The Peninsular Effect on Species Diversity and the Biogeography of Baja CaliforniaAuthor(s): Robert J. Taylor and Philip J. RegalReviewed work(s):Source: The American Naturalist, Vol. 112, No. 985 (May - Jun., 1978), pp. 583-593Published by: The University of Chicago Press for The American Society of NaturalistsStable URL: http://www.jstor.org/stable/2460125 .Accessed: 27/04/2012 13:45

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

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 support@jstor.org.

The University of Chicago Press and The American Society of Naturalists are collaborating with JSTOR todigitize, preserve and extend access to The American Naturalist.

http://www.jstor.org

Vol. 112, No. 985 The American Naturalist May-June 1978

THE PENINSULAR EFFECT ON SPECIES DIVERSITY AND THE BIOGEOGRAPHY OF BAJA CALIFORNIA

ROBERT J. TAYLOR* AND PHILIP J. REGAL

Department of Ecology and Behavioral Biology, University of Minnesota, Minneapolis, Minnesota 55455

Simpson (1964) observed that there are fewer species of mammals present at the tips of all the major North American peninsulas than at their bases, a pattern that has subsequently been found to apply to other vertebrate groups. Cook (1969) documented its existence in birds; Kiester (1971) found it to hold for amphibians and reptiles on Florida. This pattern of diversity (sensu "species richness" or "species density") is neither coincidental nor transient, according to Simpson; it is an equilibrium resulting from the peculiarities of peninsular geometry. The implications of this suggestion are intriguing inasmuch as peninsulas are common, not only the terrestrial form in bodies of water but also peninsulas of suitable habitat extending into hostile environments.

We propose in this paper to begin looking in detail at the nature of Simpson's "peninsular effect," its roots and its implications. Before the peninsular effect is used as an explanation for existing patterns (e.g., Willis 1974), its cause should be clear. Does it indeed reflect a general property of peninsular geometry, or does it merely reflect peculiarities of the North American sites for which we have data? For example, paleoecological data for Florida published since Simpson's paper appeared (Watts 1971) demonstrate that the vegetation on that peninsula is undergoing systematic changes quite apart from man's disturbances. The xeric vegetation of 5,000 yr ago has become more mesic, primarily as a result of a continuing rise in the water table. One can only speculate on the likelihood of faunal equilibria in the face of floristic instability, but the possibility remains that modern diversity patterns of vertebrates on Florida are not stable and do not therefore necessarily represent an equilibrium effect. Nor, unfor- tunately, are the patterns on Alaska and Labrador unambiguous; both peninsulas have few species and harsh climatic gradients along their lengths. Elimination of Florida, Alaska, and Labrador leaves only two major sites where Simpson's hypothesis can be pursued: Yucatan and Baja California. The decrease in diversity on Yucatan is difficult to evaluate. The fauna are not well known, and Simpson's data base (Hall and Kelson 1959) is certainly incorrect (Birney et al. 1974). Compounding this uncertainty is the fact that there are striking changes in topography and vegetation from Guatemala north- ward. If Yucatan is excluded, then Baja California is the only remaining major North American peninsula the diversity profile of which is adequately known and suggests a peninsular effect.

Are biogeographic patterns on Baja California in equilibrium or not? It looks as if they are. The peninsula has been stable geologically since the early Pliocene, with a configuration similar to that of today (Durham and Allison 1960), though the Gulf of California has widened substantially in the last 4 million yr (Larson et al. 1968). Apparently the only recent geological change of importance was siltation by the

* Present address: Department of Zoology, Clemson University, Clemson, South Carolina 29631. Amer. Natur. 1978. Vol. 112, pp. 583-593. ? 1978 by The University of Chicago. 0003-0147/78/1285-0030$01.02

583

584 THE AMERICAN NATURALIST

TABLE 1

DECREASE IN DIVERSITY OF VARIOUS VERTEBRATE GROUPS ON BAJA CALIFORNIA

Species Present Species Present Vertebrate Group at Northern End at Cape San Lucas % Decrease

Snakes ....................... 20 20 0 Lizards (terrestrial and

arboreal spp.) ......... ...... 18 14 22 Birds (breeding spp.) ........... 107 95 11 Mammals .................... 60 38 37 Heteromyid rodents ........... 12 2 83 Bats ......................... 20 14 30

SOURCES.-Friedman et al. 1957a, 1957b; Stebbins 1954; Hall and Kelson 1959; Savage 1960. NOTE.-Data from the northern end of the peninsula represent only those species found in the

eastern Sonoran Desert portion.

