Allozyme differentiation among gopher tortoises (Gopherus): conservation genetics and phylogenetic and taxonomic implications

Transcription

1 Allozyme differentiation among gopher tortoises (Gopherus): conservation genetics and phylogenetic and taxonomic implications DAVID J. MORAFKA Department of Biology, California State University, Dominguez Hills, Carson, CA , U.S.A. GUSTAVO AGUIRRE L. Institute de Ecologia, Apartado Postal 632, Durango, Durango, Mkxico AND ROBERT W. MURPHY' Deportment of Ichthyology and Herpetology, Royal Ontario Museum, 100 Queen's Park, Toronto, ON M5S 2C6, Cunudu Received February 1 1, 1994 Accepted July 27, 1994 MORAFKA, D.J., AGUIRRE L., G., and MURPHY, R.W Allozyme differentiation among gopher tortoises (Gopherus): conservation genetics and phylogenetic and taxonomic implications. Can. J. Zool. 72: The genetic diversity among 18 loci within and among four species of gopher tortoises was investigated using horizontal starch gel electrophoresis. Within species variation ranged from 2 to 4% by direct count and from 2 to 8% by Hardy -Weinberg expectation. Of the loci resolved, 11-22% expressed variation. The northern and southern populations of Gopherus flavomarginatus could not be distinguished. No fixed differences were observed between G. agassizii and G. berlundieri, as reflected in a Nei genetic distance of These latter species may be little more than allopatric populations of G. ugussizii. Differentiation between the two remaining species was not extensive; G. polyphemus was only slightly distinct from G.,flavomarginatus, being separated by a Nei genetic distance of only The two pairs of species were separated by an average genetic distance of The evolutionaly rates of divergence were observed to be unequal, especially between G. polyphemus and G. flavomarginutus. The overall genetic similarity suggests a relatively recent age of otigin. MORAFKA, D.J., AGUIRRE L., G., and MURPHY, R.W Allozyme differentiation among gopher tortoises (Gopherus): conservation genetics and phylogenetic and taxonomic implications. Can. J. Zool. 72: La diversit6 gcnctique a 18 locus chez quatre espbces de tortues terrestres a CtC CtudiCe par Clectrophorbse horizontale sur gel d'amidon. La variation entre les espbces a CtC CvaluCe 2 2-4% par dknombrement direct et a 2-8% selon les prkdictions de la loi Hardy-Weinberg. Parmi les locus identifits, 11-22% montraient des variations. Les populations du nord et du sud de G. flavomarginatus ne peuvent pas &tre distinguces. L'estimation de la distance gcnctique de Nei (0,008) ne permet pas de dcceler de diffkrences diagnostiques entre G. ugussizii et G. berlandieri. I1 se peut que ces espbces ne soient que des populations allopatriques de G. ugussizii. La diffkrence entre les deux autres espbces n'est pas considkrable; G. polyphemus se rapproche beaucoup de G. fluvomurginut~~s, puisque la distance gcnctique de Nei entre les deux n'est que de 0,006. Les deux paires d'espbces sont stpartes par une distance gtnctique moyenne de 0,200. Les taux Cvolutifs de divergence sont incgaux, particulibrement entre G. polyphem~~s et G. fluromurginatus. La resemblance gcnctique globale qui prcvaut chez ces espbces indique que leur origine remonte a une Cpoque relativement rccente. [Traduit par la RCdaction] -. - Introduction Gopher tortoises, genus Gopherus, are widely distributed across southern North America, in both the United States and MCxico. Gopherus consists of four extant species (Crumly 1994) and more than a dozen nominal fossil congeners (Auffenberg 1974). Of the extant species, G. agassizii is distributed from southwestern Utah southward to northern Sinaloa, MCxico, and eastward from the Mojave Desert to south-central Arizona. Gopherus berlandieri occurs in southeast Texas and the adjacent Gulf Coast of MCxico. Gopherus polyphemus is found from the extreme south of South Carolina to the extreme east of Louisiana and most of Florida, and G. Jlavomarginatus is restricted to the Chihuahuan Desert of MCxico (Auffenberg and Franz 1978). Recently, Ottley and..,. Velazques Solis (1989) mistakenly described a new species, G. lepidocephalus, from the Cape Region of southern Baja California Sur, MCxico. They incorrectly diagnosed a population of G. agassizii that had been introduced into a park (Crumly and Grismer 1994). No subspecies have been described for any recognized species. 