Small mammal assemblages and habitat distribution in the northern Junggar Basin, Xinjiang, China: a pilot survey

Mammalia 72 (2008): 309–319 � 2008 by Walter de Gruyter • Berlin • New York. DOI 10.1515/MAMM.2008.048

Article in press - uncorrected proof

2008/48

Small mammal assemblages and habitat distribution in thenorthern Junggar Basin, Xinjiang, China: a pilot survey

Patrick Giraudoux1,*, Hongxia Zhou2,*, Jean-Pierre Quere3, Francis Raoul1, Pierre Delattre3,Vitaly Volobouev4, Thomas Deforet1, Akira Ito5,Wulamu Mamuti6, Renaud Scheifler1 and PhilipSimon Craig7

1 Department of Chrono-environment UMR CNRS6249 usc INRA, University of Franche-Comte, 1 PlaceLeclerc, 25030 Besancon cedex, France,e-mail: [email protected] Department of Epidemiology, Public Health School,Guangxi Medical University, Nanning, Guanxi, China,e-mail: [email protected] CBGP-UMR 1062 INRA, Campus International duBaillarguet, CS 30016 34988 Montferrier sur Lez,France4 National Museum of Natural History, 16 rue Buffon,75005, Paris, France5 Department of Parasitology, Asahikawa MedicalCollege, Asahikawa, Japan6 Department of Parasitology, Xinjiang MedicalUniversity, 83000 Urumqi, Xinjiang, China7 Biomedical Sciences Research Institute, School ofEnvironment and Life Sciences, University of Salford,Manchester M5 4WT, UK

*Corresponding authors

Abstract

Small mammal assemblages were surveyed in five areasof the northern Junggar basin, Xinjiang, China, usingstandard trapping methods and index transects. In total,23 species were recorded. The relationships betweenhabitats and the distribution of the main species weredescribed at local scale in Baihaba, Altai mountains.Three types of assemblages linked to (i) forest (Myodesrufocanus, Microtus agrestis, Myodes rutilus, Apodemuspeninsulae, Sorex isodon), (ii) transitional areas and farm-land (mixed pool of species including Apodemus uralen-sis), and (iii) grassland (Ellobius tancrei, Microtusobscurus, Cricetulus migratorius) were identified. Addi-tional species, such as Microtus oeconomus, could befound along small streams. At a broader scale, speciescomposition was estimated in Kokehada, Narenhebukeand Fuhai areas and compared. Although ecological gra-dients from mountain grasslands to cold semi-desertpartly explain differences in assemblage composition ona regional scale, similar habitats at the same altitude may,however, harbour different assemblages. For example,Spermophilus erythrogenys was the dominant species ofthe Kokehada grassland and Microtus obscurus and Ello-bius tancrei were dominant in Baihaba grassland. In con-clusion, the need for multi-scale standard descriptions of

small mammal assemblage distribution in the northernJunggar basin of China is indicated.

Keywords: community; grassland; mountain; scale;semi-desert.

Introduction

Small mammals exist in virtually every terrestrial habitat.They are generally the main food resource for predatorcommunities (carnivores, birds of prey, etc.), and somespecies can be an agricultural pest. They can also bereservoir hosts for pathogen agents impacting publichealth and thus may play a critical role in disease trans-mission (Gratz 1994, Giraudoux et al. 2006). If the pres-ence of species over large continental areas may beconsidered as the result of biogeography and evolution,population dynamics and distribution patterns at finertime-space resolution are generally considered as aresponse to ecological factors, including the spatialarrangement of optimal habitats in a landscape and theresult of individual dispersion and community processes,such as predation (Hansson and Henttonen 1988,Bjornstad et al. 1999, Lidicker 2000, Huitu et al. 2004).

A minimum of 170 species of small mammals has beenrecorded in continental China. Atlases of distribution,species lists and identification keys generally documentthe distribution of small mammals at a continental orregional scale (Wang and Yang 1983, Ma et al. 1987, Luand Yan 1989, Zhu and Chen 1993, Zhang 1997, Shen-brot and Krasnov 2005), but species distribution at amore local scale has been sparsely documented (e.g.,sub-regional and habitat range). Serious economiclosses and public health problems have been reported tobe associated with small mammals in China since the1950s (Zhao 1996). Rodent outbreaks were frequentlyrecorded in deforestation areas of north-eastern China atthe end of the 1950s (Xia 1996) and involved the genusMyodes (formerly Clethrionomys). Indication of popula-tion outbreaks as a consequence of anthropogenic mod-ification of habitats was also found for Microtuslimnophilus Buchner and Cricetulus longicaudatus Milne-Edwards in deforestation areas of south Gansu, forLasiopodomys brandtii Radde in the Inner MongoliaAutonomous Region, for Mus musculus and Microtusobscurus in the extensive farmland areas of XinjiangUygur Autonomous Region and for Microtus obscurusand Spermophilus erythrogenys in the pastures of north-ern Xinjiang (Wang and Yang 1983, Zhu and Chen 1993,Giraudoux et al. 1998).

More than 85 species of small mammals have beenreported in Xinjiang Uygur Autonomous Region, the larg-est (1.7 million km2) administrative region of China (Wangand Yang 1983), with contrasted natural conditions rang-

Brought to you by | Ryerson Polytechnic University LibraryAuthenticated

Download Date | 12/2/14 10:13 PM

310 P. Giraudoux et al.: Small mammal assemblages in Northern Xinjiang


Figure 1 Northeast Xinjiang (China): location of the study sites (digital elevation model 1 km from the US National Oceanic andAtmospheric Administration, altitude classes computed by the Jenks’ method, water bodies from ESRI World Basemap Data).

ing from deserts to alpine ecosystems. Investigations offauna and flora have been conducted for years andmostly aimed to establish species lists and identificationcriteria (Ma 1981, Wang and Yang 1983, Ma et al. 1987,Halik et al. 1999). However, quantification of small mam-mal distribution amongst habitat patches on a local range(some square kilometres), and turnover of assemblagesbetween ecosystems or regions over larger range (someten thousands of square kilometres) are poorly docu-mented for China.

