Vous êtes sur la page 1sur 9

Estimating the Distribution and Abundance of a Cryptic Species, Dipodomys stephensi (Rodentia: Heteromyidae) , and Implications for Management

M. V. PRICE*
Department of Biology University of California Riverside, CA 92521, USA.

P. R. E N D 0
Department of Biology University of California Riverside, CA 92521, USA.

Abstrack A common hurdle in developing conservation plans for endangered species threatened by habitat loss is lack of basic information on distribution and factors affecting population densities. Such information is timeconsuming and expensive to obtain for cryptic species, such as nocturnal small mamma& whosepresence at a site can only be determined through intensive sampling. We illustrate how readily-available soils maps can be used to estimate the potential distribution of such a species, Stephens kangaroo rat. In this case the distribution map accurately predicts presence or absence of this species at 70% of sites Our maps reveal extensive habitat loss and pagmentation. Data from repeated censuses at several sites indicate that population densities of this species can vary more than 10fold in response to rainfallpatterns, and that temporalfluetuations in geographically separated populations are e w e lated These results suggest that habitat subdivision is unlikely to substantially increase expected time to extinction for the species as a whole; habitat conservation efforts for Stephens kangaroo rat should therefore be directed at establishing a few large, widely separated rvsems, rather than many smaller reserves.
Correspondence should be addressed to M. V. Price. Paper submitted March 28, 1988; w i s e d manuscript accepted January 16 1989.

Resumen: Un obstaculo comlin en el desamllo deplanes


de conservaci6npara especies en peligm, amenuxadaspor la perdida de habita4 es lafalta de infmnuciha bdsica sobre la distribuci6n y losfactores que afectan la densLdad depoblaci6a La obtenci6n de esta informaci6n demora much0 en conseguirse y es costosa para obtenerla para especies cripticas, tales como pequetios mamijkros nocturnos, cuya presencia solo puede ser determinudo por muestreo intensiuo. Ilustramos como mapas de suelos, que son facilmente d k ponibles pueden ser usadospara estimar la distribucidnpotencial de una de estas especies: la Rata canguro de Stephens. En este caso, el mapa de distribuci6n predice con exactitud la presencia o ausencia de esta especie en el 70% de los lugares Nueslms mapas revelan un extensa p6rdidu y flag mentacih del habitat de esta especie. Datos, de censos repetidos en various lugares, indiean que las densidades de poblaci6n de esta especiepueden variar en msiS de 10 veces en respuesta a pahones deprecipitacih, y quepuctuaciones temporales en poblaciones geograficamente separadas estzin cwelacionudus. Estos resultados sugieren que la subdivisi6n de habitat no parece aumentar substancialmente el tiempo esperado de extincidnpara la especies en general Por ello, los esfuerzos de conservaci6n de habitat para la Rata canguro de Stephens deben ser dirigidos a establecer algunas pocas reservas, grandesy bien s e p a r m en vex de muchasy pequetias reservas.

293
ConservationBiology Volume 3, No. 3, scptcmber 1989

294

Stephens hhgamo Rat Distribution

Price ti Endo

Introduction
In principle, developing a conservation plan for species facing extinction because of habitat destruction is straightforward: simply protect enough habitat to maintain one or several populations larger than the size (hereafter the critical size) that meets a specified probability of persistence over a specified time period. In practice, however, developing an effective plan is difficult because the requisite information is rarely available. Determining the critical population size, for example, requires detailed, long-term demographic and genetic information that is currently available for a handful of species at most (Shaffer 1981 ). Nevertheless, existing models do allow crude estimates, and the lack of detailed information affects our confidence in the longterm success of a plan more than it affects our ability to come up with one that should work reasonably well over the short term. Of more immediate consequence is uncertainty about the distribution of species, because targeting the most appropriate areas for preservation is difficult without a reasonably accurate distribution map. The primary effect of inadequate distributional information is that known localities become the primary focus of conservation efforts, even though they may be more expensive to acquire and less suitable as preserves than unknown alternatives. Arriving at a detailed distribution map is not easy. Direct spot-census methods are impractical because the cost of censusing enough localities to achieve reasonable resolution would be prohibitive especially for cryptic species like nocturnal small mammals whose presence or absence can be established with confidence only after intensive sampling. The only alternative is to map the habitats with which the species is associated. This may be straightforward if the species is associated with a habitat easily recognized on topographic maps or aerial photographs. In many cases, however, the habitat may be identified only with sophisticated remotesensing methods. We have been able to map the potential distribution of an endangered species of kangaroo rat (Dipodomys stepbensi), using readily available soil maps. Because our method is likely to be widely applicable for small terrestrial vertebrates and for plants, we describe here how we developed habitat maps for western Riverside County, Calif. (U.S.A.) at three points in time. We provide an estimate of the accuracy of our method and use it to quantify temporal changes in habitat availability. Finally, we discuss how the map, coupled with information on temporal fluctuations in population densities, can be used to guide development of a habitat conservation plan for this species.

