Assessment of habitat suitability of the snow leopard (Panthera uncia) in Qomolangma National Nature Reserve based on MaxEnt modeling

2018·20.1 MB·

by  Bai, De-Feng

ZOOLOGICAL RESEARCH
Assessment of habitat suitability of the snow leopard
(Panthera uncia) in Qomolangma National Nature
Reserve based on MaxEnt modeling
De-FengBai1,2,Peng-JuChen1,LucianoAtzeni1,LhabaCering3,QianLi2,KunShi1,4,*
1WildlifeInstitute,SchoolofNatureConservation,BeijingForestryUniversity,Beijing100083,China
2EverestSnowLeopardConservationCenter,RikazeXizang857000,China
3QomolangmaNationalNatureReserveAdministration,RikazeXizang857000,China
4Eco-BridgeContinental,Beijing100085,China
ABSTRACT 000 m a.s.l. influenced snow leopard preferences at
the landscape level in QNNR. We advocate further
Habitat evaluation constitutes an important and research and cooperation with Nepal to evaluate
fundamental step in the management of wildlife habitatconnectivityandtoexplorepossibleproxiesof
populations and conservation policy planning. populationisolationamongthesepatches. Furthermore,
Geographic information system (GIS) and species evaluationofsubdivisionswithintheprotectionzonesof
presence data provide the means by which such QNNRisnecessarytoimproveconservationstrategies
evaluation can be done. Maximum Entropy (MaxEnt) andenhanceprotection.
iswidelyusedinhabitatsuitabilitymodelingduetoits
Keywords: Qomolangma National Nature Reserve;
powerofaccuracyandadditionaldescriptiveproperties.
Snowleopard;MaxEnt;Habitatsuitabilityassessment;
To survey snow leopard populations in Qomolangma
Tibet
(Mt. Everest)NationalNatureReserve(QNNR),Xizang
INTRODUCTION
(Tibet), China, we pooled 127 pugmarks, 415 scrape
marks,and127non-invasiveidentificationsoftheanimal Wildlifehabitatisdefinedasthesurroundingenvironmentwhere
along line transects and recorded 87 occurrences wildanimalscanaccomplishtheirlifecycle(Cody,1987;Jianget
al.,2012).Habitatssupplyresourcesforpopulationpersistence,
throughcameratrapsfrom2014–2017. Weadoptedthe
representing a determining factor for survival and successful
MaxEntmodeltogenerateamaphighlightingtheextent
reproduction (Wang & Chen, 2004; Yang et al., 2000). The
of suitable snow leopard habitat in QNNR. Results habitat suitability index (HSI) is a measure of the ability for a
showed that the accuracy of the MaxEnt model was habitattosustainaspeciesandisanimportantindicatorforthe
excellent(meanAUC=0.921). Precipitationinthedriest qualityofagivenhabitat(Jinetal.,2008;Luetal.,2012;Song
etal.,2014).Suchevaluationconstitutesoneofthefirststepsin
quarter,ruggedness,elevation,maximumtemperature
wildlifeprotectionandmanagement,offeringascientificrationale
ofthewarmestmonth,andannualmeantemperature
fortheimprovementofconservationpolicies(Liuetal.,2013).
werethemainenvironmentalfactorsinfluencinghabitat
suitabilityforsnowleopards, withcontributionratesof
20.0%, 14.4%, 13.3%, 8.7%, and 8.2% respectively. Received: 01December2017;Accepted: 29March2018;Online: 24
The suitable habitat area extended for 7 001.93 km2, May2018
representing22.72%ofthewholereserve. Theregions Foundation items: This project was funded primarily by the Everest
borderingNepalwerethemainsuitablesnowleopard SnowLeopardConservationCenter, apartnershipinitiativeofVanke
habitatsandconsistedofthreeseparatehabitatpatches. FoundationandQomolangmaNationalNatureReserveAdministration
Our findings revealed that precipitation, temperature *Correspondingauthor,E-mail:[email protected]
conditions, ruggedness, and elevations of around 4 DOI:10.24272/j.issn.2095-8137.2018.057
SciencePress ZoologicalResearch39(6):373–386,2018 373
The snow leopard (Panthera uncia) is a feline species MaxEnt models were originally formulated to estimate the
distributed over 12 countries in Central Asia. It is estimated presence density of a target species across a landscape
thatChinacontainsapproximately60%ofthepotentialhabitat (Phillips et al., 2006), but have since been applied to model
available to snow leopards, who are reported to reside in species distribution and environmental niches, relying on
the western provinces of Xinjiang, Xizang (Tibet), Qinghai, “presence-only” data and environmental predictors, thereby
Gansu, Sichuan, and Inner Mongolia (McCarthy & Chapron, avoiding bias and leading to accurate results (Merow et al.,
2003; Riordan & Shi, 2010, 2016). The species is currently 2013; Pearson et al., 2007; Phillips et al., 2006). MaxEnt
classified as Vulnerable by the IUCN (International Union for modelsarereliable,straightforward,andallowdatatobeeasily
Conservation of Nature and Natural Resources, 2017) and obtained (Merow & Silander, 2014; Phillips & Dudík, 2008;
listed as a Class I protected animal by the China Key List Radosavljevic&Anderson,2014). Thesemodelsaresuitable
(Riordan & Shi, 2010; Wang, 1998). Snow leopard-oriented for the prediction of species habitat distribution at the whole
researchspansadiverserangeofareas,includingabundance reserve level, and an appropriate model for habitat suitability
and density (Alexander et al., 2015; Jackson et al., 2006), analysis(Haegeman&Etienne,2010; Stachura-Skierczyn´ska
homerange(Johanssonetal.,2016),diet(Wangetal.,2014), etal.,2009;Xing&Hao,2011).