The northern Junggar basin lies between the Tarbata-gai mountains to the northwest and the Altai mountainsto the northeast. It is a remote, poorly accessible areawith low human population density, and the paucity ofsmall mammal studies reported from this region reflectsdifficulties of working in this isolated region. In the currentstudy, we hypothesise that standard sampling methodsin relation to land cover and habitat patches can providea rapid assessment of small mammal communities insuch a remote area. The objective was to provide a firstassessment of small mammal communities and to deter-mine at which scales further studies of small mammalcommunities should be undertaken in northern Xinjiang.

Study sites and methods

The results presented here have been obtained duringthe course of a research programme on the epidemiology

and transmission ecology of the zoonotic cestode Echi-nococcus multilocularis Leuckart. Thus, field selectionand the duration of each stay were to a degree pre-selected, including field accessibility, time available andlogistics for ecological and medical teams (Wang et al.2001). Three main geographical areas were monitored: (1)Baihaba area in northern Altai; (2) Kokehada and Naren-hebuke area in the Tarbatagai mountains; and (3) Fuhaiarea in the northern Junggar basin (Figure 1).

Baihaba area

Baihaba village (86.78 E, 48.69 N) is located 30 km fromthe Kanas National Nature Reserve (87.07 E, 49.26 N)close to the Kazakstan border. In the Kanas NationalNature Reserve meteorological station, the mean annualprecipitation is 1000 mm, mean temperature is -168C inJanuary and 15.98C in July (Halik et al. 1999). The altitudeof the study area around the village ranges between 1210and 1780 m above sea level and the area is typical of themiddle alpine belt of the Xinjiang Altai mountains. Sixtypes of habitats were identified: (1) dense forest domi-nated by Larix sibirica, Picea obovata, Pinus sibirica,Betula pendula and Populus tremula; (2) open forestwhere the dominant species were the same but withgrass cover on the ground and undergrowth of Spiraeachamaedryfolia, Lonicera altaica, L. hispida, Rosa alberti,Ribes sp., Cotoneaster sp., Spiraea hypericifolia, Alche-milla vulgaris, Salix sp., Caragana jubata; (3) shrubland



P. Giraudoux et al.: Small mammal assemblages in Northern Xinjiang 311


with dominant species Rosa alberti, Spiraea hypericifolia,Hamamelis mollis oliver and large grass cover, includingAlchemilla vulgaris; (4) grassland, mainly extensive her-baceous pastures with Deschampsia caespitosa, Carexpediformis, Dactylis glomerata; (5) village garden, mainlypotatoes and green Chinese onion bordered with grassstrips and woodlots; and (6) farmland, mosaic of fieldsand Carex grass cropped for hay.

Small mammals were sampled in September 1998 withINRA (Institut National de Recherche Agronomique) livetraps (5=5=15 cm) (Aubry 1950), small break-back traps(SBBT, snapping bar 4.5=4.5 cm) and big break-backtraps (BBBT, snapping bar 9=12 cm). Traps were baitedwith dough consisting of a mixture of flour, peanut butterand water. Additionally, a small number of tongue-trapsand Sherman traps (10=10=25 cm) were used for trap-ping strictly subterranean species and to link species tosurface indices (e.g., runways, earth hills, faeces, etc.).

A total of 57 trap lines were set up. Each line consistedof 25 traps of the same type, 3 m distant from each other(Giraudoux et al. 1998, Raoul et al. 2006). Traps werechecked every morning, re-baited and reset, for 3 con-secutive nights. Such lines were considered ‘‘standard’’lines. Standard lines were paired, i.e., each INRA line cor-responded to a SBBT line in the same habitat. Traps wereoccasionally stolen during the field survey; others wereset for only 2 nights for logistical reasons. Those lineswere not used for statistical analysis.

Small mammals were humanely killed, weighed anddissected for determination of sex and reproductive stat-us, and parasitological examination. Heads (or the wholebody for some specimens of every species) were pre-served in 5% formalin solution. Skull and skin were pre-pared and identified at the Centre de Biologie et Gestiondes populations, Montferrier (France), using the followingreferences: Allen (1940), Corbet (1978), Hoffmann (1987),Gromov and Erbajeva (1995) and Musser et al. (1996).Tissue samples of the genus Microtus and Apodemuswere preserved for later karyotyping and DNA cyto-chrome b sequencing at the Museum National d’HistoireNaturelle of Paris and at the Centre de Biologie et Ges-tion des populations, Montferrier. Nomenclature followedWilson and Reeder (2005), except for the genus Arvicola(Wust-Saucy 1998) and the complex Microtus arvalis/obscurus where preliminary results of karyotype and DNAanalysis indicate that M. obscurus can be distinguished(Tougard et al. in preparation).

The relative density of small mammals is expressed asthe number of animals caught per trap=night number(further termed trap nights). Landscape transects wereperformed to establish a general linkage between openhabitats and small mammal indices over larger areas.Along each transect, the presence-absence of rodentindices (burrow, corridor, faeces, earth tumulus, etc.) andthe habitat types were noted within every 10 paces(approximately 8–9 m) intervals. Indices have all beenconfirmed by non-standard trapping. Ellobius tancreileaves a conspicuous fine grain earth tumulus (10–30 cmdiameter, 5–10 cm high), sometimes with a middle orside hole from which a breach is maintained on one sideof the hill. Those tumuli are quite different from those ofArvicola terrestris and Talpa sp. (see Giraudoux et al.