'Author to whom all correspondence should be addressed Gopher tortoises have a substantial fossil distribution and geological history (Auffenberg 1974; Bramble 1982; Crumly 1994) dating back to the Oligocene Epoch. The paleontological record of this genus is perhaps more complete than that of any other extant genus of North American chelonian. Osteological features of the gopher tortoises, coupled with their chin glands and behavior, yield numerous synapomorphies, affirming monophyly and sequestering their early divergence from other testudinids (Gaffney and Meylan 1988). The systematics of gopher tortoises, both fossil and living, are controversial, ranging from invalid species descriptions to the generic allocation of extant species. Bramble (1982) split the gopher tortoises into Scaptochelys spp. and Gopherus spp., with S. agassizi, S. berlandierz, G. polyphemus, and G. Javomarginatus as living representatives. Later, Bour and Dubois (1984) pointed out that Scaptochelys was a junior synonym of Xerobates. Subsequently, Crumly ( 1984, 1994) recommended suppression of Xerobates because it could not be defined by synapomorphies, but rather by symplesiomorphies. Herein, we follow Crumly 's taxonomy. Molecular investigations of genetic diversity have been pursued partly because of conservation efforts. In the past Printed in Canada 1 Imprime au Canada

2 1666 CAN.. I. ZOOL. VOL. 72, years four studies have examined aspects of genetic differentiation using molecular methods. Jennings ( 1985). Glenn et al. ( 1990), and Rainboth et al. (1989) confined their studies to G. agassizii. Rainboth et al. (1989), in the most comprehensive allozyme study, noted that characterization of the population genetics of G. agassizii is a prerequisite for appropriate captive propagation, relocation, and strategic placement of preserves to effectively maintain genetic diversity. In the fourth study, Lamb et al. (1989) characterized all four extant species in terms of mtdna restriction fragment length polymorphisms. Genetic comparisons such as these provide an important context, both historical and demographic. Extrapolations may be made from one species of Gopherus to another, especially when conservation decisions need to be made quickly and life-history data for one particular taxon are incomplete. Such evaluations seem critical. For example, Gibbons ( 1986) concluded that an inadequate baseline biology for G. agassizii precluded effective life-history and demographic modeling. However, some of this information is available for congeners. Genetic information would assist conservation planners in choosing the most appropriate species or population from which to extrapolate. This is especially true if the life-history traits of the other species were mapped onto a cladogram to obtain predictions about evolutionary ecology (Brooks and McLennan 1991 ). All four species of gopher tortoises are suffering population declines. These range from local extirpations to widespread decimation caused by epizootic disease and massive habitat fragmentation and degradation (see Berry 1989; Rose and Judd 1989; Diemer 1989; Morafka et al. 1989). Mkxico's G. flavornarginatus has been listed as an endangered species (Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES), Appendix I). Since 1987 the U.S. Fish and Wildlife Service has recognized that the Mojave Desert G. agassizii is threatened. The Arizona Fish and Game Department has moved to protect this species. The Texas Parks and Wildlife Department now protects G. berlandieri, and the Florida Fish and Game Department has implemented a protection policy for G. polyphemus. The tortoises are being protected, but protection alone may not be enough. Perhaps, as for many other threatened and endangered species, at least some species of tortoises may survive only through captive propagation and informed management that considers the maintenance of high genetic diversity, among other variables. Herein we provide an allozyme-based survey of genetic variability and differentiation among the four extant species of gopher tortoises. Our survey is based on whole blood, because it obviated the necessity for sacrificing specimens. Our investigation includes a comparison of northern and southern populations of the endangered G. flavornarginatus. Materials and methods Tissue samples consisted of 1-3 ml of whole blood, which were initially frozen in liquid N2 and subsequently stored at -75 C. Blood samples were drawn by syringe (25 gauge needle) from either the scapular vein or the brachial artery of the axillary region (Avery and Vitt 1984) or from either the carotid artery or the jugular vein of the lateral neck (Jacobson et al. 1992). All samples were taken from wildcaught individuals, except for G. agassizii and northern G. flavomarginatus. These were captives being maintained in semi-natural outdoor habitats in the vicinity (within 60 km) of their original sites of collection. Species (sample size) and collecting data are as follows: G. agassizii (6) from Tuscon, Pima County, Arizona, 16 July 