tree cover boundaries. Glob. Ecol. Biogeogr. 15, 182–191. Cameron, S.E., Williams, K.J., Mitchell, D.K., 2008. Efficiency and concordance of alternative methods for minimizing opportunity costs in conservation planning. Conserv. Biol. 22, 886–896. Corlett, R.T., 2015. The Anthropocene concept in ecology and conservation. Trends Ecol. Evol. 30, 36–41. Cotter, M., Berkhoff, K., Gibreel, T., Ghorbani, A., Golbon, R., Nuppenau, E.-A., Sauerborn, J., 2014. Designing a sustainable land use scenario based on a combination of ecological assessments and economic optimization. Ecol. Indic. 36, 779–787. Editorial Committee of Flora Reipublicae Popularis Sinicae, 1959–2004,. Flora Republicae Popularis Sinicae. Science Press, Beijing.

7

Fang, J.Y., Shen, Z.H., Tang, Z.Y., Wang, X.P., Wang, Z.H., Feng, J.M., Liu, Y.N., Qiao, X.J., Wu, X.P., Zheng, C.Y., 2012. Forest community survey and the structural characteristics of forests in China. Ecography 35, 1059–1071. FAO, 2002. Terrastat; global land resources GIS models and databases for poverty and food insecurity mapping. Land and Water Digital Media Series #20. Game, E.T., Grantham, H.S., 2008. Marxan User Manual: For Marxan version 1.8.10. University of Queensland, St. Lucia, Queensland, Australia, and Pacific Marine Analysis and Research Association, Vancouver, British Columbia, Canada. Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regression. Ecol. Lett. 84, 2809–2815. Graham, C.H., Elith, J., Hijmans, R.J., Guisan, A., Townsend Peterson, A., Loiselle, B.A., 2007. The influence of spatial errors in species occurrence data used in distribution models. J. Appl. Ecol. 45, 239–247. Hansen, M.J., Clevenger, A.P., 2005. The influence of disturbance and habitat on the presence of non-native plant species along transport corridors. Biol. Conserv. 125, 249–259. Hortal, J., Diniz-Filho, J.A.F., Bini, L.M., Rodríguez, M.Á., Baselga, A., Nogués-Bravo, D., Rangel, T.F., Hawkins, B.A., Lobo, J.M., 2011. Ice age climate, evolutionary constraints and diversity patterns of European dung beetles. Ecol. Lett. 14, 741–748. Huang, J.H., Chen, B., Liu, C.R., Lai, J.S., Zhang, J.L., Ma, K.P., 2012. Identifying hotspots of endemic woody seed plant diversity in China. Divers. Distrib. 18, 673–688. Huggins, A.E., 2005. MARXAN conservation planning decision support software tutorial. The Nature Conservancy. Kleidon, A., Mooney, H.A., 2000. A global distribution of biodiversity inferred from climatic constraints: result from a process-based modelling study. Glob. Chang. Biol. 6, 507–523. Li, D.-Z., Pritchard, H.W., 2009. The science and economics of ex situ plant conservation. Trends Plant Sci. 14, 614–621. Liu, J., Moeller, M., Provan, J., Gao, L.-M., Poudel, R.C., Li, D.-Z., 2013. Geological and ecological factors drive cryptic speciation of yews in a biodiversity hotspot. New Phytol. 199, 1093–1108. Liu, J.G., Ouyang, Z.Y., Pimm, S.L., Raven, P.H., Wang, X.K., Miao, H., Han, N.Y., 2003. Protecting China's biodiversity. Science 300, 1240–1241. Liu, C.R., White, M., Newell, G., 2011. Measuring and comparing the accuracy of species distribution models with presence-absence data. Ecography 34, 232–243. Loiselle, B.A., Jørgensen, P.M., Consiglio, T., Jiménez, I., Blake, J.G., Lohmann, L.G., Montiel, O.M., 2008. Predicting species distributions from herbarium collections: does climate bias in collection sampling influence model outcomes? J. Biogeogr. 