2001 Asiatic Herpetological Research Vol.9, pp. 113-121

Lizards of the Northern Mongolian Deserts: Densities and Community Structure

1 1 2 A. V. AND GEORGY I. SHENBROT DMITRY SEMENOV , KONSTANTIN ROGOVIN ,

'A.N.Severtzov Institute of Ecology and Evolution, Russian Academy of Science, Leninsky pr. 33 Moscow ~ 1 17071, Russia, Ramon Science Center, Ben-Gurion University of the Negev, P.O. Box 194, Mizpe Ramon, 80600, Israel

Abstract.- Spatial organization and population densities of three-species lizard community was studied in the Gobi Desert, Mongolia. To evaluate the effect of habitat variables on the distribution and abundance of each species we used the stepwise procedure of factor selection with ANOVA on each step. To describe the distribution of species' spatial niches in the space of environmental variables, we used stepwise discriminant function analysis (DFA). The number of species in 1-ha grid areas varied from to 4. Phrynocephalus versicolor was the only species distributed over the 91% of grids occupied. There was a positive relationship between distribution and local species abundance. A set of two to three habitat variables determined the abundance of each species. The result of DFA signify to the well pronounced segregation, but not even distribution of species spatial niches in the space of resources.

Key words.- Lizards, Mongolia, community, ecology, density

91* 96* 101* 106° 111* 116'

Fig 1. Map of Mongolia and location of sites where data were collected. Southern Gobi, 2-Western Gobi, 3- Eastern Gobi, 4-Barun-Churay Basin.

Introduction at sites, a low level of species turnover between habi- tats, low abundance of most species and high domi- The reptile communities of Mongolian deserts are nance of only one species, Phrynocephalus versicolor. characterized by several specific features discussed A few common species have rather broad spatial elsewhere (Ananjeva and Semenov, 1986; Borkin and niches, diverse behavioral and physiological charac- Semenov, 1984; Munkhbajar, 1976; Semenov and teristics (diverse range of thermobiological patterns, Borkin, 1986; Semenov and Shenbrot, 1988). There wide active search for food items, etc.). are few species in the fauna with low species richness Vol. 9, p. 114 Asiatic Herpetological Research 2001