Colorado River at the head of the gulf, creating a broader land bridge with the Sonoran Desert. Sonoran Desert vegetation extends the length of Baja California, breaking up in the cape region into small patches interspersed with more mesic lowland and montane vegetation. Southwestern desert vegetation seems to have remained stable for a long time. According to paleoecological evidence from the southwestern United States and northern Mexico, no major vegetational shifts have occurred within the last 30,000 yr (Martin and Mehringer 1968; Meyer 1973). Detailed evidence from fossil pack rat middens indicates that woodland vegetation 10,000-20,000 yr ago may have extended 200-600 m below its present elevation, but the Sonoran Desert undoubtedly persisted at elevations of up to 400 m (Van Devender and King 1971; Van Devender 1973).

Closer examination of the fauna of Baja California reveals the generality of the pattern among various vertebrate groups. Table 1 lists the number of species found in the Sonoran Desert portion of the northern base of the peninsula, the number in all vegetation types at the cape, and the percentage of decrease. These data are biased against a peninsular effect because the diversity of habitats in the cape region is greater than in the Sonoran Desert portion of the north. Even with that, the numbers of species are either the same or fewer at the tip. This is especially striking in the snakes, birds, and bats, because the diversity of these groups increases markedly over the same vegeta- tional transect on the Mexican mainland (Stebbins 1954; Hall and Kelson 1959; Cook 1969; Hardy and McDiarmid 1969). We could speculate on the degree to which the ecological differences among these vertebrate groups influence the variation in the peninsular pattern, but this would be premature for the purposes of this paper. Our first task is to see if the pattern in any of these groups is best explained by an equilibrium peninsular effect.

We chose the heteromyid rodents, kangaroo rats and pocket mice, as a model group for several reasons, the most important of which is that their ease of capture makes it likely that the distributional records are fairly accurate. Inasmuch as they form an ecologically homogeneous group, the Sonoran Desert species being all granivorous, their biogeographic patterns may reflect simple differences in the interactions of the species with the environment and with one another. Finally and not least important, the decrease in the diversity of Sonoran Desert heteromyidae from 12 species on the northeast gulf coast to two species near Cape San Lucas (fig. 1) accounts for nearly half the total decrease in mammals.

We identify three alternative explanations for the pattern in heteromyid diversity on Baja California. The first, advanced by Orr (1960), is that too little time has elapsed

BIOGEOGRAPHY OF BAJA CALIFORNIA 585

a11

1 2

10

9

9

7

6

5

\4 J

FIG. 1.-The species diversity of heteromyid rodents in the Sonoran Desert portions of Baja California (data from Hall and Kelson 1959).

since Baja California became Sonoran Desert to have allowed all the mammals to spread to the cape. On the basis of existing paleoecological data, Orr's estimate of 12,000 yr since the vegetation stabilized in its present form appears to be a bit late. Even if correct, however, one wonders if any mammal could not extend its range 1,100-1,200 km in 12,000 yr, given favorable ecological conditions. The dramatic extension of the range of the armadillo (Dasypus novemcinctus) northward into the United States since the mid-nineteenth century attests to the speed with which range extensions can occur (Buchanan and Talmadge [1954], cited in Udvardy 1969).

