Ren, Guo-Peng, et al. "Habitat evaluation and conservation framework of the newly discovered and

critically endangered black snub-nosed monkey." Biological Conservation 209 (2017): 273-279.

1

Habitat evaluation and conservation framework of 1

the newly discovered and critically endangered black 2

snub-nosed monkey 3

Guo-Peng Ren a,c1,Yin Yang a,b,c1, Xiao-Dong He c,d, Guang-Song Li c,d, Ying Gao a,c, 4

Zhi-Pang Huang a,c, Chi Ma a,c, Wei Wang e*,Wen Xiao a,c* 5

6

aInstitute of Eastern-Himalaya Biodiversity Research, Dali University, Dali, Yunnan 7

671003, China 8

b School of Archaeology & Anthropology, Australian National University, Canberra, 9

ACT 0200, Australia 10

c Collaborative Innovation Centre for Biodiversity and Conservation in the Three 11

Parallel Rivers Regionof China, Dali, Yunnan 671003, China 12

d Nujiang Bureau of Gaoligongshan National Nature Reserve, Nujiang, Yunnan 13

673100, China 14

e State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese 15

Research Academy of Environmental Sciences, Beijing 100012 China 16

1 Joint first authors/these authors contributed equally to this work 17

* Corresponding authors at: 18

Institute of Eastern-Himalaya Biodiversity Research, Dali University, Dali, Yunnan 19

671003, China. Email address: xiaow@eastern-himalaya.cn (W.Xiao); 20

State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese 21

Research Academy of Environmental Sciences, Beijing 100012 China (W.Wang). 22

23

24

25

2

Abstract 26

The black snub-nosed monkey (Rhinopithecus strykeri) is an IUCN-Critically 27

Endangered primate, recently discovered on the northern Sino-Myanmar border. In 28

order to identify the most urgent gaps in the conservation of the black snub-nosed 29

monkey, a hierarchical process was employed to predict the distribution and 30

alterations in its habitat over the past 15 years. Our study showed that R. strykeri 31

appeared to inhabit a range from E98°20'-98°50', N25°40'- 26°50', including high 32

quality habitat at 1420 km2, medium quality habitat at 750 km2, and low quality 33

habitats at 1410 km2. Only 21.1% of the total habitat for R. strykeri is within 34

protected areas in China. Approximately 2.6% of the entire habitat has been lost in 35

the past 15 years, 96% of which has been in Myanmar. To save this species from 36

extinction, it is urgent to establish trans-boundary conservation and management 37

networks to address the loss of habitat, and to locate and preserve key wildlife 38

corridors to link fragmented habitats between Myanmar and China. 39

40

Key words: Rhinopithecus strykeri, habitat alterations, trans-boundary conservation, 41

species distribution model 42

43

44

45

46

3

1. Introduction 47

The recently described black snub-nosed monkey (Rhinopithecus strykeri), 48

alternatively known as the Myanmar or Nujiang snub-nosed monkey, is found in the 49

high altitude forests of north-eastern Kachin state, Myanmar (Geissmann et al., 2011) 50

as well as in the middle segment of the Gaoligong Mountains, Yunnan, China (Long 51

et al., 2012; Yang et al., 2016). There are up to 14 groups (10 in China and 4 in 52

Myanmar, < 950 individuals) of black snub-nosed monkeys living in the northern 53

Sino-Myanmar border area according to interviews (Geissmann et al., 2011; Ma et 54

al., 2014). As a slow breeding species, combined with the prevalence of hunting and 55

the extensive loss of habitat, this species is likely experiencing a rapid demographic 56

decline (Geissmann et al., 2011). With the booming economy and population 57

growth of the region, deforestation for agricultural cultivation and infrastructure 58

development pose potential perils to this IUCN-Critically Endangered primate. 59

To save this species from the brink of extinction, it is important to study its habitat 60

and its distribution, but, due to the extreme ruggedness of the terrain and the long 61

wet season, almost no information on population distribution and habitat status are 62

available for conservation planning. Where the data are poor in given study areas, 63

employing MAXENT can offer reliable assessments for a species’ possible 64

distribution, and can support conservation planning (Thorn et al., 2009; Peck et al., 65

2011; Ingberman et al., 2016) by prioritizing appropriate habitats for new protected 66

areas (Urbina-Cardona & Loyola, 2008; Campos & Jack, 2013) or guiding 67

prospective land use planning and management (Illera et al., 2010), by identifying 68

least-cost corridors for habitat connectivity (Liu et al., 2016; Luo et al., 2016; 69

Schaffer-Smith et al., 2016), and by pinpointing ideal reintroduction sites (Thorn et 70

al., 2009; Cilliers et al., 2013). The function of MAXENT is to generate a model 71

across one study area based on existing information concerning environmental 72

variables which is then used in another area to make predictions and inference of 73

maximum entropy occurrence under similar environmental constraints (Phillips & 74

4

Dudík, 2008; Thorn et al., 2009; Araújo & Peterson, 2012). Compared to other 75

ecological niche algorithms, MAXENT as a present-only model can moderately 76

offset for imperfect, limited species occurrence data sets and reach near-maximum 77

accuracy levels under these circumstances (Hernandez et al., 2006, 2008; Giovanelli 78

et al., 2010). 79

Based on climatic variables and two sets of locality records obtained by 80

interview-based survey (Gessiman et al. 2011; Ma et al. 2014; Figure 1), a 81

MAXENT model was built up to map the habitat of R. strykeri. Forest cover maps 82

in the 2000s and 2015, based on Landsat images, were used to mask out 83

non-woodland areas from the habitat. We then assessed habitat changes from 2000 84

to 2015. According to the results, we identified additional areas where R. strykeri 85

may occur, proposed specific conservation priorities, and determined crucial areas 86

in which urgent protection measures should be taken to ensure the long survival of 87

this rare and little known primate. 88

2. Methods 89

2.1 Study area 90

The Gaoligong Mountains rise from the low altitude (183 m) drainage of the N'mai 91

