comparative population ecology of eleven species...

Comparative Population Ecology of Eleven Species of Rodents in the Chihuahuan DesertAuthor(s): James H. Brown and Zongyong ZengSource: Ecology, Vol. 70, No. 5 (Oct., 1989), pp. 1507-1525Published by: Ecological Society of AmericaStable URL: http://www.jstor.org/stable/1938209 .Accessed: 22/06/2011 14:44

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 unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay 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 printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof 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

http://www.jstor.org/action/showPublisher?publisherCode=esa

http://www.jstor.org/stable/1938209?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/action/showPublisher?publisherCode=esa

Ecology, 70(5), 1989, pp. 1507-1525 © 1989 by the Ecological Society of America

COMPARATIVE POPULATION ECOLOGY OF ELEVEN SPECIES OF RODENTS IN THE CHIHUAHUAN DESERT1

JAMES H. BROWN Department of Biology, University of New Mexico, Albuquerque, New Mexico 87131 USA

AND

ZONGYONG ZENG Department of Biology, University of Sichuan, Chengdu, Sichuan, China

Abstract. Comparisons of mark-recapture data on life histories and population dynamics of the 1 1 commonest species of nocturnal desert rodents inhabiting our experimental study site in the Chihuahuan Desert of extreme southeastern Arizona permitted assessment of the role of evolutionary relationships and ecological factors in the coexistence of these species. The species varied greatly in population density, extent of interannual variation in abundance, timing of reproduction, extent to which reproduction was seasonal, rate of disappearance of marked individuals, frequency and distance of lifetime dispersal movements, but perhaps less so in death rate and maximum longevity.

Most of the species showed positively correlated year-to-year fluctuations in population density, suggesting that they responded similarly to interannual variation in precipitation, primary production, and availability of food resources. In contrast, there were both positive and negative correlations in seasonal patterns of reproductive activity and population density. Lifetime dispersal movements were inversely related to body size, suggesting that energy constraints cause the smallest species to move among rich patches in a coarse- grained manner. Patterns of similarities and differences among closely related (congeneric and confamilial) species suggested that evolutionary constraints sometimes, but not always, limited variation in life history and demography. The relationship between population ecology and competition among these species was not clear.

We interpret the diversity of life histories and population dynamics in these coexisting species to be a consequence of: (a) a productive and spatially and temporally variable environment that provides a variety of resources that may be used in different ways, (b) historical biogeographic events that have made available a large regional pool of species from which potential colonists can be drawn, and (c) differences in population ecologies among the species that evolved primarily in other environmental contexts, but that permit coexistence by enabling the species to use different resources or to use the same resources in different ways.

Key words: community; demography; desert; dispersal; evolutionary constraint; field study; life history; population dynamics; reproduction; rodent; southwestern North America; survival.

INTRODUCTION

Communities are assemblages of species, and the composition and dynamics of communities reflect the structure and dynamics of the species populations. The population ecologies of the component species are just as relevant to understanding community organization as are the patterns of resource utilization and the dynamics of interspecific interactions. Yet rarely have the population ecologies of the species that coexist to form a sizeable guild or community been compared.

Such comparisons should be of particular interest because the population structures and dynamics of coexisting species, like their patterns of resource utilization, should reflect a compromise between opposing forces. On the one hand, the species will tend to be

1 Manuscript received 3 April 1988; revised and accepted 25 September 1988.

similar because they inhabit a common environment. Such similarities may represent both adaptations to similar conditions and colonization of local areas by species that already possessed similar tolerances and requirements. Also, to the extent that communities contain closely related species, similarities among them may reflect evolutionary constraints owing to common ancestry. On the other hand, species will tend to differ because they interact with the same local environment in different ways. These differences may reflect the res- olution of interspecific interactions among the species that permit or promote coexistence. Also, to the extent that communities contain distantly related species, differences among them may reflect evolutionary constraints owing to divergent ancestry.

Although the rodents of the desert regions of southwestern North America have been the subject of many studies of interspecific interactions and community or-

JAMES H. BROWN AND ZONGYONG ZENG

ganization (e.g., Brown 1987, Brown and Harney, in press), surprisingly little is known about the basic population biology of many species. Especially lacking are long-term and comparative studies to provide data on the population dynamics and life histories that enable diverse species to persist and coexist in highly variable desert environments. Here we present eight years of data comparing the population ecology of the 11 species of common heteromyid and murid (cricetid) rodents that coexist on - 20 ha of Chihuahuan Desert in southeastern Arizona. On the one hand, all of these species must cope with the same climate, soil, vegetation, and predators. On the other hand, the species differ markedly in several respects: in taxonomic affinity, with representatives of two families and seven genera; in morphology, which varies from bipedal kangaroo rats to quadrupedal mouselike forms; in body size, which ranges from 7 to 175 g; in diet, which includes granivory, folivory, carnivory, and omnivory; and in many other features of their biology.

METHODS

The study was conducted on the Cave Creek Bajada, 6.5 km east and 2 km north of Portal, Cochise County, Arizona, at an elevation of 1330 m. We present data collected from November 1977 to June 1985, a period of 7 yr and 8 mo. We did not include either 4 mo of preliminary data collected from July to October 1977 or data obtained after June 1985, although the study is continuing.

The data were collected as part of a mark-recapture study designed primarily to measure the response of rodent populations to experimental removal of granivorous rodent and ant species and to addition of supplemental seeds. The study site, experimental design, layout of the experimental plots, trapping regime, and analytical methods are described in detail in previous papers (Brown and Munger 1985, Zeng and Brown 1987a, b), so they will be treated only briefly here. The habitat is transitional between arid grassland and upper elevation Chihuahuan Desert shrubland. The terrain is relatively flat except where it is dissected by several temporary watercourses. The 20-ha study site has been fenced since 1977 to exclude livestock.

Rodents were censused on 24 experimental plots, each 0.25 ha in area (see aerial photograph in Brown and Munger 1985). All plots were fenced so as to render them potentially rodent-proof, but 16 equally spaced holes (gates) of varying sizes cut in the fences allowed access of selected rodent species to appropriate plots. The largest sized gates allowed free access of all rodent species to 16 plots (8 seed-addition, 2 ant-removal, 2 Pogonomyrmex rugosus [a large harvester ant]-removal, and 2 unmanipulated control), medium-sized gates excluded the large kangaroo rat, Dipodomys spectabilis, from 2 plots, small gates excluded three kangaroo rat (Dipodomys) species from 4 plots (2 of which also had P. rugosus removed), and no gates at all excluded vir-

tually all rodents from 4 plots (2 of which also had ants removed).

Rodents were captured during a monthly live-trapping program that was used to monitor responses of all rodent species to the experimental manipulations. Each plot was trapped for a single night during a two- or three-night trapping period timed to correspond as much as possible to the new moon. On the night of trapping, the gates in the fences were closed so as to catch only those individuals resident on that particular plot. Forty-nine Sherman live traps (23 x 8 x 9 cm) baited with millet or mixed birdseed were set at per- manent grid stakes spaced at 6.5-m intervals. The monthly trapping was sometimes interrupted by skip- ping a period in winter. Approximately once each year we trapped with the gates open to capture both resi- dents and those individuals that foraged on the plots but had their home burrows outside.

Each individual was marked when first captured with a numbered monel fingerling tag attached to an ear, except for the three smallest species (Chaetodipus penicillatus, Perognathus flavus, and Reithrodontomys megalotis) which were marked by toe-clipping. At each capture, identification number, body mass, hind foot length, and standardized data on reproductive condition (for males: testes abdominal or scrotal; for females: vagina swollen or plugged, pregnant, and/or lactating) were recorded for each individual.

The experimental design introduces some compli- cations that affect the analysis and the comparison of the results with other studies using unfenced, unmanipulated trapping grids. The use of fenced plots with gates was quite effective in restricting captures to individuals resident on the plots. Only very infrequently was the same individual caught on two different plots on successive nights. Occasional trapping with the gates open (reported in Brown and Munger 1985) showed the extent to which individuals occupying burrows outside the fences moved onto the plots at night to forage. Because of the exclusion experiments, different species had free access to different numbers of plots: D. spectabilis to 14, D. ordii and D. merriami to 16, and the other eight species to 20. Biomass was calculated using only data for the unmanipulated control plots. Popu- lation densities were calculated as the number of individuals captured on a plot in a trapping period di- vided by the area of the plot (0.25 ha), and the resulting values were then averaged over all trapping periods when the gates in the fences were closed. Population densities were determined both for all of the plots to which each species had free access and for only the two unmanipulated control plots. Although individuals captured on plots designated for removal of that species were not used to calculate population density, all other data collected from those animals were used in the analyses. N. albigula, Pm. eremicus, and R. megalotis are excellent climbers; these species quite frequently immigrated onto exclusion plots despite monthly trap-

1508 Ecology, Vol. 70, No. 5

DESERT RODENT POPULATION ECOLOGY

ping and removal. To maximize the continuity and reliability of the life-history data, marked individuals captured on plots designated for exclusion of that species were removed from the plot but released elsewhere on the study site.