1986; G. berlandieri (23) from Jim Hogg County, Texas, July 1986; G. polyphemus (26) from North Palm Beach, Florida, 5 March 1987; G.flavomarginatus (6) from Rancho Benthon, Sierra del Diablo, Chihuahua, MCxico, 8 December 1986; G.fla~~omurginatus (21) from 10 km northwest of the Laboratorio del Desierto de Instituto de 't Ecologia, Chihuahua, MCxico, 25 June - 5 July, A single specimen of Geochelone chilensis, purchased from a commercial 5 supplier, was initially included as an out-group. No voucher specimens were kept. The identity of the individuals and the location of their respective populations are not in question, and thus sacrificing was not justified. Moreover, all taxa are currently protected at state and (or) federal levels of jurisdiction. Horizontal starch gel electrophoretic methods follow those of Murphy et al. ( 1990). Electrophoretic conditions and locus nomenclature follow those of Rainboth et al. (1989), with minor modifications. We resolved enzymatic products of esterase-d (Est-D; EC ; Harris and Hopkinson 1976) on a Tris-citrate-EDTA buffer system, and the mitochondria1 form of superoxide dismutase (msod-a) on the buffer systems of Rainboth et al. (1989); Rainboth et al. (1989) did not resolve either of these loci. In addition, as well as following the methods of Rainboth et al. (1989), the enzymatic products of L-lactate dehydrogenase and hemoglobin (HB) were resolved on the Tris-citrate-EDTA buffer system, and HB was also examined on a Tris-borate-EDTA buffer system. Our allelic nomenclature follows Murphy and Crabtree (1985). Our genotypic data were evaluated using ~10sYs release 1.6 (Swofford and Selander 1989). All loci were evaluated for conformance to Hardy -Weinberg expectations, genetic polymorphism, and genetic structuring using Wright's F statistics. Genetic divergence was also examined by calculating the genetic distance coefficients of Nei (1978) and Rogers (1972), with subsequent clustering of the latter using the Distance Wagner procedure (Farris 1972). Results Thirteen enzyme systems encoded by 18 loci were resolved for all in-group taxa (Table 1) but not necessarily for all individuals. No alleles were observed to be shared between any of the four species of Gopherus and the Geochelone chilensis specimen. Consequently, the out-group data were dropped from further consideration. (In many cases the enzyme products for Geochelone chilensis were inadequately resolved on the buffer systems used for Gopherus). No fixed differences were observed between G. berlandieri and G. agassizii, and only one apparent fixed difference separates G. polyphernus and G. flavornarg~nar~ls (s~od-a'). The only variation separating the former two taxa was the presence of relatively rare alleles and resulting differences in genotypic (and gene) frequencies (Table 1). A relatively low level of genetic divergence was observed among the species. The two phenetically similar pairs were separated by two apparently fixed differences at Est-D and Pep-A, both typically highly variable loci. These differences are summarized as genetic distance coefficients of Rogers (1972) and Nei (1978) (Table 3). The extent of locus polymorphism is given in Table I. The average heterozygosity per individual is about 2-4% based on direct count, and 2-8% based on the Hardy-Weinberg equilibrium. The discrepancy indicated a heterozygote deficiency as discussed below. The level of variability is comparable to that observed for many vertebrates, especially sit-and-wait predatory lizards (Gorman et al. 1977), although lower than that reported by Rainboth et al. (1989) for G. agassizii. The estimations are neither particularly high nor low. Despite the rather small sample sizes for some taxa, the estimates would seem to be reasonably accurate based on both vb C

3 MORAFKA ET AL. TABLE 1. Genotypes of four species of gopher tortoises, genus Gopherus, and estimates of variability, including mean sample size per locus (MN) (and standard error, ), percentage of loci that are polymorphic (PLP), mean number of alleles per locus (MNAPL), and mean locus heterozygosity by both direct observation (MHO), and Hardy -Weinberg expectations (MHHW) among them Locus N S G. polypphemrrs G. berlandieri G. agassizii Genotypes aa (3) aa (3) aa (6) aa (7) aa (6) aa (6) bb (6) bb (6) aa (6) aa (26) aa (3) aa (6) bb (4) bc (2) aa(1) ab(1) bb (5) MN PLPO MNAPL MHO MHHW~ Estimates of variability O NOTE: Monoallelic loci are listed because of variance in the number of individuals from which some locus products were resolved. Numbers in parentheses are sample sizes. N, northern; S. southern. a A locus is considered polymorphic if the frequency of the most common allele does not exceed Unbiased estimated (see Nei 1978). empirical (Gorman and Renzi 1979) and theoretical (Nei 1978) grounds. The number of loci expressing variation within populations ranged from 1 to 4, or %, using a 5% frequency cutoff level (Table 1 ), or from 11.1 to 22.1 % with a 1 % cutoff level. The two populations of G. flavomarginntus were highly concordant in expressing similar amounts of variation and in having the least amount of variation among all taxa sampled. Despite comparable sample size, G.flavomarginatus displayed only one-half to one-fourth as many polymorphic loci as reported for all other species. This species has both the smallest distribution and lowest absolute numbers. The pair G. berlandieri and G. agassizii were similarly consistent in having equivalent amounts of genetic variation, although this is only half as much as that observed in G. polyphemus. Our Hardy-Weinberg estimations showed one locus to be problematic in G. agassizii, mmdh-a (p < 0.006), and two in G. polyphemus, slcdh-a (p < 0.057) and Pk-A (p < 0.000). In both cases, a deficiency in heterozygotes may be responsible. However, these results are neither surprising nor unexpected, considering the relatively small sample sizes of these two taxa and that such reports are not atypical (e.g., Aguillar-S. et al. 1988). Table 2 provides calculations of Wright's F statistics, indicating Fi,, Fit, and F,, values for all variable loci. The F statistics are hierarchical (Wright 1978; Hart1 and Clark

4 CAN. I. ZOOL. VOL. 72,1994 TABLE 2. Wright's F statistics for the genus Gopherus when the genus is considered to be conlposed of a single breeding population LOCUS Fi s Fit Fst Distance from Root G. fiavornarginoi~~ (south) G. fiavornorginafus (north) G. polyphernus Pep-A Hb-b ooo G. beriandieri Pnp-A slcdh-a G. agoss~z~i Esr-D rnmdh-a OOO FIG. 1. Phenetic similarity among five populations of gopher tortoises smdh-a from North America, genus Gopherus, as summarized by Nei's ( 1978) Pep-D genetic distance, clustered by the Distance Wagner procedure (Farris 1972) (cophenetic correlation = 1.000; Fanis' (1972) "F" = 0.008), s~od-a' and arbitrarily rooted at the midpoint of the largest distance. This Pk-A phenogram is not rooted using an out-group and as such is not a Mean defensible hypothesis of phylogenetic history. I 4 TABLE 3. Matrix of Nei's ( 1978 (below diagonal) and Rogers' ( (above diagonal) genetic distance coefficients calculated from 18 loci for the genus Gopherus 1989) in that Fis gives the average inbreeding coefficient of an individual within a breeding unit, Fit gives the average inbreeding coefficient of an individual relative to the total population, and Fst denotes the level of substructuring within the total population. F,, values will range from 0 (complete panmixia) to 1.0 (breeding units fixed for alternative alleles), while Fi, and Fit values can be either positive (heterozygote deficiency) or negative (heterozygote excess). Among the species of Gopherus, Fi, values ranged from (smdh-a) to (mmdh-a). Four values of this statistic indicated heterozygote deficiencies, whereas three others suggested a slight heterozygote excess (Table 2). Treating the five groups theoretically as demes within the same population, it may be safely assumed that gene flow exists neither between the two subgroups nor between G. polyphenzus and G. flavomarginatus. However, the relationships between G. agassizii and G. berlandieri remain problematic. Similarly, Fit values vary widely by locus, from (smdh-a) to 1.00 (Pep-A, Hb-B, Est-D, and n~mdh-a), with an average of These values also indicate that the taxa are not part of a panmictic gene pool. Table 3 gives the genetic distance values (D; Nei 1978; Rogers 1972) comparing all five populations. Northern and southern populations of G. flavomarginat14s are essentially indistinguishable, having a Nei D value of and a Rogers D value of Similarly, a Nei D value of only separates G. agassizii from G. berlandieri; G. polyphemus is slightly more distinct from G. flavomarginatus with a D value of This degree of separation is essentially the same as that observed by Rainboth et al. (1989) between Mo.jave Desert populations of G. agassizii (D = 0.007). In contrast, the two species pairs are separated by relatively high values, such as between G, agasslzii and G, flavomarginatus, and between G. ber1andrer.i and G. polyphemus. Figure 1, a phenogram of Nei's D values clustered by the FS FN P B A Distance Wagner procedure (Farris 1972), clearly depicts the relative similarities for all five groups. The tree is arbitrarily - - FS rooted at the mid~oint of the longest ~atristic distance of the u L FN network because the out-group provided no information. This P tree is not necessarily