Download Species Diversity of Seed-Eating Desert Rodents in

Species Diversity of Seed-Eating Desert Rodents in Sand Dune Habitats
Author(s): James H. Brown
Source: Ecology, Vol. 54, No. 4 (Jul., 1973), pp. 775-787
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/1935672 .
Accessed: 20/06/2011 15:26
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=esa. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.
http://www.jstor.org
SPECIES DIVERSITY OF SEED-EATING DESERT RODENTS
IN SAND DUNE HABITATS'
JAMES
H.
BROWN
Department of Biology, University of Utah, Salt Lake City 84112
Abstract. The seed-eating rodent faunas were sampled on 18 dunes of semistabilized, winddrifted sand in eastern California, Nevada, and western Utah. These dunes were similar in
general appearance and in the form and spacing of vegetation, yet the number of common
species of granivorous rodents ranged from 1 to 5. The faunas of all dunes were characterized
by a regular spacing of body sizes, and the difference in size between coexisting species was
inversely related to the number of species present. Differences in species diversity between dunes
can be attributed to ecological, biogeographic, and evolutionary processes. A common pool of
species had equal opportunity to colonize 13 of the dunes. The numbers of species inhabiting
these dunes are determined primarily by ecological factors. Species diversity is most closely
correlated with the predictable amount of annual rainfall, the best available estimate of the
abundance of seeds. At least three dunes apparently are biogeographically depauperate; biogeographic barriers have prevented colonization of particular species that probably could
exist on those dunes. Two of the dunes occur at high altitudes and these appear to be depauperate in an evolutionary sense; although these dunes seem to be capable of supporting additional rodent species, it is likely that no existing species possess the adaptations necessary to
colonize them.
INTRODUCTION
The number of species that occur in a particular
habitat depends on the interaction between ecological, biogeographic, and evolutionary processes. Speciation and immigration provide species with the
opportunity to colonize the habitat. The ability of
these potential colonists to exist and coexist in the
habitat determines the number and identity of species
which actually will be found there. In order to understand how these processes influence species diversity,
it is necessary to study natural communities where
the effects of specific variables can be analyzed by
observation or experimental manipulation. The works
of MacArthur and MacArthur (1961), MacArthur
(1964), Paine (1966), Pianka (1967), Sanders
(1969), Rosenzweig and Winakur (1969), Janzen
(1970), and others are encouraging beginnings, but
of these authors only Pianka has attempted to analyze simultaneously the effects of ecological, biogeographic, and evolutionary variables on the diversity of species in continental habitats. Pianka concluded that ecological parameters (particularly spatial heterogeneity of vegetation) were the most important determinants of species diversity of North
American lizards in desert flatland habitats. More
studies on other groups of organisms are essential if
we hope to discover general patterns of variation in
species diversity and to understand the mechanisms
which produce these patterns.
In this paper I describe factors that determine the
number of species of seed-eating rodents found in
sand dune habitats in the deserts of the southwestern
United States. These rodent communities are ideal
'Received April 7, 1972; accepted September 11, 1972.
for such a study. Like most desert ecosystems, sand
dunes have a small number of rather sparsely distributed plant and animal species; relative to the rest
of the biota, rodents are abundant and diverse, and
most of them are ecologically similar in that they
spend the day in burrows and emerge at night to
forage for the dry seeds which comprise the bulk of
their diets. From one to five or more species of seedeating rodents are found on sand dunes in the North
American deserts. Sand dunes are distinctive, geographically widespread habitats remarkably similar
in general characteristics and appearance, although
they vary greatly in parameters (such as size, degree
of isolation, climate, and number and identity of
plant species) which might be expected to affect
rodent species diversity. This paper discusses the
extrinsic ecological, biogeographic, and evolutionary
factors that determine the diversity of seed-eating
rodent species. A second paper will describe food
resource utilization and other mechanisms that enable these ecologically similar species to coexist
(Brown and Lieberman 1973).
HABITATS,
RODENTS,
METHODS
Sand dunes occur as widely dispersed patches of
distinctive habitat in the arid areas of western North
America. A special combination of soil, wind, and
topography is required for the formation of dunes.
They occur most frequently in sheltered areas on the
leeward side of dry lakes. Only large, semistabilized
dunes were chosen for the present study. These consist of hills of wind-drifted sand held in place by
isolated clumps of shrubby vegetation. Most of these
plants, regardless of their species, are remarkably
JAMES H. BROWN
776
cause of their wide geographic distribution the habitats differed in climate, particularly in the severity of
the winters and the amount of precipitation. There
were some differences in size and composition of the
sand particles, and this, together with the climate, apparently accounted for differences in vegetation between the dunes. There was considerable variation in
the identity, species diversity, and density of the perennial shrubs, but less variability in their size and
shape. In general the northern (Great Basin Desert)
dunes received more rainfall and had greater plant
species diversity than the southern dunes of the Mojave and Colorado deserts. Sarcobatus baileyi and
A triplex canescens tended to be the most abundant
species of plants on the northern dunes, and if any
species could be said to predominate on the southern
dunes it was Larrea divaricata.
14
5o~~~~~~~~~~~'
T84/ [SER
CREA
r
OS
6~~~~~
j. 7
18
Ecology, Vol. 54, No. 4
17
The seed-eating rodents
Fid 1. Map of southwestern United States showing
sand dunes sampled in relation to the major desert areas
( stippled ) and the geographic ranges of D . merriami
(enclosed by heavy dashed line) and D. deserti (light
dashed line) . Open circles (sites 1-13 ) indicate dunes
to which approximately equal numbers of species have
had access; solid circles (14-16) represent dunes in iso-
lated desert basins to which some species have not had
biogeographic access; solid triangles ( 17-18 ) denote
dunes located at elevations above the attitudinal ranges
of several species.
similar in form and height; their vegetative parts tend
to be roughly hemispherical in shape and between
15 and 125 cm in height. Between these shrubs are
extensive areas of bare, wind-drifted sand. Occasionally after one or more good rains small herbs and
grasses grow there, but most of these are ephemeral
and within a few months they die and are uprooted
by the wind or covered by drifting sand.
I have
sampled
the
rodent
communities
of
18
se1istabilized sand dune habitats in an area of the
southwestern
United
States which
includes
extensive
and Great Basin
wn-fed
,
portions of the Coloradob
deserts (Fig. 1, Table 1). The wind-drifted sand and
the form, height, and spacing of the perennial plants,
made all of these dunes quite similar in general appearance (Fig. 2), but they differed in many respects.