NATURAL HISTORY OF THE SPECIES

The 11 species considered in this paper differ greatly in taxonomic affinity and natural history. They represent seven genera and two families: Heteromyidae: Dipodomys spectabilis, D. ordii, D. merriami, Chae- todipus penicillatus, and Perognathus flavus; and Mu- ridae: Neotoma albigula, Onychomys leucogaster, 0. torridus, Peromyscus eremicus, Peromyscus maniculatus, and Reithrodontomys megalotis.

The heteromyids are endemic to arid and tropical regions in southern and western North America and northernmost South America. The three genera represented here are widespread in the southwestern deserts and possess several characteristics that have been interpreted as adaptations for arid environments. These include external cheek pouches used to collect and transport the seeds that comprise the majority of their diets, and kidneys capable of concentrating urine so that most species can subsist without drinking water on a diet consisting solely of dry seeds. In addition, the kangaroo rats, genus Dipodomys, have inflated tympanic bullae and bipedal, saltatorial locomotion, characteristics that have been interpreted as facilitating the detection and avoidance of predators in open desert habitats. Pocket mice, genera Chaetodipus and Perog- nathus, do not possess such extreme morphological and locomotor specializations, and this has been related to observations that they usually do more foraging under and up in vegetation than do kangaroo rats. Unlike the kangaroo rats, which are active all year, most pocket mice hibernate for several weeks or months. All heteromyids are nocturnal and spend the day in burrows. As indicated in Table 1, body size of the desert heteromyids varies by more than an order of magnitude, from 7 g in Pg. flavus to 123 g in D. spectabilis.

Although the heteromyids are usually considered to be the most specialized North American desert rodents, the study site is not unusual in being inhabited by even more species of murids. The family Muridae is a very large, diverse group of rodents with a world- wide distribution. The species here are members of the subfamily Cricetinae whose center of distribution, diversity, and abundance is North and South America. Although these mice are relatively unspecialized mor- phologically compared to the heteromyids, they are more diverse behaviorally and ecologically. The wood rats, Neotoma, are primarily folivorous and build large dens of sticks and debris. The grasshopper mice, Onychomys, are carnivores that feed on a variety of invertebrates and small vertebrates. The harvest mice, Reithrodontomys, like the heteromyids, feed largely on

seeds. The deer mice, Peromyscus, are the least specialized and are highly omnivorous and opportunistic, although seeds comprise a substantial fraction of their diet. These murids are nocturnal and active throughout the year, although harvest mice and deer mice can utilize torpor to avoid short periods of severe weather or food shortage. The murids also span a wide range of body sizes, from 10 g in R. megalotis to 175 g in N. albigula.

In evaluating the variety and specializations of these rodents it should be mentioned that the study site is typical of the habitats of the three genera and all five of the species of heteromyids. In contrast, although some murid species such as N. albigula, 0. torridus, and Pm. eremicus are confined to desert environments, others are much less restricted. The most extreme case is Pm. maniculatus, the most widely distributed small mammal in North America, which inhabits alpine tun- dra, coniferous and deciduous forest, and grasslands as well as deserts.

RESULTS

The data base

Our analyses are based on 9090 captures of 3039 individuals of 11 species (Table 1). Not included in the analyses are six additional species (Chaetodipus hispidus, Sigmodon hispidus, Reithrodontomysflavescens, Ammospermophilus harrisi, Spermophilus spilosoma, and Thomomys bottae) that were present but captured too infrequently (<50 times) to provide reliable data (Brown 1984, Brown and Munger 1985).

Also in Brown and Munger (1985) but not in the present paper are analyses of the responses of individual rodent species to the experimental manipulations. Addition of seeds and exclusion of other rodent species had substantial effects on the population densities of certain species. Table 1 presents data for standing-stock population densities of each species averaged for all 14-20 plots to which it had access and for only the two unmanipulated control plots. Brown and Munger (1985) compared several life history parameters between experimentally manipulated and control plots, and found virtually no significant differences. This suggests that those species that responded to the experimental per- turbations did so primarily by adjustments in population density, while other aspects of their population ecologies remained characteristic of the site as a whole.

Population density and dynamics We determined the mean population density of each

species by direct enumeration, based on the average number of individuals caught in each 0.25-ha plot for all trapping periods when the gates in the fences were closed. More than 80% of the individuals of most species known to be alive were captured each month. Since there is considerable movement, even of adults, across and beyond the study area (Zeng and Brown 1987a, b;

October 1989 1509


TABLE 1. Data on the size and composition of the populations of 11 species of desert rodents inhabiting the experimental study site near Portal, Arizona.

Number of Number of captures individuals Num- Popu- Coeff. of ber of Sex ratiot Body Bio- lation variation Fe- Fe- juve- (male: mass mass density§

Species Male male Total Male male Total niles female) (g) (g ha- ) (no./ha) mo yr

Dipodomys 1019 921 1940 273 239 512 130 1.11:1* 123.4 504.8 5.30 0.25 0.31 spectabilis (1.14:1) (4.09)

Dipodomys ordii 389 384 773 135 145 280 33 1.01:1 48.1 64.9 1.82 0.36 0.60 (0.93:1) (1.35)

Dipodomys 1847 1642 3489 511 457 968 156 1.12:1*** 42.8 505.1 7.93 0.13 0.31 merriami (1.12:1) (11.81)

Chaetodipus 59 121 180 37 52 89 20 0.49:1*** 16.0 9.0 0.39 0.85 0.26 penicillatus (0.71:1) (0.56)

Perognathus 272 184 456 118 96 214 2 1.48:1*** 7.0 3.9 0.93 0.31 0.73 flavus (1.23:1) (0.56)

Reithrodontomys 221 177 398 88 92 180 7 1.25:1* 10.2 5.7 0.65 0.74 0.93 megalotis (0.96:1) (0.56)

Peromyscus 89 52 141 36 38 74 9 1.71:1t 21.4 4.1 0.28 0.38 1.16 maniculatus (0.95:1) (0.19)

Peromyscus 150 106 256 59 49 108 23 1.42:1** 21.2 5.5 0.43 0.23 0.76 eremicus (1.20:1) (0.26)

Onychomys 249 245 494 87 81 168 44 1.02:1 34.7 64.5 0.89 0.39 0.26 leucogaster (1.07:1) (1.86)

Onychomys 265 254 519 129 120 249 55 1.04:1 24.6 28.0 0.97 0.35 0.20 torridus (1.08:1) (1.14)

Neotoma albigula 189 255 444 90 107 197 73 0.74:1t 174.7 235.8 0.95 0.46 0.36 (0.84:1) (1.35)

Total 9090 3039 562 1431.1 23.73 (20.54)

t Two values are given for sex ratio: above, based on number of captures; below, in parentheses, based on number of individuals. Statistical significance of departure from 1:1 ratio: * P < .05, ** P < .01, t P < .005, *** P < .001.

§ Two values are given for population density: above, mean for all plots to which the species had access; below, in parentheses, mean for the two control plots.

see also Dispersal, below), many of those individuals not captured but known to be alive could have been residing temporarily outside the fenced plots. Our method of counting captures within fenced plots avoids the problem that unfenced grids sample an unknown area larger than that covered by the traps. We believe our estimates of density are far more accurate than those calculated from mark-recapture data using as- sumption-laden, indirect methods (e.g., see Seber 1986).

Table 1 shows total number of captures, estimated population densities, and standing-stock biomass for the 1 1 species. Combined density of all species totaled 23.73 individuals/ha (20.54 individuals/ha for the unmanipulated control plots). Including the captures of the six additional species present on the study site would only increase this figure to 21 individuals/ha for the control plots. Rodent biomass totaled - 1.43 kg/ha. As is typical for most communities, the contributions of different species to these totals were highly uneven. Population densities ranged from <0.56 individual/ha in C. penicillatus, Pg.flavus, R. megalotis, Pm. maniculatus, and Pm. eremicus, which made up collectively only 6.7% of the total, to 11.81 individuals/ha in D. merriami, which accounted for 50% of the total. Bio- mass ranged from < 10 g/ha in each of the same five rare species to >500 g/ha for D. spectabilis and D. merriami. Although the number of species in the two

families were similar, the heteromyids dominated the community, accounting for 77.5% of the population density and 76% of the biomass, and the kangaroo rat genus Dipodomys alone accounted for 73% of the population density and 75% of the total rodent biomass.