behavior (Li et al., 2013), genetic diversity (Janecˇka et al., Understanding the distribution of snow leopard habitat in
2017), climate change impact (Aryal et al., 2014d, 2016), QNNRisimportantfortheimprovementofresearchoutcomes
human-snowleopardconflict(Aryaletal.,2014b; Chenetal., andconservationplans. Thus,ourstudyaimedto: (1)assess
2016),andtranslocationofpreyspecies(Aryaletal.,2013). the potential habitat distribution for snow leopard in QNNR,
Previous studies on snow leopard habitat use indicate two along with biotic and abiotic factors of influence using the
broad components determining such selection. First, snow MaxEntmodel; and(2)identifycriticalareasforsnowleopard
leopardoccurrenceispredictedbyseveralabioticfactorssuch conservationundertheexistingfunctionalzonesofQNNR.
asterrain(slopeandaspect)(Sharmaetal.,2015;Wolf&Ale,
MATERIALSANDMETHODS
2009), elevation(Alexanderetal.,2016a; Aryaletal.,2014c),
snowcover(Aryaletal.,2014c),distancefromrivers(Aryalet Studyarea
al.,2014c),andannualmeantemperature(Lietal.,2013).The Located in the southwest Xizang (Tibet) Autonomous Region,
secondcomponentisrepresentedbybioticfactorssuchasprey China, QNNR (N27◦48(cid:48)–N29◦19(cid:48), E84◦27(cid:48)–E88◦23(cid:48)) was
availability(Alexanderetal.,2016b; Aryaletal.,2014a; Xu& established in 1989 to protect wildlife and ecosystems along
Luo,2010)andhumanactivity(Wolf&Ale,2009). Duringthe the border of China and Nepal. The reserve covers an area
1990s,slightlymorethan100snowleopardswereestimatedto of 33 814 km2, centering on the world’s highest peak, Mt.
inhabitQomolangmaNationalNatureReserve(QNNR).Based Everest. Altituderangesfrom1440mto8844ma.s.l.. The
on three brief surveys, Jackson et al. (1994) estimated that averageannualtemperatureis2.1◦Candtotalannualrainfall
“good”habitattotaledapproximately8000km2,orabout25% is 270.5 mm. Some 81 species of mammals, 342 birds, 29
of QNNR’s area. Furthermore, in May–June 2014 and May amphibiansandreptiles,and8fishinhabitthereserve(QNNR
2015, a preliminary snow leopard presence survey (Chen et Administration, unpublished). Large mammals include the
al., 2017) and a human-carnivore conflict survey (Chen et snowleopard(Pantherauncia),wolf(Canislupus),lynx(Lynx
al., 2016) were conducted in QNNR, respectively. A more lynx), brown bear (Ursus arctos), leopard (Panthera pardus),
comprehensiveandinterdisciplinarystudywasalsoconducted blue sheep (Pseudois nayaur), wild ass (Equus kiang),
to provide an evidential basis for the formulation of effective Tibetan gazelle (Procapra piticaudata), and Himalayan tahr
conservationpoliciesandprograms(Chenetal.,2017).Based (Hemitragus jemlahicus) (Chen et al., 2017). QNNR consists
on the above studies, habitat suitability assessment was mainlyoftwobroadecosystems:thatis,semi-humidmountain
deemednecessaryforasystematicsurveyofthesnowleopard forest and semi-arid shrub along the southern and northern
population in QNNR. Several studies have been conducted parts, respectively. Within these, vegetation varies across
to estimate the extent of suitable habitat covering the whole different sub-ecosystems. As altitude increases, 10 distinct
snowleopardrange(McCarthyetal.,2016;Hunter&Jackson, sub-ecosystems can be observed progressively, including
1997; Fox, 1994). However, habitat suitability assessment at subtropical evergreen broad-leaved forest, subtropical
theregionallevelhasnotyetbeenreported,resultinginagap semi-green broad-leaved forest, subtropical evergreen
betweenresearchandlocalconservationactions. coniferous forest, warm temperature green coniferous forest,
Recently, with the development of 3S (GIS, RS, GPS) warm temperature sclerophyllous evergreen broad-leaved
techniques,multiplemodelshavebeenusedtoassesssuitable forest, subalpine temperature zone evergreen coniferous
habitat distribution, including mechanism models (Ouyang et forest, subalpine temperature zone broad-leaved deciduous
al., 2001; Xu et al., 2006), regression models (Schadt et forest, alpine sub-frigid zone shrubland and grassland, alpine