The dunes varied in size, from dune 2, which covered
only about 0U25 kM2, to dune 12, which encompassed
an area of hundreds of square kilometers. However,
these were exceptional and the other dune habitats
ranged in size from approximately 1 to 20 km2. Be-
Sampling methods.-The rodent fauna of the dunes
was sampled by trapping with "Museum Special"
dead traps. I had hoped that these traps would sample all species with approximately equal efficiency;
however, some individuals of Dipodomys desert, the
largest species, were able to break away from them.
The traps were placed in sets of four (positioned as
described by Brown and Lieberman 1973) spaced
approximately 10 m apart, in and around the perennial shrubs. Dunes were normally trapped for two
successive nights during the summer months; at least
140 traps were set each night and all traps were
moved to new locations on the second night. Most
of the dunes were sampled only once, but dune 10
was sampled twice and dune 7 four times during different years and different seasons. Although absolute
and relative abundance of the rodent species fluctuated somewhat, species composition and diversity
appeared to remain relatively constant.
The species.-During the present study 18 species
of rodents were captured (Table 2). Three of these
species are not primarily granivorous and have been
excluded from the analyses. The grasshopper mice
(Onychomys torridus and 0. leucogaster) are largely
carnivorous, preying on arthropods and less frequently on small vertebrates. The antelope ground
squirrel (Ammospermophilus leucurus) is an omnivore which feeds on green vegetation, arthropods,
small vertebrates, and carrion as well as seeds. The
remaining 15 species are largely granivorous and their
populations obviously fluctuate in response to the
abundance of seed crop. Most of the species and
individuals belong to the family Heteromyidae, the
group of North American rodents that shows the
most obvious specializations for desert life. The genera, Dipodomys (kangaroo rats), Microdipodops
(kangaroo mice), and Perognathus (pocket mice),
SPECIES DIVERSITY
Summer 1973
TABLE
777
IN DESERT RODENTS
1. Location, sampling effort, and species diversity for the sand dune localities
Rodent species
Nevada, Humboldt Co., Pronto, 29
km WSW Winnemucca, 1290 m.
Nevada, Pershing Co., 15 km NE
Lovelock, 1290 m.
Nevada, Pershing Co., 34 km WNW
Lovelock, 1250 m.
Nevada, Pershing Co., 11 km S
Lovelock, 1190 m.
Nevada, Churchill Co., Sand
Mountain, 37 km SE Fallon, 1260 m.
Nevada, Mineral Co., 21 km S Mina,
1370 m.
Nevada, Esmeralda Co., Fishlake
Valley, 21 km. N Dyer, 1490 m.
1
2
3
4
5
6
7
California, Inyo Co., Owens Lake,
7 km NW Keeler, 1190 m.
California, San Bernadino Co., 25
km SW Baker, 430 m.
California, San Bernadino Co., Kelso
Dunes, 8 km S Kelso, 670 m.
8
9
10
California, Riverside Co., 3 km NE
Indio, 20 m.
California, Imperial Co., Algodones
Dunes, 8 km W Glamis, 90 m.
California, San Diego Co., 6 km
ENE Borrego Springs, 150 m.
Nevada, Elko Co., 11 km NE
Montello, 1490 m.
Utah, Tooele Co., 2 km NE Dugway
Proving Ground, 1460 m.
Nevada, Nye Co., 13 km S Currant,
1460 m.
Utah, Kane Co., Coral Pink Sand
Dunes State Park, 16 km W Kanab,
1830 m.
California, Mono Co., NE shore of
Mono Lake, 1980 m.
11
12
13
14
15
16
1718
aH
Date
Location
Dune
=
-Zpi
loge pi where pi are proportions
Sample
trap
nights
Aug. 3-4,
1970
Aug. 1, 1970
July 30-31,
1970
July 13-14,
1971
Aug. 5-6,
1970
Sept. 1-2,
1970
Sept. 5-13,
1970
May 29-30,
1970
July 15, 1968
Feb. 19-20,
1968
Sept. 13-14,
1970
Mar. 27-28,
1971
May 8-9,
1970
May 1969
July 19-20,
1971
July 21-22,
1971
April 23-24,
1971
Aug. 19-20,
1971
Aug. 16,
1971
Aug. 30-31,
1971
Aug. 9-10,
1971
Sept. 3-4,
1970
Individuals
caught
Total
Common
(> 5%)
Diversity(H)
280
41
8
5
1.80
140
59
6
5
1.41
280
50
5
5
1.46
300
23
5
4
1.42
280
49
7
4
1.33
280
69
7
4
1.12
1050
114
6
5
1.47
400
200
94
50
7
5
5
4
1.54
1.40
250
42
5
5
1.50
300
30
3
2
0.74
300
10
3
3
0.90
300
120
29
12
2
1
1
1
0.15
0.0
300
2
1
1
0.0
300
4
1
1
0.0
300
9
1
1
0.0
280
26
3
3
0.54
140
14
3
3
0.66
280
45
4
2
0.78
280
21
5
3
1.30
280
61
7
4
1.33
of each species in the sample.
have efficient kidneys that enable them to maintain
water balance on relatively dry diets, and fur-lined
cheek pouches used for collecting and transporting
seeds. In addition the kangaroo rats and kangaroo
mice are saltatorial and largely bipedal, which apparently enables them to forage efficiently over large
areas and to avoid predators in open terrain. The
next most diverse group of rodents belongs to the
family Cricetidae (the main family of New World
rats and mice), represented by two genera, Peromyscus (deer mice; to avoid confusion with Perognathus, Peromyscus will not be abbreviated) and
Reithrodontomys (harvest mice). A single species of
chipmunk (Eutamias minimum), a member of the
family Sciuridae (squirrels and ground squirrels),
was taken on one dune.
With the exception of Eutamias minimum,which is
diurnal and frequently climbs and forages in the
branches of desert shrubs, the seed-eating rodents
are quite similar in habits and general ecology. They
spend the day inactive in burrows in the sand, and
emerge at dusk to forage on the surface of the
ground. Although some seeds are harvested from the
plants during the fruiting season, through most of
the year seeds are collected individually from the
substrate. When seeds are abundant large numbers
are collected and stored in burrows or in shallow
caches in the sand. All species except those of the
genus Dipodomys may enter torpor or hibernation
during periods of food scarcity or climatic severity.