The population densities of the species fluctuated substantially over the study period (Fig. 1). The combined densities of all species attained their highest levels in 1982 and 1983 and their lowest levels during 1979 and 1984. Most of the species showed generally similar trends, except for the two Onychomys species, which did not exhibit significantly lower numbers in 1979 and 1984 than in most other years. Of the 11 species, only D. merriami and D. spectabilis were captured in every monthly trapping period. Because the wide fluctuations included zero density for most species, the coefficient of variation provides perhaps the best measure of relative differences among them. We have calculated this statistic for two different time intervals, months and years, which give somewhat different results (Table 1). On a month-to-month time scale, D. merriami exhibits by far the least severe fluctuations, followed by Pm. eremicus and D. spectabilis. At the other extreme, C. penicillatus showed the most variation, followed by R. megalotis and N. albigula. The apparently higher densities of the two Onychomys species in the winter, compared to the summer, months



A-- D. merriami

+-+-+ D. ordii

15 - *-. D. spectabilis

10

5

0

6

5

4

3

2

0

*-- C penicillatus -- Pg. flavus

II Ii i II i li II I ii i ii I 111

* * R. megalotis

A-A N. albigula

* * Pm. maniculatus

a-nA Pm. eremicus

3r

-- 0. leucoguster

A-nLO torridus

1978 1979 1980 1981 1982 1983 1984 1985

DATE

FIG. 1. Fluctuations in population density in the 11 species of rodents inhabiting the Chihuahuan Desert study site. Note that Perognathus flavus was locally extinct for the last 18 mo of the study, whereas the two species of Peromyscus did not attain significant densities on the study site for the first 4 yr of the study.

0 -c . _ c,n

0

"0

z

C)

LUJ

z 0

I-

_J :D O_ 0 0-

3

2

O 0

I I I I I I I I I 1 1 1 1 T I I I I I I I I I I 1 11 1 1 1 1 1 i 1 I I 1 I I I 1 I I . I I I I 1

October 1989 1511

-= In A


C a D merriami /o Pm eremicus

X -, A C penicillatus / + R megalotis

1 \ pg flus + Afl 0. leucogaster

-c0 # 0 torridus

8- 8 i 20- -

J' F MAMJ J /ASON5 J FMAMJ JASOND

--

MONTH

FIG. 2. Seasonal patterns of variation in population density of the 11 species of desert rodents. Values plotted are the mean densities for each species for each month averaged over the duration of the study (1977-1985). Note that different species reach peak densities at different times of year.

0

4 0

MONTH

densities for each species for each month averaged over the duration of the study (1977-1985). Note that different species reach peak densities at different times of year.

might in part reflect increased trapability during cold seasons when insect prey are less available, but the increases in late summer and fall probably also reflect recruitment from breeding in spring and summer. Much of the month-to-month variation in density of C. penicillatus can be attributed to the fact that this pocket mouse hibernates. When assessed on a year-to-year scale, this species and the two Onychomys exhibited the least fluctuations, and Pm. maniculatus, R. megalotis, Pm. eremicus, and Pg. flavus showed the greatest fluctuations.

The general patterns of year-to-year fluctuations in the different species can be characterized as follows. All three species of kangaroo rats were reliably present, but showed substantial variation and attained high densities in 1982 and 1983. In addition, D. ordii showed a clear trend of increasing density throughout the eight years of the study. C. penicillatus was consistently present at low density every summer and absent (in hibernation) each winter. Pg. flavus increased to a peak of 3-6 animals/ha in 1982-1983 and then almost disappeared for the last two years of the study (and remained virtually absent from the study area until the summer of 1988). R. megalotis and the two Peromyscus species showed similar patterns; they were almost absent from the area during the first four years (except for a modest number of R. megalotis in late 1979), attained substantial numbers in 1982-1983, and then declined to low densities. The two Onychomys species showed substantial month-to-month variation within years, but maintained very constant populations from year-to-year. N. albigula attained its highest densities in the summers of 1982-1984.

The coefficient of variation in population density by years (Table 1) provides a good measure of interannual

fluctuations. O. torridus, 0. leucogaster, C. penicillatus, D. merriami, and D. spectabilis maintained the most constant populations over the seven years of the study, whereas Pm. maniculatus, R. megalotis, Pm. eremicus, and Pg. flavus showed the greatest year-to-year variation.

The species differed in the extent to which they showed annual population cycles and also in the time of year at which peak densities were attained. Inspec- tion of Fig. 2 suggests that those species (C. penicillatus, Pg. flavus, and Pm. maniculatus) that had the lowest average densities also showed the least seasonal fluctuation. This is misleading, however, and the coefficients of variation by month (Table 1) provide a much more accurate estimate of the magnitude of annual cycles. These show that C. penicillatus and R. megalotis had the greatest variation, whereas D. merriami, D. spectabilis, and Pm. eremicus had the least. The season of maximum population size varied from winter for D. merriami, D. ordii, R. megalotis, and 0. leucogaster, to spring for D. spectabilis and Pg. flavus, to summer for C. penicillatus and N. albigula (Fig. 2).

Reproduction

The species differed markedly in seasonality of re-

production, as indicated by the months when individuals were in reproductive condition and when juveniles were recruited into the population (Fig. 3). In all species a larger proportion of males than of females were in

reproductive condition in any one month. This is not surprising since "reproductively active" males were potentially capable of breeding, whereas reproductively active females were actually breeding. We used Shan- non's diversity index, H/Hmax where H = -2 piln p, and pi is the proportion of females in reproductive

Ecology, Vol. 70, No. 5 1512


D spectabilis 100 - *-_

+0 - +-

D ordii 100- ,

_ + -+ + +, i + I I I I I I I I I I I

D. merriami 100 -

O-++ + + + + +

C. penicillatus 100 --

0 -(3 1 1 1 1 1 1 1 1 1 1 1 al I

Pg flavus 100 -

0-+---+ +--- +----+ -+- + + + I ! I I I I I I I I I |

R megalotis 100 -

O- M-+ M t+-+ ++ --- ++D J F M A M J J A S O N D

Pm maniculatus 100 -

0 + +

Pm. eremicus 100 -

00

0. leucogaster 100 -

0--

0 torridus 100-

0+ +

N. albigula 100 -

0- J

I I I I I J

I I I I D d F M A M 3 d A S 0 N D

MONTH

FIG. 3. Percent of males (0) and females (0) in reproductive condition and the percent of individuals captured that were juveniles (+). Note that for each species the appearance of juveniles usually closely followed the peak of reproductive activity, especially in females, and that the appearance of juveniles corresponds well with the seasonal peak in population density shown in Fig. 2.

condition in the ith month, to compare species with respect to the evenness of the proportion of females in reproductive condition from month-to-month over the annual cycle (Table 2). The three Dipodomys species and the two Onychomys species had the highest values, indicating the least seasonal reproduction. The kangaroo rats in particular were very aseasonal breeders: in D. merriami both sexes were reproductively active and juveniles were recruited in every month (see also Zeng and Brown 1987a); in D. spectabilis some females apparently had three widely spaced litters in favorable years (see also Jones 1984, 1986). By contrast, Pm. maniculatus, C. penicillatus, N. albigula, and R. megalotis showed the most seasonal reproductive patterns; in particular, breeding of C. penicillatus, an apparently obligate hibernator, was confined to a few months in spring and early summer. In general, maximum appearance of new juveniles followed peak reproductive activity of females with a lag of 1-2 mo, and this timing

of recruitment corresponded well with the season when the highest population densities were attained (compare Figs. 2 and 3).

There was also substantial variation in the extent of reproductive activity in different years. We assessed this by calculating the coefficient of variation among years in the percentage of females in reproductive condition for the seven complete years of the study. Table 2 shows that Pm. eremicus, D. merriami, D. ordii, and C. penicillatus showed relatively little year-to-year variation, whereas R. megalotis, Pg. flavus, and Pm. maniculatus showed the greatest variation in reproductive activity.

Sex ratios

Sex ratios (males: females) varied from female-biased in C. penicillatus (0.49:1) and N. albigula (0.74:1) to male-biased in Pm. maniculatus (1.71:1) and Pg. flavus (1.48:1) (Table 1). These ratios were calculated on the

Z LLJ

0

October 1989 1513


TABLE 2. Year-to-year and month-to-month variation in the proportion of females of the 11 rodent species in reproductive condition. Note that low values of the coefficient of variation indicate low interannual variation, whereas high values of evenness indicate low seasonal variation in reproductive activity.

Species

Dipodomys spectabilis Dipodomys ordii Dipodomys merriami Chaetodipus penicillatus Perognathus flavus Reithrodontomys megalotis Peromyscus maniculatus Peromyscus eremicus Onychomys leucogaster Onychomys torridus Neotoma albigula

Evenness of the proportion

Coefficient of females of variation each month

in the showing proportion of reproductive females each activity year showing (H/Hma,, reproductive where H=

activity -I piln pi) 0.53 3.72 0.44 3.62 0.35 4.03 0.48 2.89 0.70 3.42 0.71 3.29 0.63 2.97 0.31 3.44 0.58 3.60 0.64 3.53 0.52 3.15

megalotis and Pg. flavus, respectively. These results indicate a high frequency of long-distance movements by adults of all species except D. spectabilis.