35, 105–116. Lopez-Pujol, J., Zhang, F.-M., Sun, H.-Q., Ying, T.-S., Ge, S., 2011a. Centres of plant endemism in China: places for survival or for speciation? J. Biogeogr. 38, 1267–1280. Lopez-Pujol, J., Zhang, F.-M., Sun, H.-Q., Ying, T.-S., Ge, S., 2011b. Mountains of Southern China as “plant museums” and “plant cradles”: evolutionary and conservation insights. Mt. Res. Dev. 31, 261–269. Ma, Y.P., Chen, G., Grumbine, R.E., Dao, Z.L., Sun, W.B., Guo, H.J., 2013. Conserving plant species with extremely small populations (PSESP) in China. Biodivers. Conserv. 22, 803–809. Margules, C.R., Pressey, R.L., 2000. Systematic conservation planning. Nature 405, 243–253. McGarigal, K., Cushman, S., Ene, E., 2012. FRAGSTATS v4: Spatial Pattern Analysis Program for Categorical and Continuous Maps. University of Massachusetts, Amherst. MEP, 2007. Planning Compendium of Biology Resource Conservation and Utilization in China (MEP). Ministry of Environmental Protection of China, Beijing. MEP and CAS, 2013. China Biodiversity Red List. Ministry of Environmental Protection of China, Beijing. Merow, C., Smith, M.J., Silander, J.A., 2013. A practical guide to MaxEnt for modeling species' distributions: what it does, and why inputs and settings matter. Ecography 36, 1058–1069. Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000. Biodiversity hotspots for conservation priorities. Nature 403, 853–858. Pearson, R.G., Raxworthy, C.J., Nakamura, M., Townsend Peterson, A., 2006. Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. J. Biogeogr. 34, 102–117. Phillips, S., Anderson, R., Schapire, R., 2006. Maximum entropy modeling of species geographic distributions. Ecol. Model. 190, 231–259. Possingham, H., Ball, I., Andelman, S. (Eds.), 2000. Mathematical Methods for Identifying Representative Reserve Networks. Springer-Verlag, New York. Qian, H., Wiens, J.J., Zhang, J., Zhang, Y., 2015. Evolutionary and ecological causes of species richness patterns in North American angiosperm trees. Ecography 38, 241–250. Qin, H.N., Zhao, L.N., 2014. China higher plants and their threatened status. In: Jiang, Z.G., Ma, K.P. (Eds.), The Principles of Conservation Biology. Science Press, Beijing, pp. 117–148. Raes, N., ter Steege, H., 2007. A null-model for significance testing of presence-only species distribution models. Ecography 30, 727–736. Raes, N., Roos, M.C., Slik, J.W.F., Van Loon, E.E., Steege, H.t., 2009. Botanical richness and endemicity patterns of Borneo derived from species distribution models. Ecography 32, 180–192. Sang, W.G., 2009. Plant diversity patterns and their relationships with soil and climatic factors along an altitudinal gradient in the middle Tianshan Mountain area, Xinjiang, China. Ecol. Res. 24, 303–314. Sang, W.G., Ma, K.P., Axmacher, J.C., 2011. Securing a future for China's wild plant resources. BioScience 61, 720–725. Sherman, R., Mullen, R., Haomin, L., Zhendong, F., Yi, W., 2008. Spatial patterns of plant diversity and communities in alpine ecosystems of the Hengduan Mountains, Northwest Yunnan, China. J. Plant Ecol. 1, 117–136. Stein, A., Gerstner, K., Kreft, H., 2014. Environmental heterogeneity as a universal driver of species richness across taxa, biomes and spatial scales. Ecol. Lett. 17, 866–880.