The degree of interest in comparative studies of transitive with no specific features of vegetation. structure and function of reptile communities Thus, only the southern (extra-arid) and northern sub- increased dramatically after seminal papers written by zones are well pronounced. E. Pianka Most of the (Pianka, 1973, 1975). ensuing The southern desert occurs mostly in the Trans- studies were devoted to the rich and diverse species Altai Gobi and is characterized by a few very dry, communities of desert lizards in Australia, south- unproductive biotopes inhabited by five lizard species western North America and southern Africa (Case, (two agamids, two gekkonids, and one Eremias spe- 1983; Fuentes, 1976; Henle, 1989; Colwell, Inger, cies). Among this group only one species, Phryno- Millado et al., 1975; Pianka,1986; Scheibe, 1977; cephalus versicolor is common in the northern Shenbrot et At the 1987; al., 1992; Simbotwe, 1984). subzone (Semenov and Borkin, 1986). same time study of species poor communities in the The northern deserts, which extend to the south- Central Asian desert can provide a significant infor- west, west, and south-east of Mongolia and along mation not only in comparison with other continents, both slopes of the Mongolian Altai and Gobi-Altai but also can help us to understand better which factors mountains, are characterized by pronounced microre- rule in reality structure and dynamic of lizard commu- lief and rich vegetation, although the main landscape nities of many species. Up to now there were only two types are the same as in the southern subzone. The examples of such studies made in China (Chang et al., vegetation in rock and gravel valleys consists of 1993;Luietal.,1992). perennial grass (Stipa), forbs, onions and succulents, The main of this was to the objective paper study and a variety of annual plant species. Shrub vegetation features of and specific spatial organization popula- is often associated with foothills and sand dunes, or is tion densities of lizard in the three-species community spread along the dry river beds (Lavrenko, 1978). The dur- Gobi Desert, Mongolia. The study was conducted difference in climate between western and eastern a research on the ing long-term program biodiversity parts of the northern Mongolian desert is not pro- of the desert and was Mongolian biota, sponsored by nounced (Murzaev, 1952); some differences exists in the Permanent Soviet-Mongolian Biological Expedi- the composition of the flora (Yunatov, 1950). tion. Lizard species Material and Methods There are four lizard species inhabiting northern Mongolian deserts: Alsophylax pipiens, Phrynoceph- desert Mongolian alus versicolor, Eremias przewalskii and E. multiocel- A map (Fig. 1) illustrates the location of desert lata. Among these, only the three last mentioned regions of Mongolia. Three desert regions to the south species are abundant and relatively widespread. of the Altai Mountains are partly separated from one another by chains of low mountains and hills. These Data collection three are Trans-Altai Gobi Alashan Gobi (South), We collected data during three field trips to the Mon- and Gobi or Basin (East) Sungarian Barun-Churay golian northern deserts in June-August 1985, 1986 Besides these deserts there (West) (Yunatov, 1950). and 1988. Forty five 1-ha grids were established in the are desert areas between the Altai and moun- Hangai Northern desert subzone (see map, Fig 1 ). Grids were cold and tains, usually called Western deserts, some distributed so that they covered the whole range of arid lands in the Lakes and Ubsu-Nur Basins Great habitats from the middle slopes of the mountains to (northwest). the clay basins and sand dunes. Each habitat type was Three arid subzones of the Mongolian desert are sampled equally. Two factors determined the number defined (Sokolov and Gunin, 1986): extra—arid of grids at a desert region: diversity of habitat types and abundance of lizards. Each was divided into desert (<50 mm of rainfall per year), real desert (50— grid smaller 20 x 20 m. The centers of 100 mm per year) and steppe—like desert (100—150 25 sample plots, were marked with 50 cm aluminum mm per year). However, the climatic border that sample plots stakes. Lizards were restricts the distribution of plants (Kazantseva, 1986; sampled by repeated, regular search of established two to four consec- Volkova et al., 1986) and animals (Podtyazhkin and grids during of their diurnal Orlov, 1986; Semenov and Borkin, 1986) exists only utive days during periods maximal all encountered lizards were between the southern part of Trans-Altai Gobi (< 50 activity. Nearly captured hand. Most of the were con- mm per year) and the northern waste belt of desert by surveys accounting ducted before the of In the rest lands with more predictable precipitation (100-200 appearance hatchlings. of the cases were not counted. Each mm per year). The narrow real desert subzone appears hatchlings cap- 2001 Asiatic Herpetological Research Vol. 9, p. 115

1. of lizard Table Density (no. ha ) species.

s. 0) O 'o w a>a c (A "(5 c 3 TON O Asiatic Research 2001 Vol. 9, p. 116 Herpetological

Results cies but with the pronounced habitat preference. The second was E. multeocellata, and the third was E. Densities and distribution through habitat przewalskii. The last species had the most restricted types habitat preference, namely sandy-loess hills in saline depressions with shrub vegetation of Nitraria sp. The results of lizard density and diversity estimations of P. versicolor was cor- on the 1-ha grids indicate the low local species diver- The abundance positively of related with its broad distribution. The of this sity in Mongolian deserts (Table 1). The number density varied from 1 to 106 individuals 1-ha species in our grid areas varied from zero to three. species per The second most abundant E. Among 45 grids there were two grids with no lizards, (Table 1). przewalskii individuals lha and the third was 21 grids with only one species. 18 grids with two spe- (44 per maximum), individuals 1-ha cies and four grids with three species. Phrynoceph- E. multiocelata (11 per maximum). alus versicolor was the only species distributed over All three species coexisted at rather high densities in with hills covered with the most number of grids (91% of grids occupied). saline depressions sandy-loess There were no Eremias przewalskii was found on 33%, E. multioce- Nitraria sp. shrubs. pronounced nega- the densities of two dominant lata on 24% and A. pipiens on 4% of grids. Regarding tive correlation between situated within this habitat distribution through the main habitat types (Table 1), species on grids type (P.