1995 for full description). Microtus obscurus generallydoes not build earth tumuli and leaves long ramified run-ways in grassland, on which characteristic dark oliveshape faeces of 3 mm length can be found. However,these indices are not specific (e.g., other Microtus andthe genus Cricetulus have quite similar droppings), thuswe grouped them under the term ‘‘Microtus size spe-cies’’. A small mammal density index was defined as theratio of intervals where one or more indices were record-ed to the total number of intervals sampled (Giraudouxet al. 1995, Quere et al. 2000).

Kokehada and Narenhebuke area

Kokehada (85.5 E, 46.95 N) is a ‘summer pasture’ usedby Mongol herdsmen for sheep and cattle grazing (Wanget al. 2001). The altitude ranges from 1700 to 2500 mabove sea level. The area consists of Alpine overgrazedgrassland and grassy steppe progressively blending intoa semi-desert at lower altitude. The closest forest is morethan 200 km distant. Vegetation consists of Carex sp.,Cobresia sp., with crawling scrub bushes of Juniperussp. and Sabina sp. Large grassy banks of Alpine cush-ions were present in the valley bottoms along streams.Small mammals were sampled in July 1996 in alpinegrassland (trap line types: 2 INRA, 2 SBBT, 1 BBBT for3 nights), bottom valley stream banks (2 SBBT, 1 INRAfor 1 night) and in a steppe area (1 BBBT for 1 night).

The surroundings of Narenhebuke village (85.67 E,46.83 N; altitude 1290 m) are used as ‘winter pastures’by Mongol herdsmen and consist of a typical semi-desertsteppe. Trap lines were set up in May 1995: (1) in asteppe dominated by Caragana soongorica, Artemisiasp., Stipa gobica and Festuca sp. (2 BBBT, 1 night), and(2) in the alpine cushions of grassy river banks surround-ing the village (1 SBBT, 2 nights).

Additionally, Spermophilus erythrogenys were caughtwith cage traps (CT, grid cages of 25=10=10 cm) inKokehada pasture.

Landscape transects were performed to establish ageneral linkage between habitats and small mammal indi-ces over large areas (see description above).

Fuhai area

Fuhai County is located near Fuhai Lake in the westernJunggar Basin. Annual rainfall ranges between 100 and150 mm (Lu and Yan 1989). Two sampling sites wereselected: (1) site 1 (87.34 E, 46.92 N; altitude 470 m) wassituated in a semi-desert at the southern edge of FuhaiLake. Sparse bushes of Ammoniptanthus nanus, Anab-asis salsa and Caragana sp. were dominant in this area.(2) Site 2 (87.55 E, 47.14 N; altitude 480 m) was situatedat 30 km to the north of Fuhai town, in a cultivated oasis.The dominant plant species there included Achnatherrumsplendens, Artemisia sp. and Calligonum sp. The speciesBlaeagnus oxycarpa was also present at lower density.

Three lines of BBBT and three lines of SBBT were setin both sites for 1 night in September 1998.

Statistics

Considering sample sizes, robust non-parametric statis-tics were used: the Wilcoxon and Mann-Whitney U-test





Figure 2 Evaluation of the trapping effort on trapping successin the six habitats of the Baihaba area.

Table 1 Species and number of animals (n) trapped in Baihaba.

Code Name n

Appe Apodemus peninsulae Thomas 16Apsp Apodemus species 1Apur Apodemus uralensis Thomas 48Myrf Myodes rufocanus Sundevall 12Myru Myodes rutilus Pallas 67Crmi Cricetulus migratorius Pallas 16Elta Ellobius tancrei Blasius 9Miag Microtus agrestis Linnaeus 7Miob Microtus obscurus Eversmann 82Mioe Microtus oeconomus Pallas 8Sina Sicista napea Hollister 3Sois Sorex isodon Turov 3Sosp Sorex species 5Sotu Sorex tundrensis Merriam 2

Figure 3 Evaluation of the trapping effort on species richnessestimate in the six habitats of the Baihaba area.

or Kruskal-Wallis test for k)2 comparisons followed bya multiple comparison according to the Siegel and Cas-tellan (1988) method when indicated. The Friedman testwas used for longitudinal data (e.g., repeated measure-ments over time), and the x2-test was used for contin-gency tables. To represent the main relationshipsbetween species and habitats, a factorial correspon-dence analysis was computed on the trap lines=speciestable where indicated. Computations were carried outusing R 2.5.1 (R Development Core Team 2007), thepackages ade4 1.4-3 (Chessel et al. 2004) and pgirmess1.3.3 (Giraudoux 2007).

Results

Baihaba

A total of 282 small mammals representing 14 taxa weretrapped (3735 trap nights) (Table 1). Microtus oeconomusidentifications were confirmed by karyotyping, Microtusobscurus by karyotyping and DNA sequencing, and Apo-demus peninsulae by DNA sequencing.

Trapping efficiency

Comparison of trapping efficiency was achieved for INRAand SBBT traps only, due to small sample sizes for othertypes of trap lines and because no animals were caughtusing the BBBT traps. The average number of animalscaptured was 5.8 animals per 100 trap nights for INRAtraps, and 6.8 for SBBT traps. Differences were not sig-nificant (Mann-Whitney U-test, ps0.93). Moreover, smallmammal species did not show differential response totrap types. In all habitats except farmland, there was asignificant decrease in the number of captures per 100trap nights according to the number of trapping nights(Friedman test, ps0.04) (Figure 2). This means that thesampling effort had a significant effect on local popula-tion density of small mammals. Figure 3 shows that thecumulative number of species trapped reached anasymptote after the second night in all habitats. Thus,the trapping effort was probably enough to sample mostof the trappable species in each habitat. These resultssuggest that 3 nights trapping was sufficient for relevantcomparisons amongst habitats.