A second explanation is that the climate, as reflected in the species composition, structure, and productivity of plant communities, determines the limits of the range of each heteromyid species. This is certainly a reasonable explanation. The desert vegeta- tion is not uniform along the length of the peninsula; Shreve and Wiggins (1964) recognized four distinct subfloras. With a couple of exceptions, heteromyid rodents are not known to depend upon particular species of plants, but they do specialize upon certain structural characteristics of the plant community (Rosenzweig and Winakur 1969; Rosenzweig 1973; Lieberman 1974; Schroder and Rosenzweig 1975). Systematic change in habitat structure along the peninsula, as the Sonoran Desert community changes to a subtropical thorn scrub, could result in an environment increasingly unsuitable for kangaroo rats and pocket mice. On the mainland a pool of rodent species is available from tropical regions to replace the heteromyids; on the peninsula the only alternative to a decrease in diversity is the evolution of greater ecological plasticity by the Sonoran Desert rodents. This is the explanation Robertson and Kushlan (1974)

586 THE AMERICAN NATURALIST

prefer for the decrease in avian diversity on Florida. In the absence of detailed data on the physical structure of Sonoran Desert plant communities on Baja California and the habitat requirements and plasticity of the particular heteromyid species involved, this second hypothesis cannot be rejected.

The third hypothesis, advanced by Simpson (1964), is that the pattern is an equilib- rium resulting from a balance between extinction and recolonization. This approach is familiar to most ecologists by its application to islands (MacArthur and Wilson 1967). Simpson proposed that smaller peninsular populations are more likely to become extinct and recolonization is increasingly difficult the farther a site is from the mainland. The resultant equilibrium would show a decrease in diversity toward the tip. The only advantage of this hypothesis over the first two is that it can be described mathematically. Accordingly, we develop next an extinction-recolonization model of the invasion process, our purposes being both to test the logic of Simpson's hypothesis and to discover the conditions under which it is valid.

THE MODEL

A model of colonization and extinction on peninsulas should be soluble by analytical methods. Our efforts in that direction have not as yet been successful, so we present here the results of a series of computer simulations of the Monte Carlo or probabilistic form. Baja California is conceived of as a linear sequence of 100 habitat "islands," on each of which each of the 12 species of heteromyids is either present or absent. Colonization of an unoccupied site can only occur from an adjacent occupied site.

The flow of the simulation is quite simple. Annual colonizing episodes alternate with periods of vulnerability to extinction. Starting with an empty peninsula and with all 12 species on the mainland, these alternate periods of colonization and extinction occur until an equilibrium is approached in which the diversity profile of the peninsula remains unchanged. Results show that an equilibrium (and only one) developed in all simulations regardless of whether the peninsula was assumed to be empty at the beginning or to contain all species at all sites; moreover, the form of the equilibrium was not dependent upon the starting conditions. Each equilibrium was then compared with figure 1 to see if the qualitative features of a real pattern were duplicated.

Simulations indicate that the equilibrium profile consists of a more or less rapid decrease to a plateau in species number, an equilibrium over space as well as time. This level of species richness is maintained until the end of the peninsula is approached, and then it drops off rapidly. Figure 2 is an example of a typical pattern. The annual colonization and extinction probabilities apparently influence both the height of the plateau and the rate of decrease to it. Our criteria for the demonstration of a peninsular effect (derived from fig. 1) require that the decrease from the peak diversity to the tip be gradual and that any plateau be low. The equilibrium pattern in figure 2, for example, would not be suitable.

The first series dealt with what can be dubbed a "noninteractive" model. All transition probabilities were the same for each of the 12 species and independent of the diversity of the community at each point. The most striking feature of this series of simulations is that the desired pattern is quite difficult to produce. In figure 3 we portray the parameter space defined by the annual probability of extinction and the annual probability of colonization of a vacant site from an adjacent occupied site. The stippled portion represents that combination of parameters producing an appropriate pattern. In the region to the upper left extinction was too likely; consequently, no species colonized the entire peninsula. The lower right region produced much too high a baseline number of

BIOGEOGRAPHY OF BAJA CALIFORNIA 587

12

1 1

10

9

z

Ln \

z 4_

L

0 50 100 DI STANCE FROM MAI NLAN D

FIG. 2.-A typical equilibrium pattern produced by the noninteractive model in which the species diversity initially drops rapidly to a plateau and then remains fairly constant until the tip of the peninsula is approached.