indicative of phyletic history. B Discussion NOTE: FS, G. flai~omarginatus, southern; FN, G. flavomarginatus, northern; Our assessment of genetic diversity and variability among P, G. polypl~~rt~ns; B, G. berlandieri; A, G. agassizii. gopher tortoises by means of blood, has provided a surprisingly complex and revealing data set while using small sample sizes obtained with a minimum of trauma or pathological risk to the animals. The blood samples were collected in conjunction with investigations of blood chemistry and cell counts for evaluations of state of health. Our study took advantage of the simultaneous collection of blood and we encourage parallel studies with other taxa at risk of extinction. From our data-base of 18 loci and the results derived from them, a number of questions may be addressed, all of which are of significance to the conservation of the gopher tortoises. Among them, the following issues are preeminent: (i) fixed allelic differences generally fail to separate gopher tortoise populations and are uncommon even between different species; (ii) differentiation of species-specific genotypes is only modestly developed, much less so than predicted by paleontological reconstructions of gopher tortoise genealogy; (iii) evolutionary rates of allozyme change may have been unequal; and (iv) G. berlandieri may be little more than an allopatric dwarf subspecies of G. agassizii. Regarding the first issue, within G. flavomalginat14s virtually no differences, even at the level of allele frequency, were detected even though the two populations are separated by nearly 100 km by the intervening dry lagunas Del Rey and Palomas. Because these playas have filled, expanded, and receded several times within the last years, terrestrial connections between northern and southern populations were probably intermittently reestablished (Morafka 1988). Even the allopatric and distinct G. berlandieri could not be separated from G. agassizii by fixed allozyme character states. Our data suggest that considerable relocation of conspecific individuals might take place as part of conservation efforts without posing a high risk of outbreeding depression. However, 4 v

5 A ET AL this finding is contradictory to those of Lamb et al. (1989); substantial differentiation among populations of G. agassizii occurs in the mitochondrial genome. Our observations have applications or relevance within an allozyme context only. Ecological, demographic, epidemiological, and ethnological considerations might also lead to conclusions that differ from our limited data set. We would not recommend relocation programs until additional data are available. Regarding points ii and iii above, Bramble (1982) suggested that the "Gopherus" (sensu stricto) stock diverged from the "Scaptochelys (Xerobates)" stock in the early Miocene, about million years ago. Scaptochelys had its origin in the middle Oligocene. This date of divergence contrasts strongly with Auffenburg's (1976) earlier conclusion that it occurred no later than middle Pleistocene. Auffenburg's minimum estimation may be too recent, given the very polyphemus-like fossils that he described (1974, 1976) from the Late Pliocene of Texas. However, even a middle Pliocene event would bring forward differentiation of the two lineages by 10 million years relative to Bramble's Miocene hypothesis. Our genetic data seem germane to this debate. It seems highly unlikely, even when compared with other vertebrates, and especially with "reptilian" mtdna distance values, that these molecular rates could be reconciled with the fossil record. Murphy (1983) found that a Nei genetic distance of 1 equaled about 19 million years of divergence, at least among lizards. If this rate of divergence is applicable to our tortoises, then the divergence of the turtles should have occurred sometime during the past 4 million years or less, i.e., around the Miocene-Pliocene boundary. A more recent date would seem likely, because most of the divergence is based on two rapidly evolving loci. However, numerous convincing arguments about the invalidity of a molecular clock (summarized by Hillis and Moritz 1990) compel us to be extremely cautious in applying molecular clocks to estimate the antiquity of speciation events in the history of the genus. A strict application of the molecular clock is eliminated, because of the unequal rates of evolution between compared lineages (Fig. 1 ). Even though strict application of the clock is not justified, the discovered divergences are small by almost any calibration of known vertebrate rates. Our findings are far more consistent with the mtdna divergence estimates of Lamb et al. (1989), who suggested that virtually all differentiation