COMMUNITY STRUCTURE OF DESERT SMALL MAMMALS: … · INTRODUCTION northern Africa (Sahara, Arabian,...

Ecology, 77(3) , 1996, pp. 746-761 O 1996 by the Ecolog~cal Society of America

COMMUNITY STRUCTURE OF DESERT SMALL MAMMALS: COMPARISONS ACROSS FOUR CONTINENTS1

DOUGLAS A. KELT* AND JAMES H. BROWN Department of Biology, University of New Mexico, Albuquerque, New Mexico 87131 USA

EDWARDJ. HESKE Illinois Natural History Survey, 607 E. Peabody, Champaign, Illinois 61820 USA

PABLO A. MARQUET Facultad de Ciencias Bioldgicas, Departamento de Ecologia, Universidad Catdlica de Chile,

Casilla 114-0, Santiago, Chile

STEPHEN R. MORTON CSIRO Division of Wildlife and Ecology, P.O. Box 84, Lyneham, ACT 2602, Australia

JULIAN R. W. REID CSlRO Division of Wildlife and Ecology, P.O. Box 2111, Alice Springs, NT 0871, Australia

KONTANT~NA. ROGOVIN Academy of Sciences of Russia, Institute of Animal Evolutionary Morphology and Ecology,

Leninskyi pr. 33, Moscow 11 7071, Russia

GEORGY SHENBROT Ramon Science Center, Ben Gurion University of the Negev, Mitpe Ramon, 80600, Israel

Abstract. Presencelabsence data for the small-mammal species at sites in seven deserts were analyzed for evidence of similarity in community structure. The deserts studied were located in North and South America (268 and 118 sites, respectively), Australia (245 sites), Israel (54 sites), and greater Eurasia (Thar, 15 sites; Turkestan, 36 sites; Gobi, 98 sites). Patterns observed in all deserts included: (1) low a diversity (2-4 species per site); (2) high P diversity (species turnover between sites); and (3) local coexistence of 20-30% of the species in the regional pool. Additionally, the number of species with which a species co-occurred increased with the number of sites at which that species occurred. Although these results suggested that some features of community structure were similar across deserts, other aspects, especially trophic structure, differed widely. Deserts in the northern hemisphere possessed more granivores, and the Turkestan Desert more folivores, than other deserts. Carnivorous small n~ammals were most strongly represented in Australia, and omnivores in South America, Australia, and the Thar. The structure of desert small-mammal communities is strongly influenced by historical factors; different taxonomic groups with distinctive trophic adaptations proliferate in different desert regions where they are subject to some common structuring processes of community assembly.

Key words: Aslan deserts; Australian deserts; community structure; desert small mammals; Gobi Desert; Negev Desert; North American deserts; South American deserts; Thar Desert; trophic structure; Turkestan Desert.

INTRODUCTION northern Africa (Sahara, Arabian, Nafud, Negev, Kara- -

The deserts of the world present natural laboratories Kum, Kyzyl-Kum, Gobi, Thar, etc.). Perhaps because of the common biotic conditions and relative structural

for the study of community structure and the processes (vegetative) simplicity of deserts, these systems have

influencing that structure. The major desert areas (Fig. played a prominent role in developing and testing ideas 1) occur in North America (Sonoran, Great Basin, Chi- in community ecology (e.g., Rosenzweig and Winaker huahuan, Mojave), South America (SechuraIAtacama, 1969, Brown 1975, Mares and Rosenzweig 1978, Bow- PunalAltiplano, Chaco, Monte, Patagonia), Australia ers and Brown 1982, Pianka 1985, 1986, Brown and (Gibson, Great Sandy, Great Victoria, Tanami, Simp- Heske 1990, Polis 1991). The morphological similar-son), southern Africa (Namib, Kalahari), and Asia and ities among some of the small mammals in these his-

torically isolated systems (e.g., Mares and Rosenzweig ' Manuscript received 21 February 1995; revised 17 July 1978, M~~~~ 1980) suggest broad-scale parallelism or1995; accepted 26 July 1995. Present address: Department of Wildlife, Fish, and Con- convergence among these tam. Much research on the

servation Biology, University of California, Davis, Davis, community ecology of desert small mammals has been California 956 16 USA. conducted in North America (Reichman 1991), and

April 1996 DESERT SMALL-MAMMAL COMMUNITY STRUCTURE

FIG. 1. Map of the world's desert regions, including some arid regions not generally considered as true desert. Areas studied in the present report are indicated with capital letters. A, North American deserts; B, AtacamaIAltiplano Deserts (denoted South America throughout text); C , Turkestan Desert; D, Gobi Desert; E, Negev Desert; F, Thar Desert; G, Australian deserts; h, other South American deserts, including Monte, Patagonia, and Chaco (a thorn scrub region); i, Caatinga, a thorn scrub region; j, Sahara Desert, including the Somali-Chalbi coastal desert; k, southern African deserts, including the Kalahari, Karroo, and Namib; 1, Arabian Desert; m, Iranian Desert.

some researchers have assumed that the patterns observed there would be applicable elsewhere, although this has increasingly been called into question.

Recent analyses have shown that morphological convergence is not necessarily associated with ecological convergence (e.g., Mares 1983, 1993a, Morton 1993, Morton et al. 1994). For example, large bipedal species (Dipodomys) in North America are predominantly granivorous, whereas similar species (e.g., Allactaga, Ja - culus) in Asian deserts generally are folivorous or omnivorous. At the community level, granivorous species are strongly represented in North American deserts, whereas carnivorous taxa are prevalent in Australia (Morton 1985, Morton and Baynes 1985), and folivorous or omnivorous species are most common in South Africa (Kerley 1989, 1992). Comparative studies of desert small-mammal biology are needed (see also Morton 1979, Marquet 1994, Morton et al. 1994, Shen- brot et al. 1994) to ascertain the extent to which patterns described in one desert are characteristic of deserts as a whole, especially with respect to ecological or morphological convergence. How general are the patterns observed in North American deserts and, therefore, how useful are these studies for understanding the evolution of desert biota? Although several other studies and symposia have touched on this issue (e.g., Reich- man and Brown 1983, Polis 1991, Mares 1993a, b, Morton and Heske 1994), none has analyzed local community structure simultaneously across many different deserts. Comparisons of desert small-mammal assemblages by Mares (1993b) and Shenbrot et al. (1994) analyzed data on species that occurred together within large regions rather than those that coexisted in local habitats. The present study differs from these earlier comparisons by focusing on how local (a)species diversity, turnover of species between local communities

(p diversity), and several kinds of community structure vary within and across several geographically and historically distinct desert regions. Such analysis allows a focus on patterns and scales of diversity and coexistence of species. Moreover, comparisons between analyses conducted at local and regional scales may provide insights into the ecological and historical factors that have produced observed patterns.

The present report was developed through the independent research programs of the authors in different regions of the world. Collaboration was undertaken to address the questions previously noted. We analyze data on community structure from seven desert areas in North and South America, Australia, and greater Eurasia. Specifically, we ask how communities of small mammals in the seven deserts compare with respect to local richness, between-site turnover, and patterns of coexistence and of trophic structure. Similarities and differences among these communities are discussed in light of the historical biogeography of these deserts (Mares 1993a, b, Ricklefs and Schluter 1 9 9 3 ~ ) .