1.0

.9

.8 z W0

z

ci3

.2

_

0 .1 . 3 . 5 . 7 8 . 1.0

PROBABI L ITY OF COLON IZAT ION

FIG. 3.-The stippled area represents that combination of the probabilities of extinc- tion and colonization which produce a diversity pattern similar to figure 1 in the noninteractive model with identical parameter values for all species

588 THE AMERICAN NATURALIST

species. The model appears to be a bit more sensitive to variation in the probability of extinction than to the probability of colonization.

The observation that the fraction of the parameter space producing a peninsular effect is small might indicate that the extinction-recolonization paradigm is inappropriate. Unfortunately we have no idea which are realistic parameter values in nature. Hetero- myids may in fact fit precisely into the region producing a peninsular effect. What certainly emerges from this model is that the constraints upon the probabilities of colonization and extinction are severe, making the model vulnerable to empirical evidence.

The first version of the model contains a number of simplifying assumptions which could influence the shape of the equilibrium. Several of these need discussion. We proceed next to explore the consequences of not assuming that all species have identical parameter values or that parameter values are independent of species diversity, of not assuming that the peninsula has 100 habitable sites rather than fewer, and, finally, of not assuming that these sites are arranged in a linear sequence but in parallel double or triple chains.

Any group of 12 species is unlikely to contain equally good colonizers or species equally vulnerable to extinction. We might expect systematic variation in the natural community, some species displaying high probabilities, others low. Measurements of these parameters do not exist, so we allowed for differences among species by assigning each an extinction probability chosen at random from a normal distribution with a given mean and a standard deviation of 0.05. This was done four or five times for each mean value. If any of these trials produced an appropriate equilibrium, we concluded that the mean value in question could produce a peninsular effect. Our expectation that a peninsular effect would be much easier to produce with variable parameters proved to be unfounded. Figure 4 shows that the region of parameter space producing an appropriate equilibrium is enlarged only slightly. We did not explore the parameter space nearly as extensively in this series as in the last because of the expense of this kind of simulation, but the earlier conclusion still seems to hold. It is difficult to produce a peninsular effect such as seen in the heteromyids; the constraints on the parameters are severe.

The kangaroo rats and pocket mice taken as a group appear to be resource specialists, indicating that competition for the common resource base may have been an important ecological and evolutionary factor. Although there is some question about the form and extent of this interaction (Schroder and Rosenzweig 1975; Brown and Lieberman 1973; Reichman and Oberstein 1977), we thought it reasonable to investigate the influence of competition upon the equilibrium pattern.

Competition was included by making the probability of extinction a function of the diversity of the heteromyid community. This is "extinction competition" in the sense of Levins and Culver (1971) and Slatkin (1974). The logic is that an increase in diversity will decrease the equilibrium population of each species and render it more vulnerable to chance extinction. Two basic functional forms were used to describe the dependence of extinction upon diversity. In the first the species number increased the probability of extinction in an additive fashion, the form used by Levins and Culver (1971) and Slatkin (1974). In the second form the probability of extinction increased exponentially with species number, the second function being the more reasonable (MacArthur and Wilson 1967). The result in either case was clear: a peninsular equilibrium mimicking figure 1 is very difficult to produce if competition has a significant influence. The additive form of the probability of extinction had both diversity-dependent and diversity-independent components. Even when parameters were chosen to make the diversity-dependent component small compared with the diversity-independent component, an appropriate

BIOGEOGRAPHY OF BAJA CALIFORNIA 589

.9

.8-

z 07

.6 x U,

L

05

0 . .2 .3 4 5 .6 .7 .8 9 1. PROBABILITY OF COLONIZATION

1IG. 4.-The combination of parameters producing an appropriate diversity pattern when the probability of extinction is allowed to vary among the species (see text).

equilibrium could not be produced. The patterns produced were consistent, exhibiting a high or low plateau, depending upon parameters, and always a very sharp decrease to that plateau. This seems reasonable, inasmuch as extinction is most likely near the mainland source where diversity is greatest. The dynamics of these simulations followed a typical pattern: a few species would colonize the entire peninsula rapidly, and the remainder would be unable to penetrate to any significant extent.