River in Myanmar to an altitude of 6318 m in the southeast of Zayü County 92

(Chaplin, 2005). The altitudinal gradient yields a vertical zonation of climate, 93

soil composition and solar radiation, generating diverse vegetation types and a rich 94

flora and fauna diversity (ibid). The vegetation in the Gaoligong Mountains, from 95

the valleys to the peaks, goes from tropical monsoon forest (< 1000 m a.s.l.), 96

monsoon evergreen broad-leaved forest (1100-1800 m), semi moist evergreen 97

broad-leaved forest (1700-2500 m) and mid-montane moist evergreen broad-leaved 98

forest above (1900-2800 m), coniferous broadleaved mixed forest (2700-3500 m), to 99

the alpine bush zone (> 3400 m) (Li et al., 2000). 100

5

According to Geissmann et al. (2011) and Ma et al. (2014), the known and potential 101

R. strykeri populations are restricted to a narrow range of 98°20'-98°49'E, 102

25°58'-26°31'N. Two insurmountable natural barriers, the N 'Mai Hka and Salween 103

Rivers, would appear to have physically blocked their expansion to other areas. 104

Thus, the study area for this project ranged from the China State Road 320 (named 105

Road NH3 in Myanmar) in the south to the provincial border between Yunnan and 106

the Tibet Autonomous Region in the north, and from the N' Mai Hka River in the 107

west to the Salween River in the east, with a total area of about 41,350 km2 108

(97.05-98.91E, 24.02-28.40N, Fig. 1). 109

2.2 Species Occurrence Extraction 110

Interview-based survey can cost-effectively and statistically access the number and 111

distribution of rare and poorly known large animals, especially in the case of primates 112

with a distinctive appearance such as the black snub-nosed monkey (Meijaard et al., 113

2011; Cui et al., 2015; Ma et al. 2015; Turvey et al. 2015). Between April 2011 and 114

December 2012, our research mission visited 68 villages near the high-altitude forests 115

and interviewed 358 old hunters or mountain villagers who entered the forests 116

frequently in Gongshan, Fugong and Lushui counties, on the west bank of the 117

Salween River in China (Ma et al., 2014). As a result, a total of 72 valid distributional 118

records was obtained from 67 respondents, and two groups of black snub nosed 119

monkeys then were confirmed in Pianma (Ma et al. 2014) and Luoma (Yang et al., 120

2016). These 72 records, which were further merged into 46 polygons with sizes 121

ranging from 16 to 1727 ha, were used as species occurrence data in China since R. 122

strykeri is not easily mistaken for any other primate given its peculiar appearance 123

(Fig.1). Three possible distribution records reported by Geissmann et al. (2011) in the 124

Maw River areas of Myanmar, with areas of 3854 ha, 4126 ha, and 6108 ha, were also 125

included in the species occurrence data. 126

6

2.3 Data acquisition 127

Data on both climate and vegetation were used to predict the distribution of the 128

black snub-nosed monkey. Thirty two Landsat ETM+/OLI images were downloaded 129

from USGS Earth Explorer (http://earthexplorer.usgs.gov/) to map forest cover over 130

the study area (see Table A.1 in Supporting Information). Climatic data which we 131

used include monthly maximum, minimum and mean temperature, monthly 132

precipitation, and 19 variables (see Table A.2). All these climatic data were acquired 133

from the WorldClim (http://www.worldclim.org, see Hijmans et al. 2005). As with 134

the Landsat images, the WorldClim (30 arcsec spatial resolution, equal to 0.00833°) 135

were transformed to a Universal Transverse Mercator coordinate system (UTM, 136

WGS84 datum, Zone 47 N) and spatially interpolated to 30 m resolution using 137

bilinear interpolation. 138

2.4 Forest mapping 139

510 sets of 30 × 30 m squares (9 squares per set) were systematically sampled as 140

“ground-truth data” from the study area by visual interpretation of high resolution 141

images on Google Earth (Google Inc, Table 1, Fig. A.1). Each square was deemed 142

woodland if the tree cover was higher than 40%; otherwise, it was determined as 143

non-woodland. The whole project area was divided into woodland area and 144

non-woodland area, based on Landsat TM/ETM+/OLI images (resolution = 30 m) 145

using a random forest algorithm with the “ground-truth data” (Breiman, 2001). The 146

classification process was implemented using R Statistics software 147

(http://www.r-project.org/) with the Random Forest package (Liaw & Wiener, 2002) 148

and Raster package (Hijmans, 2016). 149

2.5 Species distribution modelling 150

MAXENT is a popular species distribution model, with a history of superior 151

performance (Boubli & Lima, 2009; Thorn et al., 2009; Peck et al., 2011; Chetan et 152

7

al., 2014; Hernandez et al. 2006; 2008). Species occurrence sites and environmental 153

data are required to build a MAXENT model. At first, 10,000 pixels (30 m 154

resolution) were randomly selected from the study area as background data 155

(pseudo-absence data), 2000 samples of which were used as pseudo-absence data 156

for the MAXENT model, and the remaining 8000 samples were used for validation. 157

Two thirds of the species occurrence records, 33 of 49 distribution polygons, were 158

randomly selected as a training polygon, and the remaining 16 polygons were used 159

as test polygons. All the 49 distribution polygons were aggregated and rasterized at 160

30 m resolution. No > 100 pixels or 50% of the total pixels were systematically 161

sampled from each of the 49 distribution polygons. In total, 2772 samples were 162

obtained from the 33 training polygons, and 1008 samples were obtained from the 163

test polygons as test data. 164

A MAXENT model was built with the 2772 presence samples and 2000 background 165

samples based on the 67 climate variables with the default parameters setting in the 166