MATERIALSAND METHODS

Data used in this study are presencelabsence matrices compiled for the terrestrial mammal species with mass less than =500 g that were documented (by trapping and other sampling methods) at numerous sites in each desert region. North American data are based on Mor- ton et al. (1994), augmented by 67 additional sites from the Sonoran and Chihuahuan Deserts of Mexico studied by Shenbrot and Rogovin (Rogovin et al. 1991, Shen- brot et al. 1994; see Appendix for species collected). A total of 42 species occur at 268 sites throughout the Great Basin, Mojave, Sonoran, and Chihuahuan Des- erts. Of these species, 28 (67%) are found in two or more of these deserts. Four species (10%) occur in all

748 DOUGLAS A. KELT ET AL. Ecology, Vol. 77, No. 3

four deserts, seven (17%) occur in three deserts, and 17 (40%) occur in two deserts; of the latter category, 12 species are shared by the Sonoran and Chihuahuan Deserts, and three by the Great Basin and Sonoran Deserts. Because of this extensive sharing of species, we lump these deserts to form the North American desert region.

Australian data are from Morton et al. (1994), and consist of 26 species at 245 sites. These data are based upon surveys conducted in the Australian arid zone, primarily by Morton and Reid (see Appendix). Morton et al. (1994) and Rogovin et al. (1991) provide further details of Australian and North American data.

Data for South America were obtained from the literature, supplemented by data collected in the field by Marquet (Marquet 1994). In total, 22 species are recorded at 128 sites in the Atacama Desert (12 sites, 4 species) and the Altiplano (110 sites, 21 species, including three of the four Atacama species) of Peru, Bolivia, northern Argentina, and northern Chile. Only dry Altiplano sites ( 5 4 0 0 mm annual precipitation) were used in the present analyses. Although our sample represents a subset of South American arid regions (see Mares 1993a), we will refer to these sites collectively as South American.

Data for Greater Eurasia were extracted from the literature (Shenbrot et al. 1994) and supplemented with data collected in the field by Shenbrot and Rogovin (Shenbrot 1992, Rogovin and Shenbrot 1993, 1995, Rogovin et al. 1994, Krasnov et al., in press, Shenbrot and Rogovin, in press). These data include sites in the Thar Desert (S = 10 species at 15 sites), the Turkestan Desert (including the Kazakhstan, Kyzyl-Kum, Kara- Kum, and Dagesthan Deserts; S = 15 species at 36 sites), three regions of the Gobi Desert (east, central, and west; S = 18 species at 98 sites), and the Negev Desert of Israel (S = 11 species at 54 sites). These four Asian deserts are geographically distinct and have distinctive mammalian faunas. The greatest overlap occurs between the Turkestan and Gobi Deserts; across all sites for which we have data, only four of 29 species (<14%) are shared between these deserts. Additionally, Mus musculus is shared by the Negev, Thar, and Gobi Deserts. Therefore, these regions are considered sep- arately in the following analyses.

Species were allocated to four trophic categories, carnivores, folivores, granivores, and omnivores, based on published literature and our own experience with these species. Additionally, some Asian taxa classified as having a mixed diet of seeds and plant vegetative parts have been considered in analyses as 50% folivore and 50% granivore.

Some analyses require knowledge of the local pool of species that could potentially colonize a site. This pool was calculated by superimposing maps of the species' distributions over maps of site localities.

Precipitation data for sites in each desert region were

0.4 1 North America

Thsr

...

m TurkestanTurkestsn N = 3 6 S=15

ZEi c 0.0

Gobi

0.4 Australia

South America N = 118 S = 22:::

1

1 *, -, , , , , , , , ,

0.0"." 0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 1 3

NUMBER OF SPECIES PER SITE

FIG. 2. Number of small mammal species collected per desert site in all regions examined. In all figures, N is number of sites sampled, while S is number of species encountered.

taken from the "World WeatherDisc" (WeatherDisc Association, Seattle, Washington, USA).

Patterns of species distribution and abundance

The number of species recorded ranged from 10 to 42 species per desert, and was significantly related to the number of samples (linear regression: slope =

0.101, R2 = 0.84, P = 0.003, N = 7 , d f = 6).Because neither the number of samples nor the number of species recorded regressed significantly against desert area (samples: slope = 1.6 X R2 = 0.26, P = 0.24, N = 7, df = 6; species: slope = 2.3 X R2 = 0.08, P = 0.53, N = 7, df = 6), this may be considered a minimum estimate for the regional species pool avail- able to colonize these sites. It is possible that some species may not have occurred at any collecting site, or may have been missed in these censuses. Nonethe- less, it is clear that, in spite of this wide range in the size of regional species pools, local ( a )diversity was generally similar and low across all deserts. Recorded

749 April 1996 DESERT SMALL-MAMMAL COMMUNITY STRUCTURE

North America North America N =268 N=268

0.2 S = 42 S=42

0.0

Thar N = 15

Gobi N=98

Australia N = 245

0.2 S=26

0.0

South America N = 118

0.2 S=22

PERCENTAGE OF TOTAL SITES

FIG.3. Percentage of sites within a desert occupied by a proportion of all species within a desert. In North America, e.g., 45% of the species occurred at 0-5% of the sites sampled, 19% of the species occurred at 6-10% of the sites sampled, etc.

means were < 4 species per site in North America, < 3 species per site in the Asian deserts, and 5 2 . 5 species per site in the Negev and in both southern hemisphere deserts (Fig. 2). Medians were generally 2-3 species per site, except in the Gobi, where median richness was 4 species (Fig. 2). As might be expected, sampling effort affected the probability of observing extreme values. Thus, both range of local richness and maximum richness were loosely correlated with number of sites sampled (range: R2 = 0.52, P = 0.07, N = 7, df = 6; maximum richness: R2 = 0.51, P = 0.07, N = 7, df = 6) and with number of species present across all sites (range: R2 = 0.61, P = 0.04, N = 7, df = 6; maximum richness: R2 = 0.66, P < 0.03, N = 7, df =

6). Most species occurred in only a small proportion of

the sites sampled in each desert (Fig. 3). The distribution of species in the Thar, and to a lesser degree the Gobi, was more uniform than that in the other regions. Within most deserts, the vast majority (93- 98%) of species occurred at <40% of the sites. Thus, only a few species were widespread in each region; the

1 Thsr

7 . -. ez; S=15 W 29 0 I,,,

2 6 ] Gobi

,

Australia N = 245 S=26

2 0

1 South America

PERCENTAGE OF REGIONAL SPECIES POOL

FIG. 4. Percentage of all species sampled in a given desert with which a species coexisted across all sites for each region examined. In North America, e.g., one species coexisted with only 6-10% of the species in the regional pool, five species coexisted with 11-15% of the species in the regional pool, etc.

seeming anomaly for the Thar may reflect the limited number of sites sampled.

Most species in the Negev, North American, and Australian deserts coexisted at some sites with fewer than one-third of the other species in their regional species pools (Fig. 4). The Negev was particularly striking in this respect, with a typical species coexisting with 5 1 0 % of the other desert rodent species. In contrast, species in the Thar all occurred with >45% of the other possible coexisting species; most species in the Gobi and South American deserts also occurred at some sites with a high proportion of the other possible coexisting species. The number of species with which a species coexisted was strongly correlated with the number of sites at which that species occurred (Fig. 5; all P < 0.005; all R2 > 0.5, except North America: R2 = 0.46, and South America: R2 = 0.36). While we have not applied any of the published indices to quantify beta diversity, these results clearly indicate the high level of beta diversity in all deserts except the Thar.


NUMBER OF AT OCCURRED FIG,5 , ~ ~number of~ species with which a species co- t l

existed across all sites, as a function of the number of sites at which that species occurred. In every case, the relationship is highly significant and highly explanatory.

The low a diversity and high regional diversity are a result of a high average turnover of species among sample sites.

Regional diversity of trophic groups

Distribution of species across trophic groups also varied greatly among desert regions (Fig. 6). Grani-vores were the most species-rich group in North Amer- ica, but were much less diverse elsewhere. In fact, strict granivores were entirely absent from South American sites and from Australia. Folivores were moderately represented in the Turkestan, North America, and South America, less so in the Gobi and Negev, and were nearly absent from Australia and the Thar Desert. Om- nivores were most frequent in South America, followed by North America and Australia, and were poorly represented in the Negev. Finally, carnivores were relatively infrequent in all deserts except Australia, where they were very diverse.