al., 2002), and ecological niche models (Su et al., 2015; sub-frigid zone periglacial and alpine snow zone (QNNR
Xu & Luo, 2010; Wang et al., 2008; Phillips et al., 2006). Administration,unpublished). Theperpendicularwidthofeach
Themaximumentropy(MaxEnt)modelisanecologicalniche verticalecologicalsystemrangesfromhundredstothousands
model, which has been increasingly used to assess wildlife of meters, and each is very sensitive to external environment
habitatdistribution(Clementsetal.,2012;Wiltingetal.,2010). change(QNNRAdministration,unpublished).Forthedifferent
374 www.zoores.ac.cn
functionalzonesinQNNR,anydisturbanceisrestrictedinthe and hair) along the most likely routes (ridgelines, hillsides,
core zone, scientific research can be conducted in the buffer preyresource-richplaces,andhigherruggedplaces).Camera
area, andsomehumanactivitycanoccurintheexperimental stations were placed with a minimum spacing of 1 km to
zone. maximize the number of individuals caught and adequately
recapture individuals at different camera traps (Alexander et
Modelselection
al.,2015).Weplacedtwocamerasineachsitetocaptureboth
InformationscienceisthebasisofMaxEntmodels,whichare
sidesoftheanimal(Jacksonetal.,2006).Withineachgrid,two
widely used in many academic disciplines, as proposed by
ormorecamerastationswereestablished. TheZhalongstudy
Jaynes (1957a, 1957b) and more recent studies (Jaynes & area (Jilong County) was expanded from 400 km2 in 2015 to
Bretthorst, 2003; Xing & Hao, 2011). The MaxEnt model is 790 km2 in 2016. The transects used in 2015 were fixed for
baseduponecologicalnichetheory,relyingoninformationfrom
latersurveys,butcameratrapsiteswerenot(Figure1).
species“presence”datatoexplorethepossibledistributionof
For non-invasive genetics, we collected samples that
atargetspecieswithinastudyarea.In2004,MaxEntsoftware
belonged to snow leopards along transects to set up camera
was developed to assess and evaluate suitable habitat for
traps. The selection of feces was based upon shape, hair
targetspecieswithgoodpredictivepower(Phillipsetal.,2006).
content, and terrain type or was found in association with
Occupancy models and generalized linear models (GLM) are
other marking signs particular to this species. We collected
also used to explore the relationship between species habitat
scat samples using silica gel drying agent (Boitani & Powell,
selection and environmental factors (Alexander et al., 2015;
2012; Wasser et al., 1997). In the field, tubes containing
MacKenzie et al., 2006; Wang et al., 2018). However, GLM
silicadesiccantwerestoredawayfromlightinadryandcool
and occupancy models exhibit similar limitations, such as the environment and were immediately stored at −20 ◦C after
need for presence/absence or detection/non-detection data,
arrivalatthelaboratory.DNAfromscatsampleswasextracted
whichcanintroducedifficultieswhenanexpansivestudyarea
using a QIAamp DNA Stool Mini Kit (Qiagen, USA) following
isneededtodetectrarespecieslikesnowleopards(Alexander
a modified extraction protocol under a dedicated UV-light
et al., 2016a; MacKenzie et al., 2006). The MaxEnt model
laminar flow cabinet, physically separated by the pre-PCR
providesaself-checkingfunctionwithautomaticallygenerated
area. Each batch of extracted samples was processed along
receiver operating characteristic (ROC) curves. Furthermore,
with a negative control to account for possible contamination,
to understand the regional level habitat distribution of snow
with results screened on 1% agarose gel. Species
leopards in QNNR, presence records are more likely to be
identification was initially conducted through amplification of
accessible, and thus the MaxEnt model has an advantage
a 110 base pairs-portion of the cyt b (cyt b) gene using
over other types of models (Su et al., 2015; Liu et al., 2013).
carnivore-specificprimers,asdescribedinFarrelletal. (2000)
Therefore, the MaxEnt model was selected in this study to (forward: 5(cid:48)-TATTCTTTATCTGCCTATACATACACG-3(cid:48);reverse:
explore the potential suitable habitat of snow leopards in 5(cid:48)-AAACTGCAGCCCCTCAGAATGATATTTGTCCTCA-3(cid:48)). PCR
QNNR.