R package dismo (Hijmans et al. 2016). As a result, we obtained a climatic 167

suitability layer of the study area. Four levels of climatic suitability were quantified 168

according to relevant parameters of the MAXENT model for identifying core, 169

medium, low quality and none habitat (Table 2). 170

The Black snub-nosed monkey is arboreal and inhabits mainly subtropical 171

broad-leafed forest and mixed temperate forest. Forests provide shelter and ample 172

food for them; non-woodland areas could, therefore, be regarded as non-habitat. 173

We obtained habitat maps in both stages (2000s and 2015s) by masking out 174

non-woodland areas from the climatic suitability layer using forest cover maps. 175

Habitat changing from 2000 to 2015 was estimated by simply comparison of the 176

habitat maps in 2000 and 2015. 177

The AUC (area under curve) of ROC (receiver operating characteristic curves) was 178

employed to measure the accuracies of the habitat maps. The AUC values were 179

0.964 and 0.957, which indicated that our habitat evaluation models were accurate 180

(Araújo et al., 2005) 181

8

All the above procedures were performed using R Statistics software (version 3. 3. 1, 182

R Core Team, 2016) with some key packages, such as raster (Hijmans, 2016) and 183

dismo (Hijmans et al. 2016). 184

3. Results 185

3.1 Habitat distribution for the black snub-nosed monkey 186

Based on results generated from the hierarchical model, the current geographical 187

distribution for R. strykeri is predicted to be in the range of E98°20'-98°50', 188

N25°40'- 26°50', with a total area of approximately 3575 km2, of which the core 189

habitat is 1420 km2, medium quality habitat is 750 km2, and low quality habitats are 190

1405 km2 (Fig. 2). Among these suitable habitats, 2444 km2 (68.4%) are in 191

Myanmar, and 1131 km2 (31.6%) in China. Only 21.1% of the total habitat is 192

located in existing protected areas of China. The largest patch of core habitat 193

harboring most R. strykeri groups covered 1280 km2 and crossed the 194

China-Myanmar border within a range of 50 km, including 450 km2 (35.2%) in 195

China and 830 km2 (64.8%) in Myanmar. The vertical range of core habitat is 196

estimated to be 2330 to 3240 m in both countries. 197

3.2 Habitat alterations in the past fifteen years 198

In 2000 there was an estimated 3670 km2 of habitat in the Sino-Myanmar border 199

region, but by 2015 this figure had declined by 95 km2 (a destruction of 2.6% over 200

time or a mean of 0.17% yr-1). Correspondingly, the core, medium and low quality 201

habitats decreased by 50, 20, 25 km2. Of the total habitat loss, 96% was in northern 202

Myanmar. The loss of habitat within Myanmar from 2000 to 2015 was estimated to 203

be 97 km2, or 3.8% of the entire habitat of the species in Myanmar for the year 2000 204

(Fig. 3), while in China, forest cover loss (4 km2) was lower than the amount of 205

9

forest rehabilitation (6 km2); the destruction occurred mostly outside the Gaoligong 206

Mountains National Nature Reserve. 207

4. Discussion 208

4.1 Urgent conservation needs and applications 209

In general, the entire habitat area was reduced between 2000 and 2015, although the 210

overall speed of loss and recovery was irregular (Fig. 3). In Myanmar, R. strykeri 211

faces intense pressure from habitat loss and hunting, as there are no conservation 212

measures to date. Habitat loss is due not only to shifting cultivation near human 213

settlements in the valley, but also extends along criss-crossing roads because of 214

commercial logging, dam construction and mining into the central core habitat area, 215

and even peaks (Momberg et al., 2010; Geissmann et al., 2012), which may have 216

affected population continuity because of habitat fragmentation and disruption of 217

habitat connectivity (see Fig. 3, right images). In addition, hunting of this species is 218

still very severe, even after the species’ discovery in 2010 (Meyer et al., 2015); 219

combined with habitat loss it is projected that this may lead to a species decline of > 220

80% in the next fifteen years in Myanmar (Geissmann et al., 2012). In China, more 221

than half of the habitat (753 km2, 66.6%) is located outside the reserve and there has 222

been depletion of forest resources caused by human population growth. 223

Occasionally hunting for bush meat and traditional medicine has provided further 224

pressure (Ma et al., 2014; Fig. A.2.F). Considering all these factors, we propose the 225

three following major conservation procedures: 226

First, conservation gaps should be filled by the establishment of new protected areas 227

(Fig. 2). The habitat of R. strykeri slightly recovered in the Gaoligong Mountains 228

National Nature Reserve, and this indicates that the creation of new protected areas 229

might be an efficient method to maintain the habitat in this region. Long et al. (2012) 230

recommended Sino-Myanmar cooperation in delineating a trans-boundary wildlife 231

10

sanctuary to protect this species. Support by the relevant institutions, the Myanmar 232

Ministry of Environmental Conservation and Forestry and the National Democratic 233

Army of Kachin (a pro-government militia), has already led to agreement to create a 234

protected area, named Imawbum National Park, in Sawlaw Township (pers. comm. 235

Fauna & Flora International, 2016). We propose that the range of Imawbum 236

National Park should border the Gaoligong Mountains National Nature Reserve in 237

China (see Fig. 2), which covers an area of 1934 km2. At present, the proclamation 238

of new national parks in China is proceeding rapidly. The Program for 239

National Park Development in Yunnan Province 2009-2020 under the Yunnan 240

Provincial Government proposed to create a national park in Nujiang Autonomous 241

Prefecture. We hence suggest setting up the range of Gaoligong Mountains National 242