Comparisons of local trophic diversity across deserts

The mean representation at sites was significantly uneven across deserts for all four trophic categories (P

6 - 1 Folivores I

I Carnivores

FIG. 6. Number of species in the regional pool for each region examined. Two relatively specialized trophic categories (granivory and carnivory) have undergone remarkable radiations in two separate regions, whereas folivory and omnivory are much more evenly distributed across these deserts.

< 0.0001; Fig. 7, Table 1). Granivores were most important in North America, although they were still prominent at sites in the Thar, Gobi, Negev, and Tur- kestan Deserts (Fig. 7, Table 1). However, a much greater proportion of the regional granivore pool was present at sites in the latter deserts than in North Amer- ican sites (compare Figs. 7 and 6). Sites in the Tur- kestan Desert possessed significantly more folivorous species than did other deserts (Fig. 7, Table 1). Among sites in the Gobi, Thar, Negev, and South America, species richness of folivores did not differ significantly, but was significantly greater than at sites in North America or Australia. Omnivores were most abundant at sites in South America, although not significantly more than at sites in the Thar. A second grouping con- sisted of Australia, the Thar, and the Gobi. Some comparisons in Fig. 7 appear counterintuitive: the mean number of omnivorous species occuring per site was greater in Australia than in the Thar, but variability in the Thar meant that it was not significantly poorer in species than South American sites, whereas Australian sites were significantly poorer in species. The Negev

I

April 1996 DESERT SMALL-MAMMAL COMMUNITY STRUCTURE 75 1

I

FIG. 7. Number of species per site (mean + 1 SE) in each desert for each trophic category. Number of species in each trophic category is highly significantly different across the regions examined (Granivores, F = 232.1, P < 0.0001; Fo- livores, F = 70.4, P < 0.0001; Omnivores, F = 26.3, P < 0.0001; Carnivores, F = 42.1, P < 0.0001; for all tests, df = 3 , 843.

exhibited the lowest number of omnivores, but was not significantly less rich than the Thar or Turkestan Des- erts (Fig. 7). The marsupial component of Australian deserts contributed to the significantly greater fre-quency of carnivory there than in any other desert (Fig. 7). Sites in the remaining deserts possessed few carnivorous species.

Comparisons of local trophic diversity within deserts

The mean representation of trophic categories at sites within each particular desert differed significantly (all P < 0.0001; Table 2). The dominant groups, however, were not the same in all deserts. In North America and the Negev, Thar, and Gobi Deserts, sites had significantly more granivores than other trophic groups (Table 2). In these deserts, groups of secondary importance were folivores (Negev), omnivores (North America), or both groups (Thar, Gobi). In contrast, Australia and South America had communities dominated by omnivores (Table 2), with carnivores (Australia) or folivores (South America) second in importance. Folivorous taxa

dominated local assemblages only in the Turkestan Desert, followed in importance by granivores.

Distribution of trophic categories across sites

The mean number of sites at which species of different trophic groups were reported varied greatly (Fig. 8). For example, 21 granivorous species in North America occurred at a mean of 33.4 sites each. whereas the two carnivorous species (both in the genus Onych- omys) occurred at an average of 42.0 sites each. A similar situation occurred in Australia, where a single folivore (Pseudomys desertor) occurred at 23 sites, whereas 16 carnivorous species occurred at a mean of 15.4 sites each.

No trophic category contained species that were con- sistently the most widely distributed across sites in all deserts. Rather, the mean geographic distribution of species in these categories (as indexed by the mean number of sites occupied per species) varied consid- erably across deserts. There was no clear relationship between the geographic distribution of a species and its trophic categorization.

Similarities in community structure across broadly separate geographic regions may reflect several processes, which may be grouped in three categories. First, similarities may reflect underlying general characteristics of community structure, either within a taxonomic group or within a wide range of community types, including but not restricted to desert communities. If this were the case, then similarities in community structure among regions would be less interesting biologi- cally than differences among regions. Second, similar environmental conditions might result in convergence of taxa and assemblages toward common adaptive forms. This may occur in lineages with very different evolutionary origins and histories (e.g., Orians and Paine 1983, Schluter 1986, Schluter and Ricklefs 1993). For example, small-mammal communities in arid regions may possess common features (e.g., of coloration and morphology) that are direct conse-quences of selection on performance in similar environments. Finally, similarities in community structure may reflect common evolutionary histories. Related taxa frequently possess similar niche dynamics (Mares 1980, 1983, 1993a, b, Brown and Zeng 1989, Brown and Harney 1993), and such similarity may be ex-pressed at a community level in similar patterns of morphological, trophic, or other structure.

Differences in the structure of small-mammal communities from geographically separate regions may be due to differences in their contemporary environments. For example, deserts occurring at high and low elevation or high and low latitude may have distinctive climates and, as a result, very different vegetation and substrate. Alternatively, differences may reflect differ-


TABLE1. Means, standard errors, and results of ScheffC a posteriori tests on number of species of each major trophic category per site (N , number of sites). Significant comparisons are denoted with asterisks. All comparisons across deserts are significant (ANOVA, P < 0.0001, df = 6, 833).

Desert N

Granivores (F= 226.4) N. America 268 Gobi 98 Turkestan 36 Thar 15 Negev 54 S. America 128 Australia 245

Folivores (F = 61.5) N. America 268 Gobi 98 Turkestan 36 Thar 15 Negev 54 S. America 128 Australia 245

Omnivores (F = 26.0) N. America 268 Gobi 98 Turkestan 36 Thar 15 Negev 54 S. America 128 Australia 245

Carnivores (F= 40.8) N. America 268 Gobi 98 Turkestan 36 Thar 15 Negev 54 S. America 128 Australia 245

No. species

Mean 1 S E NA

2.62 1.42 1.07 1.83 1.65 0.00 0.00

0.45 0.95 1.82 0.83 0.52 0.84 0.09

0.81 0.99 0.50 1 .00 0.07 1.61 1.27

0.31 0.49 0.08 0.00 0.19 0.06 1.01

ent histories either of the lineages occurring there or of the places themselves. The Gobi and Turkestan Des- erts, for example, share many higher faunal elements, being dominated by rodents in the families Muridae and Dipodidae (Mares 1 9 9 3 ~ ) . These may be expected to exhibit similar suites of ecological traits, which may in turn influence multispecies interactions and, therefore, community structure. With respect to places, rain shadow desert climates, caused by mountain ranges and, thus, by historic orogenic events, may differ from those formed by arid climates that develop within the interior of large continents. Deserts that have been strongly influenced by Pleistocene climatic and vege- tational changes may differ from those that have per- sisted without such major changes.

Similarities

Small-mammal communities in these deserts share certain features. Many correspond with patterns documented for small-mammal communities of North American deserts (e.g., Brown and Kurzius 1987). Lo- cal (a)species diversity is relatively low at most sites, even in the face of high P diversity and varying regional

ScheffC tests comparing desert regions

GO TU TH NE S A AU

species pools (Fig. 2); as a result, most species occur at relatively few of the sites sampled (Fig. 3). Finally, the number of species with which a given species co- occurs depends strongly upon the number of sites at which that species is found (Fig. 5). The significant relationships between the number of sites occupied and (a) the number of species with which a species co- occurs (Fig. 5 ) , and (b) the species-specific patterns in the number of sites occupied (Fig. 3), indicate that species respond individualistically to environmental and community characteristics, as suggested by Glea- son (1926; see also Brown and Kurzius 1987). If sets of species were to respond to abiotic characteristics as integrated units that replace each other across the landscape (as proposed by Clements 1936), then we would not expect the regressions in Fig. 5 to deviate significantly from zero. Brown and Kurzius (1989) have noted the resemblance of these large-scale patterns to the patterns predicted from the hypothesis of core and sat- ellite species (e.g., Hanski 1982). However, Brown and Kurzius (1989) emphasize that the presence or absence of species across different sites appears to reflect not stochastic variation in local abundances across time and


TABLE2. The mean number of small mammal species ( 21 SE) in each trophic category present per site, across all deserts. Within deserts, similar superscripts are not significantly different, as determined by a Scheffk a posteriori test. Dominant categories within deserts are presented in boldface type. These are the same data as in Fig. 7, but presented by desert rather than by trophic category.