None of the species are restricted in their distributions to sand dune habitats, although D. deserti, M.
Ecology, Vol. 54, No. 4
JAMES H. BROWN
778
,.
,.!:Substant,.)
..:......
.. .
:.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.:
. : : : , : . ... . . : ' : .:
. . . ..
:.:
...
:
*
~~~~~~~~~~
... ::.... ::.:.
::.
..
...
....'.....
..
....
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.:.
........
como
spce
of sedetn
roens
Blow
dun 8 in eatr
Clifri
whreto
..
seiswrcomn
:
SPECIES DIVERSITY
Summer 1973
TABLE
Species
Dipodomys deserti
D. panamintinus
D. microps
D. ordi
D. merriani
Microdipodops pallidus
M. megacephalus
Perognathus parvus
P. penicillatus
P. longimembris
Peromyscus maniculatus
Peromyscus crinitus
Peromyscus truei
Reithrodontomys megalotis
Eutamias minimus
Total seed-eating rodents
Onychomnys torridus
0. leucogaster
Anunospernmophilus leuicuruis
Total rodents
Body
weight
(g)
100.6
80.4
58.5
48.9
37.6
12.5
12.0
16.9
12.8
7.1
18.1
15.8
23 .8
8 .6
33 .5
779
IN DESERT RODENTS
2. Composition of the rodent samples of the sand dune localities
Dune
1
2
3
4
5
6
7
8
9
3
1
8
8
4
2
21
20
1
10
40
11
12
13
2
4
9
14
15
16
17
18
1
1
1
9
14
11
6
10
7
5
2
29
6
47
7
6
1
98
68
22
9
11
1
22
10
8
1
17
1
7
29
3
12
2
6
1
4
6
26
21
5
3
4
6
1
3
2
2
1
67
38
1
1
21
2
9
1
41
1
59
6
42
I
66
50
4
23
10
1
3
49
4
69
2
1
299
2
30
2
301
I
33
1
2
2
10
2
41
1
2
4
9
26
14
45
3
21
2
3
61
4
54
33
53
71
pallidus, and P. penicillatus are normally found only
on friable soils. The species range in body size from
P. longimembris, which weighs about 7 g, to D. deserti, approximately 100 g. A variety of mammalian,
avian, and reptilian predators are known to take these
rodents, but predation does not appear to have much
effect on the rodent populations.
Measurement of species diversity.-All students of
species diversity are faced with the problem of how
to measure or represent diversity. The usual solution
to this problem is either simply to count the total
number of species present or to use some diversity
index that gives the moderately abundant species in
a sample more weight than the rare ones. Because I
was interested in the number of ways the seed resources were utilized by rodents in each habitat, my
initial inclination was to use simply the total number
of species. However, it was obvious that samples
from some dunes contained one or more species
so rare that they were either transients dispersing
through the habitat or resident species so scarce that
they were not important constituents of the community. These rare species were excluded by counting
only those species, except for D. desert and D. panamintinus, which comprised 5% or more of the individuals of seed-eating rodents in a sample. Dipodomys deserti, the largest species, was counted whenever it was present for three reasons: ( 1) as one
would expect from its size, its populations were often
less dense than those of smaller species; (2) as mentioned earlier, some individuals escaped from the
traps, so that, relative to the smaller species, D.
deserti was underrepresented in the samples; (3) D.
deserti occurs only on sandy soils, and there is little
doubt that it is a permanent and important member
Total
123
1
8
100
192
114
29
4
6
109
140
2
1
21
3
853
37
4
2
12
42
2
4
9
26
14
48
25
61
896
of any dune community where it is present. D. panamintinus, which occurred on only one dune, is almost
as large as D. deserti and was treated similarly. I
feel that the number of common species derived as
described above is an unambiguous and biologically
reasonable measure of species diversity. However,
it should be mentioned that I have computed Shannon-Wiener diversity indices (H = -1pi loge pi;
where pi is the proportion of the sample that are
members of species i) for the faunas of all of the
dune habitats (Table 1) and these indices are closely
correlated (r - 0.94) with the number of common
species.
Environmental parameters
climatic data used for each dune
Climate.-The
locality are those collected at the nearest U.S.
Weather Bureau station at a similar elevation. Proximity to a weather station was one criterion used in
the selection of dunes. All available records since
1951 were used to provide data for the analyses; in
all cases the statistics used are based on 14 to 20
years of complete records.
Perennial shrubs.-Measurements of plant density,
foliage height diversity, and plant species diversity
were obtained by the following sampling procedure.
Either 5 or 10 straight-line transects, each 25 m in
length, were dropped at random in the habitat sampled by trapping. The identity and height of each
plant touching the line was recorded. Because the
seed-eating rodents are primarily terrestrial, a logarithmic series of height categories (0-8, 8-16, 1632, 32-64, 64-128, > 128 cm) was used for the
analysis of foliage height diversity. The ShannonWiener diversity index was used to measure foliage
height diversity and plant species diversity. It is im-
JAMES H. BROWN
780
portant to emphasize that all measurements of plant
parameters are for perennial shrubs only. Significant
numbers of herbs and grasses were present only on
those dunes where there had been good recent rains,
and these were ignored.
Soils.-Measurements of soil hardness and grain
size distribution were obtained for most of the dunes.
It was found that they showed little variability and
bore no relation to rodent species diversity, so they
were discontinued.
RESULTS AND DISCUSSION
Distribution of species and body sizes
Each of the dunes sampled was inhabited by from
I to 5 common species of seed-eating rodents (Table
1). The first insights into the determinants of diversity are provided by the recurring patterns of species
composition and body size distribution which characterized the faunas of the dunes.
Seven species. D. desert, D. ordi, D. merriami,
M. pallidus, P. longimembris, Peromyscus inaniculatus, and R. megalotis, were common on four or
A
I rj
6
.6
2
r^
56
m
r`1 r
Jo
i6
6 rL r~l m
m
o6
r^
v6 ii
_6
rr
8 16
w
31012 -
13
14-
v6
v
6
16- r^
17
18-
r1,
Wm
,r%,
m6
r4m
A
^
do
di 1,
t
BODYWEIGHT
(grams)
FIG. 3. Distribution of body sizes in the seed-eating
rodent faunas of the various sand dunes. Shaded symbols
represent common species; unshaded symbols, rare species. The width of the symbols is simply for illustration
and does not represent variation in body weight. The
body weights indicated are the means for all individuals
of each species from all localities. Brown and Lieberman
( 1973 ) found no evidence of geographic variation in
body weight (character displacement) in relation to number or identity of coexisting species. Species identities
can be read from body weights and abundances in Table
2.