Standard sampling

Population relative densities of Apodemus peninsulae,Apodemus uralensis and Myodes rutilus were higher inforests, and shrubland for the two species of Apodemus,with intrusions in village gardens, and no capturesoccurred in grassland (Table 2). Microtus agrestis andSorex isodon were found in forest exclusively. By con-trast, Cricetulus migratorius relative densities were higherin grassland, farmland and shrubs, and Microtus obs-curus was found in all habitats except dense forest (Table2). The trapping of the eight Microtus oeconomus in openforest and garden was actually linked to the vicinity ofsmall streams. A correspondence analysis was carriedout with the species having shown differences in relativedensities between habitats, except M. oeconomuswhose distribution was linked to streams. A factorial mapillustrates how species and habitats were ordered (Figure





Figure 4 Factorial map of the correspondence analysis of thematrix trap line=species. Trap lines belonging to the same habi-tat are linked to the barycentre of the habitat by segments. df,dense forest; of, open forest; ga, village garden; gr, grassland;sh, shrubland; fa, farmland. Species codes are given in Table 1.

4). The two factorial axes (65% of the total inertia) showa species gradient from dense forest to grassland, farm-land and shrubland, with open forest and village gardenhabitats in an intermediate position.

Index transects

A total of 2835 intervals of 10 paces were walked(approximately 25 km). Figure 5 shows the index distri-bution of Ellobius tancrei and of smaller small mammals(Myodes, Microtus and Cricetulus faeces could not bedifferentiated). Under the assumption of independencybetween sample units, the density index of E. tancrei ingrassland and shrubland was significantly higher than inother habitats (x2s89.4, dfs1, p-0.00001). The densityindex of smaller species of small mammals in grassland,shrubs and forest was higher than in forest pastures(x2s6.5, dfs1, ps0.01).

Kokehada and Narenhebuke area

Considering the small number of trap lines and the dif-ferences in the number of night traps amongst habitats,relative densities could not be compared and were notcomputed here. Thus, the results are expressed as thenumber of trapped animals. Evidence of species pres-ence will be discussed on this basis.

In total, 33 animals (561 trap nights) representing sixspecies were trapped in the Kokehada area (Table 3).Apodemus uralensis, Cricetulus eversmani, C. migrato-rius, Meriones meridianus and Spermophilus erythroge-nys were caught in pastures, while Microtus gregalis wastrapped in the Alpine cushions of stream banks.

In total, 15 rodents representing four species weretrapped in the semi-desert around the Narenhebuke vil-lage (60 trap nights). Allactaga sibirica, Stylodipus telumand Spermophilus erythrogenys were caught in the

steppe areas around the village, and Microtus gregalis inthe Alpine cushions of a stream bank.

Figures 6 and 7 show the distribution of small mammalindices along transects walked in Kokehada and Naren-hebuke, respectively. In Kohehada, significantly differentfrequencies between habitats were observed for Sper-mophilus erythrogenys indices (x2s28.2, dfs1, p-0.0001) with larger numbers in slope grassland and shrubareas. No significant differences between habitats wererecorded for smaller size species and the larger speciesMarmota bobak.

In Narenhebuke, no differences between habitats werefound. Furthermore, Spermophilus indices were morethan six times lower than in the Kokehada summerpastures.

Fuhai area

In the semi-desert, five rodents representing four specieswere trapped (150 trap nights): Allactaga sibirica,Meriones meridianus, Stylodipus telum and Mus mus-culus (Table 3). In the oasis, a total of 41 rodents repre-senting five species were trapped (150 trap nights):Apodemus uralensis, Cricetulus migratorius, Merionestamariscinus, Microtus oeconomus and Mus musculus.

Discussion

The logistics of small mammal studies in the northernJunggar basin are reflected by the paucity of studiesreported from this region. To our knowledge, only lists oftrapped species in large regions of Xinjiang have beenpublished (see Ma et al. 1987), which did not report thesampling design nor the stratification of small mammalspecies per habitat. Here, we present for the first timeresults of small mammal surveys carried out in northeastXinjiang in relation to land cover with a quantitative anal-ysis at the community level. Obviously, a communityassessment would be more complete if temporal varia-tion was incorporated (e.g., several species trapped inthe present study belong to genera which include cyclicspecies, such as Microtus, see below). However, we areaware of unavoidable biases and incompleteness thatmust lead to considering results cautiously and a call formore comprehensive studies. Nevertheless, we think thatusing standard methods in small mammal communityassessment may be a first step towards better habitatand community definition of small mammal species ofChina and community comparisons (Giraudoux et al.1998, Raoul et al. 2006).

Limitations of the study

In one study site (Baihaba), 3-night standard trappingcould be carried out and the effect of sampling on pop-ulation estimated. This was not the case for the otherareas where trapping was carried out over a shorter dura-tion and with a smaller number of trap lines due to cir-cumstantial constraints of a larger programme aiming atscreening human populations for echinococcoses (Wanget al. 2001). However, we think that in this case, occa-sional trapping could be targeted to a first assessment





Tab

le2

Trap

pin

gre

sults

inth

eB

aiha

ba

area

exp

ress

edas

the

num

ber

ofan

imal

str

app

edp

er10

0tr

apni

ghts

.

Line

cod

eH

abita

tS

pec

ies

Ap

pe

Ap

spA

pur

Myr

fM

yru

Crm

iM

iag

Mio

bM

ioe

Sin

aS

ois

Sos

pS

otu

01D

ense

-for

est

(a)

5.3

1.3

1.3

1.3

JA1.

32.

713

.32.

71.

302

1.3

JB12

.0

03O

pen

-for

est

(b)

1.3

1.3

JC1.

31.

304

5.3

JD4.

01.

31.

310

1.3

4.0

5.3

1.3

2.7

JJ1.

36.

71.

311

2.7

2.7

JK5.

312

.02.

7

12S

hrub

land

(c)

JL4.

013

4.0

JM1.

316

5.3

4.0

20.0

JP1.

312

.04.

08.

0

14V

illag

e8.