No. Granivores Folivores Omnivores Carnivores

Desert sites F df Mean 1 SE Mean 1 SE Mean 1 SE Mean 1 SE

North America 268 391.7 3, 1071 2.62a 0.005 0.45' 0.002 0.81h 0.000 0.3 lC 0.002 Gobi 9 8 25.8 3. 391 1.42 0.010 0.95b 0.007 0.99b 0.000 0.49c 0.006 Turkestan 36 60.2 3,143 107b 0.011 1.8Za 0.021 0.50d 0.001 0.08< 0008 Thar 15 20.9 3, 59 1.83a 0.043 0.83b 0.016 l.OOb 0.005 0' 0 Negev Australia

54 245

139.0 197.4

3, 215 3, 391

1.65" 0'

0.012 0

0.52b 0.09c

0.001 0.001

0.07L 1.27a

0.005 0.004

0.19C l .Olh

0.008 0.004

South America 118 107.4 3, 471 Od 0 O.Wb 0.007 1.67" 0.010 0.06C 0.002

space (as suggested by Hanski 1982), but rather the These similarities do not appear to be attributable to individualistic response of different species to essential unique features of deserts, although two possible ex- resources, as well as the tendency for species from ceptions may be noted. Low a diversity may be related different functional groupings (e.g., quadrupedal vs. to the low productivity of arid regions, and high P bipedal heteromyids) to co-occur more frequently than diversity may reflect some unique form of spatial het- expected by chance (Fox and Brown 1993). erogeneity that characterizes many arid regions. For

example, soil type and structure are known to influence plant community structure and composition, which in

80 . North America turn affect the small-mammal species that may exist in 6 0 - 1 I a given locality. Local and even geographic distribu-

tions of many species of small mammals in arid regions are strongly and directly limited by substrate (e.g., Schmidly et al. 1993, Shenbrot et al. 1994). The impact of substrate heterogeneity is less direct in grassland or forest ecosystems, where increased plant and litter bio-

1 1 1 4 mass ameliorates physical and other differences among ////A ",,,,,,- r / , / ; , / , A soil types. At any rate, low a and high P diversityU

3 w

40 Gobi appear to be fundamental characteristics of arid zone small-mammal communities, and they are relatively in- variant across deserts of varying history, faunal composition, geomorphology, and climatic regimes.

Thar Previously, we have shown, for several of these 2 0 -

4 2 regions, that local richness of desert small-mammal 0 V/ / / / / /A l/////'m 0 6

communities is relatively independent of regional richness (see Brown and Kurzius 1987, North America; Morton et al. 1994, Australia; and Marquet 1994. South American sites below 3000 m elevation). This also appears to hold for small mammals across a range of

W 40 / South America I 1 environmental conditions in southern Chile (Kelt et al.

1995), and may be a general characteristic of small- mammal communities. If so, this argues against local richness being a relatively linear function of regional enrichment (e.g., Ricklefs 1987, Ricklefs and Schluter Australia

60 1 9 9 3 ~ ) . Instead, it implies greater regulation of local species diversity and composition by local ecological processes, e.g., competition, predation, etc. (Brown and Kurzius 1987, Kelt et al. 1995; see chapters in Ricklefs and Schluter 1993b for examples and counterexamples

Granivores Folivores Carnivores Omnivores to this pattern in other taxa). This is not meant to imply, however, that regional and historical processes do not

FIG. 8. Number of sites (mean + 1 SE) occupied by spe- affect local community composition (e.g., South Amer- cies in each trophic group. Numbers above the bars are the number of species of a given trophic group existing in the ican sites above 3000 m elevation; Marquet 1994). As regional pool. we will discuss, these may have major influences on


community composition and trophic organization. Al- though competitive interactions have a strong influence on small mammal community structure in North Amer- ican deserts (Brown 1975, Brown and Heske 1990, Heske et al. 1994, Valone and Brown 1995), too few studies have been conducted in other deserts to evaluate the importance of competition on community structure, especially the potential role of competitive exclusion in creating the high P diversity observed (but see Abramsky and Sellah 1982, Moss and Croft 1988, Fox and Gullick 1989, Mitchell et al. 1990, Brown et al. 1994, Meserve et al. 1995).

Finally, the frequency of local occurrence of species from a given trophic category, in all deserts, was not predictable from the proportional representation of that trophic category in the regional pool (compare Figs. 5 and 6). In North America, for example, our species pool included two carnivorous species and nine folivorous species, but both groups were equally represented locally (Fig. 7) because the carnivorous taxa occurred at a greater number of sites, on average, than did the folivorous species (Fig. 8). Similarly, the Thar Desert possessed a large number of omnivorous taxa (Fig. 6): which may reflect the restricted area and relatively recent anthropogenic origin of this desert, as well as the fact that many species here are most closely allied with species from the forested Oriental region (Prakash 1963, 1974, Rogovin et al. 1994, Shenbrot et al. 1994). Even in the Thar, however, granivores were clearly more important than omnivores at the local level (Fig. 7).

Differences

Many striking differences in community structure are evident across these deserts. Brown and Kurzius (1987) reported that most species in North American deserts co-occurred with relatively few members of the re-gional pool (see also Morton et al. 1994). Although this pattern also held true for the Negev and Australia, such a conclusion was not well supported for the Thar and Turkestan Deserts, and did not appear to hold for the Gobi or South America (Fig. 4).

Perhaps the most striking dissimilarities across these deserts, however, concern the relative abundances of taxa exploiting different trophic strategies. In particular, the dominant (in terms of species richness, population density, and total biomass) mammalian granivores that have received much attention in North Amer- ica (e.g., Brown et al. 1979, Reichman and Price 1993) were less common elsewhere. Strictly granivorous species were not observed in either southern hemisphere desert (Fig. 7). In other deserts that have been studied, ants, e.g., in the Karoo Desert of South Africa (Kerley 1991), the Monte Desert of South America (Mares and Rosenzweig 1978), and deserts in Australia (Morton and Davidson 1988, Morton 1993), as well as birds, e.g., in Australia (Morton and Davies 1983), were important granivores. Carnivory was common in Austra-

lia, omnivory in Australia and South America, and folivory was pronounced in the Turkestan Desert.

These differences may be broadly explained by differences in contemporary environments and by the different histories of the deserts and the mammalian lineages occurring in them.

With respect to the contemporary environment, two principal features may have profoundly influenced mammalian communities. First, these deserts vary both in latitude and elevation. The Gobi Desert and the Great Basin of North America both occur at high temperate latitudes ( > 3 j 0 N), whereas the Thar Desert and much of the Australian and South American arid regions are essentially subtropical (<25" S). The Negev, lower So- noran, and Atacama Deserts all occur near or along oceanic coastlines, whereas the Gobi, Turkestan, Great Basin, and especially the Altiplano, all occur at high elevation (>I500 m). These geographical differences result in very different patterns of insolation and precipitation over both seasonal and die1 time scales. Such climatic differences in turn influence vegetation. All else being equal. therefore, floristics, life-forms, and life histories of the natural vegetation of these regions differ in important ways that almost certainly have important implications for small mammals.