The change in the intensity of competition on Baja California may not be nearly as dramatic in reality as even the weakest version of the model. It is. unlikely, for example, that more than five of the 12 species on the north Gulf Coast occur together in any one local area. Several may be insignificant competitors wherever they occur. Nevertheless, the model is remarkably sensitive to competition in the form in which we described it.

The division of Baja California into 100 habitable sites was originally chosen as a compromise between reality, which is that the peninsula may be nearly continuously habitable for heteromyids, and the difficulty and expense of simulation, which increases enormously as the peninsula is chopped into more and smaller segments. There are animal groups, however, for which Baja California certainly does not provide contin- uously suitable habitat; obvious examples are the birds and mammals of montane habitats. To explore the difficulty of producing a peninsular effect with fewer than 100 habitable sites, we repeated the simulation with identical parameter values, without competition, and using only 25 sites. We reasoned that if the abrupt decreases in diversity from the mainland to the plateau and from the plateau to the peninsular tip were brought close enough together, they could swamp the effects of a high plateau and produce an appropriate peninsular pattern for a larger set of parameter values. The results in figure 5 bore this out. It was reasonably easy to produce a peninsular effect given only that the probability of extinction was low. While a string of 100 habitable sites seems more realistic than 25, we cannot rule out the possibility that subtle ecological factors result in a lower figure.

590 THE AMERICAN NATURALIST

1.0-

.9-

n .8-

.7-

z x .6-

co at .43 C .3

0 0 1 2 .3 4 5 6 7 8 9 1.0

PROBABILITY OF COLONIZATION

FIG. 5.-The parameter space producing an appropriate peninsular effect when the peninsula is divided into only 25 habitable sites.

The final assumption we discuss is the use of only one chain of habitable sites rather than two or more adjacent chains. The robustness of the model to this assumption clearly depends upon which of two types of colonizers one considers. The simpler case deals with active dispersers such as vertebrates who maintain control over the direction of their dispersal. These organisms will reflect off a barrier, such as the ocean, and return to the peninsula. A little reasoning leads one to the conclusion that the diversity pattern for a single chain of habitable sites will be the same, whether it occurs alone or adjacent to any number of other chains of suitable sites. The increase in the number of sources of colonists will be exactly balanced by an increase in the number of target sites.

The situation is a bit more complicated with passive dispersers. A propagule leaving a point along an isolated linear string of sites clearly stands a poor chance of getting to another habitable site. But if this one chain lies adjacent to another, then the angle of departure leading to a hospitable site forms a very much larger proportion of a circle. We conclude, therefore, that the effect of relaxing the assumption of a simple one- dimensional model may be thought of for passive dispersers as equivalent to an increase in the probability of colonization and for active dispersers as nothing at all.

One of the few advantages of a simulation model over an analytical model solved for equilibrium conditions is that it provides an idea of the time taken to approach equilibrium. The length of the transient phase is important for comparing the equilib- rium and disequilibrium hypotheses. Even when the equilibrium pattern is appropriate, the model must be rejected if the transient phase is too long. If it is sufficiently short, the disequilibrium hypothesis can be rejected.

With this in mind, then, we note that in no simulation did the time to equilibrium exceed 2,000 yr; in most cases it was substantially less. The length of the transient phase of colonization depends upon several factors, the starting conditions, the dispersal abilities of the species involved, and the lag between the time a site is colonized and when it begins to send out colonists in turn.