02.

72.

75.

3ga

rden

(d)

22Fa

rmla

nd(e

)1.

32.

72.

74.

04.

0JV

1.3

2.7

07G

rass

land

(f)JG 08

2.7

09 211.

3JU

2.7

1.3

231.

32.

7JW

4.0

05 062.

7

Typ

eof

test

MW

MW

KW

KW

KW

KW

KW

MW

(a–c

/d–f

)(a

–e/f

)(a

,b/c

–f)

Pro

bab

ility

ofH

o0.

03ns

0.05

ns0.

0004

0.04

0.01

0.05

0.00

5ns

0.03

nsns

Mul

tiple

com

par

ison

––

––

a/f

nsns

nsns

––

––

(p-

0.05

)

Sp

ecie

sco

de

isgi

ven

inTa

ble

1.M

W:

Man

n-W

hitn

eyU

-tes

t(le

tter

ssh

oww

hich

cont

rast

sha

veb

een

used

,e.

g.,

a–c

/d–f

ind

icat

esha

bita

tsa,

b,

cp

oole

dco

mp

ared

tod

,e,

fp

oole

d);

KW

:K

rusk

al-

Wal

liste

st(fa

ctor

leve

lsar

eha

bita

ts).

Whe

rem

ultip

leco

mp

aris

ons

wer

ein

dic

ated

:let

ters

show

the

hab

itats

that

wer

esi

gnifi

cant

lyd

iffer

ent;

ns:

nosi

gnifi

cant

diff

eren

ce.





Figure 5 Index distribution of Ellobius tancrei and Microtus sizespecies along transects in the open habitats of the Baihaba area.

Table 3 Species and number of animals trapped in Fuhai, Narenhebuke and Kokehada.

Kokehada Narenhebuke Fuhai 1 Fuhai 2(semi-desert) (oasis)

Allactaga sibirica Forster 10 1Apodemus uralensis Thomas 3 3Cricetulus eversmani Brandt 2Cricetulus migratorius Pallas 3 20Meriones meridianus Pallas 1 1Meriones tamariscinus Pallas 14Microtus gregalis Pallas 7 1Microtus oeconomus Eversmann 2Mus musculus Linnaeus 1 2Stylodipus telum Lichtenstein 1 2Spermophilus erythrogenys Brandt 17 3

of species present on a comparative basis (see below).Although little is known about species sensitivity to traptypes, it is received wisdom that species are not equallyresponsive to trapping (for instance ground squirrels aredifficult to catch with BBBT traps set up in a standardline). Subsequently, relative densities computed as num-ber of animals per 100 trap nights do not allow directcomparisons between species, but permit comparisonsbetween habitats on an ordinal scale for each trappablespecies. Methods grounded on surface indices allowmonitoring at a smaller scale (over large areas). Thiswould not be possible with standard trapping (Giraudouxet al. 1995, Quere et al. 2000). Furthermore, some spe-cies are difficult to catch with surface traps (e.g., the sub-terraneous Ellobius tancrei); moreover, SBBT and INRAtraps do not capture species over 70 g and BBBT trapsover 200 g. These species can, however, be easilydetected from their indices; Ellobius tancrei leaves con-spicuous earth hills, different from those of Arvicola

terrestris Linnaeus, Talpa sp., Marmota sp. and Sper-mophilus dens are characteristic. Many studies havebeen carried out with these methods and resultsobtained are reasonably consistent if the limits of indexspecificity are known and taken into account (see e.g.,Giraudoux et al. 1998, Delattre et al. 1999, Duhamel etal. 2000, Hansson 2002, Michelat and Giraudoux 2006,Raoul et al. 2006).

Assemblages of small mammals and habitats

Species sampled in the present study are in general con-formity with regional and national species lists of China(Ma 1981, Zhang 1997).

Baihaba area

The Baihaba area was the only place where standardtrapping could be achieved allowing reasonable com-parisons of small mammal species relative densitiesbetween habitats. Based on results obtained here, wepropose a first scheme for the northern Junggar Basin,recording small mammal species assemblages at localscale according to a gradient from forest to grassland(Figure 8). Three assemblages could be recognised cor-responding to three main habitat classes: (I) forest habi-tats, characterised by Myodes rufocanus, Microtusagrestis, Myodes rutilus, Apodemus peninsulae andSorex isodon; (II) grassland areas, characterised by Ello-bius tancrei, Microtus obscurus, Cricetulus migratorius;and (III) transitional areas and farmland, which includesadditional species, such as Apodemus uralensis. Addi-tionally, the presence of Microtus oeconomus wasrecorded in the vicinity of small streams in forest as wellas in gardens. Species richness was higher in forest andin transitional areas and farmland, i.e., assemblages I andIII (9 and 8 species, respectively) and lower in grasslands(3 species). Sampling pressure was comparable amongstthe three assemblages (12, 9 and 10 trap lines, respec-tively). This is likely to limit sampling-related bias in thecomparison of richness among communities (Gotelli andColwell 2001).

However, this scheme has several biases. Firstly, com-mensal small mammals (rats, mice) and large rodent spe-cies (e.g., Marmota bobak Muller) have not beensampled. Secondly, a representation has been obtainedfrom the very short time span of a 2-week survey, which





Figure 6 Index distribution of Spermophilus erythrogenys, Microtus size species (black bars) and Marmota bobak (grey bars) alongtransects in Kokehada.