Another indication of the importance of adult dispersal is the large percentage of individuals that were first captured as adults (>74% in all species except N. albigula; Table 3). This indicates that the majority of individuals were not recruited into the adult population on the same plot where they were born and weaned, but instead dispersed to their present home range as an adult. Note that this does not imply any differential movement onto or off the experimental plots in response to differences in recruitment or survival. In fact, Brown and Munger (1985) found virtually no consistent differences in reproduction or persistence among experimental treatments. It is just the case that if there is a great deal of movement by individuals of all ages, most of the recruitment will consist of immigrating adults, most of the disappearances will be emigrating adults, and at any given time most of the individuals will be residing at some distance from where they were born.

basis of number of captures; sex ratios calculated on the basis of number of individuals showed qualitatively similar, but less extreme, deviations from 1.0:1 for most species (Table 1). This wide range suggests potentially interesting differences among the species in sex-specific mortality, dispersal, and home range size, and/or in social structure and breeding systems.

Dispersal We used the frequency distributions of the distances

between captures of the same individual to assess short- term and lifetime movements in these rodents. Figs. 4 and 5 show the data, respectively, for distances between successive captures 1 mo apart and distances between first and last capture for those individuals that lived >4 mo. Nearly all of these distributions show a con- centration of values at distances of <50 m and a tail of values out to much greater distances. Naturally, the lifetime data show a greater frequency of long-distance movements (Table 3).

These raw data underrepresent the frequency of long- distance dispersal because, with increasing distance from each capture site, a smaller proportion of the area into which individuals could potentially move is actually sampled by traps. We have developed a method to correct for this bias and accurately estimate the real frequency distribution of dispersal movements (Zeng and Brown 1987b). We applied this method and calculated the median lifetime dispersal distance for each species (Table 3). These values range from lows of 32, 53, and 70 m in D. spectabilis, D. ordii, and D. merriami, respectively, to highs of 286 and 202 m in R.

Mortality

Mortality of individuals was assessed in two ways. One was simply to calculate the time between first and last capture for each marked individual. By combining data for all individuals of each species we obtained a persistence curve that is similar to a survivorship curve with two important exceptions. First, persistence is measured from first capture, which usually occurred as an adult (Table 3), rather than from birth. For most species we had too few juveniles to measure survival as a function of age, but newly caught juveniles had a probability of being recaptured in the next trapping period that was slightly lower than or indistinguishable from first-captured adults (e.g., see Zeng and Brown 1987a). Second, the disappearances of marked individuals included those that had dispersed into areas not sampled by traps as well as those that had died.

Bearing these caveats in mind, the persistence curves were nearly linear when the proportion of individuals recaptured was plotted on a logarithmic scale, suggesting that each species had approximately a constant probability of disappearing per unit time (Fig. 6). The slopes of these persistence curves varied among species by a factor of >2, from -0.055 and -0.060 in C. penicillatus and D. spectabilis, respectively, to -0. 134, -0.130, and -0.129 in R. megalotis, Pm. eremicus, and Pm. maniculatus, respectively (Table 4). These correspond to disappearance rates that varied from 0.699 to 0.885 yr-' (Table 4). These slopes were estimated only for the first 10 mo following initial capture, because only small numbers of individuals of some species persisted for longer periods, making calcula- tions over a longer period inaccurate. For the same reason, the data on maximum survivorship of marked individuals (Table 4), which exhibited great variation,



400 -

320 -

240 -

160 -

80 -

0 - . . . . . , . . . , .

o O o In o

100 -

80 -

60 -

40 -

20 -

D spectabilis n =972

0 0 0 00O000 n 0 ) O L D 00J (M- 0 c- j c C\ J r )r rro r -

D. ordii

n =273

Pm maniculatus

n =37

10 -

8-

0

10 -

8-

6-

4-

2-

.i. d .

0 0 0 LO 0 Lo U3 O .,,

......

0

0 0 0 0 o o o in o cN cJ r

Pm. eremicus n =71

0

500

400

L 300

T 200

-- 100

< 0

L LL.

0 0 0 0 0 O o o o o o ° UL 0 n 0 LO O0

^~~~~~~~- *~D merriami

:~- ~~~~~n =1504

0 0 0 0 0 0 000 Lu 0 u~ 0o or

_ - - \j c r_ ) re) 0 rt 20-

m 16 -

2 12-

s8 -

4

o o LU

C penicillatus n =47

o o o i

o 0 o in n. n.

40 -

32-

24 -

16 -

8-

0

o 0 0 rN)

Pg flavus n =119

o 0 0 0 0 0 C in 0 in 0 in C

n\j N rN

20-

16-

12 -

8-

4-

0 0 0 0 0 - -

cj

30 -

24 -

18 -

12 -

6 - 6-J J

. ...

0 o

cJ

0 0 r0

0. leucogaster n =140

0 0 0 o o o o Ln o i o in o - cM cM r)

20 - 0. torridus 16 - n =128

12-

8-

4-L 0 . . . . . . . . . . . . . . . . . . . . , . . . . . . .

o 0 o o o o o i> o 0 o in o - - - cCJ (cM rN

3D D CO " cr

o

R megalotis n =92

N albigula n= 140

50 -

40 -

30 -

20 -

10 -

0 0 0 0 0 0 0000 0 0 0 0 0 0 0 n o0 l 0 in Oc0\j O ) i 0 o 0 in o

-- - - oC\J c fOrn;- -.- - c\j cN

DISTANCE (m) FIG. 4. Frequency distributions for the 11 rodent species of the distance between successive captures of the same individuals

in successive trapping periods 1 mo apart. Although these are raw data, not adjusted for the distribution of traps, note the high frequency of relatively short-distance movements and compare with lifetime movements in Fig. 5.

should be interpreted with caution. We suspect that above, and Zeng and Brown 1987b). This can be done some individuals of all species may live 3-5 yr in the only for a species population as a whole, because the field. fates of individuals that have disappeared cannot be

The other way we assessed mortality was by using determined. It is also subject to error unless the num- the estimated dispersal distances to correct the dis- bers of individuals are large and the dispersal distances appearances for the proportion of individuals that dis- are modest with respect to the size of the study area. persed into areas not sampled by traps (see Dispersal, For these reasons we have separated the disappearance

October 1989 1515

WI

' .

' .

.


60 -

45 -

I

D. spectabiiis n= 162

mmm ~ m m.

4

3

2

0 0 0 0 0 0 0 0 0 0

to 0 LO 0 i) 0 L5 0 C-- -C \J J r') r) I-

D. ordii

Pm. maniculatus n =5

iiii 0 0 0 0

LO 0 LO

I 0 0 0 0 0 o tn 0 LO 0 Cj C\ rf) re) <

n =78

4- Pm. eremicus

n 18 3

*C*:ii It ii 0 O 0 o 0 0 0o 0 o I

,--,o..II I '. . 1 1 1 1 U merriam i l .,. ,,..