Ecology, Vol. 54, No. 4
more of the dunes. These species formed the basic
pool from which the faunas of the dunes were derived. They range in body size from 7 to 100 g. One
of the most conspicuous features of the faunas of the
dunes was the remarkably regular spacing of body
sizes which occurred on each dune (Fig. 3). With
remarkably few exceptions, common species that coexisted on the same dune differed in weight by a
ratio (larger/smaller) of at least 1.5. Species differing in weight by a ratio of less than 1.5 occurred
together less frequently than expected if coexistence
were independent of body size, and pairs of species
for which the weight ratio is greater than 1.5 coexisted more frequently than expected by chance
(;t,2 = 7.24; P < 0.01). Also, the difference between
pairs of adjacent species in the body size spectrum
was inversely related to the number of species present. Thus when the number of common species inhabiting a dune was 5, 4, 3, or 2, the adjacent pairs
of species differed in size by a ratio which averaged
2.0, 2.3, 2.7, or 2.7 respectively. Even those species
which were rare (less than 5 percent of the total
sample) tended to fall within size ranges not represented by other species on that dune.
The ability of rodent species to coexist in sand
dune habitats appears to depend on their being of
quite different sizes, and the fewer species a dune
can support, the more different in size they must be.
These patterns suggest that coexistence is achieved
by the division of some environmental resource on
the basis of size. The most reasonable hypothesis is
that the rodents differentially utilize the sizes of seeds
available in their habitats. We have tested and verified this hypothesis (Brown and Lieberman 1973);
small rodents feed on small seeds and large ones feed
on large seeds. However, there is a great deal of
overlap in the sizes of seeds taken by pairs of coexisting species of similar size, and subdivision of
the seed resources on the basis of size alone is probably not sufficient to account for coexistence.
The faunas of the dunes also showed regular patterns of species composition (Table 2). On this basis
the dunes could be divided into two groups. Dunes
14 to 18 were similar in that their faunas consisted
of D. ordi, Peromyscus maniculatus, and sometimes
one or two additional species. All of these 5 dunes
lie outside the geographic ranges of D. merriami and
D. deserti (Fig. 1), and their faunas are quite different from those of the other dunes, which occur
within the ranges of these two species. Dunes 1 to
13 were all inhabited by D. deserti and, if two or
more species were common, by D. merriami (however, on dune 5 D. merriami, although present, was
rare). These were also the only dunes where M. pallidus occurred, and it was found on all but two of
the dunes within its geographic range. Later I will
Summer 1973
SPECIES DIVERSITY
IN DESERT RODENTS
show that these two groups of dunes, which could
be distinguished on the basis of differences in species composition, also differed in other respects related to rodent species diversity. Biogeographical or
evolutionary explanations are required to account for
the differences in the composition and diversity of
rodent species between dunes 1 to 13 on the one
hand, and 14 to 18 on the other.
Two other patterns were apparent as a result of
simply examining the rodent faunas of the dunes.
First, on dunes I to 13 species were absent or
"dropped out" in a fairly regular sequence as rodent
species diversity decreased. Reithrodontomnys mnegalotis and D. ordi were the first species to drop out;
neither was found on dunes with fewer than four
species. These were followed by M. pallidus, Peronyscus naflniculaltus,and P. Iongimemubris. Finally
D. m'nerriatni
dropped out, leaving only D. deserti, the
largest species, which was found on all of the dunes.
Second, in several cases a species that was absent was
-replaced" by another species of similar size. The
most spectacular cases involved three species weighing 12 to 13 g. Microdipodops pallidus was found
on most of the dunes within its geographic range,
where it was the only species of this size, but on three
dunes outside its range other species of the same size
were found; M. inegacephalus occurred on dunes I
and 18 and P. pen icillatus was present on dune 9.
(80 g) apparently "replaced" D.
D. punainintinus
(lcserti (100 g) on dune 18. A similar situation involved P. ion ginenembris (7.1 g) and R. inegalotis
(8.6 g). which were never common on the same
dune. These observations support the previous genTABLL
eralizations about the influence of body size on the
composition of these desert rodent communities. Apparently the coexistence of seed-eating rodents in
sand dune habitats depends more on the body sizes
of species than on the identity of the species.
Ecological determinants of species diversity
Dunes I to 13 lie within the altitudinal and geographic ranges of P. ion gitnemnbris, R. inegalotis,
Peroinyscus inaniculatus, D. inerriatni, and D. deserti. Even when these species were absent from
dunes they were usually abundant in habitats adjacent to them or only a few miles away. For example,
P. ion gimnembrisand D. inerrialni were abundant on
the harder soils immediately surrounding many of
the dunes in southern California which they did not
inhabit. Microdipodops pallidus and D. ordi, which
were common on several of the dunes in the Great
Basin, do not have geographic ranges that include
the dunes in the Mojave and Colorado deserts of
southern California. However, other species of the
genera Perognath us, Dipodomnys, and Peromnyscus
occur in the deserts of Southern California but not
in the Great Basin. These species have had the biogeographic potential to colonize the southern dunes
and one of them has done so: P. pen icillatus was
common on dune 9, where it was apparently a geographic "replacement" for Al. pallidus. I conclude
that at least six species, including a wide spectrum
of body sizes, have had the opportunity to colonize
all of these 13 dunes. The absence of some of these
species from several of the dunes must be attributed
to ecological characteristics of the dunes, rather than
3. Some characteristics of the sand dune habitats
Annual precipitation (mm)
Common
Perennial plants
--
-
Dune
rodent
species
.