Figure 7 Index distribution of medium size species (Spermo-philus, Dipodidae, etc.) (black bars) and Microtus size species(grey bars) along transects in Narenhebuke.

does not take into account possible multi-annual chang-es of population densities. Ellobius tancrei indices werefound at frequencies as large as those observed for thesubterranean vole species Arvicola terrestris during pop-ulation outbreaks in Europe (Giraudoux et al. 1995, Duha-mel et al. 2000). Furthermore, Microtus obscurus, atypical grassland species, was found in almost everyhabitat, except dense forest. Thus, such cyclic speciescan drastically change assemblage structure over timebecause populations are limited to optimal habitats atlow density and may spill over to any sub-optimal habitatduring high density peaks. Delattre et al. (1992), Raoul etal. (2001) and Giraudoux et al. (2008) observed such pat-terns for Microtus arvalis Pallas in Europe, a species sim-ilar to Microtus obscurus wthey are even considered thesame species in Wilson and Reeder (2005)x. Microtus

agrestis is also known to undergo large multi-annual fluc-tuations of populations in Scotland (Bierman et al. 2006),Myodes rufocanus in Hokkaıdo, Japan (Stenseth et al.1996, Stenseth and Saitoh 1998a,b) and in northeasternChina (Xia 1996, Zhang 1996), Myodes rutilus in Scan-dinavia (Haukisalmi et al. 1988), Microtus oeconomus inAlaska and northern Siberia (Rausch 1995).

Northern Junggar assemblages

In total, five sites were sampled, although standard trap-ping could not be carried out everywhere and samplingpressure varied. All sites were situated at a maximumdistance of 216 km from each other (Kokeha-da–Baihaba). In total, five assemblages could be recog-nised on a regional range: (i) assemblages of coldsemi-desert with Allactaga, Stylodipus, Meriones, etc. inFuhai and Narenhobuke; (ii) assemblages of alpine grass-land of two types: (iia) one is a mixed forest landscape,described above at fine grain (Baihaba) with Ellobius tan-crei, Microtus obscurus, Myodes sp. etc., (iib): the otherwith just grassland (Kokehada) with Spermophilus ery-throgenys (Marmota bobak was recorded in the two are-as); (iii) assemblages of stream banks and wetlandscharacterised either by the presence of (iiia) Microtusoeconomus (Baihaba, Fuhai) or (iiib) Microtus gregalis(Kokehada, Narenhebuke).

However, one can suspect that this list is still incom-plete. For instance, other assemblages including Sper-mophilus undulatus Pallas, Ochotona sp., Arvicolaterrestris (reported as A. amphibius) have been describedin northern Xinjiang (Ma et al. 1987). These authors havealso reported Apodemus sylvaticussuralensis, Mus mus-culus, Sicista concolorstianshanicasnapea and Ellobiustalpinusstancrei in the Tarbatagai mountains, a commu-





Figure 8 Draft distribution of the main species along the forest-grassland gradient in the Baihaba area.

nity resembling that recorded in Baihaba in the presentstudy.

This highlights the extreme diversity of small mammalassemblages within a regional range in the northernJunggar area. The distribution of these communities andspecies has never been described at a local scale withstandard methods (e.g., linking local habitats and spe-cies), nor the transitions between communities on aregional scale. Semi-desert rodent communities havebeen found in Narenhebuke and Fuhai at very differentaltitudes and the range of some semi-desert rodents(e.g., Meriones meridianus) reached the degraded alpinegrasslands of Kokehada pastures where Marmota bobakwas recorded. Ellobius tancrei and Microtus obscurusseemed to be the dominant species (in terms of biomass)of the assemblages of the summer pastures of Baihabain Altai, and Spermophilus erythrogenys dominated theassemblage of the Kokehada pastures in the Tarbatagaimountains at the same altitude, where Marmota bobakwas also present. The origin of such contrasts and diver-sity is still not clear. Altitudinal gradients from desert tomountain zones are not the only factor that may explainassemblage distribution. Soil conditions which may facil-itate or jeopardise digging should also be considered.Precipitation may differ according to aspect and grassproduction may also be a key factor in small mammaldistribution. Overgrazing by sheep, goats and cattle inmountain grassland and the vicinity of the Junggar semi-desert may in part explain the extension of semi-desertspecies to the Kokehada pastures. Furthermore, the factthat similar habitats, such as grasslands, are dominatedby S. erythrogenys when others at the same altitude aredominated by Ellobius talpinus and M. obscurus in thesame region is still unexplained.

On a local scale, small mammal assemblages varied atvery short distance according to habitat. For instance, inFuhai, Microtus oeconomus was present in wetlands ofcultivated areas, within only some hundreds of meters ofa typical semi-desert assemblage. This species was alsopresent along small streams in forest and the vicinity ofgardens in Baihaba. Microtus gregalis was linked to riverbanks and Alpine cushions close (some metres) to asemi-desert assemblage (Narenhebuke), and to Sper-

mophilus erythrogenys populations in mountain grass-lands (Kokehada).

Conclusions

Many small mammal communities co-exist in the north-ern Junggar Basin (China) over a range of 20,000 km2,and they form a mosaic of assemblages whose compo-sition dramatically varies over space. Considering thiscomplexity, species distribution and assemblagedescription should be undertaken on at least two scales:

– at a local scale of some square kilometres, the aimwould be to standardise observation and describe therelationships between species and habitats at finegrain. Here, the study undertaken in Baihaba was ofthis type, but due to the short time span of the studyit failed to address the issue of multi-annual variationsof small mammal populations. However, the timedimension may be essential to understand processesgoverning the ecological system locally;

– on a regional scale of some ten thousands of squarekilometres, the target would be to standardise obser-vations and describe the transitions between com-munities at a coarse grain and understand the factorsresponsible for changes in the composition of assem-blage mosaics and the species turnover, given the factthat, over this range, similar habitats can harbour dif-ferent assemblages.

Some years ago this type of study would have be moredifficult to undertake due to lack of an easy-to-use spatialreference system. The increasing availability of globalpositioning system tools (GPS) and satellite data (digitalelevation models, multispectral imaging, etc.) should helpto guide small mammal studies on a regional scale. Theywould be essential not only for understanding the basicecology of small mammals in this area, but also to pro-vide a baseline to understand the effect of environmentalvariables on small mammal population time-space dis-tribution, to predict where population outbreaks are likelyand to estimate the risk of infectious disease transmis-sion from small mammal reservoirs.