0 0 0 0 0 0 0 n =306 ,n o ,n o to o ,n

10 - 0. leucogaste n254

0 o0 0 0 0 0 0 01 Ln ©. L. 0 Ln 0 LO 0 --

- ca c\a rn rN- B5-

~~~- C. penicillatus n =21

0 0 0 0 0 0 t 0 u) 0 Ct

O O O O M CM

10- Pg flovus

~1 07~n =46

5-

0__iL IiI m I I o o r- > o o o 0 0 c

L 0 L 0 L O- LO 0 C- - rC , r) t -

R. megalotis n =35

I

10

5

0

I0

5

0

0 0 lr-

o 0 o o Lo O I~( rI'

0. torridus

n =44

LO 0 to o LO o Lo 0 CM CM r o r 'I

N. albigula n =28

0 0 0 0 0 0 0 0 0 0 to 0 OO O 0 o 0 r n - - CM CM ro ro ^-r

DISTANCE ( m )

30

15

0

20

15

10

5

0

80

60

40

20

0

CO _J

:D

z

LL 0

cr LJ

:D

:7

7Z

10

5

i 0 I II I o 0 0 0 0 0 0 0 0

o 0 Lto 0 L 0 Lto 0 CM C\ r( rtO 1-

4

3

2

0 'ih ,.vML 0 0 0 0 0 0 0

to 0 to 0 o o 0 - \CM ro

I J I . & I L I .. . . . . ... ( iI, I J & I 1 J I, J J I i -

& 7 & . . . L_J I , i I . 1. . . I . J .. , . . . ... . . . . .............. 4

1-

. . . .. .

... ... I I__i__..~_J


.r

D ))


TABLE 3. Data on month-to-month and lifetime movements of individuals of the 11 species of rodents.

X distance between Median lifetime dispersal Percent first X no. Percent first X no. recaptures (m) distance (m)

captured recaptures Species as adult per individual 1 mo apart Lifetime Observed Calculated*

Dipodomvs spectabilis 74.6 3.79 21 36 20 32 Dipodomvs ordii 88.2 2.76 24 51 25 53 Dipodomvs merriami 83.9 3.60 24 52 26 70 Chaetodipus penicillatus 77.5 2.02 33 77 30 150 Perognathusflavus 99.1 2.13 37 87 40 202 Reithrodontomys megalotis 96.1 2.21 87 164 95 286 Peromvscus maniculatus 74.3 1.91 65 91 75 116 Peromvscus eremicus 78.7 2.37 76 117 110 148 Onvchomys leucogaster 73.8 2.94 71 91 72 133 Onvchomvs torridus 77.9 2.08 70 95 73 132 Neotoma albigula 62.9 2.25 23 63 23 109

* This is the best estimate of distances moved over a lifetime. See Zeng and Brown (1987b) for method of calculation and other details.

rate into dispersal and death rates only for the three Dipodomys species (Table 5). Note that 70-75% of the individuals of each species disappeared each year, and about half of these disappearances can be attributed to dispersal off the trapping grids and the other half to mortality. Dispersal appears to be less pro- nounced in D. spectabilis, which inhabits large per- manent mounds that may be occupied by parents and offspring for many generations (Jones 1984, 1986).

Although we did not feel that we could accurately calculate death rates for the other species by this method, we note that most of them had substantially longer dispersal distances than the three kangaroo rat species (Table 3). Therefore we would expect a larger proportion of the disappearances to be due to dispersal rather than to mortality. Thus the steeper slopes of the persistence curves for these species, especially the murids and Pg. flavus (Fig. 6), may not mean that these species necessarily have higher mortality than the kangaroo rats. This is reinforced by the fact that maximum longevities of N. albigula, R. megalotis, and 0. torridus were comparable to those of the three Dipodomys species (Table 4).

DISCUSSION

Trade-offs

Most life history theory assumes a trade-offbetween reproductive investment and survivorship. We tested whether there was a significant cost of reproduction that was expressed in terms of reduced survivorship of those individuals that had made a reproductive commitment compared to those that had not engaged in reproductive activity. For each species we compiled a

2 x 2 contingency table comparing whether those individuals that were reproductively active or inactive were either recaptured or had disappeared. None of these tables showed significant nonrandomness (all Ps > . 1), except for Pm. maniculatus (. 1 > P > .05), which showed a greater tendency for those individuals that had not reproduced to disappear. Since one result significant at the <.1 level is expected in the 11 tests, we can attribute this result to chance. It is also consistent with the interpretation made below (see Dispersal) that individuals facultatively move to better sites as they become available.

We emphasize that this should not be construed as a strong test for postulated trade-offs between present reproduction and survival to future reproduction. A rigorous test for such a trade-off would hold all variables except magnitude of reproductive commitment constant: year, season, age, sex, body size, and so on. However, we do not have sufficient data for even the most abundant species (see also Zeng and Brown 1987a) to perform such an analysis. Also, it would be important to ensure that those individuals that disappeared had died rather than dispersed. We believe that two important lessons can be drawn from these results. First, it may not be easy to document postulated trade- offs between reproduction and survival at the level of individuals within populations. Although selection causing life history evolution is presumed to operate at this level, most of the "tests" of life history theory have involved comparisons between different populations or different species (e.g., Stearns 1976, 1977). Second, our data do imply that if there is a significant cost of reproduction, those individuals that make reproductive efforts do so at times when the environment

FIG. 5. Frequency distributions for the 11 rodent species of lifetime movements (distance between first and last capture) of all individuals that were captured at least 4 mo apart. Although these raw data have not been adjusted for the decreasing number of traps at increasing distances from the initial capture site, compare with Fig. 4 and note the substantial proportion of individuals that moved > 150 m.

October 1989 1517


05

02-

01 -

005 -

002 -

001

0.005 .

0002

0001

C

0.5,

)>- z

D Ln

:_ LL

cr _J

LU'

-- R. megalotis

-o-o Pm maniculatus

+-+-+ Pm. eremicus

0.2

0.1

005

002

001

0005

0002

0001

28 32 36 40 44 48

-.- O0 leucogaster

o-o 0. torridus

+-+-+ N. albigula

05

02

0.1

005

002- \

001 -. . ....

0.005- \- ++

0002 -

0.001-1 I I I ,,,,,,,,,,, I I I I I I I I I I I I I I I I I I I I I I I I I I I I I ,

0 4 8 12 16 20 24 28 32 36 40 44 48

NO. MONTHS

FIG. 6. Disappearance curves for the 1 rodent species. The proportion of individuals recaptured is plotted (on a logarithmic scale on the ordinate) as a function of the number of months since first capture (on the abcissa). These are analogous to survivorship curves, except that dispersal as well as death may account for the disappearance of individuals. Note that the small murid species (middle panel) had higher rates of disappearance than the heteromyids (upper panel) or large murids (lower panel), but these small murids also tended to disperse longer distances (see Table 3).

Ecology, Vol. 70, No. 5 1518


TABLE 4. Longevity and rates of disappearance of the 11 species of desert rodents.

Species

Dipodomys spectabilis Dipodomys ordii Dipodomys merriami Chaetodipus penicillatus Perognathus flavus Reithrodontomys megalotis Peromyscus maniculatus Peromyscus eremicus Onychomys leucogaster Onychomys torridus Neotoma albigula

Maximum longevity

(mo) 50 35 43 25 35 40 19 19 24 35 45

is sufficiently favorable and their own condition is sufficiently good so that risks of mortality are minimized. This is the standard explanation for the highly seasonal reproduction often exhibited by rodents in other environments.

We also tested for a relationship between present nutritional status of individuals and their probability of survival. We reasoned that we might expect individuals that had been losing body mass (excluding pregnant females that gave birth) to suffer higher mortality than those that had been gaining mass, in part because they might spend more time foraging and ex- perience more intense predation. We tested for this using 2 x 2 contingency tables for each species to compare whether individuals that either gained or lost mass between two successive trapping periods either were recaptured or disappeared. All of these tests were indistinguishable from random (all Ps > .2), except for C. penicillatus in which individuals that gained mass showed a greater tendency to disappear (.1 < P < .05). This pattern might be expected for a hiberator such as C. penicillatus, but if so it implies substantial mortality or dispersal associated with hibernation. Since one such result would be expected by chance in 11 tests, we are reluctant to attach any significance to it.

We find the apparent lack of relationships between change in mass and disappearance to be interesting, since it suggests that a recent history of mass loss does not necessarily imply that a particular individual is any less fit than members of the population as a whole.

Coefficient of variation

Slope of persistence Disappearance of disappearance rate curve from 0 to 10 mo rate (yr-1) year-to-year

-0.060 0.70 0.18 -0.085 0.75 0.13 -0.070 0.72 0.09 -0.055 0.79 0.12 -0.099 0.77 0.12 -0.134 0.89 0.07 -0.129 0.86 0.10 -0.130 0.85 0.09 -0.082 0.79 0.24 -0.101 0.83 0.16 -0.097 0.76 0.18

Although we did not present data on changes in body mass in the populations of these 11 species (but see

Zeng and Brown 1987a), there were tendencies for individuals to gain and lose mass in synchrony, apparently in response to temporarily favorable and unfa- vorable environments. The analyses suggest that such periods of mass change were not strongly associated with altered risk of mortality. They further imply that if predation is a major cause of mortality, either the rodents do not respond to temporary energy deficits

by increasing the amount of time spent foraging, or

during periods of food stress foraging time can be increased without increasing the risk of predation, perhaps because other rodents behave similarly and the risk is shared by the population as a whole.

Dispersal

One of the striking features of our results is the high frequency and long distance of adult dispersal in all

species (Fig. 5, Table 3). More than 50% of the individuals that lived at least 4 mo moved >30 m from the site of their initial capture, and since the experimental plots measured 50 x 50 m, this means that most of them had dispersed to a new plot. These observations contrast with much of the traditional dogma regarding dispersal in small mammals. Dispersal has been viewed primarily as a phenomenon of newly weaned, recently independent juveniles or of compet- itively inferior adults in poor condition (e.g., Lidicker 1975, Shields 1982, Waser 1985). Movements of young