occurred within the last 5 million years, than they are with the fossilbased estimation of middle Oligocene made by Bramble (1982). (T. Lamb, 1994, personal communication, states that revised estimates may place the divergence of the two species pairs in the Miocene, about 10 million years ago.) Admittedly, it is treacherous to calibrate mtdna distances against standard vertebrate rates, especially when reptile taxa divergences are estimated against a mammalian standard. Higher metabolic oxidation rates might accelerate mutation considerably in mammals (Martin et al. 1992; Avise et al. 1992), and similar changes might require a time interval of a greater order of magnitude in poikilothermic vertebrates (such as sharks). Given the arbitrary midpoint rooting of the largest distance of our tree (Fig. 1 ), and the theoretical implications about the inability to interpret distance data phylogenetically (e.g., Farris 1981), we do not consider this summary of similarities to be a phylogenetic hypothesis. Rather, we believe that our distance analysis resulted in a phenetic tree free from the assumption of equal rates of change among all lineages. Moreover, gene frequencies, the source of all genetic distance coefficients, are not heritable characteristics and should not be used in the reconstruction of genealogical relationships (Murphy 1993). Consequently, our tree is a phenogram, one that more accurately depicts the similarity values in Table 3. Notwithstanding this philosophical interpretation, the separation of the two pairs of species is clear. It appears that rates of allozyme evolution have been unequal when different lineages are compared. If we assume that the midpoint root of the tree is correct, then there has been a relatively slow rate of change in G.,flavomarginatus relative to G. polyphemus and the other Gopherus spp. However, it is equally likely that there has been a particularly rapid rate of change in G. polyphemus and that the root of the tree occurs at the midpoint between G.,flavomarginatus and the G. herlandieri - G. agassizii pair; our data do not differentiate between these two hypotheses. Finally, the taxonomic relationship of G. herlandieri to G. agassizii has been discussed rarely. Mertens and Wermuth (1955) suggested the subjugation of all extant species of Gopherus as subspecies of G. polyphemus, a position not specifically defended in their text, and one generally rejected by subsequent workers. Yet, morphologically the Mojave Desert G. agassizii is little more differentiated from G. herlandieri than it is from Sonoran Desert conspecifics (see Germano 1989). These variances include subtle differences in carapace ground color, the bifurcation of the male gular shields, hind foot and head shape, and adult shell size and depth. Lamb et al. (1989) found that two of five G. agassizii lineages and G. herlandieri show comparable, nearly identical genetic distances based on mitochondrial DNA restriction fragment data. Furthermore, it has long been known that captive G. agassizii and G. herlandieri can "hybridize" (Woodbury 1952; Mertens 1964; Watson 1993). Apparently, the 500-km gap between their ranges (from Benson, Arizona, to Del Rio, Texas) was reduced or even closed as recently as years ago. At this time, G. agassizii ranged into the Lake Wisconsin of eastern New Mexico and western Texas (VanDevender et al. 1976). Gopherus herlandieri was probably not displaced to the south during this time but rather remained proximate to Texas G. agassizii (Morafka 1977, 1994). The overall similarity of Sonoran Desert populations of G. agassizii and G. herlandieri, combined with the greater differentiation of Mojave Desert G. agassizii, suggests that the former species may not be monophyletic. This was also noted by Lamb et al. (1989). In order to resolve the taxonomic implications of these data, i.e., the synonymization of G. berlandieri into G. agassizii, a detailed geographically broad-based phylogenetic investigation is required. The practical implications of these close relationships are severalfold. The robust G. herlandieri populations of south Texas might provide some insight into optimal environments for G. agassizii. This might include whether or not the Pleistocene habitats are preferable to those to which G. agassizii is currently restricted. Similarly, ecological, demographic, and ethnological information that has already been established for one of the two species might well be extrapolated to the other when direct evidence is not available. However, such attempts should be employed only when absolutely necessary and with consideration for fundamental differences in climate, shelter, primary productivity, and body size differences.