by C. H. Merriam (1907).A summary of the literature on this species can be found in Bleich (1977).As is common among the kangaroo rats (genus Dipodomys), D. stepbensi is a seed-eating,nocturnal, burrowing rodent. Its entire geographic range, estimated at 287,000 ha (Kramer 1 9 8 7 ) , is centered in the SanJacinto and Perris Valleys of western Riverside County, California,with minor extensions south into San Diego county and north into San Bernardino county. This is an unusually small range for rodents in general and kangaroo rats in particular (cf. Burt & Grossenheider 1964). Within its range the species occurs at lower elevations in flat or gently rolling annual grassland (Bleich 1 9 7 7 ) . It is replaced on steeper slopes and in shrublands by a morphologically similar kangaroo rat, D. ugfZk In 1987 the U.S. Fish and Wildlife Service proposed that the species be given endangered status under the Endangered Species Act of 1973 (Kramer 1987) because of extensive habitat loss to agricultural and urban development; in 1988 (Kramer 1988) the species w a s listed. The literature on Stephens kangaroo rat is sparse, consisting primarily of treatments of its systematic status and records of its presence or absence at particular sites from short-term live-trappingstudies (Bleich 1977; Kramer 1987,1 9 8 8 ) . Much of the distributional information is available only in the form of unpublished reports to private landowners or public agencies such as the Bureau of Land Management or California Department of Fish and Game. The most extensive demographic studies are reported in unpublished undergraduate (Moore-Craig 1984) or masters theses (Bleich 1973;Thomas 1975;Bontrager 1 9 7 3 ) . While the available information on this species is sufficient to justify endangered status, it provides a meager basis for planning a conservation strategy. Our purpose in undertaking this mapping project was to obtain quantitative estimates of the distribution of Stephens kangaroo rat in western Riverside County prior to urban and agricultural development, and of the amount and distribution of potential habitat still remaining.

Methods
Many terrestrial vertebrates have relatively strict habitat requirements, and kangaroo rats are no exception. The basis for these habitat a n i t i e s is known in only a few cases, but the fact that they exist provides an extremely useful tool for predicting species distributions (e.g., Price 1978; Price & Waser 1984). In reviewing the literature on Stephens kangaroo rat, we found that it is reported to occur below about 61 0m elevation, in flat or gently sloping, often degraded, annual grasslands with sparse cover of perennial shrubs. On the San Jacinto Wildlife Area in Riverside County, Moore-Craig (1984)trapped D. stepbensi on 1 6 of 28 sampled sites. The 1 6 sites all had slopes of 7-lo%, less than 15% shrub cover, and nonclay soils. Moore-Craigs

The Species
Dfpodomys stepbensi, Stephens kangaroo rat, is a species in the rodent family Heteromyidae first described
Conservation Biology Volume 3, No. 3, September 1989