analysis was conducted under a dedicated UV-sterilized
laminarflowhoodusingPremixTaq(ExTaqVersion2.0;Takara
Snowleopard“presence”data
Biotechnology)toafinalvolumeof20µLcontaining0.2mmol/L
In2014,apreliminarysurveywasconductedinfourareaswith
ofeachdNTP,2mmol/LofMg2+,0.4µmol/Lofeachprimer,7.4
33cameratrapsaroundthevillagesofZhalong(JilongCounty),
µLofDNAase/RNAase-freeultrapurewater(TiangenBiotech,
Dacang (Dingjie County), Riwu and Qudang (Dingri County)
Beijing),and1µLofextractedDNA.PCRconditionsconsisted
(Chen et al., 2017). Other study areas were later selected
ofaninitialdenaturationstepof10minat94◦C,35cyclesat94
to estimate snow leopard abundance and density with high
◦Cfor30s,57◦Cfor45s,and72◦Cfor45s,followedbyafinal
intensity. Ourstudyareawasdividedintoseveral4km×4km
elongationstepat72◦Cfor10min.Allamplificationsincluded
grids. We selected sampling transects and camera locations
one positive snow leopard reference sample and a blank
according to our knowledge of the ecological requirements
control to account for contamination. Results were screened
of snow leopards and the experience of our local guide to
by1%agarosegelelectrophoresis. SuccessfulPCRproducts
maximize the detectability of animals in each grid. In 2015,
were sequenced through an Applied Biosystems ABI 3730XL
following the preliminary survey, we selected Zhalong (Jilong
system by SinoGenoMax Limited Company (Beijing, China).
County)andQudang(DingriCounty)asstudyareas. Intotal,
ReadingswereassembledinGeneious10.1(BiomattersLtd.)
83 and 23 camera traps (Ltl Acorn) were systematically set
and matched against the NCBI database using BLAST to
up over 400 km2 in the Zhalong study area and 208 km2 in
obtainthepercentageofsimilarity.Onlycompletelyassembled
theQudangstudyareabetweenOctoberandNovember2015,
sequenceswereconsideredforthedataset. Samplesyielding
respectively. Insummer2016(ApriltoMay),112cameratraps
incompleteassemblies, alongwithunsuccessfulPCRsforcyt
(LtlAcorn)weresetupover480km2intheZhalongstudyarea.
b andnon-carnivorespeciesidentifications,werere-screened
Inwinter2016,anadditional68cameratrapsitesweresetup
targetinga126basepairs-portionoftheATP6 geneusingthe
over790km2inanotherpartoftheJilongstudyarea.Inwinter
carnivore-specific primers described in Chaves et al. (2012)
2017, asnowleopardsurveywasconductedover750km2 in
(DF3: 5(cid:48)-AACGAAAATCTATTCGCCTCT-3(cid:48); DR1: 5(cid:48)-CCA
theZhaxizongstudyarea(DingriCounty). Werecordedsnow
GTATTTGTTTTGATGTTAGTTG-3(cid:48)).Theprimersdescribed in
leopard signs (scats, tracks, scent marks, scrapes, bedding,
ZoologicalResearch39(6):373–386,2018 375
Farrell et al. (2000) were used to target prey species DNA protocol used for cyt b, as were the PCR conditions, except
(Chaves et al., 2012). PCR volumes were identical to the fortheannealingtemperature(51◦Cfor45s).
E87°18ʹ0″ E87°24ʹ0″ E85°6ʹ0″ E85°12ʹ0″ E85°18ʹ0″ E85°24ʹ0″
N28°48ʹ0″
Camera trap positions Camera trap positions
400 m Contour 400 m Contour
16 km2 grid cell 16 km2 grid cell
N28°12ʹ0″ E<VleAvaLtUioEn-13>80060 -- 12880000 mm N28°12ʹ0″ N28°42ʹ0″ E<VleAvaLtUioEn-13>80060 -- 12880000 mm N28°42ʹ0″
2800 - 4200 m 2800 - 4200 m
4200 - 5000 m 4200 - 5000 m
5000 - 8648 m 5000 - 8648 m
N28°36ʹ0″ N28°36ʹ0″
N28°6ʹ0″ E87°18ʹ0″ E87°24ʹ0″ E87°30ʹ0″ N28°6ʹ0″ E85°6ʹ0″ E85°12ʹ0″ E85°18ʹ0″ E85°24ʹ0″ E85°30ʹ0″
E85°0ʹ0″ E85°6ʹ0″ E85°12ʹ0″ E85°18ʹ0″ E85°24ʹ0″
E85°6ʹ0″ E85°12ʹ0″ E85°18ʹ0″ E85°24ʹ0″
N28°54ʹ0″
N28°48ʹ0″ 4C0a0m mer aC tornapto puorsitions N28°48ʹ0″ C40a0m mer aC toranpto puorsitions N28°48ʹ0″
16 km2 grid cell 16 km2 grid cell
Elevation Elevation
N28°42ʹ0″ <VALUE-13>80060 -- 12880000 mm N28°42ʹ0″ N28°42ʹ0″ <VALUE>-1380060 -- 12880000 mm N28°42ʹ0″
2800 - 4200 m 2800 - 4200 m
4200 - 5000 m
4200 - 5000 m
N28°36ʹ0″ 5000 - 8613 m N28°36ʹ0″ N28°36ʹ0″ 5000 - 8613 m N28°36ʹ0″
E85°6ʹ0″ E85°12ʹ0″ E85°18ʹ0″ E85°24ʹ0″ E85°30ʹ0″ E85°0ʹ0″ E85°6ʹ0″ E85°12ʹ0″ E85°18ʹ0″ E85°24ʹ0″ E85°30ʹ0″
Figure1LocationofcameratrapsforsnowleopardsurveysinQomolangmaNationalNatureReserve,Xizang(Tibet)
A:QudangStudyArea,DingriCounty,winter2015.B:ZhalongStudyArea,JilongCounty,winter2015.C:ZhalongStudyArea,JilongCounty,summer2016.D:
ZhalongStudyArea,JilongCounty,winter2016.