1
2
3
4
5
6
7
8
9
I)
11
12
13
14
15
16
17
18
5
5
5
4
4
4
4.75
2
3
1
1
1
1
3
3
2
3
4
191
138
138
138
119
114
120
125
58
58
69
52
80
182
170
105
302
321
-a
-2
131
91
91
91
77
73
66
57
23
23
28
21
37
119
116
52
214
203
72
45
45
45
36
31
11
-11
-12
-12
-14
-10
- 6
56
62
- 1
125
85
a
-
-
Sand
_--
Species
PSDa
Density',
FHDc
Hardness1l
6
6
4
4
4
3
3
3
3
3
3
3
2
3
4
1
4
5
1.36
1.03
1.04
0.86
1.26
0.57
0.62
0.76
0.66
0.35
1.00
0.48
0.66
0.95
1.22
0
1.22
1.18
7.6
4.6
3.4
5.0
2.6
7.0
4.9
10.5
5.8
8.6
3.4
3.7
1.6
4.0
12.2
4.2
5.6
8.0
0.67
0.95
0.45
0.83
1.03
0.96
1.18
1.07
1.15
0.65
0.88
0.73
1.03
1.26
1.23
1.06
1.55
1.00
2.16
2.14
2.31
2.72
2.67
3.74
2.82
2.79
2.60
aPlant species diversity = -) pi loge pi.
of plants intersected by a straight line 25 u-. in lengtlh
eFoliage height diversity - -"pj loge pi; heigiht categories as in text
'1100m/mi penetration by stand.trd pointed metal stake
e lmesh size; measured with gr.aded series of sieves
I Numibcr
781
2.72
3.25
2.85
2.63
2.05
Grain
sizeO
2.99
3.39
3.64
3.35
3.61
3.30
3.94
3.67
3.07
--
3.16
3.69
3.18
JAMES H. BROWN
782
to biogeographic barriers that have prevented colonization.
If we consider only these 13 dunes where approximately equal numbers of species have had the opportunity to colonize, we should discover the environmental parameters that determine how many rodent
species they can support. Several variables have been
tested for their relationship to rodent species diversity; these include dune size, soil characteristics,
measurements of the amount and predictability of
precipitation, diversity of plant species, plant density,
and foliage height diversity (Table 3). These were
chosen on the basis of my own intuition or because
other students of desert rodent ecology (especially
Rosenzweig and Winakur 1969) suggested that they
influence species diversity. Somewhat surprisingly.
the number of rodent species was independent of dune
size. In fact, the smallest dune (dune 2) was inhabited by five common and one rare species, and only
a single species was present on the largest dune (dune
12). This lack of correlation can be attributed to
the fact that none of the dune-dwelling rodents are
found only in sand dune habitats. Therefore, although
the dunes have a patchy distribution, most species
have ready access to them and diversity is not related to area as it is on islands. Neither soil particle
size, soil strength, vegetation density, nor foliage
height diversity were correlated with the number of
common rodent species.
Those environmental parameters with which the
number of common rodent species were correlated,
were measures of annual precipitation and plant species diversity. The number of rodent species was
equally well correlated (r = 0.84, Fig. 4) with two
measures of the predictable amount of annual rain5
c0
0
0
n 4-
o00
0
A*-203
lii
Ecology, Vol. 54, No. 4
fall (K -- a and x - 2a of annual precipitation, or the
amount of rain which can be expected 5 years out of
6 and 39 years out of 40 respectively). This correlation was somewhat better than that relating the number of rodent species to mean annual rainfall (r =
0.80). The number of rodent species was also questionably correlated with plant species diversity (r =
0.56). However, when multiple regression analysis
was used to eliminate the variability in rodent species
diversity owing to rainfall, then there was no relation
On the
-0.17).
to plant species diversity (rrz.,
other hand, when the variability owing to plant species diversity was removed, the number of rodent
species was still correlated with x - a of annual rainfall (r,. - 0.79). This suggests that rainfall influences the diversity of both seed-eating rodents and
perennial shrubs, but that the diversity of shrubs has
no direct effect on the diversity of rodents.
The apparent effect of precipitation on the diversity of seed-eating desert rodents is not as surprising
as it seems at first. Germination and seed production
by desert annuals and seed production by perennials
is known to be highly dependent on rainfall (Went
1948, 1955, Went and Westergaard 1949, Beatley
1967 and 1969). Thus, the amount and predictability of annual rainfall should provide an accurate
estimate of the size and predictability of the annual
seed crop, which it was impractical to measure directly. Diversity of coexisting seed-eating rodents in
these sand dune habitats apparently depends largely
on the annual production of seeds in the habitat. The
diversity of rodents is also influenced by the unpredictable nature of desert rains and the fact that in
years of low precipitation there is little or no seed
production. Many areas of the Mojave and Colorado
Deserts receive sufficient precipitation to produce a
good crop of annuals only every 2nd or 3rd year
on the average. For example, in 2 years out of the
last 20, Brawley, California, the Weather Bureau station nearest dune 12, received less than 2.5 mm of
precipitation.
aZ
C,,
~. 0000
0
150
100
50
PRECIPITATION
(millimeters)
X-oa of ANNUAL
FIG. 4. Relation of rodent species diversity to a measure of the predictable amount of precipitation-in this
case the mean minus one standard deviation of annual
precipitation,or the amount of rain that can be expected
5 years out of every 6. Symbols as in Fig. 1.
0
It is not difficult to account for the observed variability in rodent species diversity as a function of
the abundance and predictability of the seed crop.
Coexisting seed-eating rodents on these sand dunes
apparently subdivide the seeds and avoid competitive
exclusion mostly by feeding on seeds of different
sizes and by foraging in different places (Brown and
Lieberman 1973). As the seed production of a habitat decreases, the availability of seeds of appropriate
sizes and in appropriate places for a particular rodent
species also decreases, and eventually some threshold
is reached below which that species can no longer
persist in the habitat. The exclusion of species from
a habitat as a result of diminution of resources need
not depend on interspecific competition, but com-
Summer 1973
SPECIES DIVERSITY
petition between species will facilitate their exclusion.
Competition does seem to have an important effect
on the coexistence of seed-eating rodent species on
sand dunes. The best evidence for this is the variation
in body sizes as a function of species diversity; the
difference in body size between the most similar
pairs of species increases as the number of coexisting
species decreases (Fig. 3).
The predictability of the seed crop seems to affect
rodent species diversity by determining the minimum
levels of seed availability. Thus the mean annual
rainfall minus some measure of its variability (standard deviation) is highly correlated with the number
of rodent species, and measures of variability per se
show no relation to rodent species diversity. Populations of seed-eating desert rodents apparently reach
minimal levels after seasons of low rainfall and little
seed production. If the population of a species in
some habitat is so low that it frequently becomes
locally extinct in years of low productivity, then that
species should evolve habitat selection to avoid recolonization of that habitat, even during periods of
high productivity. Thus it is the minimum predictable
level of resources which should (and apparently
does) have the greatest effect in determining the
number of species which can coexist in a habitat.