Price & Endo

Stephens Kangaroo Rat Distrjbution

295

results are completely consistent with other reports (Thomas 1975; Lackey 1967; Bleich 1 9 7 7 ) . If one assumes that the present-dayhabitat association of D. stephensi reflects its original habitat association, then the original distribution of sparse annual grassland provides an estimate of the speciesdistribution in westem Riverside County after Spanish ranching had established annual grasslands, but before modern urban and agricultural development (hereafter originaldistribution). Presumably Stephens kangaroo rat inhabited the perennial bunchgrass vegetation that once occupied the sites of present-day degraded annual grassland. As estimate of the current potential distribution can then be obtained by subtracting urban and agricultural lands from the original distribution. This was the basic procedure we employed. The Soil Survey of Western Riverside County Area (USDA 1971)provides all the information necessary to map the potential extent of annual grassland. The survey classifies soils by elevation and slope as well as by features such as texture and depth; provides descriptions of vegetation supported by each soil type; and maps all recognized soil categories onto a series of aerial photographs. We compiled a list of soils that potentially support Stephens kangaroo rat habitat by first listing all soils described as supporting annual grasses mixed with forbs (cf. Thomas 1975) and shrub species characteristic of sparse California coastal sage scrub vegetation rather than chapparral or oak woodland. We then excluded heavily alkaline or clay soils (generally in flood plains), excessively rocky soils (generally on steep slopes), soils less than about 20 inches deep (kangaroo rat burrows are often 1 8 or more inches deep; Thomas 1 9 7 5 ) ,those occurring only above 610 m elevation, and those having greater than 25% slope. This left us with a list of soils potentially supporting Stephens kangaroo rats. The list is available upon request. We used acetate overlays of the soil survey photographs to delimit the extent of all suitable soils, and transferred these to a 1:100,000acetate overlay of a base map (Bureau of Land Management 1 :100,000scale series of Santa Ana and Palm SpMgs quadrangles) using a Kargl Reflecting Projector (Keuffel & Esser Co.) that allows scale transformation o f images. The total area of suitable soils was estimated by passing cutouts from Xerox copies of the soil survey photographsthrough a Li-CorModel 3 100Area Meter. The size distributions of habitat patches were obtained by tracing the outline of each patch of contagious habitat (from the 1:100,OOObase map) on a digitizing pad (Summagraphics Bit Pad 11) with direct input into a Basic program written for an IBM-PC-ATcomputer to calculate areas of closed figures. Maps of land use in western Riverside County were available for 1938,from a 1:125,000 scale map of veg etation and land use (Weislander 1 9 3 8 ) , and for 1984, from 7 ?h minute topographic quadrangles prepared by the California State Department of Water Resources.

These maps were transferred to 1:100,000 acetate overlays of the base map, using the Kargl. We estimated the amount of potential D. stephensi habitat remaining in 1938 and 1984 by subtracting areas of urban and agriculturaldevelopment from the o r i w habitat map. The sizes of remaining habitat patches were determined with a digitizing pad, as described earlier. These procedures yielded maps and quantitative estimates of past and present distribution of Stephenskangaroo rat. To evaluate the accuracy of our estimates we took advantage o f recent surveys in which the trapping lines or grids could be located precisely on soil survey photographs. For 69 trapping sites we determined what soil types were present, recorded whether by our method we would have predicted presence or absence ofD. stephens4 and compared our predictions with census results. Most of the trapping data were derived from topographic maps compiled by the Bureau of Land Management from studies of Thomas (1975) and Hicks & Cooperrider ( 1977); the rest were taken from MooreCraigs (1984) study or from a collecting trip of Dr. Troy Best (Auburn University, unpublished data). The primary limitation on the number of sites we could include in the evaluation was our ability to locate the trapping site precisely on the soil survey photographs; often, locality information is reported to quarter section at best. It should be emphasized that we compiled our soils list from vegetation data, not from information on soil types present at known D. stephensi localities. For this reason, our evaluation procedure provides an unbiased estimate of the accuracy of predicting Stephenskangaroo rat distributions from soils data. To develop a habitat conservationplan one needs not only the sorts of distributionalinformation we compiled by the methods just described, but also some idea of how much land in how many separate parcels needs to be set aside. It is of course virtually impossible to specify with any precision what minimum population size would achieve persistence goals for any particular species (cf. Gilpin & Soule 1986; Shaffer l g S l ) , even if one had tremendous amounts of demographic and genetic information. Nevertheless, even simple models indicate that populations of 50-100 individuals are much more likely to persist in the face of demographic and environmental stochasticity than populations of 10 individuals (MacArthur & Wilson 1967; Gilpin & Soul6 1986; QuiM & Hastings 1987; Pimm et al. 1 9 8 8 ) , and that a species divided into several populations is more likely to persist, in the face of certain types of genetic and environmental stochasticity,than one consisting of a single larger population (Gilpin & Soul6 1986;Leigh 1975,1981;Quinn & Hastings 1 9 8 7 ) . These models give us a population size target, but how much land would be required to support a population of that size? To answer this question one needs information on temporal and spatial variation in population densities.
Conservation Biology Volume 3, No. 3, September 1989