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007

8

M.-G. Zhang et al. / Biological Conservation xxx (2016) xxx–xxx

Stewart, R.R., Possingham, H.P., 2005. Efficiency, costs and trade-offs in marine reserve system design. Environ. Model. Assess. 10, 203–213. Svenning, J.-C., Fitzpatrick, M.C., Normand, S., Graham, C.H., Pearman, P.B., Iverson, L.R., Skov, F., 2010. Geography, topography, and history affect realized-to-potential tree species richness patterns in Europe. Ecography 33, 1070–1080. Tambosi, L.R., Martensen, A.C., Ribeiro, M.C., Metzger, J.P., 2014. A framework to optimize biodiversity restoration efforts based on habitat amount and landscape connectivity. Restor. Ecol. 22, 169–177. Tang, Z.Y., Wang, Z.H., Zheng, C.Y., Fang, J.Y., 2006. Biodiversity in China's mountains. Front. Ecol. Environ. 4, 347–352. Tscharntke, T., Tylianakis, J.M., Rand, T.A., Didham, R.K., Fahrig, L., Peter, B., Bengtsson, J., Clough, Y., Crist, T.O., Dormann, C.F., Ewers, R.M., Fruend, J., Holt, R.D., Holzschuh, A., Klein, A.M., Kleijn, D., Kremen, C., Landis, D.A., Laurance, W., Lindenmayer, D., Scherber, C., Sodhi, N., Steffan-Dewenter, I., Thies, C., van der Putten, W.H., Westphal, C., 2012. Landscape moderation of biodiversity patterns and processes — eight hypotheses. Biol. Rev. 87, 661–685. Tsoar, A., Allouche, O., Steinitz, O., Rotem, D., Kadmon, R., 2007. A comparative evaluation of presence-only methods for modelling species distribution. Divers. Distrib. 13, 397–405. van Proosdij, A.S.J., Sosef, M.S.M., Wieringa, J.J., Raes, N., 2015. Minimum required number of specimen records to develop accurate species distribution models. Ecography http://dx.doi.org/10.1111/ecog.01509. Wan, J.Z., Wang, C.J., Han, S.J., Yu, J.H., 2014. Planning the priority protected areas of endangered orchid species in northeastern China. Biodivers. Conserv. 23, 1395–1409. Wang, Z.H., Fang, J.Y., Tang, Z.Y., Lin, X., 2012. Relative role of contemporary environment versus history in shaping diversity patterns of China's woody plants. Ecography 35, 1124–1133.

Wang, F.-Y., Ge, X.-J., Gong, X., Hu, C.-M., Hao, G., 2008. Strong genetic differentiation of Primula sikkimensis in the East Himalaya-Hengduan Mountains. Biochem. Genet. 46, 75–87. Weng, Y.-C., 2007. Spatiotemporal changes of landscape pattern in response to urbanization. Landsc. Urban Plan. 81, 341–353. Wisz, M.S., Hijmans, R.J., Li, J., Peterson, A.T., Graham, C.H., Guisan, A., 2008. Effects of sample size on the performance of species distribution models. Divers. Distrib. 14, 763–773. Yang, X., Xu, M., 2003. Biodiversity conservation in Changbai Mountain Biosphere Reserve, northeastern China: status, problem, and strategy. Biodivers. Conserv. 12, 883–903. Ying, T.-S., Zhang, Y.-L. (Eds.), 1994. Endemic Genera of Seed Plant Species in China. Science Press, Beijing. Zhang, Z.J., He, J.-S., Li, J.S., Tang, Z.Y., 2015. Distribution and conservation of threatened plants in China. Biol. Conserv. 192, 454–460. Zhang, M.-G., Slik, J.W.F., Ma, K.-P., 2016. Using species distribution modeling to delineate the botanical richness patterns and phytogeographical regions of China. Sci. Rep. 6, 22400. http://dx.doi.org/10.1038/srep22400. Zhang, M.-G., Zhou, Z.-K., Chen, W.-Y., Cannon, C.H., Raes, N., Slik, J.W.F., Bradley, B., 2014. Major declines of woody plant species ranges under climate change in Yunnan, China. Divers. Distrib. 20, 405–415. Zhang, M.-G., Zhou, Z.-K., Chen, W.-Y., Slik, J.W.F., Cannon, C.H., Raes, N., 2012. Using species distribution modeling to improve conservation and land use planning of Yunnan, China. Biol. Conserv. 153, 257–264. Zhu, H., 2013. Geographical elements of seed plants suggest the boundary of the tropical zone in China. Palaeogeogr. Palaeoclimatol. Palaeoecol. 386, 16–22.

Please cite this article as: Zhang, M.-G., et al., Priority areas for the conservation of perennial plants in China, Biological Conservation (2016), http://dx.doi.org/10.1016/j.biocon.2016.06.007