P. versicolor was also the most distributed widely spe- versicolor-E. przewalskii: R"=0.04, n=20, ns).

Table 2. Designation and description for the 23 habitat variables included in the analysis.

Mnemonic Variable Unit

RCK Content of rocks in the soil %%

GRW Content of gravel in the soil %%

CLY Content of clay in the soil %%

SCS Sand cover area %%

SCH Sand cover height cm

WDS Dry river bed area %%

WDD Dry river bed depth cm

NRB Number of rodent burrows no/sq.m.

FRB Abundance of annual forbs no/sq.m.

AGR Abundance of annual grasses no/sq.m.

ANN Overall abundance of annual grasses and forbs no/sq.m.

ALL Perennial Allium covet %%

PGR Perennial grass cover %%

MIC Cover of microphyllous shrubs %%

HAL Cover of halophytuos shrubs %%

sue Cover of small succulent shrubs %%

HLX Cover of Haloxylon %%

NIT Cover of Nitraria /o /o

o/ o/ SHC Overall shrub cover /o /o

SV1 Perennial plant crown volume at the level 0-25 cm %%

SV2 Perennial plant crown volume at the level 25-50 cm %%

0/ O/ SV3 Perennial plant crown volume at the level 0.5-1 m /o /o

o/o. SV4 Perennial plant crown volume at the level 1-2 m Vo% 2001 Asiatic Herpetological Research Vol. 9, p. 117

AGR NIT

1 1 0.5 0.5

-0.5

-0.5 -1 -1

SHV2 SHV2

1 0.5

-0.5

-1 CLY

1 GRV 0.5

1 -0.5

0.5 -1

-0.5

-1 Fig. 3. Habitat use profiles of E przewalskii. For expla-

1 nation see Fig. 2.

Fig 2. Habitat use profiles of P. versicolor. For each FRB habitat variable rank, mean use by species is plotted as a difference between capture frequency on plots of this rank and capture frequency of all plots. The size of histograms therefore denote the selectivity (positive or negative) of habitat use. For mnemonics see Table 2

Spatial niches

There was a statistically significant influence of habi- tat variables on the distribution and abundance of all MIC three studied lizard species (Table 3). For each of these species we extracted a set of two to three habitat variables, determining 6.6-28.7% of observed vari- ance in abundance. Densities of two species (P. versi- color, E. przewalskii) were moderately affected by habitat variables, whereas density of E. multiocellata was weakly affected. Fig. 4. Habitat use profiles of E multiocellata. For There was a statistically significant influence of explanation see Fig.2. habitat variables on the distribution and abundance of all three studied lizard species (Table 3). For each of Patterns of habitat usage based on selected vari- these species we extracted a set of two to three habitat able sets for each species are presented on Figures variables, determining 6.6-28.7% of observed vari- 2-4. Phrynocephalus versicolor clearly avoided ance in abundance. Densities of two species (P. versi- microsites with low gravel content, very low and very color, E. przewalskii) were moderately affected by high annual grass abundance, moderate and high habitat variables, whereas density of E. multiocellata shrub crown volume in the level 0.25-0.5 m and pre- was weakly affected. ferred microsites with moderate gravel content, mod- erate annual grass abundance and very low shrub Vol. 9, p. 118 Asiatic Herpetological Research 2001

Table 3. Summary of ANOVA analyses of influence of habitat variables on individual species' abundance. Values are proportions of total dispersion determined by given variable. Total proportion of variance determined by a set of variables may be greater than sum of influences of individual variables as a result of high-order interactions.