TABLE 5. Disappearance rate broken down into its component death rate and dispersal rate for the three kangaroo rat species. The method of Zeng and Brown (1987b) was used to estimate the proportion of those individuals that disappeared each year that had dispersed into areas not sampled by traps, and the remainder were assumed to have died. Coefficients of variation for death rate and dispersal rate from year-to-year are also given.

Species

Dipodomys spectabilis Dipodomys ordii Dipodomys merriami

X disappearance rate (yr-1)

0.70 0.74 0.72

X dispersal rate (yr-1)

0.39 0.39 0.37

X death rate (yr-1)

0.31 0.35 0.35

Coefficient of variation

Dispersal rate Death rate

0.56 0.43 0.46 0.35 0.43 0.35

October 1989 1519


E 300

Li

z 200 I- or)

J < 100 cr) 012 L! c) 60

z < 40

R. m

Pg9 f Pm. m.

0· 0 _ . C Op 0

t. --

Pm. e

D. m.®0 D D.o

Do

6 10 20 40 60 BODY MASS (g)

FIG. 7. Median lifetime dispersal distance of body mass plotted on logarithmic axes. The tance has been adjusted for the probability of re( to the placement of traps (see the last columr Note that an inverse relationship between disp and body size appears to characterize all but the 1

individuals have been interpreted as an a avoid competition with parents or siblings site or to prevent deleterious inbreeding. adults has usually been considered to be presumably because of the disadvantages familiar site and an established territory or

Although these traditional ideas ma, characterize some dispersal, especially th movements of immature individuals ofce' we doubt that they apply to most of the we have observed in these desert rodent Instead we suspect that individuals ofbotl they have become independent, move qu enhance their fitness. They are familiar area surrounding the heavily used core re home range. When better food, den site opportunities become available because ( the availability of resources, the death o of a previous resident, or the ability to disi individual from a superior site, individua istically shift the center of their activitie vantage of these changes. This view of disp more consistent with the distance and movements we have documented and wi vation that these shifts are often made b robust, healthy, and sometimes breedinl of both sexes. Other recent studies of srr populations have also begun to question tt dogma about dispersal and to advance in similar to ours (e.g., Waser and Jones 19 1985, Waser 1985, Gaines and Johnson

The relationship between body size a distance (Fig. 7) is of particular interes seems to challenge traditional thinking ; lometric scaling of home range size and (e.g., McNab 1963, Schoener 1968, Peters 1984). Our data suggest that the relatior linear: species with body sizes in the ran

g make the shortest lifetime movements, and both smaller and larger rodents disperse substantially far- ther. Although the linear relationship between dispersal distance and body mass is marginally significant (r = -0.54; .1 < P < .05), the linear fit becomes highly

N. a significant if both variables are log-transformed and the value for N. albigula, the largest species, is omitted (r = -0.93; P < .01). The best fit to the entire data set is given by a parabola in whose equation both the mass term and the square of mass contribute highly

D.s. significantly (both Ps < .01) to accounting for the ob- 100 200 served variation in lifetime dispersal distance.

Interestingly, Brown and Maurer (1987) have pre- as a function dicted such a curvilinear relationship between individ-

dispersal dis- ual movements and body mass in endothermic ver- capture owing tebrates based on the relationship between population

of Table 3) size and body size in birds. The fact that no species of )ersal distance argstTeciese. very small birds attained such high local population

densities as species in the size range of 50-150 g suggested that intense energy requirements constrain the

Ldaptation to smallest endotherms to move between the richest at their natal patches of habitat, using them in a coarse-grained fash- Dispersal of ion. These data on desert rodents support this inter- e infrequent, pretation. The three kangaroo rat species that move of leaving a the shortest distances also have higher population den- home range. sities (Table 1) and tend to use the environment in a y accurately more fine-grained way in the sense that they have been ie sex-biased captured in a larger proportion of experimental plots, rtain species, trap sites, and trapping periods (Brown and Kurzius, movements in press) than the smaller heteromyids and murids. The

populations. fact that individuals of the small granivorous species h sexes, once (Pg. flavus, C. penicillatus, R. megalotis, Pm. manicu- ite readily to latus, and Pm. eremicus) travel distances of more than with a large 100 m to utilize sequentially different plots from which gion of their kangaroo rats have been experimentally removed (see s, or mating Brown and Munger 1985) further supports the inter- )f changes in pretation that increasingly severe energy constraints )r movement cause the smallest rodents to use multiple, widely place another spaced, exceptionally rich patches over their lifetimes. ils opportun- s to take ad- Interspecific variation )ersal is much These data provide an almost unique opportunity frequency of to compare the population ecologies in a diverse as- th the obser- semblage of coexisting species of varying biological )y apparently similarity and taxonomic affinity. To what extent are g individuals the life histories and population dynamics of these lall mammal species similar because of the constraints of an identical he traditional environment and close taxonomic relatedness, and to terpretations what extent are they different because of interspecific )83, Lidicker competition, resource allocation, and distant related- 1987). ness? Lnd dispersal We applied time series analysis to the data on pop- ;t, because it ulation densities to calculate cross correlation coeffi- about the al- cients (r; for zero lag time, which is equal to the Pearson l movements product-moment correlation coefficient) between each 1983, Calder pair of species for: (a) year-to-year variation in density, iship is non- using the average density of each species in each year; ge of 40-140 and (b) seasonal variation in density, using the average


) -

) -


TABLE 6. Significant cross correlations (0 time lag) in population density between pairs of species from year to year (N = 9 yr). Data are broken down by rodent family; pairs of species are indicated by the first letters of the generic and specific names (see Table 1). Note that all correlations are positive.

Heteromyid/heteromyid Murid/murid Heteromyid/murid D.. -D.o. 0.73 P.e.-R.m. 0.87 D.o.-P.m. 0.70 D.o.-C.p. 0.70 P.m.-R.m. 0.83 D.o.-R.m. 0.67 D.s. -P.f 0.68 P.m.-P.e. 0.68

density of each species for each month as plotted in Fig. 2. To increase the power of these analyses we used a more complete data set that included nine full years of censuses (1977-1986). In each analysis, the number of pairwise interspecific comparisons was 55, but the sample sizes were modest (N = 9 yr and N = 12 mo, respectively) so high values of the correlation coefficients are required for statistical significance.

For the year-to-year variation, all eight of the significant cross correlation coefficients were positive (Ta- ble 6). We call attention to three interesting results of this analysis. First, the facts that all significant correlations were positive, that these involved 8 of the 11 species, and that these 8 are the largely granivorous species suggest that the species with the most similar resource requirements responded similarly to interannual variation in their common environment. A good year for one species was good for most of the others. This is consistent with the fact that the seeds consumed by all these species are produced sporadically in desert environments in response to precipitation sufficient to wet the soil and result in significant primary production (e.g., Went 1949, Tevis 1958, Beatley 1969, 1974, 1976, Brown et al. 1979). Second, the lack of any strong negative relationships suggests that the interannual fluctuations in density cannot be attributed to strong competitive exclusion and/or resource division on a temporal basis (but see Hallett 1982). Third, some of the species that showed very similar year-to-year fluctuations in abundance were closely related (e.g., congeneric D. ordii and D. merriami, r = 0.73), but others were in completely different families (e.g., D. ordii and P. maniculatus, r = 0.70). This suggests that interannual fluctuations in population density are more strongly affected by convergent similarities in diet than by phy- logenetically conservative aspects of history and demography.

The patterns of seasonal fluctuations in density were much more variable. Out of 55 comparisons, there were almost equal numbers of each sign, 9 significantly positive and 7 significantly negative (Table 7). We call attention to three aspects of these results. First, there is no best time of year for reproduction and recruitment of all species. This is reinforced by Figs. 2 and 3, which show that peaks of reproductive activity and population sizes can occur at any time from midwinter to midsummer depending on the species. A wide variety of reproductive patterns appears to be capable of main-

taining rodent populations in this environment. Sec- ond, closely related species showed both some of the most similar seasonal patterns (e.g., D. ordii and D. merriami, r = 0.81; and 0. leucogaster and 0. torridus, r = 0.70) and some of the most different patterns (e.g., D. spectabilis and D. merriami, r = -0.49; and R.

megalotis and N. albigula, r= -0.90). Distantly related

species exhibited both seasonally synchronous (e.g., D. merriami and 0. leucogaster, r = 0.90) and asynchro- nous (e.g., C. penicillatus and 0. leucogaster, r = -0.87) patterns of seasonal population fluctuation. Clearly, closely related species are neither evolutionarily con- strained to respond to seasonal environmental variation in similar ways, nor do they necessarily exhibit marked differences associated with competition and/ or resource allocation. Third, there are no obvious re-

lationships between the seasonality of recruitment and