Feces, scent marks, and killings could be attributed to thesamesurveyeffortasoutsidetheJilongarea. Asonly59
carnivoresotherthansnowleopards;however,pugmarksand recordswereincludedforanalysisoutsidetheJilongarea(east
scrapesareeasilyrecognized. Therefore,pugmarks,scrapes, QNNR),weonlyincluded59recordsfromtheJilongarea(west
and feces identified as belonging to snow leopards were QNNR).
selected to create a suitable habitat distribution simulation in
Environmentvariables
QNNR (Jackson & Hunter, 1996; Janecˇka et al., 2008). We
We tested the relationship of snow leopard presence data
used 1-km2 grid cells for the MaxEnt niche-based modeling
with possible influencing factors. We identified two main
(Phillips et al., 2006). One snow leopard presence per cell
categories related to suitable habitat distribution, namely
was used and repeated location data were removed. A total
abiotic factors (elevation, slope, aspect, ruggedness, land
of 222 GPS snow leopard presence points were included for
use type, and distance factors) and bioclimatic and biotic
further analysis. However, due to the higher number of field
factors(preyspecies,humanrelatedfactors)(Alexanderetal.,
studiesintheJilongarea,recordswereheavilygeographically
2016a, 2016b; Sharma et al., 2015). We selected elevation,
unbalanced,withtheJilongareacontainingmorethan50%of
slope, aspect, ruggedness, land use type, and bioclimatic
all records. We thus further reduced the number of records
factors for landscape level analysis. We used snow leopard
in the Jilong area by randomly selecting records to produce
presence points and extracted 19 bioclimatic variables of our
376 www.zoores.ac.cn
studyareafromWorldClim(www.worldclim.org)data(Table1). obtained from the University of Louvain (UCLouvain) and
In ArcGIS, elevation, aspect, slope, and ruggedness were European Space Agency (ESA). Distance factors (distances
determined using a digital elevation model (DEM) layer, with to roads, settlements, and rivers) and biotic factors were
all variables clipped to our study areas. Ruggedness was excludedduetothesmall-scalesnowleopardandpreyspecies
extracted based on the method of Riley et al. (1999). The presencedata. Wethenextractedthevaluesofeachvariable
TopographicRuggednessIndex(TRI)wasusedtoexpressthe corresponding to the presence locations of snow leopards to
elevationdifferencebetweenadjacentcells(alleightfirst-order performcorrelationanalyses(SupplementaryTableS1). After
neighbors within a quadratic grid) of a 90-m-resolution DEM removing 10 highly correlated variables (>0.85), we used the
on a scale from 1 (level) to 7 (extremely rugged). Land remaining14variablesforfinalanalysis.