296

Stephens Kangaroo Rat Distribution

Price & Endo

We obtained preliminary information about temporal density variation by periodically retrapping one of Moore-Craigs (1984) census sites (her site 14), and by comparing the results with those we have obtained over the same time period from long-term censuses of a nearby population of a similarly sized kangaroo rat species (0.agilis) on the University of Californias Motte Reserve, 23 km south of Riverside, Calif. The trapping protocol for the D. agilis population is described in Price & Waser (1984). The permanent trapping grid consists of five parallel lines 15 m apart, with 20 trapping stations per line spaced 10 m apart and two traps per station. Censuses have been conducted for three nights in late fall of each year since 1980. The grid covers an area of 190 X 45 m, but we estimate from maximum average between-station, within-census movements of individual D. agilis that it actually samples at least an additional 15 m peripheral zone (mean maximum movements during a sampling period = 14.4 m, SD = 16.42 m, n = 123 records with one or more recaptures per sampling period). Hence, we estimate the effective grid size as 220 X 75 m, or 1.6ha. This may very well be an underestimate, given the propensity of kangaroo rats to return to a known food source. For Moore-Craigssite 14 we established a live trapping grid of six to eight parallel lines of 10 stations each with 15 m between stations and two live-traps per station. We trapped this grid in March 1985 (six lines, 120 traps each night, three nights), June 1985 (six lines, 120 traps each night, two nights), February 1987 (eight lines, 160 traps each night, four nights), and March 1988 (eight lines, 160 traps each night, two nights). MooreCraig ( 1984) had trapped in January 1984 (two nights) and March 1984 (two nights) for a total effort of 198 trap-nights. Standard trapping procedures were followed: traps were opened at dusk, baited with rolled oats, and checked and closed the next morning. Captured individuals were ear-tagged or fur-clipped for individual identification. Population sizes were estimated by direct enumeration, and densities were estimated by dividing numbers known alive by the area of the grid, augmented by a peripheral zone 1 5 m wide (see above). We did not have enough recapture information for D. stephensi to estimate movement distances accurately for this species; but since the few observed values were consistent with those obtained for D. agiZis on the Motte Reserve, we used the latter data to determine the peripheral zone.

(37%) of habitat remained (Fig. 2). About the same total amount, 50,518 ha, remained in 1984 (Fig. 3), although

shifts in land use caused the spatial distribution of remaining habitat to change. We included abandoned agricultural lands as potential Stephenskangaroo rat habitat, because the species is known to invade cultivated fields in suitable locations once plowing ceases (Thomas 1975; Moore-Craig 1984). Since 1984, accelerating urban development has made substantial inroads into remaining habitat (personal observation). Of perhaps even greater significance than changes in the total amount of habitat remaining is habitat fragmentation. Visual comparison of Figures 1, 2, and 3 indicates that by 1938 the potential habitat had become greatly fragmented. Patch size distributions remained more or less constant from 1938 to 1984, in both cases skewed toward a higher frequency of tiny patches as compared to the original habitat distribution (Table 1). By 1938, 83% of habitat patches were smaller than 1 square km (100 ha) in size, with a mean patch size of 0.77 km2 (77 ha). The proportion (84%) and the mean (73 ha) were essentially the same in 1984. Hence, in 1984 only 8,588 ha remained in patches larger than 1 square km. Accuracy of the Distribution Estimates In Table 2 we indicate in how many of 69 trapping localities we would have correctly or incorrectly predicted D. stephensi to be present or absent. Of the 51 sites where D. stephensi was present, our method would have led to an accurate prediction in 36 (7 1 % ) cases. Of the 18 sites where D. stephensi was absent, our method
3400

33045

Results
33-30

Estimated Distribution of Stephens Kangaroo Rat


We estimate that 124,775 ha of western Riverside County originally supported habitat suitable for Stephenskangaroo rat (Fig. 1). By 1938, only 45,569 ha

Figure 1. Distribution of potential Stephens kangaroo rat habitat before modern development. Habitat is in black PR = Perris Reservoit; LM = Lake Mathews; LE = Lake Elsinore

Conservation Biology volume 3, No. 3, September 1989

Price & Endo


1 1 7 ' 3 0 ' 117*15' 1 1 7 ' 0 0 '

Stephens' Kangaroo Rat Distribution

297

34-00'

Table 1. Size distribution o f D. stephemi habitat patches remahim in 19.78 and 1984.
Patch Size (ha
<5 5-9 10-19 20-39 40-79 80-99 100-149 150-199 200-299 300-399 400-499 >499

Number (cumulative %) of patches

1938
92 (16) 106 (34) 109 (52) 99 (69) 67 (80) 16 (83) 30 (88) 19 (91) 16 (94) 12 (96) 8 (97) 15 (100)

1984

33.45'

107 (16) 123(33) 131 (52) 101 (67) 101 (82) 19 (84) 35 (90) 15 (92) 23 (95) 9 (96) 6 (97) 19 (100)