Topographic relief also varies greatly across the deserts. The Basin and Range topography of southwestern North America presents alternating mesic and xeric conditions, with numerous low- to high-elevation mountain ranges separating intervening desert basins. Such topography is also found in the Negev and parts of the Gobi Desert. Similarly, the Andes provide a wide range of habitats by influencing precipitation patterns in the deserts occurring both within the Andes and on either side of the range. The coastal Atacama and Sechura Deserts are extremely arid because the Andes intercept moisture-laden air flowing westward from the Amazo- nian Basin, and the cold upwelling waters along the Pacific coast help to further dry the air. Additionally, weathering and erosional processes produce alluvial de- posits at the bases of mountains, contributing to spatial heterogeneity of substrates. In contrast, deserts of Aus- tralia and most deserts in Asia present broad, open ex- panses with little topographic relief. Their soils have been exposed to weathering and leaching for many mil- lennia. As a result, they often are poor in nutrients and support a very different vegetation than soils with strong allochthonous inputs (e.g., Lindsay 1985, Stafford Smith and Morton 1990, Lamont 1995).

The amount and predictability of precipitation also vary among deserts. This may affect the predictability of certain food resources and, thereby, representation of trophic strategies. Morton (1993) argued that the lack of strict granivores in Australia may partly reflect low predictability of annual precipitation and, therefore, of a reliable seed crop. This argument has been extended to explain termite abundance in Australia, and myrmecophagy in lizards and mammals (Abensperg-

April 1996 DESERT SMALL-MAMMAL COMMUNITY STRUCTURE 755

MEAN ANNUAL PRECIPITATION (cm)

700 Z 9$ 600

EFIG 9 Mean annual preclpltatlon vs co- 500

effic~entof varlatlon (CV) ~n annual preclplta- W tlon for 28 statlons across seven deserts Note & that Australla has moderate annual preclpltatlon 4 400 but a cons~stentlyh ~ g hCV for annual preclpl- 2 tatlon In contrast, South Amer~canstatlons ex- 300hlblt a range of preclpltatlon wlth generally low d CVs Most sltes show low p r e c ~ p ~ t a t ~ o nbut w ~ d e v a r ~ a b ~ l ~ t y~n CVs between statlons O 200,5

Traun 1994). When annual precipitation is plotted against the coefficient of variation (CV) in annual pre-cipitation for various deserts (Fig. 9 ) , Australian sites clump relatively tightly with respect to these variables, having high yearly variability for their moderate mean annual precipitation. The Thar Desert receives more rainfall but is also highly unpredictable. Other deserts overlap broadly, exhibiting the highest CVs in sites where mean precipitation is very low (<20 mmlyr). In South America, sites in the Atacama Desert receive very little precipitation, whereas sites located in the Altiplano receive moderate precipitation. In more me-sic parts of the Altiplano, precipitation is greater and less variable; however, only dry Altiplano sites ( 5 4 0 0 mm annual precipitation) were selected for this study. Variation in means and CVs among sites within these deserts indicates that species inhabiting areas with ex-treme unpredictability in annual precipitation may be "rescued" via metapopulation or sourcelsink dynamics frorn nearby sites with more predictable precipitation and food resources. The number of granivorous taxa, however, does not correlate well with either mean an-nual precipitation or its CV, suggesting that this may only partly explain granivore species abundance in North America.

The deserts studied also possess historically distinct faunal lineages. Perhaps most apparent, Australian des-erts have a large marsupial component, with strongest representation in the arid regions by the Dasyuridae (e.g., Antechinomys, Dasyuroides, Ningaui, Sminthop-sis). Additionally, Australia possesses several groups of murid rodents, including two of Australia's most species-rich genera, Notomys and Pseudomys. Asian deserts are dominated by two families of rodents: the Muridae, including gerbils and jirds (Gerbilinae, e.g., Dipodillus, Desmodillus, Gerbillus, Meriones, Rhom-bomys, Tatera),and the Dipodidae, including 1 1 genera of jerboas and their relatives (e.g., Allactaga, Cardio-cranius, Dipus, Jaculus, Pygeretmus, Salpingotus). North America has the endemic family Heteromyidae, which includes several extremely desert-adapted gen-

I I

- . A

* - v A A

4 A * South Amerlca

- m . v

- 1

A * v

v

a v 1

era (e.g., Chaetodipus, Dipodomys, Perognathus). This group is unique in possessing external fur-lined cheek pouches, an evolutionary innovation that may have been central to the subsequent specialization on seeds. Along with members of this family, a suite of cricetine rodents (e.g., Neotoma, Onychomys, Peromyscus, Rei-throdontomys) exhibits varying degrees of adaptation to arid conditions. In South American deserts, small mammals are mostly of two groups, the very old ca-viomorph rodents (e.g., Octodon, Octodontomys) and the younger, but much more diverse, cricetine radiation (e.g., Akodon, Auliscomys, Eligmodontia, Phyllotis). Hence, each major desert region may be broadly char-acterized as having relatively distinct mammalian lin-eages. Overlap in genera is nonexistent, and is quite low even at the family level.

The time span over which taxa have been present in these deserts also differs greatly, which may help to explain the differing degrees of adaptation to xeric con-ditions. Current consensus is that the heteromyids of North America diverged from savanna-dwelling an-cestors in the Oligocene (Hafner and Hafner 1983, Haf-ner 1993, Wahlert 1993). The Asian dipodids and cri-cetids are thought to have radiated within northern Asian deserts in the middle Miocene (Pavlinov et al. 1991, Shenbrot et al., in press), whereas the gerbilines evolved in the Sahara-Sindian Deserts (Sahara, Arabia, Negev, Thar) in the Miocene; subsequent colonization of northern Asia by the southern taxa, and vice versa, occurred in the middle-upper Pliocene. However, the highly desert-adapted dipodids evidently never suc-cessfully colonized the Thar Desert. Marsupials have been present in Australia since at least the early Ter-tiary, whereas the conilurine rodents, descendents of mesic-adapted taxa of southeast Asia, arrived by the Pliocene (Lee et al. 1981) and have been present through a long history of Australia's arid and lacustrine phases (Morton 1993). Two later colonizations brought additional murid rodents (see Lee et al. 1981, Baver-stock et al. 1983, Baverstock 1984, Hand 1984). Mar-supials and caviomorph rodents have been present in


South America during much of the Andean uplift and resultant aridification (since early late cretaceous and Oligocene, respectively), whereas the sigmodontine rodents arrived with the closing of the Panamanian Isth- mus (or slightly earlier; Hershkovitz 1962) some 3-5 X lo6 yr ago. Considered by most authors to have originated from pastoral or sylvan species, this lineage subsequently radiated spectacularly within South America (Reig 1984, 1986).

The deserts clearly vary in their age, geographical area, persistence through geological time, and type of habitats occurring on their borders, which contain pools of potential colonists. The Miocene was a time of global aridification and desertification (e.g., Alpers and Brimhall 1988, Singh 1988, Rundel et al. 1991). The deserts of central Asia are thought to be relatively old (known from the Cretaceous, although not widespread until the Miocene; Sinitzin 1962). Those of Australia (Pliocene or Pleistocene; Bowler 1982) and of South (probably Holocene; Arroyo et al. 1988) and North America are younger. North American deserts are generally thought to have formed in the Pleistocene (Ax- elrod 1958, Webb 1977, Van Devender and Spaulding 1979, Wells 1979, Thompson and Mead 1982), although recent work suggests an origin as early as late Miocene (Riddle 1995). An evident exception to the Asian pattern is the Thar Desert, in which most real sand desert landscapes are the result of human activites in historical times (Wadia 1960, Prakash 1963). Rel- ative ages of these deserts and the implications for their mammalian faunas are considered in greater detail by Morton (1993), Marquet (1994), and Shenbrot et al. (1994).