Species V a r 2001 Asiatic Herpetological Research Vol. 9, p. 119

Fig. 5. Seventy-five percent confidence ellipses for the species observations on two discriminant axes (DF1 and - DF2). Em- Eremias multiocellata, Ep- Eremias przewalskii, Pv Phrynocephalus vesicolor.

fa Q

DF1 dance with food abundance and a species-specific tac- here means not the size of the species range area, but tic of antipredator behavior. It preferred microsites the number of sites where each species was found. with moderate to high Nitraria cover and moderate to Phrynocephalus versicolor was encountered on 41 high shrub crown volume at 0.25-0.5 m (Fig. 3). In grid areas and had the highest population density summer E. przewalskii feed predominantly on berries (mean: 37.3, median: 26 ind/ha, maximum: 106 ind/ and young green twigs of Nitraria, and also find pro- ha). With the edge-effect correction (Semenov, 1991) tection under the dense cover of the crowns of low maximum density was 70.02 ind/ha (175.5 g/ha biom- spiny shrubs. In contrast to E. przewalskii, E. multio- ass). Eremias przewalskii was found on 1 5 grid areas cellata avoided microsites with very high annual plant and was the second abundant species (mean: 9.5, abundance and shrub cover, but also preferred micro- median: 4 ind/ha, maximum: 44 ind/ha). Eremias mul- sites with moderate and rather high values of these tiocellata was found on 1 1 grid areas and its maxi- variables (Fig. 4). This difference can be interpreted mum density was 12 ind/ha (mean: 3.6, median: 2 ind/ in accordance with thermobiological and size charac- ha). Alsophilax pipiens was met on two grids with teristics of two Eremias species. The larger species, E. density 3 ind/ha. is not so as the smaller E. multiocel- przewalskii quick Brown (1995) explains this rather common rela- lata. The first one well in soft soil, and ther- digs tionship by the "Hutchinsonian niche model" (see on or under the shrub moregulates climbing escaping also for one species, Brown 1984), suggesting that active the Small E. periodically, being throughout day. "the species that is slightly more tolerant of some abi- multiocellata that habitats with low occupy relatively otic conditions or biotic interaction or is slightly bet- cover small vegetation (with sparsely-distributed ter able to use some resource should not only be able shrubs) must cross sites in search for food items open to occur in more places but also to attain higher abun- and attacks into escape predator's by quickly rushing dance in some of those places." In the approach we small shrubs or burrows. This reduces heat species by use here niche breadth reflects microhabitat require- into burrow the escaping during day-time. ments of each species, namely the range of microcon- Another result of the above comparison is the pos- ditions where each lizard species occurred. Diversity itive relationship between distribution and local abun- of these microhabitats is not necessarily correlated dance among the species considered. Distribution with the diversity of macrohabitats as well as with the Vol. 9, p. 120 Asiatic Herpetological Research 2001

Table 4. Summary of discriminant analysis of the habi- tats of lizard species. DF1 and DF2 are the first two components (all are significant, P<0.001). Mnemonics for habitat variables are from Table 1 . 2001 Asiatic Herpetological Research Vol. 9, p. 121

Brown J. H. 1995. Macroecology. The University of Institute of Animal Evolutionary Morphology and Chicago Press. Chicago and London. 269 pp. Ecology. Moscow. (In Russian).

Case, T. J. 1983. Niche overlap and the assembly of Scheibe, J. S. 1987. Climate, competition, and the island lizard communities. Oikos 41(3):427-433. structure of temperate zone lizard communities. Ecol- 1424-1436. Chang Cheng, Quang Chen, and Nai-Fa Liu. 1993. ogy 68(5): [The coexisting mechanism of two species of lizards Semenov, D. V. 1991. [Variations of characteristics in the same Journal of community]. Lanzhou Univer- of animal spatial distribution in Phrynocephalus ver- Natural Scienes 180- 183. sity, 29(3): (In Chinese). sicolor populations (Reptilia; Agamidae]. Ekologiya, Sverdlovsk Fuentes, E. R. 1976. Ecological convergence of lizard 1991(3):77-81 (In Russian). communitiesin Chile and California. Ecology 57:1- Semenov, D. V, and L. J. Borkin. 1986. [Amphibians 17. and reptiles]. Pp. 1 14-119. In V E. Sokolov and P. D. Gunin deserts of the Transaltai Gobi. The Henle, K. 1989. Ecological segregation in an assem- (eds.). [The natural blage of diurnal lizards in arid Australia. Acta Oeco- features, ecosystems, and regionalization]. Nauka, Moscow. logica 10(1): 19-35. (In Russian).