33'30'

Figure 2. Potential Stephens' kangaroo rat habitat remaining in 1938. Conventions are as in Figure 1.
would have led to an accurate prediction in 12 (67%) cases. The overall probability of a correct prediction was 70%.This is sigmficantly higher (X2> 9.517, d.f. = 1, P < 0.005) than the accuracy expected under two other prediction protocols. Accuracy would have been 51% if we had randomly chosen 41% of the sites to contain the species, based on knowledge that 41% of the area of western Riverside County contains soils supporting habitat for the species. Accuracy would have been worse (26%) if we had simply guessed that no sites would contain the species because its habitat was in the minority. The only strategy that would have been superior to ours (74% accuracy) would have been to
7 ' 3 0 ' 34'00' 1 1 7 ' 0 0 '

predict that all trapped sites contain the species (based on knowledge that the areas were censused because there was a suspicion the species was present). From this analysis we conclude that the soil-based method for identifying areas o f potential Stephens' kangaroo rat habitat is reasonably accurate. It could be improved by inspecting patterns of incorrect prediction. For example, most of the error involved sites where we incorrectly predicted the species would be absent. Most of these instances involved soils that had been excluded from our list because they seemed too shallow, or because the slopes seemed too steep. A careful adjustment of the list of included s o i l s would improve accuracy.

Estimates of Temporal Variation in Population Densities


Results of censuses on Moore-Craig's (1984) site 14 are summarized in Table 3. She reported a success rate of 0.172 D. stephensi captures and 0.131 individuals per trap-night, and 0.152 D. agilis captures and individuals per trap-nightduringJanuary-March 1984 (Table 3). In March and June 1985 we trapped the same site and recorded no D. stephens4 and 0.008 D. agilis captures (.004 individuals) per trap-night. In February 1987 we recorded 0.025 D. stephensi captures (0.014 individuals) per trap-night and 0.031 D. agilis captures (0.012 individuals) per trap-night.In March 1988 we recorded 0.025 D. stephensi captures (0.022 individuals) per trap-night and 0.019 D. agilis captures and individuals
Table 2. Accuracy o f the soils-based method i n predicting the f Stephens' kmgamo rat on sites in presence or absence o w e s t e r n Riverside County. Jhch site is classified i n t o one of four cells on the basis o f the predicted and actual presence or absence o f the species.
Pwdlcted Occurrence

33'45'

33-30

Observed Occurrence

hsmt
36

Absent 15 12

Figure 3. Potential Stephens' kangaroo rat habitat remaining in 1984. Conventions are as in Figure 1.

Present Absent

298

Stephens' Kangaroo Rat Distribution

Price & Endo

Table 3. Numbers of D. stephensf and D. Sgius individuals and captures recorded on Moore-Craig's (1984) site 14, San Jacinto Wildlife Area, Riverside County, California, over a three-year period.
Cmusperiod Jan-Mar 1984 Mar 1984 Mar 1985 Jun 1985
Feb 1987 Mar 1988

Number of trap-nights
198 79 360 240 640 320

D. stephensi

D. agilis

# individuals
26

# captures

# individuals

# captures

0 0

34

12

12

9
7

0 0 16 8

3
1 8

3
1 20 6

per trap-night. By the fourth night of trapping in February 1987 the proportion of unmarked individuals caught had decreased to 1/5 D. stephensi and 2/11 D. agiliq indicating that we had censused a large fraction of the resident individuals on the grid. We estimate densities ofD. stephensi of 5.2 and 4.9 individualsper ha from the February 1987 and March 1988 censuses, respectively. These results suggest substantialvariation in the number of D. stephensi or D. agilis captures or individuals recorded per trap-night, statistics correlated with population densities. The highest capture rates per trapnight, obtained in early 1984, were 5-7 times higher than those in the next highest period (February 1987), and 19-38 times higher than those in the low 1985 sample periods for D. agilis These temporal fluctuations paralleled those we observed at the Motte Reserve (Table 4, Fig. 4). Numbers of individuals of two seed-eatingheteromyid species (0. agilis and Perognathus fallax) recorded in these censuses have fluctuated concordantly in response to yearto-year variation in rainfall, with highest densities in fall of 1983 (52.5 D. agilisha) after an exceptionally wet winter of 1982-1983, much lower densities in fall of 1984 (3.7 D. agilisha) after a year of subnormal winter rainfall, and intermediate densities in fall of 1986 (13.1
Table 4. Numbers of individual D. @&is and Perognathhus &flax recorded on a 1.6 ha permanent study site on the U. C. Motte Bcologlcnl Reserve, 23 km south of Riverside, California. Each census consisted of 600 trap-nights (200 trapdnight x three nights). Raillfall is total recorded precipitation (in em) for the Riverside Fire Station, for the period Nov 1 to Mar 31 preceeding each census, with the deviation from average in parentheses.
Rain fall for previous rainy season (deviation from normal)
39.84 ( 18.75) 13.34 ( - 6.86) 26.70 ( 6.50) 37.92 ( 16.76) 10.41( - 9.37) 18.71 ( - 1.07) 24.00 ( 4.21) 11.83( - 7.95) 12.47( - 7.32)