Species that occur in several types of habitat in
areas where resources are abundant should inhabit
fewer kinds of habitats as resources diminish. This
is precisely what happens to the desert rodents. In
the high-rainfall (and presumably productive) Great
Basin Desert species such as P. longimembris, R.
megalotis, Peromyscus maniculatus, and D. inerriami
are common in a number of habitats including sand
dunes. In the Mojave and Colorado Deserts, which
receive less precipitation and presumably produce
fewer seeds, these species still occur but they are
restricted to a much smaller range of habitats and
are absent from nearly all of the dunes, where D.
deserti is usually the only common species. In the
Mojave and Colorado Deserts the low diversity of
rodent species within a single habitat type (ac diversity) is accompanied by a high degree of habitat specificity and a relatively high turnover of species between habitats (P diversity). The net result is that
areas of the Colorado and Mojave Deserts large
enough to contain many kinds of habitats have about
the same number of species of seed-eating desert
rodents as comparable areas in the Great Basin,
where the diversity of species within single habitat
types is several times greater.
Biogeographic and evolutionary influences
on species diversity
The diversity of seed-eating rodents on five dunes
(14 to 18) was not discussed in detail in the pre-
IN DESERT RODENTS
783
vious section because there was good evidence that
several rodent species occurring on similar dunes in
other areas had not had the opportunity to colonize
these five. Something other than the local ecology
of the dune habitats seems to be required to explain
the composition of the rodent faunas of these dunes.
Three of these dunes (14, 15. and 16) are similar
in latitude, altitude, climate, and vegetation to the
13 dunes discussed in the previous section. Although
dunes with comparable rainfall and plant species diversity normally supported four or five rodent species.
dunes 14, 15, and 16 were inhabited by only two or
three species, and D. merriami, D. deserti, and M.
pallidus were conspicuously absent. As mentioned
earlier, all three dunes are outside the geographic
ranges of D. merriami and D. deserti and dunes 14
and 15 are outside the range of M. pallidus. This is
good evidence that dunes 14, 15, and 16 are biogeographically depauperate. The geographic location of
these dunes has prevented several species from colonizing them, and has resulted in a lower species diversity than the dunes are apparently capable of supporting. This contrasts markedly with the situation
on dunes 8 to 13, which are also inhabited by three
or fewer species but which do not receive enough
rainfall or produce enough seeds to support more,
even though potential colonists have access to them.
It is not difficult to account for the failure of D.
merriami and D. deserti to colonize the depauperate
dunes. Both species prefer lowland areas with sandy
soils. The depauperate dunes lie in isolated basins
separated from the ranges of the two species of kangaroo rats by rocky hills and mountains of unsuitable
habitat. I have shown elsewhere (Brown 1971) that
small terrestrial mammals are very poor at crossing
altitudinal and habitat barriers only a few miles in
extent. The ranges of these and other desert-dwelling
species are probably more extensive now than they
were during those periods of the Pleistocene when
the climate of the southwest was significantly cooler
and wetter than it is at present. Whether D. merriami
and D. deserti once occurred in the isolated basins
but have become extinct as a result of habitat changes
during the Pleistocene, or whether these species never
inhabited the basins is immaterial. What is important
is that at present there are large areas of apparently
suitable habitat which these species have not had the
opportunity to exploit. D. merriami, D. desert, and
M. pallidus are absent from the entire Bonneville
Basin, which includes much of western Utah and
part of eastern Nevada. It contains several sets of
dunes that are almost identical in general appearance,
climate (including rainfall), and vegetation to dunes
several hundred kilometers to the west, where D.
merriami, D. deserti, and M. pallidus are abundant
and most of the dunes are inhabited by four or five
784
JAMES H. BROWN
common species. In contrast, dunes 14 and 15. in the
Bonneville Basin, have only three common species:
D. ordi, Perornyscusyinaniculatus, and either P. iongiinembnrisor R. mnegalotis.
The unequivocal test of whether the rodent faunas
of these dunes are depauperate for biogeographic
reasons is a series of introduction experiments to determine whether the number of species can be increased. Such experiments would be easy to perform
and should yield interesting results. In the 1930's a
small number of D. ordi was introduced into sand
dunes on the south shore of Lake Erie in northern
Ohio, more than 1,000 km east of the natural geographic range of D. ordi and all other heteromyid
rodents. The kangaroo rats became established, increased. and persisted for at least a decade. They
may still be present; if not, their extinction was probably caused by alteration and destruction of the dune
habitat by man. Thus, some sand dunes seem to be
capable of supporting more species of seed-eating
rodents than occur there naturally.
Two other dunes (17 and 18) which several rodent species did not have the opportunity to colonize
were at higher elevations, and consequently received
more rainfall, than any of the other dunes sampled.
It is possible, perhaps likely, that even if the missing
species M. pallidus, D. merriami, and D. desert, had
the opportunity to colonize these dunes they would
be unable to do so. The climate, particularly the cold,
snowy winters, might be too severe for these lowland species, even if the dunes produced sufficient
seeds to support them. In this case the absence of
species must be explained not simply in terms of lack
of access to species already capable of living there,
but rather in terms of the lack of species capable of
colonizing the habitat even if access were provided.
Thus. these dunes may be depauperate in an evolutionary sense, rather than in a strictly biogeographic
sense; that is, speciation has failed to produce additional species capable of inhabiting these dunes even
though there are sufficient resources to permit the
coexistence of other species. Of course this is difficult
to test, but one kind of evidence does suggest that
these high altitude dunes could support more species
of seed-eating rodents. In addition to the three or
four common species, the samples of these dunes
contained several rare species (two on dune 17,
three on dune 18). The presence of these species,
usually found in other kinds of habitats, suggests that
the dunes produce sufficient seed resources to support small populations of these marginal species. If
this is so. I see no reason why the dunes could not
support at least one or two additional common species
if they were of appropriate size, and adapted to the
cold climate and open, sandy habitat.