Semenov, D. V, and Shenbrot G. I. 1988. den- Inger, R. F, and R. K. Colwell. 1977. Organization of [Lizard sities in some Natural features contiguous communities of amphibians and reptiles Mongolian regions]. and resources of some of in Thailand. Ecological Monographs 47:229-253. regions Mongolian People's Republic. [Abstr.] (In Russian). Kazantseva, T I. 1986. [Distribution and dynamics of Shenbrot, G. I. 1988. of factors that the overground phytomass]. Pp. 106-114 In V.E. [Evaluation affect lizard distribution and abundance in deserts of Sokolov and PD. Gunin (eds.). [Deserts of Trans- the South of Bukhara Oblast Altai Gobi]. Nauka, Moskva. (In Russian) (Uzbek SSR, USSR)]. Ekologiya, Sverdlovsk 1988 (5):51-56. (In Russian). Lavrenko, E. M. 1978. [On vegetation of steppes and Shenbrot, G.I. 1991. On the environmental factors deserts of Mongolian People's Republic]. Problemy desert lizard distribution and abundance. 6th osvoeniya pustyn. Ashkhabad 1:3-18. (In Russian). affecting Ordinary General Meeting of Societas Herpetologica Liu Nai-Fa, Ren-De Li, and Xiao-Cheng Liang. Europaea. Budapest. [Abst.]. 1992. [Community structure of desert lizard in Gansu Shenbrot, G. I., K. A. and A. Surov. Province]. Acta Zoologica Sinica 38(4):377-384. (In Rogovin, V China). 1992. Comparative analysis of the spatial organization of the desert lizard communities in Middle Asia and in Millado, J., F Amores, F F. Parreno, and F Hiraldo. Mexico. Pp. 147-188. In V Sokolov, G. Halffter and 1975. The structure of a Mediterranean lizard com- A. Ortega (eds). Vertebrate ecology in arid zones of munity. Donana Acta Vertebrata 2 (2): 145- 160. Mexico and Asia. Inst, de Ecologia A.C., Centro de Munkhbayar, Kh. 1976. [Amphibians and of reptiles Investigaciones Biologicas de Baja California Sur Ulaanbaatar. 167 Mongolian People's Republic]. pp A.C., and MAB-UNESCO. (In Mongolian). Simbotwe, M. P. 1984. Resource partitioning in a Murzaev, E. M. 1952. [Mongolian People Republic]. woodland reptile community of the Kafue flats, Zam- 472pp. Geografgiz. Moskva. (In Russian) bia. African Journal of Ecology 22:281-287. Pianka, E. R. 1973. The structure of lizard communi- Sokolov, V E. and P. D. Gunin (eds.). 1986. [Deserts ties. Annual Review of Ecology and Systematics of Trans- Altai Gobi]. Nauka. Moskva. 207 pp. (In 4:53-74. Russian). Pianka, E. R. 1975. Niche relations of desert lizards. Volkova, G.A., E. I. Rachkovskaya, and I. T. Feodor- Pp. 292-314. In M.L. Cody and J.M. Diamond (eds). ova. 1986. [General features of plant distribution]. Pp. Ecology and Evolution of Communities. Harvard 84-95. In V. E. Sokolov and P. D. Gunin (eds.). University Press. [Deserts of Trans-Altai Gobi]. Nauka. Moskva. (In Pianka, E.R. 1986. Ecology and natural history of Russian). desert lizards. Princeton University Press. Princeton, Yunatov, A. A. 1950. [The main features of vegetation New Jersey. 208 pp. of Mongolian People's Republic]. Proceedings of Podtyazhkin, O. I. and V. N. Orlov. 1986. [Faunoge- Mongolian Comission 39. AS USSR Publishers. Mos- netic of rodents in arid grouping Mongolian zone]. cow, Leningrad. 223 pp. (In Russian). Pp. 124-136. In V.E. Sokolov (ed.). [Zoogeographical regionalization of Mongolian People's Republic].