D. agilisha), after a year of approximately average winter rainfall. On the Motte Reserve, the correlation between abundances of D. agilis and P. fallax across years was high (r = 0.85, d.f. = 7, P < 0.01), and the correlation between abundance of D. agilis and P. fallar and the deviation of rainfall from normal was 0.63 (t = 2.146, d.f. = 7, P = 0.07) and 0.69 (t = 2.522, d.f. = 7, P < 0.05), respectively. Since rainfall is a good indicator of annual seed production in arid regions, this pattern is not surprising, and has been documented for

Total number of individuals


D.agilis
24 38

OJ' 4
I . l ' 1 ' I ' l ' I ~ l ' I

C e n S l 4 . S

Perid
Oct Oct Nov Nov Nov Nov Nov Nov Nov 1980 1981 1982 1983 1984 1985 1986 1987 1988

P.fallax
30 38 71 102 1 28 68 15 15

-8

-4

+4

+8

+I2

+I6

+20

+
+ +

DEVIATION F R O M AVERAGE R A i N F A L L ( C M )

35
84 6

13
21 11

13

Figure 4. Relationship between deviation jivm normal rainfall and numbers of individuals of two hetmmyid rodent species mcorded during standard fall censuses of a permanent study plot on the Motte Reserve. Numbers refer to year (eg., 0 = 1380, 1 = 1381, etc); lines connect successive years. See text and Table 4 for deuils

Conservation Biology Volume 3, No. 3, September 1989

Price & Endo

Stephens Kangaroo Rat Djsmbution

299

heteromyid rodent populations throughout the southwestern United States (Munger et al. 1983). The fact that heteromyid populations at two localities (Motte Reserve and Moore-Craigssite 14) separated by about 20 km show parallel temporal density patterns suggests that D. stephensi populations in western Riverside County can be expected to show correlated, > 10-fold temporal density fluctuations in response to regional rainfall patterns.

Discussion
The Utility of Soil Maps
In this paper we have illustrated how soil maps can be used to estimate the potential distribution of a cryptic species, Dzpodomys stephensi. By overlaying maps of land use patterns on the habitat map, we could also arrive at a quantitative estimate of habitat remaining in 1938 and 1984. The soils-based method for identlfying potential habitat for this kangaroo rat species correctly predicted its presence or absence for 48 out of 69 (70% ) sites samples by live-trapping. Refinement of the list of soils that support suitable habitat would allow even better prediction. We believe that soil-basedmethods of identlfylng habitat of threatened species should be of general utility. Soils are likely to be good habitat indicators for burrowing animals, for animals associated closely with types of vegetation recognized in soil surveys, or for plants that have strict soil requirements. Furthermore, soil surveys are most likely to be available for regions inhabited by threatened species, because a common threat is intensive habitat destruction for agriculture or urban uses. A final virtue of soil-based methods is that they permit identification of areas that potentially support suitable habitat, but that do not at present because of human disturbance or some other reason. If such areas can be reclaimed as wildlife habitat, they expand the number o f sites that could be included in a habitat conservation plan.

cultural lands (Thomas 1973, 1975;Moore-Craig 1984). Urban development is irreversible. Habitat fragmentation has been extensive. Most of the original habitat of this species was contiguous (Fig. 1). We did not quantify the size distribution of habitat patches in Figure 1, but we counted only about 350 separate fragments, far fewer than the approximately 600 fragments present in 1938 and 1984 (Figs. 2, 3, Table 1). By 1938, over 80% of fragments were smaller than 1 square km, the minimum size (see below) we consider compatiblewith reasonably long-term survival o f an isolated population of Stephenskangaroo rats. From these observations,we agree with the U.S. Fish and Wildlife Services judgment that Dipodomys stephensi is threatened with extinction unless an &ective habitat conservation plan is implemented in the near future.