Ecology, Vol. 54, No. 4
GENERAL DISCUSSION
The results of the present study and some of the
previous empirical and theoretical work on species
diversity should be examined for patterns of general
significance. The relationship between the predictable
amount of resources and diversity undoubtedly is a
general one, although few of the existing empirical
studies suggest that productivity or availability of
resources has an important effect on species diversity
in most groups of animals or plants. However, the
quantity of resources must be one of the primary
variables that determines how many resource-limited
species can subdivide the resources and coexist. Differences in availability of resources in a geographic dimension must be the major explanation of the relation between insular area and species diversity on
islands (MacArthur and Wilson 1967). As pointed
out earlier, resources that are unpredictable are
scarce at certain times, and this has the effect of
reducing the number of coexisting species which they
can support.
The literature of empirical studies of species diversity suggests that the effects of abundance and predictability of resources are often negated or at least
obscured by the effects of the way in which the resources are distributed. Resources can be apportioned
among species on the basis of inherent characteristics,
such as variations in their size, shape, or chemical
composition, or on the basis of extrinsic properties,
such as their distribution in space and time. The
greater the distribution of a resource, both with respect to number of parameters and variation in a
single parameter, the greater possibilities for subdivision of the resource and coexistence of species. The
influence of the distribution of resources on species
diversity is emphasized by numerous papers which
report correlations between species diversity and
some aspect of habitat structure (e.g., MacArthur and
MacArthur 1961, MacArthur 1964, Pianka 1967,
Rosenzweig and Winakur 1969). However, in most
of these studies the abundance and predictability of
resources was not measured, and so it is not known
to what extent resource abundance is correlated with
habitat structure and is therefore an important part
of the explanation of diversity.
The patterns of body sizes in rodent communities
of varying diversity are similar to those reported by
others who have studied closely related, coexisting
species. Regular distributions of dimensions of the
body or of trophic structures is a conspicuous feature
of many groups of coexisting species using particulate
resources (e.g., Hutchinson 1959, Shoener 1965,
Rand and Williams 1969, Rosenzweig 1966, McNab
1971). The significance of these patterns of body
size will be discussed further in another paper (Brown
and Lieberman 1973). The increased habitat specific-
Summer 1973
SPECIES DIVERSITY
ity and turnover of species between habitats ( 3 diversity) which appears to accompany decreased resource abundance and species diversity within habitats
(x diversity) may also be a general pattern, but there
is little evidence for this phenomenon in the literature.
Ecological parameters such as abundance, predictability, and distribution of resources determine how
many species a habitat can support. but significantly
fewer species may actually be present. Two kinds of
depauperate habitats may be distinguished. First, some
habitats lack species because biogeographic barriers
have prevented colonization. The fact that introduced
species frequently have become established in mainland or insular habitats without causing the extinction of native species provides a posteriori evidence
of the existence of such habitats. With sufficient
knowledge of appropriate habitats and species it
should be possible to predict which habitats are biogeographically depauperate, and to test these predictions by means of introduction experiments. Second,
some habitats may be evolutionarily depauperate in
the sense that they are capable of supporting additional species, but no existing species possess the
adaptations necessary to colonize them. This sort of
evolutionary influence on species diversity is difficult
to detect and almost impossible to verify scientifically,
yet it may account for significant variability in numbers of species between natural habitats. Environments characterized by unusual physical conditions,
small size, or impermanence in evolutionary time
(such as hot springs, salt lakes, and caves) frequently
contain remarkably few species. It seems likely that
speciation has simply not produced sufficient species
to fill these habitats to their ecological capacities.
Apparently species diversity in these habitats is determined by an equilibrium between speciation (or
adaptation) and extinction, rather than by an equilibrium between immigration of existing species and
local extinction as it is in most other habitats.
Two empirical studies of species diversity of terrestrial vertebrates in desert habitats warrant detailed
comparison with the present study. Pianka (1967)
studied the species diversity of lizards in flatland
habitats in the same deserts as my sand dune localities. Geographic variation in the diversity of lizard
species follows a pattern almost opposite that observed in the seed-eating rodents. Only four to five
species of lizards are found in flatland habitats in
the Great Basin Desert, and about twice that number
occur in comparable habitats in the Mojave, Colorado, and Sonoran Deserts. Pianka attributed this
variation in species diversity to biogeographic barriers to dispersal, length of growing season, amount
of warm season productivity, and spatial heterogeneity. He concluded that spatial heterogeneity of
IN DESERT RODENTS
785
the vegetation was the most important factor. Since
most lizards are insectivorous (some feed on green
vegetation or small vertebrates, but none on seeds),
poikilothermic, and active only during the warm part
of the year, it is hardly surprising that they show a
different pattern of species diversity than that of the
rodents, which eat mostly seeds, are homoeothermic,
and can be active all year. Unfortunately, no information is available on the abundance, diversity, and
predictability of food resources in the habitats of
these lizards, so it is impossible to evaluate the effect
of this factor on species diversity. However, Pianka's
work indicates that the structure of the vegetation
certainly has some direct effect on diversity, because
the trees and yuccas which are found only in the
southern deserts are utilized by specialized arboreal
species (Pianka 1966).
Rosenzweig and Winakur (1969) recently have
studied the ecology of seed-eating rodents in southern
Arizona and have attempted to account for variation
in rodent species diversity between different kinds of
habitats. Their results were quite different from mine;
they found that foliage height diversity, vegetation
density, and soil strength were the variables that most
affected rodent species diversity. Much of the discrepancy between the two studies is probably the
result of differences in design. Rosenzweig and Winakur studied several kinds of habitats within a local
area, whereas I worked in the same type of habitat
(with respect to vegetation and soils) over a wide
geographic area. Rosenzweig and Winakur did not
measure seed abundance in the habitats they studied,
and it is entirely possible that this was a proximate
variable which accounted for some of the variability
in species diversity. Certainly one would expect that
within a local area, all of which receives approximately the same amount of rainfall, seed production
might be correlated with such things as foliage height
diversity, vegetation density, and soil type. Also, some
of the habitats in southern Arizona contained much
more tall, woody vegetation (including small trees)
than did the sand dune habitats I studied. Rosenzweig
and Winakur suggested that in these habitats seeds
may be subdivided among rodent species partly on
the basis of vertical foraging zones. This is unlikely
in sand dune habitats where all of the vegetation is
low and roughly similar in form (Brown and Lieberman 1973). If additional parameters are involved
in apportioning resources among rodent species in
southern Arizona, one might expect species diversity
to be influenced by factors other than the predictable
amount of seed production.