Park (328 km2) to link the southern and central segments of the Giaoligong 243

Mountains National Nature Reserve in Lushui County (Fig. 2). 244

Second, management plans need to consider the linkages between the protected 245

areas on both sides of the border and the importance of maintaining habitat corridors 246

that presently connect them. Based on the habitat distribution pattern, two wildlife 247

corridors with management measures are proposed here to contribute habitat 248

connection and connectivity maintenance. In particular, the most important 249

trans-boundary habitat corridor should be located in the northern edge of the 250

fragmented habitats to connect the two largest core habitats in Myanmar (Fig. 2, 251

black circled area C1) in order to enable population dispersal and gene flow and 252

increase the probability of the species’ persistence. Timber extraction, shifting 253

cultivation and road construction are imminent threats to this corridor. Any further 254

deforestation should be prohibited there, and the lost habitats must be speedily 255

restored by natural rehabilitation. We subsequently propose that the remaining 256

vegetation in Myanmar must be strictly protected where the area borders the habitat 257

at Gangfang Village of Pianma Township in China (Fig. 2, black circled area C2). 258

The habitat there was dramatically disjointed by commercial harvesting, shifting 259

cultivation, road construction and mining; a small cross-border population, however, 260

11

may inhabit this general area (Ma et al., 2014; a newly dead adult female specimen 261

was collected from a hunter in Gangfang of Myanmar in November, 2015). Natural 262

recovery assisted by artificial restoration of trees in this area will quickly restore a 263

habitat corridor on the Sino-Myanmar border and increase the chances for survival 264

for the threatened population. 265

Third, we urgently recommend a Sino-Myanmar joint-patrol law enforcement team 266

for implementation of the trans-boundary zoning and management plan to combat 267

hunting and illegal trans-boundary logging activities. Thus, we can ensure that 268

effective National Park management will be in place and that both forest 269

departments can enforce the law against hunting and illegal logging in the border 270

area, and to avoid a “paper park” becoming established like many others in 271

Myanmar (see Rao et al., 2002; Aung, 2007). As the Sino-Myanmar border is a 272

relatively poor and backward remote area, hunting wildlife has become a source of 273

protein supplementation and income for local minorities; it is therefore important to 274

China and Myanmar to set up an intergovernmental investment fund for payments 275

for ecosystem services, developing sustainable alternative livelihoods, and 276

community co-management projects, improving financial support to the protected 277

areas for their effective operation and capacity building, and enhancing 278

environmental education for local people and park staffs. 279

As a supplementary note we also suggest that the distribution maps produced here 280

should be superimposed with existing and future land-use maps for both the 281

development departments of China and Myanmar to prevent future damage of the 282

stretches of habitat that are key for the conservation of R. strykeri. 283

4.2 Determining new investigation areas 284

Another contribution of our study is to determine where the monkeys may occur for 285

future field survey. Further population investigations should be conducted in the 286

core habitats which have not been investigated and monitored. According to 287

12

MAXENT, for example, field census or interview based surveys should be 288

conducted in the core habitat between the proposed corridors of C1 and C2 (Fig. 2) 289

which has a great potential to sustain some R. strykeri populations. Besides, the 290

peripheral population in marginal habitats are probably more subject to habitat loss 291

and more vulnerable to extinction before conservationists can find them. 292

Accordingly, the northernmost habitat (Fig. 2, black circled area NO. M1), the 293

Imawbum Mountain (black circled area NO. M2), and the southernmost habitat 294

(black circled area NO. M4) need immediate field censuses to confirm the potential 295

existence of peripheral populations. 296

4.3 Approach caveats 297

There probably are some prediction errors resulting from insufficient information on 298

habitat. The elevational range of the core habitats is strongly associated with the 299

mid-montane moist evergreen broad-leaved forest and coniferous broadleaved 300

mixed forest (Li et al., 2000). We found that R. strykeri prefers the abovementioned 301

forest types but avoids secondary deciduous broad-leaved forest, secondary bamboo 302

forest, artificial pine forest and coppice lands. These non-preferred forests have 303

mostly been degraded and fragmented by large scale logging, road creation and 304

burning or Lisu and Law Waw peoples' forest-crop rotation on these lands (eg. Fig. 305

3, black circled area NO. M3). We did not consider any of these forests in this study 306

due to the limitation of ground-truth data for forest mapping. Additionally, the 307

woodland being used in forest mapping includes shrub land, thus contains some 308

non-habitats for the arboreal R. strykeri. Therefore, the entire habitat range could 309

eventually be overrated and the habitat loss rate underestimated. A fine-scale 310

vegetation type map of the distribution of R. strykeri, where possible, should be 311

considered. Moreover, very informative materials are also essential to catch the 312

typically patchy distribution of intense anthropogenic impacts, which in our study 313

case include dam construction, mining sites, road opening and human settlements. 314

315

13

5. Conclusion 316

As the black snub-nosed monkey is endemic to the junction of two biodiversity 317

hotspot regions, Indo-Burma and Mountains of Southwest China (Marchese, 2015), 318

we believe that, to protect this critically endangered species, all of its actual and 319

potential distribution area should be strictly protected by banning hunting, logging, 320

agricultural reclamation and other improper activities. As an umbrella and global 321

flagship species, the effective conservation of R. strykeri will also protect other 322

sympatrically threatened biota (see Tab. A.3). Results from this work highlighted 323

that strengthening current conservation practices should ensure the long survival of 324

R. strykeri in China while establishment of cross-border conservation and the rapid 325

elimination of the threat in Myanmar are necessary for re-emergence of this species 326

in that country and across the border. The conservation framework proposed in this 327

paper therefore can serve as a basic prop for conservation department to formulate 328

more specific strategies to protect R. strykeri. 329

330

331

332

333

334

335

336

337

14

ACKNOWLEDGMENTS 338

This research complied with the policies and guidelines of the Australian National 339

University and Dali University for the ethical treatment of primates. Funding was 340

provided by the Zoological Society for the Conservation of Species and Populations 341