6 1670 CAN. J. ZOOL. \. Acknowledgments This research was supported in part by the Natural Sciences and Engineering Research Council of Canada (NRC) grant A3148 and by the generous financial assistance of the Royal Ontario Museum (ROM) Department of Volunteers and the ROM Foundation to R.W.M. Most laboratory analyses were carried out in the ROMKJniversity of Toronto Department of Zoology joint Laboratory of Molecular Systematics established by a generous grant from NRC. D.J.M. acknowledges the support of Southern California Edison, Inc., the World Wildlife Fund, the U.S. Army National Training Center at Fort Irwin, and the American Museum of Natural History Center for Biodiversity and Conservation. G.A.L. received partial support from Consejo Nacional Ciencias y Tecnologia grant PCCBBNA Permits for the transportation of blood were provided by the U.S. Fish and Wildlife Service, Agriculture Canada, and MCxico's wildlife service (Secretaria Dessorrollo Urbana y Ecologia (DUE), Permit No ). Several colleagues provided us with critical samples. In particular, we thank R. Burke, J. Flanagan, J.V. Jarchow, and G. Adest. Valuable assistance was provided by G. Adest and P.M.S. Dagney. The manuscript was improved by the comments of T. Lamb, R. MacCulloch and one anonymous reviewer. This research is part of Mtxico's contribution to Man and Biosphere - United Nations Educational Scientific, and Cultural Organization Program of Biosphere Reserves. Aguillar-S., M.A.A., Sites, J.W., Jr., and Murphy, R.W Genetic variability and population structure in the lizard genus Petrosaurus. J. Herpetol. 22: Auffenberg, W Checklist of the fossil land tortoises (Testudinidae). Bull. Fla. State Mus. Biol. Sci. 18: Auffenberg, W The genus Gopherus (Testudinidae): I. Osteology and relationships of extant species. Bull. Fla. State Mus. Biol. Sci. 20: Auffenberg, W., and Franz, R Gopherus Rafinesque. Cat. Am. Amphib. Reptiles Avery, H.W., and Vitt, L.J How to get blood from a turtle. Copeia, 1984: Avise, J.C., Bowen, B.W., Lamb, T., Meylan, A.B., and Bermingham, E Mitochondria1 DNA evolution at a turtle's pace: evidence for low genetic variability and reduced microevolutionary rate in testudines. Mol. Biol. Evol. 9: Berry, K.H Gopherus agassizi. Desert tortoise. In The conservation biology of tortoises. Edited by I.R. Swingland and M.W. Klemens. International Union for Conservation of Nature and Natural Resources (IUCN) Species Survival Commission Occas. Pap. No. 5. IUCN, Gland, Switzerland. pp Bour, R.? and Dubois, A Xei.obates: a synonym older than Scaptochelys (Reptilia, Chelonei, Testudinidae). Bull. Mens. Soc. Linn. Lyon, 53: Bramble, D.M Scaptochelys: generic revision and evolution of gopher tortoises. Copeia, 1982: Brooks, D.R., and McLennan, D.A Phylogeny, ecology, and behavior: a research program in comparative biology. University of Chicago Press, Chicago. Crumly, C.R A hypothesis for the relationships of land tortoises genera (in Testudinidae). First International Symposium on Fossil Turtles. Studia Geologica Salmanticensia, Vol. 1 (Studia Palaeocheloniologica I), pp Crumly, C.R The phylogenetic systematics of gopher tortoises: evidence for their classification. In Biology of North American tortoises. Vol. 13. Edited by R.B. Bury and D.J. Germano. U.S. Department of Interior, Fish and Wildlife Research. pp Crumly, C.R., and Grismer, L.L Validity of the tortoise, Xerobates lepidocephalus Ottley and Velazques, in Baja California. In Biology of North American tortoises. Vol. 13. Edited by R.B. Bury and D.J. Germano. U. S. Department of Interior, Fish and Wildlife Research. pp Diemer, J.E Gopherus polyphemus. Gopher tortoise. In The i. conservation biology of tortoises. Edited by I.R. Swingland and M.W. Klemens. International Union for Conservation of Nature * and Natural Resources. (IUCN) Species Survival Commission Occas. Pap. No. 5. IUCN, Gland, Switzerland. pp Farris, J.S Estimating phylogenetic trees from distance matrices. Am. Nat. 106: Farris, J.S Distance data in phylogenetic analysis. In Advances in Cladistics: Proceedings of the First Meeting of the Willi Hennig Society, Lawrence, Kansas, October Edited by V.A. Funk and D.R. Brooks. New York Botanical Garden, Bronx. pp Gaffney, E.S., and Meylan, P.A A phylogeny of turtles. Syst. Assoc. Spec. Vol. 35: Germano, D.J Growth and life histories of North American tortoises (Gopherus) with special emphasis on the desert tortoise (G. agassizi) Ph.D. thesis, University of New Mexico, Albuquerque. Gibbons, J.W Movement patterns among turtle populations: applicability to management of the desert tortoise. Herpetologica, 42: Glenn, J.L., Straight, R.C., and Sites, J.W., Jr A plasma protein marker for population genetics studies of the desert tortoise. Great Basin Nat. 50: 1-8. Gorman, G.C., and Renzi, J., Jr Genetic distance and heterozygosity estimates in electrophoretic studies: effects of sample size. Copeia, 1979: Gorman, G.C.