BIOGEOGRAPHY OF BAJA CALIFORNIA 591

In most of these simulations, the peninsula was assumed to be devoid of heteromyids at the outset. While this admittedly artificial assumption had no effect upon the form of the equilibrium, it certainly influenced the length of the transient period. The possibility that we underestimated the time taken to approach equilibrium could be tested by simulations beginning with all species present at all locations. This- was done, and in every case the same equilibrium was achieved substantially more rapidly; hence the estimate of the length of the transient phase one gets by assuming an initially empty peninsula is probably conservatively long.

The dispersal abilities of heteromyids are not very well documented. In the model a dispersing animal was assumed to be able to make it to the adjacent site with a given probability. Division of Baja California into 100 points would set this dispersal distance at about 10 km. This is probably excessive for pocket mice but may not be unreasonable for the larger kangaroo rats. Schroder and Rosenzweig (1975) trapped one male Dipodomys ordii at two sites separated by 9.5 km of rough terrain and another D. ordii that had covered 5 km in only a week.

The third assumption, that population density need not be considered, must result in a shortened transient phase. Since it must take a population at least 2-3 yr to build to the point that it sends out colonists, the species would undoubtedly march down the peninsula more slowly than predicted.

In spite of these difficulties, we think it unlikely that the model is in error by more than a factor of 5 or 6. This means that the transient colonizing phase on Baja California was probably not greater than 10,000-12,000 yr. The vegetation has apparently been stable for this length of time, leading us to conclude that the disequilibrium hypothesis is not an adequate explanation of the current diversity pattern. This conclusion is buttressed by the fact that birds, with their potentially rapid dispersal abilities, also demonstrate a peninsular effect.

CONCLUSIONS

The extinction-recolonization model can generate a great many patterns in which there are fewer species at the tip of the peninsula than at the base; however, approxima- tion of one available documented pattern (that exhibited by heteromyid rodents) is difficult and requires severe constraints upon parameters and assumptions. The extinction-recolonization hypothesis is not necessarily threatened by these limits; the heteromyidae may meet the conditions the model imposes. Patterns for other less well documented vertebrate groups could prove to be somewhat different, for the most part not showing nearly so severe a decrease and perhaps not changing gradually over the entire length of the peninsula. Models of these groups will possibly not require such strict constraints upon parameters.

The most reasonable alternative to the model is that systematic changes in climate and vegetation sequentially eliminate all but Perognathus spinatus and D. merriami, the two species that penetrate all the way to Cape San Lucas. Such an explanation of the peninsular effect on Baja California would be substantially less general in its application to other sites than Simpson's extinction-recolonization hypothesis.

SUMMARY

A simulation model is presented which explores the extinction-recolonization explana- tion of the decreased number of animal species on peninsulas. Using as an example the pattern of species density of heteromyid rodents on Baja California, we show that severe constraints on the values of the parameters are necessary for the model to produce an

592 THE AMERICAN NATURALIST

appropriate pattern. When competition, modeled as influencing the probability of extinction, is included in the simulation, an appropriate equilibrium pattern is difficult to produce. A peninsular effect results more easily if the habitable sites are few and widely spaced. The rapid speed with which equilibria were approached, even when adjusted for inaccurate assumptions, makes an explanation based upon a faunal disequilibrium unlikely.

ACKNOWLEDGMENTS

We thank Robert Armstrong, Richard McGehee, and Ann and Jay Goldman for many stimulating discussions. Computational costs were borne by the University of Minnesota Computer Center.

LITERATURE CITED

Birney, E., J. Bowles, R. M. Timm, and S. L. Williams. 1974. Mammalian distributional records in Yucatan and Quintana Roo, with comments on reproduction, structure, and status of peninsular populations. Bell Mus. Nat. Hist. Occasional Papers 13:1-25.

Brown, J. H., and G. A. Lieberman. 1973. Resource utilization and coexistence of seed-eating desert rodents in sand dune habitats. Ecology 54: 788-797.