(Germany, 7.Rhinopithecus.MMR.2015) and National Science Foundation of China 342

(31560599). The authors were supported during this project by the Nujiang Forestry 343

Department and the Gaoligong Mountains National Nature Reserve. Thanks are due 344

to Emeritus Professor Colin Groves for his valuable suggestions on this paper and his 345

help in English editing; three anonymous reviewers provided constructive comments 346

to improve the original manuscript, and Frank Momberg and Guy Michael Williams 347

(FFI, Myanmar) shared with us their knowledge and research outcomes on the species 348

in Myanmar. Our field guides, Mr. Dong Shaohua and Mr. Kenhuaniuba, contributed 349

materially to this study. 350

REFERENCES CITED 351

Aung, U.M., 2007. Policy and practice in Myanmar's protected area system. J. 352

Environ. Manage. 84, 188-203. 353

Araújo, M.B., & Peterson, A.T., 2012. Uses and misuses of bioclimatic envelope 354

modeling. Ecology, 93, 1527-1539. 355

Araújo, M.B., Pearson, R.G., Thuiller, W., Erhard, M., 2005. Validation of 356

species–climate impact models under climate change. Glob Chang Biol. 11, 357

1504-1513. 358

15

Breiman, L., 2001. Random forests. Mach Learn. 45, 5-32. 359

Boubli, J.P., De Lima, M.G., 2009. Modeling the geographical distribution and 360

fundamental niches of Cacajao spp. and Chiropotes israelita in 361

Northwestern Amazonia via a maximum entropy algorithm. Int. J. 362

Primatol. 30, 217-228. 363

Campos, F.A., Jack, K.M., 2013. A potential distribution model and conservation 364

plan for the critically endangered Ecuadorian capuchin, Cebus albifrons 365

aequatorialis. Int. J. Primatol. 34, 899-916. 366

Chaplin, G., 2005. Physical geography of the Gaoligong Shan area of southwest 367

China in relation to biodiversity. Proc. Calif. Acad. Sci. 56, 527-556. 368

Chetan, N., Praveen, K.K., Vasudeva, G.K., 2014. Delineating Ecological 369

Boundaries of Hanuman Langur Species Complex in Peninsular India Using 370

MAXENT Modeling Approach. PloS one 9, e87804. 371

Cilliers, D., Evans, S., Coetzee, H., van Rensburg, L., 2013. Developing a site 372

selection tool to assist reintroduction efforts for the Southern 373

Ground-Hornbill Bucorvus leadbeateri. Ostrich 84, 101-111. 374

Cui, L.W., Li, Y.C., Li, J.F., He, X.Y., Ma, C., Scott, M.B., Li, D.H, Sun, J., Sun, 375

W.M. Xiao, W., 2015. Distribution and conservation status of Shortridge's 376

capped langurs Trachypithecus shortridgei in China. Oryx 50: 732-741. 377

Geissmann, T., Lwin, N., Aung, SS., Aung, TN., Aung, ZM., Hla, TH., Grindley, 378

M., Momberg, F., 2011. A new species of snub-nosed monkey, genus 379

Rhinopithecus Milne-Edwards, 1872 (Primates, Colobinae), from northern 380

Kachin state, northeastern Myanmar. Am. J. Primatol. 73: 96-107. 381

Geissmann, T., Momberg, F. Whitten, T., 2012. Rhinopithecus strykeri. The IUCN 382

Red List of Threatened Species 2012: 383

e.T13508501A13508504. http://dx.doi.org/10.2305/IUCN.UK.2012-1.RLTS384

.T13508501A13508504.en (accessed 16.03.24). 385

16

Giovanelli, J.G.R., Siqueira, M.F.D., Haddad, C.F.B., Alexandrino, J., 2010. 386

Modeling a spatially restricted distribution in the neotropics: how the size of 387

calibration area affects the performance of five presence-only methods. Ecol. 388

Modell. 221, 215-224. 389

Hernandez, P.A., Graham, C.H., Master, L.L., Albert, D.L., 2006. The effect of 390

sample size and species characteristics on performance of different species 391

distribution modeling methods. Ecography 29, 773-785. 392

Hernandez, P.A., Franke, I., Herzog, S.K., Pacheco, V., Paniagua, L., Quintana, 393

H.L., Soto, A., Swenson, J.J., Tovar. C., Valqui, T.H., Vargas, J., Young, 394

B.E., 2008. Predicting species distributions in poorly-studied landscapes. 395

Biodivers. Conserv. 17, 1353-1366. 396

Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. Jarvis, A., 2005. Very high 397

resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 398

25, 1965-1978. 399

Hijmans, R.J., 2016. Raster: Geographic Data Analysis and Modeling. R package 400

version 2.5-8. 401

Hijmans, R.J., Phillips, S., Leathwick, J., Elith, J., 2016. dismo: Species Distribution 402

Modeling. R package version 1.1-1. 403

Illera, J.C., Von Wehrden, H., Wehner, J., 2010. Nest site selection and the effects 404

of land use in a multi-scale approach on the distribution of a passerine in an 405

island arid environment. J. Arid Environ. 74, 1408-1412. 406

Ingberman, B., Fusco-Costa, R. and de Araujo Monteiro-Filho, E.L., 2016. A 407

Current Perspective on the Historical Geographic Distribution of the 408

Endangered Muriquis (Brachyteles spp.): Implications for 409

Conservation. PloS one. 11. p.e0150906. 410

Li, H., Guo, H.J., Dao, Z.L., 2000. Flora of Gaoligong mountains. Science Press, 411

Beijing. pp. 6-46. 412

17

Liaw, A., Wiener, M., 2002. Classification and regression by random forest. R 413

news, 2, 18-22. 414

Liu, F., McShea, W.J., Li, D., 2016. Correlating habitat suitability with landscape 415

connectivity: A case study of Sichuan golden monkey in China. Ecol. Model. 416

doi:10.1016/j.ecolmodel.2016.09.004 417

Long, Y., Momberg, F., Ma, J., Wang, Y., Luo, Y., Li, H., Yang, G., Li, M., 2012. 418