Acknowledgements

We would like to acknowledge Professors Yang Wen Guangvice-president of the Xinjiang Medical University, Wang Yun Haiand Wen Hao for their dedicated support to this work, to Drs.Karine Bardonnet and Tamotsu Kanazawa for their help in thefield. We are grateful to the two anonymous referees who helpedto improve the manuscript. Financial support from the EuropeanUnion (TS3-CT94-0270), the French-Chinese Programme ofAdvanced Research (PRA E95-01), the Rotary Internationalthrough the Rotary Club of Dijon, the Conseil general de Haute-Saone, the French Ministere de la Jeunesse et des sports (DefiJeune) and the US National Institutes of Health and NationalScience Foundation (EID TW001565-05) from the Fogarty Inter-national Centre are acknowledged.

References

Allen, G.M. 1940. The mammals of China and Mongolia. Amer-ican Museum of Natural History, New York. 620 pp.

Aubry, J. 1950. Deux pieges pour la capture de petits rongeursvivants. Mammalia 14: 174–177.

Bierman, S.M., J.P. Fairbairn, S.J. Petty, D.A. Elston, D. Tidharand X. Lambin. 2006. Changes over time in the spatiotem-poral dynamics of cyclic populations of field voles (Microtusagrestis L.). Am. Nat. 167: 583–590.

Bjornstad, O., R. Ims and X. Lambin. 1999. Spatial populationdynamics: analyzing patterns and processes of populationsynchrony. Trends Ecol. Evol. 14: 427–432.

Chessel, D., A.B. Dufour and J. Thioulouse. 2004. The ade4package-I: one-table methods. R News 4: 5–10.

Corbet, G.B. 1978. The mammals of the Palearctic region: ataxonomic review. British Museum (Natural History), CornellUniversity Press, London. pp. 314.

Delattre, P., P. Giraudoux, J. Baudry, D. Truchetet, P. Musard, M.Toussaint, P. Stahl, M.L. Poule, M. Artois, J.P. Damange andJ.P. Quere. 1992. Land use patterns and types of commonvole (Microtus arvalis) population kinetics. Agric. Ecosyst.Environ. 39: 153–169.

Delattre, P., B. De Sousa, E. Fichet, J.P. Quere and P. Giraudoux.1999. Vole outbreaks in a landscape context: evidence froma six year study of Microtus arvalis. Landscape Ecol. 14:401–412.

Duhamel, R., J.P. Quere, P. Delattre and P. Giraudoux. 2000.Landscape effects on the population dynamics of the fos-sorial form of the water vole (Arvicola terrestris sherman).Landscape Ecol. 15: 89–98.

Giraudoux, P. 2007. Pgirmess: data analysis in ecology. R Foun-dation for Statistical Computing, http://www.R-project.org.Accessed 30.08.07.

Giraudoux, P., B. Pradier, P. Delattre, S. Deblay, D. Salvi and R.Defaut. 1995. Estimation of water vole abundance by usingsurface indices. Acta Theriol. 40: 77–96.

Giraudoux, P., J.P. Quere, P. Delattre, G. Bao, X. Wang, D. Shi,D. Vuitton and P.S. Craig. 1998. Distribution of small mam-mals along a deforestation gradient in south Gansu, China.Acta Theriol. 43: 349–362.

Giraudoux, P., D. Pleydell, F. Raoul, J.P. Quere, Q. Wang, Y. Yang,D. Vuitton, J. Qiu, W. Yang and P.S. Craig. 2006. Transmis-sion ecology of Echinococcus multilocularis: what are theranges of parasite stability among various host communitiesin China? Parasitol. Int. 55: 237–246.

Giraudoux, P., D. Pleydell, F. Raoul, A. Vaniscotte and P.S. Craig.2008. Why multidisciplinary and multiscale are essential ininfectious disease ecology: the Echinococcus multiloculariscase. Trop. Med. Health 35: 293–299.

Gotelli, N.J. and R.K. Colwell. 2001. Quantifying biodiversity:procedures and pitfalls in the measurement and comparisonof species richness. Ecol. Lett. 4: 379–391.

Gratz, N.G. 1994. Rodents as carrier of disease. In: (A.P. Buckleand R.H. Smith, eds) Rodent pests and their control. CABInternational, Wallingford, Oxon. pp. 85–108.

Gromov, I.M. and M.A. Erbajeva. 1995. The mammals of Russia.Russian Academy of Sciences, St. Petersburg. pp. 520.

Halik, M., N. Ohtaishi, A. Abudukadir and E. Hayashida. 1999.Characteristics of mammalian fauna and its distribution in theKanas National Nature Reserve in Xinjiang Altai, China. In:(M. Cheni and I. Sheydhidin, eds) Modern research in hightechnology. Japan Silk Road Club, Tokyo. pp. 118–124.

Hansson, L. 2002. Rodent dynamics, population regulation andpredation in changeable landscapes: importance for Echi-nococcus transmission. In: (P.S. Craig and Z. Pawlowski, eds)Cestode Zoonoses: echinococcosis and cysticercosis. Anemergent and global problem. IOS Press, Amsterdam. pp.267–285.

Hansson, L. and H. Henttonen. 1988. Rodent dynamics as com-munity processes. Trends Ecol. Evol. 3: 195–200.

Haukisalmi, V., H. Henttonen and F. Tenora. 1988. Populationdynamics of common and rare helminths in cyclic vole pop-ulations. J. Anim. Ecol. 57: 807–825.

Hoffmann, R.S. 1987. A review of systematics and distributionof Chinese red-toothed shrews (Mammalia, Soricinae). ActaTheriol. Sinica 7: 100–139.