The deserts of Australia and central Asia (Gobi, greater Turkestan) are extensive, with low topographic relief. The Thar Desert is much smaller and is relatively isolated from other desert regions. Its small size, long history of human occupation, and proximity to forested regions in India have been implicated as factors influencing the morphological and trophic structure of small-mammal communities there (Rogovin et al. 1994, Shenbrot et al. 1994). The desert areas of South America also are relatively small. The Atacama is isolated from the Monte and Chaco arid regions by the Andean massif, but the biogeographic effect of this barrier remains unclear (see, e.g., Caviedes and Iriarte 1989, Marquet 1989, Meserve and Kelt 1990). The proximity and faunal affinities with the central Chilean Mediterranean zone (Meserve and Glanz 1978), as well as evidence for recent gene flow between some taxa in the Altiplano and the Atacama (Walker et al. 1984), suggest that the Atacama may not be as isolated as previously thought.

Desert conditions in Asia are thought to have been relatively persistent since their formation (Sinitzin 1962). In contrast, both Australian and North American desert regions have undergone significant fluctuations in the Pleistocene. In North America, the contraction

and expansion of arid intermontane basins with each pluvial/interpluvial cycle (Schmidly et al. 1993) may have increased the geographic isolation among related taxa, and, therefore, the possibilities for allopatric speciation, to a greater extent than has occurred in Asian deserts (Shenbrot et al. 1994). Australia has also ex- perienced multiple arid and lacustrine phases in the past 500000 yr (Bowler 1982). Glacial cycles seemingly have impacted South American deserts less strongly than those in the northern hemisphere (although these events may have been pivotal in southern South Amer- ica and the high Andes; Clapperton 1993). The western side of the Andes may have been wetter than the eastern side during the Miocene, with subsequent drying into the Pliocene (Arroyo et al. 1988). The presence of the Andean massif and development of the Humboldt Cur- rent with the closing of the Tethys Sea have had very strong impacts on the climatic patterns of the Altiplano and Atacama.

Estimates of ages of deserts and their faunas are continually being revised (e.g., Riddle 1995); thus, we hesitate to develop historical scenarios in great detail. This discussion provides a backdrop, however, against which the salient differences between the small-mam- ma1 faunas of the worlds' deserts may be summarized.

In North America, pluvial/interpluvial phases oc-curring in Basin and Range topography, plus the ser- endipitous presence of the seed-specialized heteromyid rodents, have led to the overwhelming importance of granivory at both local and regional scales (see Heske et al. 1994). The presence of external, fur-lined cheek pouches in ancestral heteromyids may have preadapted these species to their foraging mode, fostering evolution toward strongly granivorous habits. The presence of forested montane regions scattered throughout the North American desert region probably has provided species pools of taxa less adapted to deserts, and these have colonized to become members of desert com-munities there.

In contrast, the low relief and extensive area of central Asian deserts have not produced frequent isolation between populations and, therefore, have not favored extensive speciation. Rather, this topography has fos- tered species with large geographic ranges and corre- spondingly low between-site turnover (P diversity) and high patterns of species co-occurrence. Interestingly, the main ancestral stock in northern Asian deserts con- sisted of granivorous taxa; these subsequently shifted towards folivory or omnivory in the Pliocene, and these strategies remain dominant today. Most of these species are well adapted to arid conditions and, in fact, the majority of the genera are endemic to desert habitats and regions (Shenbrot et al. 1994). The exception is the Thar Desert, which is relatively small, bounded by forested areas, subtropical, young, and largely anthro- pogenically derived; perhaps as a result of these characteristics, it is conspicuously lacking in species ex-


hibiting specialized adaptations for an arid environment.

Australia has also undergone a series of pluvial and interpluvial phases. However, as in Asia, this has occurred over a landscape lacking in notable topographic relief, and probably has not facilitated extensive allop- atry. The arid regions of Australia are dominated by omnivores and carnivores. This may largely reflect phylogenetic conservatism, since dasyurid marsupials generally are carnivorous and conilurine rodents generally are omnivorous. Additionally, conilurine and other murine rodents in Australia are derived from tropical forest species of southeast Asia, where omnivory and folivory are common trophic habits. The extreme unpredictability of precipitation (and therefore of seed production) in Australia may have precluded the evolution of granivory in these species.

The South American deserts considered here show clear dominance by omnivores and folivores at both local and regional scales. There are no fully water- independent species in South America, although some species (e.g., Eligmodontia) appear to be developing such features. Strict granivores are lacking, although some taxa are highly granivorous in some areas (e.g., Oligoryzornys longicaudatus and Phyllotis danvini in central Chile, Meserve and Glanz 1978, Meserve 1981). Carnivory is restricted almost exclusively to the marsupial Thylamys elegans (Meserve 1981). Given that the more extensive arid regions east of the Andes lack strict granivores (Mares and Rosenzweig 1978), it may not be surprising that these are also absent in the Altiplano and the Atacama. It remains unclear why many species of South American rodents overlap with so many other species. This may partially reflect their relatively recent radiation (Reig 1984), but also may be a particular characteristic of the South American arid regions we have studied. In particular, the Alti- plano is known for its rich mammalian fauna. It would be quite interesting to compare the results obtained here with similar data from other arid regions of South America, such as the Chaco and Monte.

Early attempts to explain regional patterns of community organization on the basis of local processes met with some success (reviewed in Reichman 1991). How- ever, the inability of such approaches to fully explain the distribution of species and the structure of communities in similar environments led some more recent authors to a reevaluation of the importance of regional and historical processes (e.g., Ricklefs and Schluter 1993b). Here, we take an intermediate course, and suggest that similarities and differences across geographically isolated desert faunas reflect the contemporary geography and physiognomy of the deserts, as well as what Brown (1995) has termed the "history of place" and the "history of lineage." As ecologists pursue studies at greater spatial scales, replication of study sites becomes increasingly difficult. There is only one Australia and there is only one Atacama Desert. Gould

(1989) has lamented that it simply is not possible to rerun the tape of life. Barring such idealized experi- mentation, we are forced to evaluate alternative hy- potheses based on their relative abilities to provide par- simonious and reasonable explanations for large-scale ecological patterns. We emphasize that geographic variations such as those reported herein may be most readily explained with careful integration of contemporary influences and the dynamics of history.

Financial support for this project has come from a variety of sources. D. A. Kelt and J. H. Brown acknowledge NSF grants DEB-9221238 and DEB-9318096 (to J. H. Brown). E. J. Heske acknowledges the Eppley Foundation and the Na- tional Geographic Society. P. A. Marquet thanks FONDECYT Grant Number 0585, Biodiversity Support Program Grant Number 7578, the American Society of Mammalogists, Sigma Xi, the Lincoln Park Scott Neotropical Fund, and NSF INT- 9223313 (to J. H. Brown). S. R. Morton and J. R. W. Reid acknowledge Australian Geographic, the Australian Nature Conservation Agency, the Conservation Commission of the Northern Territory, SANTOS Limited, and the South Aus- tralian Department of Environment and Natural Resources are appreciated. K. A. Rogovin and G. Shenbrot acknowledge the Indian-Russian agreement on scientific cooperation for facilitating their work in the Thar Desert; the Mexico-USSR agreement on scientific cooperation for research in the lower Chihuahuan Desert; the scientific program of the Soviet- Mongolian Complex Biological Expedition for work in the Gobi Desert; and the special program of the Ministry of Sci- ence, Art, and Technology of Israel for Immigrant Scientists for Negev research. We gratefully acknowledge the able field assistance of L. Baker, R. Brandle, L. Doddridge, M. Gillam, J. Gillen, K. Jones, A. Kerle, and P. Masters in Australia, A. V. Surov and M. Idris in India, A. V. Surov in Mexico, D. V. Semenov in Mexico, and B. R. Krasnov in Israel. This project would never have been realized were it not for the initial efforts of M. A. Kurzius in organizing the North American data set.