Rhinopithecus strykeri found in China!. Am. J. Primatol. 74, 871. 419

Luo, Z., Yu, H., Pu, Y., Yang, J., Mei, H., Wang, D., Zhu, Z., Zhao, M., Wu, H., 420

2016. Assessment of Habitat Fragmentation and Corridors for an Isolated 421

Subspecies of the Sichuan Golden Snub-Nosed Monkey, Rhinopithecus 422

roxellana hubeiensis. Int. J. Primatol. 37, 438-459. 423

Ma, C., Luo, Z., Liu, C., Orkin, J.D., Xiao, W., Fan, P., 2015. Population and 424

Conservation Status of Indochinese Gray Langurs (Trachypithecus 425

crepusculus) in the Wuliang Mountains, Jingdong, Yunnan, China. Int. J. 426

Primatol. 36, 749-763. 427

Ma, C., Huang, ZP, Zhao, XF., Zhang, LX., Sun, WM., Scott, MB., Wang, XW., 428

Cui, LW., Xiao, W., 2014. Distribution and conservation status of 429

Rhinopithecus strykeri in China. Primates 55, 1-6. 430

Marchese, C., 2015. Biodiversity hotspots: A shortcut for a more complicated 431

concept. Global Ecol. Conserv. 3, 297-309. 432

Meijaard, E., Mengersen, K., Buchori, D., Nurcahyo, A., Ancrenaz, M., Wich, S., 433

Atmoko, S.S.U., Tjiu, A., Prasetyo, D., Nardiyono, Hadiprakarsa, Y., 434

Christy, L., Wells, J., Albar, G., Marshall, A.J., 2011. Why don't we ask? A 435

complementary method for assessing the status of great apes. PloS one. 6, 436

e18008. 437

18

Meyer, D., Matauschek, C., Momberg, F., Roos, C., 2015. Preliminary Status 438

Review of Leaf Monkeys in Myanmar. Fauna Flora International, Yangon, 439

Myanmar & German Primate Center, Göttingen, Germany. 38P. 440

Momberg, F., Lwin, N., Geissmann, T., 2010. Report on a snub-nosed monkey and 441

biodiversity field survey, April 24–May 8, 2010. Yangon: Myanmar Primate 442

Conservation Program.16p. 443

Peck, M., Thorn, J., Mariscal, A., Baird, A., Tirira, D., Kniveton, D., 2011.Focusing 444

conservation efforts for the critically endangered brown-headed spider 445

monkey (Ateles fusciceps) using remote sensing, modeling, and playback 446

survey methods. Int. J. Primatol. 32, 134-148. 447

Phillips, S.J., Dudík, M., 2008. Modeling of species distributions with MAXENT: 448

new extensions and a comprehensive evaluation. Ecography 31, 161-175. 449

R Core Team, 2016. R: A language and environment for statistical computing. R 450

Foundation for Statistical Computing, Vienna, Austria. 451

URL https://www.R-project.org/. 452

Rao, M., Rabinowitz, A., Khaing, S.T., 2002. Status Review of the Protected-Area 453

System in Myanmar, with Recommendations for Conservation Planning. 454

Conserv. Biol. 16 360-368. 455

Schaffer-Smith, D., Swenson, J.J., Bóveda-Penalba, A.J., 2016. Rapid conservation 456

assessment for endangered species using habitat connectivity models. 457

Environ. Conserv. 43, 221-230. 458

Thorn, J.S., Nijman, V., Smith, D., Nekaris, K.A.I., 2009. Ecological niche 459

modelling as a technique for assessing threats and setting conservation 460

priorities for Asian slow lorises (Primates: Nycticebus). Divers. Distrib. 15, 461

289-298 462

Turvey, S.T., Trung, C.T., Quyet, V.D., Nhu, H.V., Thoai, D.V., Tuan, V.C.A., Hoa, 463

D.T., Kacha, K., Sysomphone, T., Wallate, S., Hai, C.T.T., Thanh, N.V., 464

19

Wilkinson, N.M., 2015. Interview‐based sighting histories can inform 465

regional conservation prioritization for highly threatened cryptic species. J. 466

Appl. Ecol. 52, 422-433. 467

Urbina-Cardona, J.N., & Loyola, R.D., 2008. Applying niche-based models to 468

predict endangered-hylid potential distributions: are neotropical protected 469

areas effective enough. Trop. Conserv. Sci. 1, 417-445. 470

Yang, Y., Tian, Y., He, C., Huang, Z., Dong, S., Huang, Z., Wang, B., Li, G., Xiang, 471

Z., Long, Y., Xiao, W., 2016. The Critically Endangered Myanmar 472

snub-nosed monkey Rhinopithecus strykeri found in the Salween River 473

Basin, China. Oryx 1-3. doi:10.1017/s0030605316000934 474

Biosketches 475

Guopen REN is a College Associate Professor based at the Institute of 476

Eastern-Himalaya Biodiversity Research, Dali University. His work specializes in 477

remote sensing, GIS, spatial ecology and their applications in biodiversity and 478

conservation management. 479

Yin YANG is a PhD student at the School of Archaeology & Anthropology, 480

Australian National University with research interests in understanding the ecology 481

and conservation needs for the black snub-nosed monkey in the Gaoligong Mountains 482

Region. 483

Author contributions: WX designed the study. GPR, YY, WW and GY conducted 484

data analysis. XDH, GSL, ZPH and CM collected species occurrence data. YY wrote 485

20

the first draft, WX, GPR and YY revised the manuscript, and all authors made 486

extensive contributions to the text. 487

488

Table 1. Hierarchical definition of forests

Name Definition

Fs: forest + shurb Tree crown cover degree > 40%，including forest and shrub.