> Kim, Y.J., and Taylor, C.E Genetic variation in irradiated and control populations of Cnemidophorus tigris (Sauria, Teiidae) from Mercury, Nevada with a discussion of genetic variability in lizards. Theor. Appl. Genet. 49: Harris, H., and Hopkinson, D.A et seq. Handbook of enzyme electrophoresis in human genetics. North-Holland, Amsterdam. Hartl, D.L., and Clark, A.G Principles of population genetics. 2nd ed. Sinauer Associates, Inc., Sunderland, Mass. Hillis, D.M., and Moritz, C An overview of applications of molecular systematics. In Molecular systematics. Edited by D.M. Hillis and C. Moritz. Sinauer Associates, Inc., Sunderland, Mass. pp Jacobson, E.R., Schumacher, J., and Green, M Field and clinical technique for sampling and handling blood for hematologic and selective biochemical determinations in desert tortoises (Xerobates agassizi). Copeia, 1992: Jennings, R.D Biochemical variation of the desert tortoise, Gopherus agassizi. M.Sc. thesis, University of New Mexico, Albuquerque. Lamb, T., Avise, J.C., and Gibbons, J.W Phylogeographic patterns in mitochondrial DNA of the desert tortoise (Xerobates agassizi) and evolutionary relationships among North American gopher tortoises. Evolution, 43: Martin, A.P., Naylor, G.J.P., and Palumbi, S.R Rates of mitochondrial DNA evolution in sharks is slow compared with mammals. Nature (London), 357: Mertens, R Uber Reptilienbastarde, 111. Nat. Mus. Forschungsinst. Senckenb. 45: Mertens, R., and Wermuth, H Die rezenten Schildkroten, Krokodile, Bruckenechsen. Zool. Syst. Jahrb. Abt. Oekol. Geogr. F Tiere, 83: Morafka, D.J A biogeographical analysis of the Chihuahuan Desert through its herpetofauna. Biogeographica. 9. Morafka. D.J Historical biogeography of the Bolson tortoise. Ann. Carnegie Mus. 57: Morafka, D.J., Aguirre, G., and Adest, G.A Gopherus jzavomarginatus. Bolson tortoise. Irz The conservation biology of tortoises. Edited by I.R. Swingland and M.W. Klemens.

7 MORAFKA ET AL International Union for Conservation of Nature and Natural Resources (IUCN) Species Survival Commission Occas. Pap. No. 5. IUCN, Gland, Switzerland. pp Morafka, D.J., Adest, G.A., Reyes. L.M., Aguirre, G., and Lieberman, S.S Differentiation of North American deserts: a phylogenetic evaluation of a vicariance model. Tulane Stud. Zool. Bot. Suppl. Publ. No. 1. pp Murphy, R.W Paleobiogeography and genetic differentiation of the Baja California herpetofauna. Occas. Pap. Calif. Acad. Sci. No pp Murphy, R.W The phylogenetic analysis of allozyme data: invalidity of coding alleles by presence/absence and recommended procedures. Biochem. Syst. Ecol. 21: Murphy, R.W., and Crabtree, C.B Evolutionary aspects of isozyme patterns, number of loci, and tissue-specific gene expression in the prairie rattlesnake, Crotalus viridis viridis. Herpetologica, 41: Murphy, R.W., Sites, J.W., Jr., Buth, D.G., and Haufler, C.H Proteins I: isozyme electrophoresis. In Molecular systematics. Edited by D.M. Hillis and C. Moritz. Sinauer Associates, Inc., Sunderland, Mass. pp Nei, M Estimation of average genetic heterozygosity and genetic distance from a small number of individuals. Genetics, 89: Ottley, J.R., and Velazques Solis, V.M An extant, indigenous tortoise population in Baja California Sur, Mexico, with the description of a new species of Xerobates (Testudines: Testudinidae). Great Basin Nat. 49: Rainboth, W.J., Buth, D.G., and Turner, F.B Allozyme variation in Mojave populations of the desert tortoise. Copeia, 1989: Rogers, J.S Measures of genetic similarity and genetic distance. Studies in Genetics. VII. Univ. Tex. Publ. No pp Rose, F.L., and Judd, F.W Gopherus berlandieri. Texas tortoise. In The conservation biology of tortoises. Edited by I.R. Swingland and M.W. Klemens. International Union for Conservation of Nature and Natural Resources (IUCN) Species Survival Commission Occas. Pap. No. 5. IUCN, Gland, Switzerland. pp Swofford, D.L., and Selander, R.B BIOSYS 1.6: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systen~atics. Illinois Natural History Survey, Urbana. VanDevender, T.R., Moodie, K.B., and Harris, A.H The desert tortoise (Gopherus agassizi) in the Pleistocene of the northern Chihuahuan Desert. Herpetologica, 32: Watson, W.G Possible Gopherus agassizii X G. berlandieri interspecific hybridization. Tortuga Gazette, 29: 8-9. Woodbury, A.M Hybrids of Gopherus berlandieri and Goplzerus agussizi. Herpetologica, 8: Wright, S Evolution and the genetics of populations. Vol. 4. Variability in and among natural populations. University of Chicago Press, Chicago.