Cook, R. E. 1969. Variation in species density of North American birds. Syst. Zool. 18: 63-84. Durham, J. W., and E. C. Allison. 1960. The geologic history of Baja California and its marine

fauna. Syst. Zool. 9:47-91. Friedmann, H., L. Griscom, and R. T. Moore. 1957a. Distributional check-list of the birds of

Mexico. I. Pacific Coast Avifauna 29: 3-202. 1957b. Distributional check-list of the birds of Mexico. II. Pacific Coast Avifauna 33:3-436.

Hall, E. R., and K. R. Kelson. 1959. The mammals of North America. Ronald, New York. Hardy, L. M., and R. W. McDiarmid. 1969. The amphibians and reptiles of Sinaloa, Mexico. Univ.

Kansas Pub. Mus. Nat. Hist. 18:39-252. Kiester, A. R. 1971. Species diversity of North American amphibians and reptiles. Syst. Zool.

20:127-137. Larson, R. L., H. W. Menard, and S. M. Smith. 1968. Gulf of California: a result of ocean floor

spreading and transform faulting. Science 161:781-784. Levins, R., and D. Culver. 1971. Regional coexistence of species and competition between rare

species. Proc. Nat. Acad. Sci. 68:246-248. Lieberman, G. A. 1974. Species diversity and abundance of desert rodents in several southwestern

deserts. Ph.D. diss., Princeton University. MacArthur, R. H., and E. 0. Wilson. 1967. The theory of island biogeography. Princeton

University Press, Princeton, N.J. Martin, P. S., and P. J. Mehringer, Jr. 1968. Pleistocene pollen analysis and biogeography of the

Southwest. Pages 433-451 in H. E. Wright, Jr., and D. G. Frey, eds. The Quaternary of the United States. Princeton University Press, Princeton, N.J.

Meyer, E. R. 1973. Late-Quaternary paleoecology of the Cuatro Cienegas Basin, Coahuila, Mexico. Ecology 54: 982-995.

Orr, R. T. 1960. An analysis of the recent land mammals. Syst. Zool. 9:171-178. Reichman, 0. J., and D. Oberstein. 1977. Selection of seed distribution types by Dipodomys

merriami and Perognathus amplus. Ecology 58: 636-643. Robertson, W. B., Jr., and J. A. Kushlan. 1974. The southern Florida avifauna. Environments South

Florida: present past, Mem. 2:414-452. Rosenzweig, M. L. 1973. Habitat selection experiments with a pair of coexisting heteromyid rodent

species. Ecology 54: 111-117. Rosenzweig, M. L., and J. Winakur. 1969. Population ecology of desert rodent communities:

habitats and environmental complexity. Ecology 50: 558-572. Savage, J. M. 1960. Evolution of a peninsular herpetofauna. Syst. Zool. 9: 184-212.

BIOGEOGRAPHY OF BAJA CALIFORNIA 593

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

Shreve, F., and I. L. Wiggins. 1964. Vegetation and flora of the Sonoran Desert. Stanford University Press, Stanford, Calif.

Simpson, G. G. 1964. Species density of North American recent mammals. Syst. Zool. 13: 57-73. Slatkin, M. 1974. Competition and regional coexistence. Ecology 55: 128-134. Stebbins, R. C. 1954. Amphibians and reptiles of western North America. McGraw-Hill, New York. Udvardy, M. D. F. 1969. Dynamic zoogeography. Van Nostrand Reinhold, New York. Van Devender, T. R. 1973. Late pleistocene plants and animals of the Sonoran Desert: a survey of

ancient packrat middens in southwest Arizona. Ph.D. diss. University of Arizona. Van Devender, T. R., and J. E. King. 1971. Late pleistocene vegetational records in western

Arizona. J. Ariz. Acad. Sci. 6:240-244. Watts, W. A. 1971. Postglacial and interglacial vegetation history of southern Georgia and central

Florida. Ecology 52:676-690. Willis, E. 0. 1974. Populations and local extinctions of birds on Barro Colorado Island, Panama.

Ecol. Monogr. 44:153-169.