Implications for Habitat Conservation Plan Development


There is only meager information on mean and variance in population densities of D. stephensi within occupied areas. From repeated censuses on one site over a fouryear period, we conclude that population densities can fluctuate 10-fold.This conclusion is reinforced by observations of a population of D. agile a kangaroo rat species similar in size and ecology to D. Stephens4 at a nearby study site. Altogether, our information suggests that, under weather conditions prevailing during the last decade,population densities at a single site can fluctuate from a low value of about 3 individualha, to a high value of about 53 individualdha. These figures are consistent with the range of values reported in the literature (Kramer 1987). The minimum value of 3 individualha would be an overestimate of densities under conditions of more prolonged drought than occurred during the study period. We focus on minimum population densities, because these are more likely to dictate long-term survival probability than maximum or even average population densities. It is during low-density periods that populations are especially vulnerable to demographic or environmental stochasticity,and the erosion of genetic diversity that occurs during population declines is restored only after very long periods of time. For these reasons an effective conservation plan should be based upon minimum population densities, not averages. If we wish to maintain populations of at least 100 D. stephensi per reserve, each reserve should contain at least 100/3 = 33 ha of suitable habitat, judging from conditions over the past decade. To account for error in estimating densities and longer-term climatic fluctuations, we would increase that amount by a factor of two or three, for a minimum reserve size of about 100 ha (1 km2) of habitat. For birds in less variable environments than those in southern California, it would appear that

Status of Stephens Kangaroo Rat


We estimate that by 1938 the potential habitat of D. stephensi in western Riverside County was reduced by 60% and greatly fragmented relative to the probable post-Spaniard configuration. Since this area includes most of the geographic range of the species, it is reasonable to consider that the total species population suffered about a 60% reduction as well. The situation did not change appreciably between 1938 and 1984, but accelerating urbanization has been further eroding habitat since 1984. Urban development poses a greater threat to the species than does agriculture,because kangaroo rat populations can recolonize abandoned agri-

consemation Biology Volume 3, No. 3, September 1989

300

Stephens Kangaroo Rat Distribution

Price & Endo

t h i s population size should allow virtually 100%persis-

tence for at least 30 years (Pimm et al. 1988). The prognosis is likely to be much worse, however, for small mammals like D. stephensi because they are less vagile than birds (Brown 1978) and have much larger temporal population fluctuations (Pimm et al. 1988). The number of separate reserves that should be maintained depends on the probability of simultaneous extinction of isolated populations, which in turn depends on which extinction-causingfactors are most important for this species. If demographic stochasticity or localized catastrophic events (such as disease or flood) are the primary threat, then increasing the reserve number even a small amount greatly increases the species chance of long-term survival (Quinn & Hastings 1987). However, if widespread environmental stochasticity, such as climatic fluctuation, is a primary threat, then events on separate reserves are correlated. From our observations of concordant variation in populations of seed-eating rodents in Riverside County, and from observation that temporal fluctuations are highly correlated with rainfall,we would conclude that widespread environmental stochasticity is an important threat to persistence of Stephens kangaroo rat populations. Hence, our expectation is that a few relatively large individual reserves, placed as far apart as possible within the range of the species,would be a more effective management strategy than many small reserves, especially if the latter were clustered in a homogeneous region of the range. Once the desired number and size of reserves has been chosen, a distribution map such as ours becomes extremely useful for targeting potential reserve sites. Available lands that contain potential habitat can be located easily and censused selectively to confirm that the species is present, thus minimizing the amount of timeconsuming fieldwork necessary during the final stages of conservation plan development.

with fieldwork; Nick Waser provided encouragement and critical input during all stages of the study. The project was supported by University of California Academic Senate Intramural research grants and NSF grant BSR 84-07602.
Literature Cited
Bleich, V. C. 1977. Dipodomys stepbensi. Mammalian Species 73:l-3. Bleich, V.C. 1973. Ecology of rodents at the United States Naval Weapons Station, Seal Beach, Fallbrook Annex, San Diego County, California.Unpublished masters thesis, California x + 102 pp. State University, Long Beach. i Bontrager, D. R. 1973. Rodent ecology of the Santa Rosa Plateau, Riverside County, California. Unpublished masters thesis, California State University, Long Beach. 11 5 pp. Brown J. H. 1978. The theory of insular biogeography and the distribution of boreal birds and mammals. Great Basin Naturalist Memoirs 2:209-227. Burt, W. H., and R. P. Grossenheider. 1964.A field guide to the mammals. Houghton Mifnin Company, Boston, Massachusetts. Gilpin, M. E., and M. E. Soule. 1986. Minimum viable populations: processes of species extinction. Pages 19-34 in M. E. Soule, editor. Conservation biology. Sinauer Associates, Sunderland. Massachusetts. Hicks, D., and A. Cooperrider. 1977.Wildlife habitat inventory for the Stephens kangaroo rat (Dipodomys stepbensi).United appenStates Bureau of Land Management Report. 1 1 pp. dices.