useandlandcoverwereextractedfrom“GLOBCOVER2009”
Table1Variablesusedformodeling
SN Variable Description Source
1 BIO1(Annualmeantemperature) Continuous
2 BIO2(Meandiurnalrange) Continuous
3 BIO3(Isothermality) Continuous
4 BIO4(Temperatureseasonality) Continuous
5 BIO5(Maxtemperatureofwarmestmonth) Continuous
6 BIO6(Mintemperatureofcoldestmonth) Continuous
7 BIO7(Temperatureannualrange) Continuous
8 BIO8(Meantemperatureofwettestquarter Continuous
9 BIO9(Meantemperatureofdriestquarter) Continuous
http://www.worldclim.org/bioclim
10 BIO10(Meantemperatureofwarmestquarter) Continuous
(Hijmansetal.,2005)(1–19)
11 BIO11(Meantemperatureofcoldestquarter) Continuous
12 BIO12(Annualprecipitation) Continuous
13 BIO13(Precipitationofwettestmonth) Continuous
14 BIO14(Precipitationofdriestmonth) Continuous
BIO15(Precipitationseasonality)
15 Continuous
(Coefficientofvariation)
16 BIO16(Precipitationofwettestquarter) Continuous
17 BIO17(Precipitationofdriestquarter) Continuous
18 BIO18(Precipitationofwarmestquarter) Continuous
19 BIO19(Precipitationofcoldestquarter) Continuous
Categorical(Irrigatedcroplands,Rainfed
croplands,MosaicCroplands/Vegetation,Mosaic
Vegetation/Croplands,Closedtoopenbroadleaved
evergreenorsemi-deciduousforest,Closed
broadleaveddeciduousforest,Closedneedle-leaved GLOBCOVER2009
evergreenforest,Closedtoopenmixed formUniversityof
20 Landcover broadleavedandneedle-leavedforest,Mosaic Louvain(UCLouvain)
Forest-Shrubland/Grassland,MosaicGrassland, andEuropeanSpace
Forest-Shrubland,Closedtoopenshrubland, Agency(ESA)(20)
Closedtoopengrassland,Sparsevegetation,
Closetoopenvegetationregularly
flooded,Bareareas,Water
bodies,Permanentsnowandice)
Categorical(Flat;North:0◦–22.5◦;
Northeast:22.5◦–67.5◦;East:67.5◦–112.5◦;Southeast:
21 Aspect 112.5◦–157.5◦;South:157.5◦–202.5◦;Southwest:
202.5◦–247.5◦;West:247.5◦–292.5◦;Northwest:
GMTED2010(21–24)
292.5◦–337.5◦;North:337.5◦–360◦)
Categorical(0–4.375,4.375–9.310,9.310–14.216,
22 Slope 14.216–19.223,19.223–24.395,24.395–29.823,
29.823–35.773,35.773–43.812,43.812–72.092)
23 Elevation Continuous(m)
24 Ruggedness Continuous
SN:Serialnumber.
ZoologicalResearch39(6):373–386,2018 377
Simulationprocedure a carnivore species (either forward or reverse strand). We
Snow leopard presence data and selected variables were then amplified 64 samples for the ATP6 gene, yielding 31
adaptedtotheformatrequiredforMaxEntsoftware(v3.3.3k) positive carnivore identifications, with 26 belonging to snow
(Phillips et al., 2006). We selected 75% of snow leopard leopards.Insummary,102samples(75.5%)weresuccessfully
presencedatatobuildthemodel,withtheremaining25%used identified, with snow leopards representing 36.2% of the total
formodelverification.Weincluded10replicatesinouranalysis. (49identifications).Forwinter2016,weidentified24carnivores
Weusedajackknifeestimatortodetecttheimportanceofeach outof72samplesthroughcytb,11ofwhichwereidentifiedas
variable. Sensitivity analysis was done for each variable with snowleopard. Theremaining48, amplifiedthroughtheATP6
logistic output format. Results of the MaxEnt model were gene, yielded 19 carnivores, with snow leopards identified 16
verified by ROC values: that is, rejected with a ROC value times. In total, we identified 43 carnivores in 43 occasions
0.5–0.6; poor with 0.6–0.7; normal with 0.7–0.8; good with (60% of total), and snow leopards represented 37.5% of total
0.8–0.9; and excellent with 0.9–1.0. According to the expert fecalsamples. Poolingallsamplestogetheroverthefourfield
experiencemethod(Swets,1988),theoutputresultswereused seasons, we collected 345 fecal samples and identified 233
inArcGIS10.2toreclassifythesuitablesnowleopardhabitat (66%)asbelongingtoapredatorspecies.Snowleopardswere
distributionmap. UsingtheMaxEntmodeltoevaluatehabitat identified127times, representing36.8%ofthetotalcollected
suitability in QNNR, the ASCII format file was imported into scatsand54.5%ofthesuccessfulidentifications. Resultsare
ArcGIS 10.2 for transformation into floating raster data. The summarizedinTable2.
floatingrasterwasreclassifiedintolowgradehabitat(0–0.14),
Table2Identificationresultsfrommolecularscatology
moderately suitable habitat (0.14–0.42), and highly suitable
habitat (0.42–1). Focal statistics in ArcGIS were used to FieldSeason Scats Cytb+ ATP6+ SLs Failed
smooth the raster map to obtain the suitable snow leopard
Summer2014 52 29 15 36 8
habitatdistributionmapinQNNR.
Winter2015 84 43 1 15 42
RESULTS Summer2016 135 71 31 49 33
Winter2016 72 24 19 27 29
Snowleopardpresencedatarecords
Total 343 167 66 127 112
In2014,werecordedfivesnowleopardpugmarksitesand65
scrapesites,and17cameratrapsitescapturedsnowleopard Cytb+andATP6+indicatesuccessofidentificationusingthetwo
photographs. In winter 2015, we recorded 38 pugmark sites,
markers.SLs:Snowleopards.