Abensperg-Traun, M. 1994. The influence of climate on patterns of termite eating in Australian mammals and lizards. Australian Journal of Ecology 19:65-7 1.

Abramsky, Z., and C. Sellah. 1982. Competition and the role of habitat selection in Gerbillus allenbyi and Meriones tris- trami: a removal experiment. Ecology 63: 1242-1 247.

Alpers, C. N., and G. H. Brimhall. 1988. Middle Miocene climate change in the Atacama Desert, northern Chile: evidence from supergene mineralization at La Escondida. Geological Society of America Bulletin 100: 1640-1656.

Arroyo, M. T. K., F. A. Squeo, J. J. Armesto, and C. Vilagr6n. 1988. Effects of aridity on plant diversity in the northern Chilean Andes: results of a natural experiment. Annals of the Missouri Botanical Garden 75:55-78.

Axelrod, D. I. 1958. Evolution of the Madro-Tertiary geo- flora. Botanical Review 24:433-509.

Baverstock, P. 1984. Australia's living rodents: a restrained explosion. Pages 913-919 in M. Archer and G. Clayton, editors. Vertebrate zoogeography and evolution in Aus- tralasia. Hesperian Press, Victoria Park, Western Australia, Australia.

Baverstock, P. R., C. H. S . Watts, M. Gelder, and A. Jahnke. 1983. G-banding homologies of some Australian rodents. Genetica 60: 105-1 17.

Bowers, M. A,, and J. H. Brown. 1982. Body size and coexistence in desert rodents: chance or community structure? Ecology 6 3 : 3 9 1 4 0 0 .


small mammal assemblage in semiarid Chile. Ecology 77: 133-148.

Meserve, P. L., and D. A. Kelt. 1990. The role of aridity and isolation on central Chilean small mammals: a reply to Caviedes and Iriarte (1989). Journal of Biogeography 17: 681-689.

Mitchell, W. A,, Z. Abramsky, B. P. Kotler, B. Pinshow, and J. S. Brown. 1990. The effect of competition on foraging activity in desert rodents: theory and experiments. Ecology 71:844-854.

Morton, S. R. 1979. Diversity of desert-dwelling mammals: a comparison of Australia and North America. Journal of Mammalogy 60:253-264.

. 1985. Granivory in arid regions: comparison of Aus- tralia with North and South America. Ecology 66:1859- 1866.

. 1993. Determinants of diversity in animal com-munities of arid Australia. Pages 159-169 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological communities: historical and geographical perspectives. University of Chicago Press, Chicago, Illinois, USA.

Morton, S. R., and A. Baynes. 1985. Small mammal assemblages in arid Australia: a reappraisal. Australian Mam- malogy 8: 159-169.

Morton, S. R., J. H. Brown, D. A. Kelt, and J. R. W. Reid. 1994. Comparisons of community structure among small mammals of North American and Australian deserts. Aus- tralian Journal of Zoology 42:501-525.

Morton, S. R., and D. W. Davidson. 1988. Comparative structure of harvester ant communities in arid Australia and North America. Ecological Monographs 58:19-38.

Morton, S. R., and P. H. Davies. 1983. Food of the zebra finch (Poephila guttata), and an examination of granivory in birds of the Australian arid zone. Australian Journal of Ecology 8:235-243.

Morton, S. R., and E. J. Heske, editors. 1994. Comparative ecology of desert-dwelling small mammals. Australian Journal of Zoology 42:405-555.

Moss, G. L., and D. B. Croft. 1988. Behavioral mechanisms of microhabitat selection and competition among three species of arid zone dasyurid marsupials. Australian Journal of Ecology 13:485-493.

Orians, G. H., and R. T. Paine. 1983. Convergent evolution at the community level. Pages 431-458 in D. J. Futuyma and M. Slatkin, editors. Coevolution. Sinauer Associates, Sunderland, Massachusetts, USA.

Pavlinov, I. A,, A. Dubrovsky Yu, 0 . L. Rossolimo, and E. G. Potapova. 1991. Gerbils of the world fauna. Nauka Press, Moscow, Russia.

Pianka, E. R. 1985. Some intercontinental comparisons of desert lizards. National Geographic Research 1:490-504.

. 1986. Ecology and natural history of desert lizards. Princeton University Press, Princeton, New Jersey, USA.

Polis, G. A. 1991. The ecology of desert communities. Uni- versity of Arizona Press, Tucson, Arizona, USA.

Prakash, I. 1963. Zoogeography and evolution of the mammalian fauna of Rajasthan Desert, India. Mammalia 27: 342-35 1.

. 1974. The ecology of vertebrates of the Indian Des- ert. Pages 369-429 in M. S. Mani, editor. Ecology and biogeography in India. Dr. W. Junk bv, Publishers, The Hague, The Netherlands.

Reichman, 0 . J. 1991. Desert mammal communities. Pages 31 1-347 in G. A. Polis, editor. The ecology of desert communities. University of Arizona Press, Tucson, Arizona, USA.

Reichman, 0 . J., and J. H. Brown, editors. 1983. Biology of desert rodents. Great Basin Naturalist Memoirs 7.

Reichman, 0 . J., and M. V. Price. 1993. Ecological aspects of heteromyid foraging. Pages 539-574 in H. H. Genoways

and J. H. Brown, editors. Biology of the Heteromyidae. Special Publication, American Society of Mammalogists 10.

Reig, 0 . A. 1984. Distribu~go geogrifica e historia evolutiva dos roedores muroideos sulamericanos (Cricetidae: Sig- modontinae). Revista Brasileira de GenCtica 7:333-365.

. 1986. Diversity patterns and differentiation of An- dean rodents. Pages 4 0 5 4 3 9 in F. Viulleumier and M. Mon- asterio, editors. High altitude tropical biogeography. Ox- ford University Press, New York, New York, USA.

Ricklefs, R. E. 1987. Community diversity: relative roles of local and regional processes. Science 235:167-171.

Ricklefs, R. E., and D. Schluter. 1993a. Species diversity: regional and historical influences. Pages 350-363 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological communities: historical and geographical perspectives. University of Chicago Press, Chicago, Illinois, USA.

Ricklefs, R. E., and D. Schluter, editors. 19938. Species diversity in ecological communities: historical and geographical perspectives. University of Chicago Press, Chicago, Illinois, USA.

Riddle, B. R. 1995. Molecular biogeography in the pocket mice (Perognathus and Chaetodipus), and grasshopper mice (Onychomys): the Late Cenozoic development of a North American aridlands rodent community. Journal of Mammalogy 76:283-301.

Rogovin, K. A,, and G. I. Shenbrot. 1993. Structural aspects of community organization of terrestrial verterbrates: an example of Mongolian desert rodents. Uspechi Sovremen- noy Biologii (Advances in Current Biology) 113:198-222 (in Russian).

Rogovin, K. A,, and G. I. Shenbrot. 1995. Geographical ecology of Mongolian desert rodent communities. Journal of Biogeography 22:1163-1180.

Rogovin, K. A,, G. I. Shenbrot, and A. V. Surov. 1991. Anal- ysis of spatial organization of a desert rodent community in Bolson de Mapimi, Mexico. Journal of Mammalogy 72: 347-359.

Rogovin, K. A,, G. I. Shenbrot, A. V. Surov, and M. Idris. 1994. Spatial organization of a rodent community in the Western Rajasthan desert (India). Mammalia 58:234-260.

Rosenzweig, M. L., and J. Winaker. 1969. Population ecology of desert rodent communities: habitats and environmental complexity. Ecology 50:558-572.

Rundel, P. W., M. 0 . Dillon, B. Palma, H. A. Mooney, S. L. Gulmon, and J. R. Ehleringer. 1991. The phytogeography and ecology of the coastal Atacama and Peruvian deserts. Aliso 13: 1-49.