Kramer, K 1987. Endangered and threatened wildlife and plants; determination of endangered status for Stephens kangaroo rat. Federal Register 52(223):44,453-44,456. Kramer, K 1988. Endangered and threatened wildlife and plants, determination of endangered status for Stephenskangaroo rat. Final rule. Federal Register 53(190):38,465-38,469. Lackey, J. A. 1967. Biosystematics of b m n n i group kangaroo rats in southern California. Transactions of the San Diego Society of Natural History 14313-344. Leigh, E. G., Jr. 1975. Population fluctuations, community stability and environmental variability. Pages 5 1-73 in M. L. Cody and J. M. Diamond, editors. Ecology and evolution of communities. Harvard University Press, Cambridge, Massachusetts. Leigh, E. G., Jr. 1981. The average lifetime of a population in a varying environment. Journal of Theoretical Biology 90:219239. MacArthur, R H., and E. 0.Wilson. 1967. Island biogeography. Princeton University Press, Princeton, New Jersey. Merriam, C. H. 1907. Descriptions of ten new kangaroo rats. Proceedings of the Biological Society of Washington 20: 75-79.

Acknowledgments
We are grateful to many people for helping us. The California Desert District of the Bureau of Land Management, USDA Soil Conservation Service,and the Riverside County Planning Department have been extremely generous in providing maps and access to unpublished information on Stephenskangaroo rat; the geography program, University of California, Riverside, kindly allowed us to use their image-processing equipment; Dr. Elizabeth Lord provided the digitizing equipment and programs; Dr. Troy Best and Narca Moore-Craig allowed us to use their unpublished data; work with D. stephensi was carried out under a memorandum of understanding with the California Department of Fish and Game; Kevin Heinz and Bill Longland, among many others, assisted

consemation Biology Volume 3, No. 3, September 1989

Price & Endo


Moore-Craig, N. A. 1984. Distribution and habitat preference o f Stephens kangaroo rat on the San Jacinto Wildliie Area. Unpublished senior undergraduate thesis, University of California, Riverside. Munger, J. C., M. A. Bowers, and W. T. Jones. 1983. Desert rodent populations: factors decting abundance, distribution, and genetic structure. Great Basin Naturalist Memoirs 7:91116.

Stephens hhgamo Rat Distribution

301

Shder, M. L. 1981. Minimum population sizes for species conservation. BioScience 31:13 1-1 34. Thomas, J. R., Jr. 1973. Stephens kangaroo rat survey. California Department of Fish and Game. Special wildlife investigations, W-54-R,Job 11-56. Sacramento, California. Final Report, 49 PP. Thomas, J. R., Jr. 1975. Distribution, population densities, and home range requirements of the Stephens kangaroo rat (Dipodomys stephensf).Unpublished masters thesis, California State Polytechnic University, Pomona. 64 pp.
US. Department of Agriculture (USDA). 1971. S o i l Survey of Western Riverside Area, California. USDA S o i l Conservation Service and University of California Agricultural Experiment Station. U.S. Government Printing Office, Washington.

Pimm, S. L., H. L. Jones, and J. Diamond. 1988. On the risk of extinction. American Naturalist 132:757-785. Price, M. V. 1978. The role of microhabitat in structuring desert rodent communities. Ecology 59:9 10-92 1. Price, M. V., and N. M. Waser. 1984. On the relative abundance of species: postfire changes in a coastal sage scrub rodent community. Ecology 5 4 1161-1 169. Quinn, J. F., and A. Hastings. 1987. Extinction in subdivided habitats. Conservation Biology 1:198-208.

Weislander, A. E. 1938. Vegetation type maps of California and western Nevada. USDA Forest Service, California Forest and Range Experiment Station, Berkeley, California.

Conservation Biology Volume 3, No. 3, September 1 9 8 9

Vous aimerez peut-être aussi