Fw: forest = Fc + Fo Tree crown height >= 3m

Fc: closed-forest Tree crown cover degree >= 70%

Fo: open-forest 40% =< Tree crown cover degree < 70%

Sh: shrub Plant cover degree >= 40%, but not fulfilling the other

conditions of forests

489

490

491

492

493

494

495

496

497

498

499

500

501

502

21

503

504

505

Table 2. Habitat quality levels description of black snub-nosed monkeys

Quality Predicted

value Description

Core > 0.50 Equal training sensitivity and specificity (0.50)

Medium 0.41-0.50 Maximum training sensitivity plus specificity (0.41)

Edge 0.04-0.41 Minimum training presence data omission (0.04)

Non-habitat < 0.04 Less than 0.1% of training presence data predicted (0.04)

506

507

508

509

510

511

22

512

Figure 1. Map of the study area and distribution of R. strykeri in the Sino-Myanmar 513

border region. 514

515

516

517

518

519

520

521

23

522

Figure 2. Predictive habitat distribution and quality map with proposed 523

conservation and management areas for R. strykeri in the Sino-Myanmar border 524

region according to climate niche analysis, forest cover in 2015 and MAXENT. 525

526

527

24

528

Figure 3. Habitat alterations (2000-2015) in the Sino-Myanmar border region, 529

measured as the percentage of habitat cover change within 1 × 1 km2 grid cells. The 530

locations of habitat loss are marked with red colours and rehabilitation with blue 531

colours. The right-hand images show an example of habitat loss, fragmentation and 532

insularization in the study area during 2000 and 2015. 533

534

25

Supporting Information (When using data in this part, please cite the original 535

publication) 536

1. Modeling Setting: 537

The default setting was used for the climatic and species distribution parameters in 538

the R package dismo: 539

maxent.sample = maxent (x = trainvals.sample, p = species.sample) 540

in which, trainvals.sample is a 4772 × 67 matrix, and species.sample is a vector 541

of length 4772 . 542

543

Table Appendix 1. Path, row numbers and acquisition dates of Landsat images 544

used in this study. 545

Landsat 7, 2000s Landsat 8, 2015s

path row date path row date

132 40 2000360 132 40 2014022

132 41 2000360 132 41 2014022

132 42 2002013 132 42 2014022

132 43 2002013 132 43 2014022

133 40 2000047 133 40 2015016

133 41 2000047 133 41 2015016

133 42 2000047 133 42 2015016

133 43 2000047 133 43 2015048

132 40 2002333 132 40 2015105

132 41 2002333 132 41 2015105

132 42 2002333 132 42 2015073

132 43 2001330 132 43 2015073

133 40 2001337 133 40 2015080

133 41 1999300 133 41 2015080

133 42 1999316 133 42 2015080

133 43 1999316 133 43 2015080

546

26

Table Appendix 2. Bioclimatic variables used in the model

BIO1 Annual mean temperature

BIO2 Mean diurnal temperature range [mean of monthly (max

temp–min temp)]

BIO3 Isothermality (P2/P7) (×100)

BIO4 Temperature seasonality (standard deviation×100)

BIO5 Max temperature of warmest month

BIO6 Min temperature of coldest month

BIO7 Temperature annual range (P5–P6)

BIO8 Mean temperature of wettest quarter

BIO9 Mean temperature of driest quarter

BIO10 Mean temperature of warmest quarter

BIO11 Mean temperature of coldest quarter

BIO12 Annual precipitation

BIO13 Precipitation of wettest month

BIO14 Precipitation of driest month

BIO15 Precipitation seasonality (coefficient of variation)

BIO16 Precipitation of wettest quarter

BIO17 Precipitation of driest quarter

BIO18 Precipitation of warmest quarter

BIO19 Precipitation of coldest quarter

547

548

549

550

551

552

553

554

555

27

Table Appendix 3. List of threatened mammals monitored in Rhinopithecus 556

strykeri habitat in the Gaoligong Mountains. Location: MW, western slopes of 557

Gaoligong Mountain, northeastern Kachin state, Myanmar; WC, western slopes of 558

Gaoligong Mountain, Yunnan, China; EC: eastern slopes of Gaoligong Mountain, 559

Yunnan, China. Habitat types: LE: low land subtropical evergreen rain forest; SM: 560

semi moist evergreen broad-leaved forest; ME: mid-montane moist evergreen 561

broad-leaved forest; TC, temperate coniferous forest; BB, bamboo bushes. 562

Monitoring method: CT, camera trapping; FO, field observation; ST, specimens of 563

skulls and tails collected from hunters; IT, interview. IUCN RED List: CR, critically 564

endangered; EN, endangered; VU: vulnerable; NT, near threatened. CITES 565

Appendix: APP. I, Appendix I; APP. II, Appendix II. 566

Species Location

Elevation

(m) /

Habitat

types

Monitori

ng

method

Conserva

tion

status

Reference

Macaca

arctoides

MW, WC,

EC

2500-3100 /

ME, TC,

BB

CT, IT,

FO

VU,

APP. II

Momberg et al.,

2010; Chen et

al., 2016; Fig.

A.2.D

Macaca

assamensis

MW, WC,

EC

2500-3100 /

ME, TC,

BB

CT; FO,

ST

NT,

APP. II

Momberg et al.,

2010; Chen et

al., 2016; Fig.