Schluter, D. 1986. Tests for similarity and convergence of finch communities. Ecology 67:1073-1085.

Schluter, D., and R. E. Ricklefs. 1993. Convergence and the regional component of diversity. Pages 230-240 in R. E. Ricklefs and D. Schluter, editors. Species diversity in ecological communities: historical and geographical perspectives. University of Chicago Press, Chicago, Illinois, USA.

Schmidly, D. J., K. T. Wilkins, and J. N. Derr. 1993. Bio- geography. Pages 319-356 in H. H. Genoways and J. H. Brown, editors. Biology of the Heteromyidae. Special Pub- lication, American Society of Mammalogists 10.

Shenbrot, G. I. 1992. Spatial structure and niche patterns of a rodent community in the south Bukhara desert (Middle Asia). Ecography 15:347-357.

Shenbrot, G. I., and K. A. Rogovin. In press. Temporal variation in spatial organization of a rodent community in the Southwestern Kyzylkum Desert (Middle Asia). Ecography.

Shenbrot, G. I., K. A. Rogovin, and E. J. Heske. 1994. Com- parison of niche-packing and community organization in Asia and North America. Australian Journal of Zoology 42:479-499.

Shenbrot, G. I., V. E. Sokolov, V. G. Heptner, and M. Ko-

7 60 DOUGLAS A. KELT ET AL. Ecology, Vol. 77, No. 3

walskaya Yu. In press. Mammals of Russia and adjacent Wadia, D. N. 1960. The post-glacial dessication of central regions. Dipodid rodents. Nauka Press, Moscow, Russia (in Asia. Monograph, National Institute of Science (India): 1- Russian). 25.

Sinitzin, V. M. 1962. Palaeogeography of Asia. USSR Acad- Wahlert, J. H. 1993. The fossil record. Pages 1-37 in H. H. emy of Sciences Press, Moscow, Russia (in Russian). Genoways and J. H. Brown, editors. Biology of the Het-

Singh, G. 1988. History of aridland vegetation and climate: eromyidae. Special Publication, American Society of Mam- a global perspective. Biological Reviews 63:159-195. malogists 10.

Stafford Smith, D. M., and S. R. Morton. 1990. A framework Walker, L. I., A. E. Spotorno, and J. Arrau. 1984. Cytoge- for the ecology of arid Australia. Journal of Arid Environ- netics and reproductive studies of two nominal subspecies ments 18:255-278. of Phyllotis danvini and their experimental hybrids. Journal

Thompson, R. S., and J. I. Mead. 1982. Late Quaternary of Mammalogy 65:220-230. environments and biogeography in the Great Basin. Qua- Webb, S. D. 1977. A history of savanna vertebrates in the ternary Research 17:39-55. New World. Part I: North America. Annual Review of Ecol-

Valone, T. J., and J. H. Brown. 1995. Effects of competition, ogy and Systematics 8:355-380. colonization, and extinction on rodent species diversity. Wells, P. V. 1979. An aquable glaciopluvial in the west: Science 267:880-883. pleniglacial evidence of increased precipitation on a gra-

Van Devender, T. R., and W. G. Spaulding. 1979. Devel- dient from Great Basin to the Sonoran and Chihuahuan opment of vegetation and climate in the southwestern Unit- Deserts. Quaternary Research 12:3 11-325. ed States. Science 204:701-710.

APPENDIX Species of small mammals tallied in seven deserts on four continents. Included are codes for trophic categories: C,

carnivore; F, folivore; 0, omnivore; G, granivore.

Trophic Trophic Species code Species code

North America North America (continued) Baiomys taylori 0 Peromyscus maniculatus 0 Chaetodipus baileyi G Peromyscus pectoralis 0 Chaetodipus fallax G Peromyscus truei 0 Chaetodipus hispidis G Reithrodontomys fulvescens 0 Chaetodipus intermedius G Reithrodontomys megalotus G Chaetodipus nelsoni G Reithrodontomys montanus 0 Chaetodipus penicillatus G Sigmodon arizonae F Dipodomys deserti G Sigmodon hispidus F Dipodomys merriami G Sigmodon ochrognathus F Dipodomys microps Dipodomys nelsoni Dipodomys ordii Dipodomys panamintinus Dipodomys spectabilis Lagurus curtatus

F G G G G F

South America Akodon albiventer Akodon andinus Akodon boliviensis Akodon sp.

0 0 0 0

Microdipodops megacephalus G Andinomys edax F Microdipodops pallidus Microtus longicaudus Microtus montanus

G F F

Auliscomys boliviensis Auliscomys pictus Auliscomys sublimis

0 F F

Neotoma albigula Neotoma lepida Neotoma micropus Onychomys leucogaster Onychomys torridus Perognathus amplus Perognathus javus Perognathus formosus Perognathus longimembris Perognathus parvus Peromyscus boylii Peromyscus crinitus Peromyscus eremicus Peromyscus leucopus

F F F C C G G G G G 0 0 0 0

Bolomys amoenus Bolomys lactens Calomys lepidus Calomys sorellus Chinchillula sahamae Chroeomys jelskii Eligmodontia typus Phyllotis xanthopygus Neotomys ebriosus Octodontomys gliroides Phyllotis magister Phyllotis osilae Punomys sp. Thylamys elegans

0 0 0 0 F 0 0 0 F F F F 0 C

April 1996 DESERT SMALL-MAMMAL COMMUNITY STRUCTURE

APPENDIX. Continued

Trophic Species code

Turkestan Alactodipus bobrinski Allactaga elator Allactaga major Allactaga severtzovi Allocricetulus eversmani Cricetulus migratorius Dipus sagitta Meriones libicus Meriones meridianus Meriones tamariscinus Paradipus ctenodactylus Pygeretmus platiurus Pygeretmus pumilio Spermophilopsis leptodactylus Stylodipus telum

Gobi Allactaga balikunica Allactaga bullata Allactaga sibirica Allocricetulus curtatus Cardiocranius paradoxus Cricetulus migratorius Cricetulus obscurus Dipus sagitta Eolagurus przewalskii Euchoreutes naso Meriones meridianus Meriones unguiculatus Mus musculus Phodopus roborovskii Pygeretmus pumilio Salpingotus crassicauda Salpingotus kozlovi Stylodipus andrewsii

Negev Acomys cahirinus Acomys russatus Eliomys rnelanurus Gerbillus dasyurus Gerbillus gerbillus Gerbillus henleyi Jaculus jaculus

Soecies Trophic

code

Negev (continued) Meriones crassus Mus musculus Psammomys obesus Sekeetamys calurus

Thar Funnambulus pennanti Gerbillus gleadowi Gerbillus nanus Meriones hurrianae Millardia gleadowi Mus booduga Mus musculus Mus platitrix Rattus cichikos Tatera indica

Australia Dasycercus cristicauda Dasyuroides byrnei Leggadina forresti Mus domesticus Ningaui ridei Ningaui yvonnae Notomys alexis Notomys cervinus Notomys fuscus Notomys rnitchellii Notoryctes typhlops Planigale gilesi Planigale tenuirostris Pseudantechinus macdonnellensis Pseudomys australis Pseudomys desertor Pseudomys herrnannsburgensis Rattus villosissimus Sminthopsis crassicaudata Sminthopsis dolichura Sminthopsis hirtipes Sminthopsis laniger Sminthopsis macroura Sminthopsis ooldea Sminthopsis psammophila Sminthopsis youngsoni

COMMUNITY STRUCTURE OF DESERT SMALL MAMMALS: … · INTRODUCTION northern Africa (Sahara, Arabian,...

Documents

Transcript of COMMUNITY STRUCTURE OF DESERT SMALL MAMMALS: … · INTRODUCTION northern Africa (Sahara, Arabian,...