Huitu, O., K. Norrdahl and E. Korpimaki. 2004. Competition, pre-dation and interspecific synchrony in cyclic small mammalcommunities. Ecography 27: 197–206.

Lidicker, W.Z. 2000. A food web/landscape interaction model formicrotine rodent density cycles. Oikos 91: 435–445.

Lu, P. and S. Yan. 1989. Xinjiang forest. People Press, Urumqi,China. pp. 577.

Ma, Y. 1981. The geographical distribution of rodents in northernXinjiang. Acta Zool. Sinica 27: 180–188.

Ma, Y., F.G. Wang, S.K. Jin and S.H. Li. 1987. Glires (rodentsand lagomorphs) of Northern Xinjiang and their zoogeo-graphical distribution. Science Press, Academia Sinica,Beijing. pp. 247.

Michelat, D. and P. Giraudoux. 2006. Synchrony between smallmammal population dynamics in marshes and adjacentgrassland in a landscape of the Jura plateau, France: a tenyear investigation. Acta Theriol. 51: 155–162.

Musser, G.G., E.M. Brothers, M.D. Carleton and R. Hutterer.1996. Taxonomy and distributional records of oriental andEuropean Apodemus, with a review of the Apodemus-Syl-vaemus problem. Bonner Zool. Beit. 46: 143–190.

Quere, J.P., F. Raoul, P. Giraudoux and P. Delattre. 2000. Anindex method applicable at landscape scale to estimate rel-ative population densities of the common vole (Microtusarvalis). Rev. Ecol. (Terre Vie) 55: 25–32.

R Development Core Team. 2007. R: a language and environ-ment for statistical computing. R Foundation for StatisticalComputing, http://www.R-project.org. Accessed 30.08.07.

Raoul, F., R. Defaut, D. Michelat, M. Montadert, D. Pepin, J.P.Quere, B. Tissot, P. Delattre and P. Giraudoux. 2001. Land-scape effects on the populations dynamics of small mammalcommunities: a preliminary analysis of prey-resource varia-tions. Rev. Ecol. (Terre Vie) 56: 339–352.

Raoul, F., J.P. Quere, D. Rieffel, N. Bernard, K. Takahashi, R.Scheifler, A. Ito, Q. Wang, J. Qiu, W. Yang, P.S. Craig and P.Giraudoux. 2006. Distribution of small mammals in a pastorallandscape of the Tibetan plateaus (Western Sichuan, China)and relationship with grazing practices. Mammalia 42:214–225.

Rausch, R.L. 1995. Life-cycle patterns and geographic distri-bution of Echinococcus species. In: (R.C.A. Thomson andA.J. Lymbery, eds) Echinococcus and hydatid disease. CABInternational, Wallingford, Oxon. pp. 89–134.

Shenbrot, G.I. and B.R. Krasnov. 2005. An atlas of the geograph-ic distribution of the arvicoline rodents of the world (Roden-tia, Muridae: Arvicolinae). Pensoft Publishers, Sophia. pp.336.

Siegel, S. and N.J.J. Castellan. 1988. Non parametric statistics





for the behavioral sciences. McGraw-Hill InternationalEditions, New York. pp. 399.

Stenseth, N.C. and T. Saitoh, eds. 1998a. The population ecol-ogy of the vole Clethrionomys rufocanus. Japanese Societyof Population Ecology, Kyoto. pp. 158.

Stenseth, N.C. and T. Saitoh. 1998b. So, what do we know andwhat do we need to know about the population ecology ofthe vole Clethrionomys rufocanus? Res. Popul. Ecol. 40:123–125.

Stenseth, N.C., O.N. Bjornstad and T. Saitoh. 1996. A gradientfrom stable to cyclic populations of Clethrionomys rufocanusin Hokkaido, Japan. Proc. R. Soc. London B 263: 1117–1126.

Wang, S.B. and G.Y. Yang. 1983. Rodent fauna of Xinjiang.Xinjiang People’s Publishing House, Urumqi. pp. 223.

Wang, Y.H., M.T. Rogan, D.A. Vuitton, H. Wen, B. Bartholomot,C. Macpherson, P. Zhou, X. Ding, X. Zhang, J. Luo, H. Xiong,Y. Fu, A. McVie, P. Giraudoux, W. Yang and P.S. Craig. 2001.Cystic echinococcosis in semi-nomadic pastoral communi-ties in north-west China. Trans. R. Soc. Trop. Med. Hyg. 95:153–158.

Wilson, D.E. and D.M. Reeder. 2005. Mammal species of theworld. Johns Hopkins University Press, Baltimore, MD.pp. 1207.

Wust-Saucy, A.G. 1998. Polymorphisme genetique et phylogeo-graphie du Campagnol terrestre Arvicola terrestris. Universitede Lausanne, Lausanne, Suisse. pp. 118.

Xia, W.P. 1996. Rodent-like pest control and ecological equilib-rium. In: (Z.W. Wang and Z.B. Zhang, eds) Theory and prac-tice of rodent pest management. Science Press, Beijing.pp. 2–18.

Zhang, Y.Z. 1997. Distribution of mammalian species in China.China Forestry Publishing House, Beijing. pp. 280.

Zhang, Z.B. 1996. Population fluctuation and regulation of smallmammals. In: (Z.W. Wang and Z.B. Zhang, eds) Theory andpractice of rodent pest management. Science Press, Beijing.pp. 145–165.

Zhao, G.Z. 1996. Rodent pest control and plant protection. In:(Z.W. Wang and Z.B. Zhang, eds) Theory and practice ofrodent pest management. Science Press, Beijing. pp. 19–37.

Zhu, S.K. and A.G. Chen. 1993. The ecology features anddetects of Mus musculus. Science Press, Beijing. pp. 339.



Documents

Small mammal assemblages and habitat distribution in the northern Junggar Basin, Xinjiang, China: a pilot survey