any aspect of the biology of these 11 species, including body size, morphology, physiology, diet, geographic distribution, and taxonomy. The analysis suggests that there may be processes that promote both similarities and differences in life histories and population dynamics among the coexisting species in this assemblage, but

they do not act in such a way that we can predict a

priori which species will occur in the habitat, what characteristics of population ecology they will possess, and how the resulting community will be organized.

This wide variation and lack of clear patterns seem to hold for most aspects of population ecology. Thus, the heteromyids include the most and least seasonal breeders (C. penicillatus and D. merriami, respectively) and the highest and lowest sex ratios (Pg. flavus and C. penicillatus, respectively), the murids include the most and least variable populations from year-to-year (Pm. maniculatus and 0. torridus, respectively), and

dispersal appears to be influenced by body size, but

relatively unaffected by taxonomy, mode of locomo-

tion, diet, or other aspects of population ecology. Al-

though the literature might suggest that the hetero-

myids have more "K-selected" and the murids more "r-selected" life histories (e.g., Whitford 1976, Conley et al. 1977), a critical examination of our data provides little support for this generalization. Slopes of the disappearance curves would suggest such a trend (Fig 6), but the facts that many of the disappearances probably represent dispersal rather than death and that the maximum longevities are comparable (Table 4) suggest that these may be misleading. Although we have not mea-

October 1989 1521


TABLE 7. Significant cross correlations (0 time lag) in average population density between pairs of species from month to month (N = 12 mo). Data are broken down by sign of correlation and by rodent family; pairs of species are indicated by the first letters of the generic and specific names (see Table 1).

Heteromyid/heteromyid Murid/murid Heteromyid/murid Positive

D.o.-D.m. 0.81 O.l.-O.t. 0.70 D.m.-O.l. 0.90 P.e.-R.m. 0.69 D.o.-O.l. 0.86

P.f-P.m. 0.78 C.p.-N.a. 0.72 D.m.-O.t. 0.70 D.o.-R.m. 0.69

Negative D.o.-C.p. -0.88 R.m.-N.a. -0.90 C.p.-O.l. -0.87 D.m.-C.p. -0.78 C.p.-R.m. -0.78

C.p.-P.e. -0.66 D.o.-N.a. -0.59

sured litter sizes, the data in the literature suggest that they are very similar, averaging 2.0-4.3 in these species of murids (Hoffmeister 1986) and 2.0-4.7 in the heteromyids (Hoffmeister 1986, Jones, in press). The apparent ability of some of the murids, such as Pero- myscus and Reithrodontomys, to produce litters more rapidly may be compensated for by the ability of the heteromyids to reproduce under more adverse conditions (see Whitford 1976).

Taken together, the data on comparative population ecologies of these 11 species suggest great interspecific variation. In fact, the data presented here probably seriously underestimate the total variation among all the rodent species that coexist in this Chihuahuan Des- ert habitat. Not included in this study, because of small sample sizes, are ground squirrels (Ammospermophilus harrisi and Spermophilus spilosoma, family Sciuridae), pocket gophers (Thomomys bottae, family Geomyi- dae), and cotton rats (Sigmodon hispidus, family Mu- ridae), each of which may have characteristics of life history and population dynamics as different from the species we have studied as the latter are from each other.

Population ecology and community structure

The desert rodent community at our study site was comprised of at least 11 reasonably common and 6 rarer species, but this is an overgeneralization. Com- munity composition varied temporally, both seasonally and from year-to-year, as species increased and decreased in different patterns. Some of the more common species disappeared completely for substantial periods. Thus, C. penicillatus entered hibernation and dropped out of the active community each winter, and Pg. flavus hibernated some winters but remained active and even reproduced during others. Pg. flavus, Pm. maniculatus, Pm. eremicus, and R. megalotis were completely or virtually absent from the study site for years at a time, yet when they were present they reproduced, successfully recruited juveniles, and were among the most abundant species. This temporal vari-

ation in community composition is analyzed in more detail elsewhere (Brown and Kurzius, in press). Here it is sufficient to emphasize that the "community" was very dynamic, varying even in its species composition as some species colonized or went locally extinct in response to the factors affecting their local population dynamics.

How should the variation in population ecology among species be interpreted, and what is its relationship to coexistence of species in this diverse assemblage? We consider three possibilities: coevolution, in- teractive sorting, and individualistic assembly. There is no reason to think that any of the patterns represent the results of coevolution of particular pairs or larger sets of species to reduce overlap in use of limiting resources and thus to facilitate coexistence. For one thing, the modest intraspecific variation in the population ecologies of these species does not appear to be related to occurrence with certain other species. Al- though population densities vary considerably among habitats (e.g., Brown and Kurzius 1987), all features of life history and demography of the populations inhabiting our study site apparently fall within the range of variation reported for these species at other sites (e.g., Reynolds 1958, 1960, Egoscue 1960, Eisenberg 1963, Chew and Butterworth 1964, MacMillen 1964, French et al. 1967, 1968, 1974, 1975, Bradley and Mauer 1971, Kenagy 1973, Maza et al. 1973, O'Farrell 1974, Smith and Jorgensen 1975, Whitford 1976, Con- ley et al. 1977, Wagner 1981, Petryszyn 1982, Munger et al. 1983, Jones 1985, in press, Kenagy and Bartholo- mew 1985, Hoffmeister 1986). Furthermore, there is so much variation in the combinations of species that occur together at different times or in different places (Brown and Kurzius 1987, in press), that it is difficult to imagine that coevolution could be sufficiently fine tuned to cause adaptations in one species in response to coexistence with another (Schroder 1987, but see Schroder and Rosenzweig 1975).

It is much more likely that the variety of population ecologies represented by these species are the outcome



of some kind of sorting for compatibility (what Janzen [1985] calls "ecological fitting"). There is abundant evidence that interspecific competition plays a significant role in the structure and dynamics of desert rodent assemblages, including this one (e.g., Munger and Brown 1981, Brown and Munger 1985, Brown et al. 1986, Bowers et al. 1987). Additional evidence shows that species of desert rodents that differ in body size, morphology, and taxonomic affinity are more likely to coexist in local assemblages than species that are similar in these attributes (Bowers and Brown 1982, Brown 1987, Hopf and Brown 1986, Brown and Harney, in press). Since differences in life histories and population dynamics are likely to be correlated with differences in resource use (the cause/effect relationship can go both ways), it might be conjectured that the variation in population ecologies reflects differences in resource use that promote or at least permit coexistence. Unfortu- nately this is a very difficult proposition to test. Not enough is known about the life histories and demographies of all those species that might be considered to constitute the pool of potential colonists from which local communities are assembled. This makes it im- possible to build and test null models. The lack of clear patterns, especially among the congeneric species, suggests that community organization in desert rodents is less apparent in the population ecologies of the coexisting species than it is in their body size, morphology, and taxonomy.

A third possibility is that the observed variation reflects the relatively independent assembly of species. Since this particular assemblage of 11 species (like any other combination) occurs infrequently in the North American deserts (Brown and Kurzius 1987), it can be assumed that the unique combination of traits possessed by each species evolved in response to physical and biotic environments somewhat different from those at our study site. But the life history and demographic attributes of each species enable them to exist and coexist in this local environment. This interpretation does not exclude the possibility that interspecific interactions play a major role in determining the relative abundances, microspatial distributions, or even pres- ence or absence of certain species with particular combinations of traits (see also Schroder 1987). However, it suggests that potentially competing species are features of the local environment that are not necessarily any more important in determining which life histories and demographies will permit existence of each species in that environment than are other features such as physical conditions, predators, and food resources.

Our own opinion is that the population ecologies of these 11 species reflect some combination of interac- tive sorting and independent assembly of species that have evolved their attributes primarily in other contexts. Some of the similarities and differences among species reflect evolutionary constraints of close and distant phylogenetic relationships, respectively. Other

similarities reflect the environmental constraints of the spatial and temporal variation in limiting factors at this locality, whereas other differences reflect adaptations, even in closely related species, to different environmental conditions in other parts of their present or past range. We suspect that at least some of the differences in life histories and population dynamics promote coexistence in the sense that they enable the species to use different resources or to use the same resources in different ways. Together with the high productivity and the large spatial and temporal variation that make a variety of resources available in this habitat and the historical biogeographic events that have produced a large regional pool of species potentially able to colonize the habitat (Findley 1969), these in- terrelated differences in population ecology and resource utilization probably account for the large number of rodent species that coexist at this site.

ACKNOWLEDGMENTS