Desert rodents are renowned for being diverse and
abundant in ecosystems characterized by few other
species and little biomass (Rosenzweig and Winakur
1969, Hall 1946, p. 1-2). It is less well known that
786
Ecology, Vol. 54, No. 4
JAMES H. BROWN
rodents are probably more diverse ill desert and semiarid grassland habitats than in any other habitats in
temperate North America (Baker 1971 ). It is not
uncommon to find between 8 and I () species, of which
5 or more are seed-eaters, in a single acre of uniform
desert habitat (Hall 1946, Hoffmeister and Goodpaster 1954). A comparable area of homogeneous
temperate forest, mesic grassland, or marsh usually
supports no more than 5 or 6 species. The spectacular diversity of rodents in the desert is largely the
result of the coexistence of several seed-eating species and can be attributed to two characteristics of
desert seeds. First, seeds are always available. Many
of the seeds are produced by annuals that exist as
vegetative plants only for a short period following
heavy rains. For the majority of their lives (for many
months and often for several years) the desert annuals lie scattered and dormant as seeds. These plants
have adapted to the unpredictability of the desert
climate by producing large numbers of seeds and by
evolving mechanisms which insure that only a portion of them germinate each season. This is essential
because many desert areas receive sufficient rainfall
for the annuals to produce a new crop of seeds only
every 2nd or 3rd year on the average. Each species
of annual has evolved a reproductive strategy which
insures that there are always sufficient seeds scattered in the habitat to produce a new generation.
This means that regardless of rainfall or the presence
of succulent vegetation rodents can always find seeds.
The rodents themselves help to insure the continued
availability of seeds by gathering and storing them
when they are abundant. Because seeds are always
present, some rodent species (especially some species
of Perot-nat1itus) have been able to specialize to the
point that they utilize virtually no other source of
food or water. In more mesic habitats it has not been
possible for species to become such specialized granivores; in the spring most of the seeds either germinate or else they are so well hidden or buried that
the rodents cannot find them and for several months
must rely on other kinds of food. The second characteristic of seeds that permits the coexistence of rodent species is their discrete, particulate nature, which
enables them to be apportioned among species on the
basis of size. Many other kinds of resources do not
lend themselves to being partitioned among species.
For example several species of the rodent genera
Microtus and Siginodoni feed on the leaves of grasses
and forbs, but it is rare to find more than one species in the same habitat (Baker 1971). Seeds, because they are omnipresent, particulate resources of
varying size, are the key to the diversity of rodent
species in desert ecosystems. The seeds are also primary resources for two other groups of animals that
probably compete to some extent with the rodents
seed-eating birds, which winter in the deserts in large
mixed-species flocks, and insects, particularly ants,
which are spectacularly abundant and diverse in many
desert habitats.
ACKNOWLLDGM
ENTS
To my wife, Astrid, I am indebted for invaluable assistance and encouragement during all phases of the
work. H. J. Thompson kindly identified the plants. Numerous colleagues and students contributed important
discussions and useful suggestions. G. A. Lieberman,
M. L. Rosenzweig, K. T. Harper, D. J. Futuyma, and
C. D. Barbour have been particularly helpful and have
kindly read and criticized the manuscript. Special thanks
is due also to R. H. MacArthur for many hours of critical and stimulating discussion. The work was supported
in large part by N.S.F. Grant GB 8765.
LITE RATURE
CITED
Baker, R. H. 1971. Nutritional strategies of myomorph
rodents in North American Grasslands. J. Mammal.
52: 800-805.
Beatley, J. C. 1967. Survival of winter animals in the
Mojave Desert. Ecology 48: 745-750.
1969. Dependence of desert rodents on winter
animals and precipitation. Ecology 50: 721-724.
Brown, J. H. 1971. Mammals on mountaintops: nonequilibrium insular biogeography. Am. Nat. 105: 467478.
Brown, J. H., and G. A. Lieberman. 1973. Resource
utilization aind coexistence of seed-eating desert rodents
in sand dune habitats. Ecology (in press).
Hall, E. R. 1946. Mammals of Nevada. Univ. of California Press, Berkeley and Los Angeles. 710 p.
Hoffmeister, D. F., and W. Goodpaster. 1954. The mammals of the Huachuca M\ountains, southeastern Arizona. Ill. Biol. Monogr. 24: 1-152.
Hutchinson, G. E. 1959. Homage to Santa Rosalia, or
Why are there so many kinds of animals? Am. Nat.
93: 145-159.
Janzen, D. H. 1970. Herbivores and the number of tree
species in tropical forests. Am. Nat. 104: 501-528.
MacArthur, R. H. 1964. Environmental factors affecting
bird species diversity. Am. Nat. 98: 387-398.
MacArthur, R. H., and J. M. MacArthur. 1961. On bird
species diversity. Ecology 42: 594-598.
MacArthur, R. H., and E. 0. Wilson. 1967. The theory
of island biogeography. Princeton Univ. Press, Princeton, N.J.
McNab, B. K. 1971. The structure of tropical bat faunas.
Ecology 52: 352-358.
Paine, R. T. 1966. Food web complexity and species diversity. Am. Nat. 100: 65-75.
Pilnka, E. R. 1966. Convexity, desert lizards, and spatial
heterogeneity. Ecology 47: 1055-1059.
1967. On lizard species diversity: North American flatland deserts. Ecology 48: 333-351.
Rand, A. S., and E. E. Williams. 1969. The anoles of
La, Palma: aspects of ecological relationships. Brevoria
327: 1-19.
Rosenzweig, M. L. 1966. Community structure in sympatric carnivore. J. Mammal. 47: 602-612.
Rosenzweig, M. L., and J. Winakur. 1969. Population
ecology of desert rodent communities: habitats and
environmental complexity. Ecology 50: 558-572.
Summer 1973
SPECIES DIVERSITY
IN DESERT RODENTS
787
tions on germination in the Joshua Tree National MonSanders, H. L. 1969. Benthic marine diversity and the
ument, California. Ecology 29: 242-253.
stability-time hypothesis, p. 71-81. In Diversity and
. 1955. The ecology of desert plants. Sci. Am.
stability in ecological systems. Brookhaven Symposia
192: 68-75.
in Biology No. 22.
Shoener, T. 1965. Evolution of bill size differencesamong Went, F. W., and M. Westergaard. 1949. Ecology of
desert plants. III. Development of plants in the Death
sympatric, congeneric species of birds. Evolution 19:
Valley National Monument, California. Ecology 30:
189-213.
26-38.
Went, F. W. 1948. Ecology of desert plants. I. Observa-