111scrapesites,and27cameratrapsiteswithphotographs.In
summer2016,werecorded35pugmarksites,131scrapesites, Intotal,127pugmarksites,415scrapesites,and127molecular
and44cameratrapsiteswithphotographs.Inwinter2016,we identifications of snow leopard were recorded. In addition, 120
recorded 39 pugmark sites, 73 scrape sites, and 32 camera cameratrapsitescapturedsnowleopardimages(Figure2).
trap sites with photographs. In winter 2017, we recorded 10
MaxEntpredictionevaluation
pugmarksitesand35scrapesites.
The ROC results (Figure 3) showed a mean AUC value of
Wecollectedatotalof52,84,135,and72fecalsamplesin
0.921,indicatingthatthepredictionsobtainedfromtheMaxEnt
summer 2014, winter 2015, summer 2016, and winter 2016,
modelwereexcellent.
respectively. In 2014, we positively identified 29 samples
throughcytb,21ofwhichbelongedtosnowleopards. The23 Suitable snow leopard habitat distribution with
unsuccessful samples, including two incomplete assemblies, environmentfactors
produced 15 snow leopard identifications through the ATP6 According to the suitable habitat distribution map, we
gene. In total, we identified 84% of samples (44), with determined that suitable habitat was located mainly along
snow leopard identifications accounting for 69% of the total mountainareasneartheNepaleseborder. Weidentifiedthree
(36). Of the 86 samples collected in winter 2015, 43 were distinct separated patches: (1) Zhalong and Gongdan areas
positively identified as a carnivore species using cyt b, of in Jilong county; (2) Chabuling area in Nielamu county and
which 14 samples belonged to snow leopard. The remaining Rongxia area in Dingri county; and, (3) Qudang area and
43 were screened using the ATP6 gene; however, out of perimeterzoneinDingricounty.
the positive PCRs, identification was only possible for one The jackknife estimator results (Figure 4; Table 3) showed
sample, which belonged to a snow leopard. In total, for that precipitation in the driest quarter (BIO17; 20.0%),
samples collected in 2015, we identified 44 samples (51%), ruggedness(14.4%),elevation(13.3%),maximumtemperature
15 of which were snow leopards (17% of total). Species of the warmest month (BIO5; 8.7%), and annual mean
identification of the 135 samples collected in spring 2016 temperature (BIO1; 8.2%) were the main factors contributing
yielded 71 assembled predator species identifications, 23 tosnowleopardhabitatselection. Theimportanceratesinthe
of which were snow leopards. Out of the 64 samples MaxEnt model prediction indicated that ruggedness (34.4%),
selected for a second amplification, 55 failed to amplify, five mean diurnal range (BIO2; 16.9%), maximum temperature
belonged to blue sheep (Pseudois nayaur), one to yak (Bos of the warmest month (BIO5; 11.3%), annual precipitation
grunniens), and three were partial assemblies belonging to (BIO12; 8.8%), and precipitation in the driest quarter (BIO17;
378 www.zoores.ac.cn
7.9%)werethefivemainfactorsaffectingsnowleopardhabitat and 5 000 m a.s.l.. The ruggedness range most suitable
preferences(Table3;Table4). for snow leopards was between 1 000 m and 1 300 m a.s.l.
Sensitivityanalysisdeterminedtheinfluenceofeachfactor wheretheterrainwasextremelyrugged. Consequently, snow
onsnowleopardhabitatdistribution(Figure5). Thealtitudein leopardspreferredtousemoreruggedterrainatelevationsof
areassuitableforsnowleopardhabitatwasbetween3900m around4000ma.s.l..
Figure2SnowleopardpresencedatacollectedinQNNR
Average sensitivity vs.1 - specificity for snow leopard
1.0 Mean (AUC = 0.921)
Mean ± SD
0.9 Random prediction
0.8
e)
at0.7
n r
o
missi0.6
o
1 - 0.5
nsitivity (0.4
e
S0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1 - Specificity (fractionsl predicated area)
Figure3ROCverificationofdistributionofsuitablesnowleopardhabitatinQNNR
ZoologicalResearch39(6):373–386,2018 379
Jackknife of regularized training gain for snow leopard
Without variable
Aspect
With only variable
Bio_1 With all variables
Bio_12
Bio_13
e Bio_14
bl
a
ari Bio_17
v
al Bio_2
nt
me Bio_3
n
viro Bio_5
n
E Bio_6
Elevation
Landcover
Ruggedness
Slope
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Regularized training gain
Figure4JackknifetestofenvironmentalvariablesintrainingdatabyMaxEnt
DistributionofsuitablesnowleopardhabitatinQNNR DISCUSSION
From the reclassified map, the area of each habitat class Environmental factors influencing snow leopard habitat
(low grade habitat, moderately suitable habitat, and highly selection
suitable habitat) was calculated. The area of highly suitable Ourresultsshowedthatprecipitationinthedriestquarter(BIO17),
habitatinQNNRwas1756.55km2,moderatelysuitablehabitat ruggedness, elevation, maximum temperature of the warmest
was 5245.38 km2, and low-grade habitat was 23 814.05 month(BIO5),andannualmeantemperature(BIO1)werethefive
km2. Thus, the area of relatively good snow leopard habitat mainfactorsinfluencingsnowleopardhabitatsuitabilityinQNNR.