This manuscript benefitted from the helpful comments of G. Ceballos, L. Hawkins, E. Heske, A. Kodric-Brown, M. Skupski, and two anonymous reviewers. We are indebted to the people, too numerous to list individually, who have con- tributed to setting up and maintaining the experiments, trapping the rodents, and maintaining and analyzing the data. From the beginning this research has been supported by the U.S. National Science Foundation, most recently by Grant BSR-8506729 to J. H. Brown. A foreign study award from the People's Republic of China made possible the collabo- ration of Z. Zeng.

LITERATURE CITED

Beatley, J. C. 1969. Dependence of desert rodents on winter annuals and precipitation. Ecology 50:721-724.

1974. Phenological events and their environmental triggers in Mohave Desert ecosystems. Ecology 55:856- 863.

1976. Environments of kangaroo rats (Dipodomys) and effects of environmental change on populations in southern Nevada. Journal of Mammalogy 57:67-93.

Bowers, M. A., and J. H. Brown. 1982. Body size and coexistence in desert rodents: chance or community structure. Ecology 63:391-400.

Bowers, M. A., D. B. Thompson, and J. H. Brown. 1987. Spatial organization of a desert rodent community: food addition and species removal. Oecologia 72:77-82.

Bradley, W. G., and R. A. Mauer. 1971. Reproduction and food habits of Merriam's kangaroo rat, Dipodomys merriami. Journal of Mammalogy 52:497-507.

Brown, J. H. 1984. Desert rodents: a model system. Acta Zoologica Fennica 172:45-49.

1987. Variation in desert rodent guilds: patterns, processes, and scales. Pages 185-203 in J. H. R. Gee and P. S. Giller, editors. Organization of communities: past and present. Blackwell Scientific, New York, New York, USA.

Brown, J. H., D. W. Davidson, J. C. Munger, and R. S. Inouye. 1986. Experimental community ecology: the desert gran- ivore system. Pages 41-61 in J. Diamond and T. J. Case, editors. Community ecology. Harper and Row, New York, New York, USA.

Brown, J. H., and B. A. Harney. In press. Population and community ecology of heteromyid rodents in temperate habitats. In H. H. Genoways and J. H. Brown, editors. The biology of the rodent family Heteromyidae. American So- ciety of Mammalogists Special Publication.

October 1989 1523


Brown, J. H., and M. A. Kurzius. 1987. Composition of desert rodent faunas: combinations of coexisting species. Annales Zoologici Fennici 24:227-237.

Brown, J. H., and M. A. Kurzius. In press. Spatial and temporal variation in guilds of North American desert rodents. In Z. Abramsky, B. J. Fox, D. W. Morris, and M. R. Willig, editors. Ecology of small mammal communities. Special Publication of the Museum, Texas Tech University, Lubbock, Texas, USA.

Brown, J. H., and B. A. Maurer. 1987. Evolution of species assemblages: effects of energetic constraints and species dynamics on the diversification of the North American avi- fauna. American Naturalist 130:1-17.

Brown, J. H., and J. C. Munger. 1985. Experimental ma- nipulation of a desert rodent community: food addition and species removal. Ecology 66:1545-1563.

Brown, J. H., O. J. Reichman, and D. W. Davidson. 1979. Granivory in desert ecosystems. Annual Review of Ecology and Systematics 10:201-227.

Calder, W. A., III. 1984. Size, function, and life history. Harvard University Press, Cambridge, Massachusetts, USA.

Chew, R. M., and B. B. Butterworth. 1964. Ecology of rodents in Indian Cove (Mojave Desert), Joshua Tree Na- tional Monument, California. Journal of Mammalogy 45: 203-225.

Conley, W., J. D. Nichols, and A. R. Tipton. 1977. Repro- ductive strategies in desert rodents. Pages 193-214 in R. H. Wauer and D. H. Riskind, editors. Symposium on the Biological Resources of the Chihuahuan Desert Region. United States Department of Interior, National Park Ser- vice, Transactions and Proceedings Series 3:1-658.

Egoscue, H. J. 1960. Laboratory and field studies of the northern grasshopper mouse. Journal of Mammalogy 41: 99-110.

Eisenberg, J. F. 1963. The behavior of heteromyid rodents. University of California Publications in Zoology 69:1-114.

Findley, J. S. 1969. Biogeography of Southwestern boreal and desert mammals. Pages 113-128 in J. K. Jones, Jr., editor. Contributions in mammalogy. University of Kansas Museum of Natural History, Miscellaneous Publication 51.

French, N. R., B. G. Maza, and A. P. Aschwanden. 1967. Life spans of Dipodomys and Perognathus in the Mohave Desert. Journal of Mammalogy 48:537-548.

French, N. R., B. G. Maza, H. O. Hill, A. P. Aschwanden, and H. W. Kaaz. 1974. A population study of irradiated desert rodents. Ecological Monographs 44:45-72.

French, N. R., D. M. Stoddart, and B. Bobek. 1975. Patterns of demography in small mammal populations. Pages 73- 103 in F. B. Golley, editor. Small mammals: their productivity and population dynamics. Cambridge University Press, London, England.

French, N. R., T. Y. Tagami, and P. Hayden. 1968. Dis- persal in a population of desert rodents. Journal of Mam- malogy 49:272-280.

Gaines, M. S., and M. L. Johnson. 1987. Phenotypic and genotypic mechanisms for dispersal in Microtus populations and the role of dispersal in population regulation. Pages 162-179 in B. D. Chepko-Sade and Z. T. Halpin, editors. Mammalian dispersal patterns. University of Chi- cago Press, Chicago, Illinois, USA.

Hallet, J. G. 1982. Habitat selection and the community matrix of a desert small-mammal fauna. Ecology 63:1400- 1410.

Hoffmeister, D. F. 1986. Mammals of Arizona. University of Arizona Press, Tucson, Arizona, USA.

Hopf, F. A., and J. H. Brown. 1986. The bull's-eye method for testing randomness in ecological communities. Ecology 67:1139-1155.

Janzen, D. H. 1985. On ecological fitting. Oikos 45:308- 310.

Jones, W. T. 1984. Natal philopatry in bannertailed kangaroo rats. Behavioral Ecology and Sociobiology 15:151- 155.

1985. Body size and life history variables in heteromyids. Journal of Mammalogy 66:128-132.

1986. Survivorship in philopatric and dispersing kangaroo rats (Dipodomys spectabilis). Ecology 67:202-207.

In press. The social systems of heteromyid rodents. In H. H. Genoways and J. H. Brown, editors. The biology of the rodent family Heteromyidae. American Society of Mammalogists Special Publication.

Kenagy, G. J. 1973. Daily and seasonal patterns of activity and energetics in a heteromyid rodent community. Ecology 54:1201-1219.

Kenagy, G. J., and G. A. Bartholomew. 1985. Seasonal reproductive patterns in five coexisting California desert rodent species. Ecological Monographs 55:371-397.

Lidicker, W. Z., Jr. 1975. The role of dispersal in the demography of small mammals. Pages 103-128 in F. B. Gol- ley, editor. Small mammals: their productivity and population dynamics. Cambridge University Press, London, England.

1985. Dispersal. Pages 420-454 in R. H. Tamarin, editor. Biology of New World Microtus. American Society of Mammalogists Special Publication Number 8.

MacMillen, R. E. 1964. Population ecology, water relations, and social behavior of a southern California semidesert rodent fauna. University of California Publications in Zo- ology 71:1-66.

Maza, B. G., N. R. French, and A. P. Aschwanden. 1973. Home range dynamics in a population of heteromyid rodents. Journal of Mammalogy 54:405-425.

McNab, B. K. 1963. Bioenergetics and the determination of home range size. American Naturalist 97:133-140.

Munger, J. C., M. A. Bowers, and W. T. Jones. 1983. Desert rodent populations: factors affecting abundance, distribution, and genetic structure. Great Basin Naturalist Memoirs 7:91-116.

Munger, J. C., and J. H. Brown. 1981. Competition in desert rodents: an experiment with semipermeable exclosures. Sci- ence 211:510-512.

O'Farrell, M. J. 1974. Seasonal activity patterns of rodents in a sagebrush community. Journal of Mammalogy 55:809- 823.

Peters, R. H. 1983. Ecological implications of body size. Cambridge University Press, London, England.

Petryszyn, Y. 1982. Population dynamics of nocturnal desert rodents: a nine-year study. Dissertation. University of Arizona, Tucson, Arizona, USA.

Reynolds, H. G. 1958. The ecology of Merriam's kangaroo rat (Dipodomys merriami) on the grazing lands of southern Arizona. Ecological Monographs 28:111-127.

1960. Life history notes on Merriam's kangaroo rat in southern Arizona. Journal of Mammalogy 41:48-58.

Schoener, T. W. 1968. Sizes of feeding territories among birds. Ecology 49:123-131.

Schroder, G. E. 1987. Mechanisms for coexistence among three species of Dipodomys: habitat selection and an alter- native. Ecology 68:1071-1083.

Schroder, G. E., and M. L. Rosenzweig. 1975. Perturbation analysis of competition and overlap in habitat utilizations between D. ordii and D. merriami. Oecologia 19:9-28.

Seber, G. A. F. 1986. A review of estimating animal abundance. Biometrics 42:267-292.

Shields, W. M. 1982. Philopatry, inbreeding, and the evolution of sex. State University of New York Press, Albany, New York, USA.

Smith, H. D., and C. D. Jorgensen. 1975. Reproductive biology of the North American desert rodents. Pages 305- 330 in I. Prakash and P. K. Ghosh, editors. Rodents in



desert environments. W. Junk, The Hague, The Nether- lands.

Steams, S. C. 1976. Life-history tactics: a review ofthe ideas. Quarterly Review of Biology 51:3-47.

1977. The evolution of life history traits: a critique of the theory and review of the data. Annual Review of Ecology and Systematics 8:145-171.

Tevis, L. 1958. A population of desert annuals germinated by less than one inch of rain. Ecology 39:688-695.

Wagner, F. 1981. Population dynamics. Pages 125-168 in Arid land ecosystems: structure, functioning and manage- ment. Volume 2, International Biological Program 17. Cambridge University Press, London, England.

Waser, P. M. 1985. Does competition drive dispersal? Ecol- ogy 66:13-21.

Waser, P. M., and W. T. Jones. 1983. Natal philopatry among solitary mammals. Quarterly Review of Biology 58: 355-390.

Went, F. W. 1949. Ecology of desert plants. II. The effect of rain and temperature on germination and growth. Ecol- ogy 30:1-13.

Whitford, W. G. 1976. Temporal fluctuations in density and diversity of desert rodent populations. Journal of Mam- malogy 57:351-369.

Zeng, Z., and J. H. Brown. 1987a. Population ecology of a desert rodent: Dipodomys merriami in the Chihuahuan Desert. Ecology 68:1328-1340.

Zeng, Z., and J. H. Brown. 1987b. A method for distin- guishing dispersal from death in mark-recapture studies. Journal of Mammalogy 68:656-665.

October 1989 1525

comparative population ecology of eleven species...

Documents