A.2.C

Macaca leonina MW < 2700/ LE,

SM, ME IT, FO

VU,

APP. II

Momberg et al.,

2010

Macaca mulatta MW, EC

1300-3800/

LE, SM,

ME, TC

IT, ST APP. II

Momberg et al.,

2010; Xiao et

al., 2013a

Trachypithecus

phayrei

MW, WC,

EC

1200-2600

SM, ME FO; IT

EN,

APP.I

Meyer et al.,

2015; Xiao et

al,. 2013a

Trachypithecus

shortridgei MW, EC

500-2000 /

LE, SM,

ME

FO, ST EU,

APP. I

Meyer et al.,

2015;

Momberg et al.,

28

2010; Xiao et

al,. 2013a

Hoolock

tianxing MW, WC

400-2000 /

LE, SM,

ME

FO, ST,

IT

EN,

APP. I

Geissmann et

al., 2013; Fan

et al., 2016

Budorcas

taxicolor

MW, WC,

EC

2700-3200 /

ME, TC,

BB

CT, FO,

IT, ST

VN,

APP. II

Momberg et al.,

2010; Xiao et

al., 2013b

Elaphodus

cephalophus WC

2500-3300 /

ME, TC,

BB

CT, FO NT

Momberg et al.,

2010; Chen et

al., 2016

Capricornis

milneedwardsii

MW, WC,

EC

2500-3100 /

ME, TC,

BB

CT, ST NT,

APP. II

Chen et al.,

2016; Fig.

A.2.E

Ailurus styani MW, WC,

EC

2300-3500 /

ME, TC,

BB

CT, FO EN,

APP. I

Momberg et al.,

2010; Chen et

al., 2016; Fig.

A.2.B

Pardofelis cf.

marmorata EC

<2800 / SE,

ME CT

NT,

APP. I Fig. A.2.A

Ursus

thibetanus MW, WC

<3200 /

ME, TC,

BB

CT VN,

APP. I

Momberg et al.,

2010; Chen et

al., 2016

Helarctos

malayanus MW

< 2700 m /

SE, ME FO, ST

VN,

APP. I

Momberg et al.,

2010

Cuon alpinus MW Uncertain Iron Trap EN,

APP. II

Momberg et al.,

2010

References cited:

Chen, Y., Xiao, Z., Li, M., Wang, X., He, C., He, G., Li, H., Shi. S., Xiang, Z., 2016.

Preliminary survey for the biodiversity of mammal and bird using camera

traps in the west slope of mid-section Mt. Gaoligong. Acta Theriol Sinica. 36,

123-128.

Fan, P.-F., He, K., Chen, X., Ortiz, A., Zhang, B., Zhao, C., Li, Y.-Q., Zhang, H.-B.,

Kimock, C., Wang, W.-Z., Groves, C., Turvey, S.T., Ross, C., Helgen, K.M.,

Jiang, X.-L., 2017. Description of a new species of Hoolock gibbon

(Primates: Hylobatidae) based on integrative taxonomy. Amer. J. Primatol.

2017; 9999:22631

Geissmann, T., Grindley, M., Lwin, N., Aung, SS., Aung, TN., Htoo, SB., Momberg,

29

F., 2013. The Conservation Status of Hoolock Gibbons in Myanmar. Gibbon

Conservation Alliance, Zürich, Switzerland, xii + 157 pp.

Meyer, D., Matauschek, C., Momberg, F., Roos, C., 2015. Preliminary Status Review

of Leaf Monkeys in Myanmar. Fauna Flora International, Yangon, Myanmar

& German Primate Center, Göttingen, Germany. 38pp.

Momberg, F., Lwin, N., Geissmann, T., 2010. Report on a snub-nosed monkey and

biodiversity field survey, April 24–May 8, 2010. Yangon: Myanmar Primate

Conservation Program.16pp.

Xiao, W., He, X., Ma, C., Zhang, L., 2013a. Report on primate distribution survey in

Lisu Autonomous Prefecture of Nujiang, Yunnan. The Institute of Eastern

Himalayan Resources and Environment, Dali University, China.

Xiao, W., Yang, G., He, X., Ma, C., Zhang, L., 2013b. Report on population and

distribution status of Gaoligong Takin (Budorcas taxicolor taxicolor) in Lisu

Autonomous Prefecture of Nujiang, Yunnan. The Institute of Eastern

Himalayan Resources and Environment, Dali University, China.

567

568

569

570

571

572

573

574

575

576

577

578

579

30

580

Figure Appendix 1. High resolution images from Google Earth, aggregated to 30 m 581

resolution. The left-hand image has several land cover categories, including forest, 582

shrub, building and farmland. Zoom-in in image 1 shows scattered build-ups; 583

images 3 and 8 correspondingly show forest cover and dense shrubs; farmland is 584

presented in window 5. 585

586

587

588

589

590

591

592

593

594

595

596

31

597

598

599

600

601

602

603

604

605

606

607

608

609

610

611

Figure Appendix 2. Mammals and human threats were monitored by camera traps 612

at Rhinopithecus strykeri’s habitats on the eastern slope of the Gaoligong Mountains, 613

China from April 2015 to January 2017. A: Pardofelis marmorata (Credit to Wang 614

Bin); B: Ailurus styani; C: Macaca assamensis; D: Macaca arctoides; E: 615

Capricornis milneedwardsii; F: two hunters with a shotgun. 616

617

618

619

F

A B

C D

E F

32

620

Figure Appendix 3. Probability distribution curves (six examples: A-F) show how 621

climate variables affect the MAXENT prediction of habitat suitability for 622

Rhinopithecus strykeri in the study area. Blue areas correspond to climate variables 623

in the study area while red areas correspond to climate variables in the distribution 624

range of R. strykeri. Among 67 climate variables, the seasonal variation of 625

temperature (B: bio4), the temperature annual range (D: bio7), the average 626

precipitation in November (F: pre11) and the maximum temperature in May (E: 627

tmax5) contributed most to the MAXENT established in this study, reflecting a 628

A B

C D

E F

33

strong selectivity of R. strykeri for the above-mentioned variables. A: bio1, annual 629

mean temperature; C: bio5, maximum temperature of the warmest month. 630