totaled 7001.93 km2 in QNNR, accounting for 22.72% of the Wolf&Ale(2009)conductedresearchintheSagarmathaNational
overall area of QNNR (Figure 6). According to the existing Park (area of 1 148 km2) of Nepal and reported that terrain and
functional zones (core, experimental, and buffer zones) of human activity were the main factors determining snow leopard
QNNR (Figure 7), only 23.24% (1 627.52 km2/7001.93 km2) spatial distribution, whilst prey species had a moderate effect.
of good habitat lies in the core zone, 36.06% (2 525.22 In Nepal’s Annapurna Conservation Area, Aryal et al. (2014c)
km2/7001.93 km2) lies in the buffer zone, and 40.52% indicatedthatcliffs,grassland,andshrublandathighelevations(3
(2837.40km2/7001.93km2)liesintheexperimentalzone. 000–5000ma.s.l.) werepreferredhabitatsofsnowleopardsin
thestudyarea(about2025km2). AstudyconductedinMongolia
Table3Contributionandpermutationimportancevaluesof reported that distribution of prey resources and rugged terrain
environmentalvariables largelyexplainedchangesinsnowleopardhabitatuse(McCarthyet
al.,2005).InIndia,investigationinintensivelygrazedareasshowed
Environmentalvariables Contribution(%) Permutationimportance(%)
that snow leopard habitat-use mainly relied on wild prey species
Aspect 1.3141 1.38 density(Sharmaetal.,2015).InChina,verylittleresearchhasbeen
BIO1 8.1786 0.4244 conductedonsnowleopardhabitatuse.Awinterhabitat-usesurvey
BIO12 6.8994 8.8319 of snow leopards in Tomur National Nature Reserve highlighted
BIO13 4.214 2.6905
thatpreyresourcesandprincipalgeographicfeatures(ruggedness,
BIO14 0.7669 2.2132
basesofcliffs,andstreambeds)werethemainfactorsinfluencing
BIO17 19.9978 7.8506
snowleopardhabitatusewithinthe3000km2studyarea(Xuetal.,
BIO2 6.3171 16.8578
2012).InSanjiangyuanNationalNatureReserve,alandscapelevel
BIO3 4.4267 3.8942
analysisofsnowleopardhabitatusingtheMaxEntmodelindicted
BIO5 8.7044 11.2613
that annual average temperature and ruggedness were the two
BIO6 4.1076 0.508
main factors influencing habitat selection (Li et al., 2013). From
Elevation 13.3151 1.6187
theabovestudies,wecanconcludethatthedeterminingfactorsof
Landcover 2.3265 3.9645
snowleopardhabitatselectionmaydifferindifferentareas. Thus,
Ruggedness 14.3777 34.3892
localsurveysonsnowleopardhabitatselectionarecriticaltoadapt
Slope 5.154 4.1157
conservationneedsaccordingtothelocalcontext.
380 www.zoores.ac.cn
Table4Cumulativeandlogisticthresholdsandcorrespondingomissionratesusedformodeling
Fractional Training
Cumulativethreshold Logisticthreshold Description
predictedarea omissionrate
1.000 0.015 Fixedcumulativevalue1 0.641 0.000
5.000 0.056 Fixedcumulativevalue5 0.382 0.014
10.000 0.113 Fixedcumulativevalue10 0.262 0.047
3.590 0.042 Minimumtrainingpresence 0.464 0.000
18.200 0.203 10percentiletrainingpresence 0.173 0.099
24.048 0.261 Equaltrainingsensitivityandspecificity 0.128 0.128
30.946 0.330 Maximumtrainingsensitivityplusspecificity 0.096 0.133
Balancetrainingomissionpredictedarea
5.371 0.060 0.376 0.006
andthresholdvalue
Equateentropyofthresholdedand
13.893 0.159 0.210 0.086
originaldistributions
Figure5ResponsecurveofselectedvariablesforsnowleopardhabitatsuitabilityinQNNR
ZoologicalResearch39(6):373–386,2018 381
Figure6DistributionofsuitablesnowleopardhabitatinQNNR
Figure7DistributionofsuitablesnowleopardhabitatindifferentfunctionalzonesinQNNR
382 www.zoores.ac.cn