Printer Friendly

Mitochondrial DNA phylogeography of Perognathus amplus and Perognathus longimembris (Rodentia: Heteromyidae): a possible mammalian ring species.

Key words.--Biogeography, cytochrome-b, evolutionary genetics, mitochondrial DNA, nucleotide sequence, Perognathus, restriction fragments, ring species, speciation.

Of continuing interest to evolutionary biologists is the question of how new species arise. Numerous modes of speciation have been proposed (White 1978; Mayr 1982), and their attributes and relative frequency in the several taxa of animals and plants have been, at times heatedly, debated. A consensus appears to have been reached, in the case of terrestrial vertebrates at least, that the primary mechanism of speciation is geographic (or allopatric) speciation by one or both of its two major modes: the Dumbell Model supposes that a broadly distributed species is divided by an extrinsic geographic barrier into two roughly equal halves; under Peripatric Speciation, small peripheral isolates form either by colonization of previously unoccupied habitat, or by climatic deterioration resulting in isolation (Mayr 1970, 1982). Genetic differences between the isolates accumulate with time under both of these modes, and speciation is judged to be complete when the isolates reestablish sympatry and fail to interbreed.

A special case of geographic speciation is seen in ring species (polytypic species demonstrating circular overlap). In Mayr's (1970, p.291) opinion, "the perfect demonstration of speciation is presented by the situation in which a chain of intergrading subspecies forms a loop or overlapping circle whose terminal links have become sympatric without interbreeding, even though they are connected by a complete chain of intergrading or interbreeding populations" [italics added]. It is curious that Mayr should include circular overlaps as examples of geographic speciation given that a complete chain of intergrading populations is nowhere extrinsically separated. Circular overlaps, if they exist, demonstrate speciation by distance: a mode of speciation for which Mayr (1970, p. 320) finds little or no evidence. Mayr (1970, pp. 291-293) resolved this paradox by close examination of the purported cases of ring species available to him, and concluded that in nearly all of them the supposed "complete chain of intergrading populations" is broken in one or more places, thus making these cases examples of geographic speciation. In my opinion, the value of ring species, to students of speciation, lies not so much in their demonstration of any particular mode of speciation, but in their clear demonstration of the sequence of events leading to differentiation.

For example, Moritz et al. (1992) studied the mitochondrial cytochrome-b phylogeny of the salamander Ensatina eschscholtzii, arguably one of the clearest cases of a ring species (Stebbins 1949). They found very high levels of nucleotide sequence divergence among populations within the presumed ancestral subspecies occurring in northern California and southern Oregon. These ancestral populations were consistently basal in the phylogenies, rendering them paraphyletic with respect to the named subspecies. Further south in California, the samples from the west slope of the Sierra Nevada, and from the coast ranges formed two separate monophyletic lineages that were interpreted as consistent with two independent stepwise colonizations from the North to the South. The southernmost (and thus most terminal or derived) populations of these two lineages occur in sympatry, and are reproductively isolated from each other. Moritz et al. (1992) conclude that these features of their mitochondrial phylogenies support the interpretation of E. eschscholtzii as a ring species. Whether E. eschscholtzii is freely intergrading around the entire circumference of the ring is debatable in my opinion. If it is not, then the possibility exists that geographic or allopatric speciation has occurred, allowing sympatric coexistence in southern California. This point is important for determining the correct names and status of the various taxa of E. eschscholtzii, but does not diminish the clarity of the history of its evolution around the ring.

The pocket mice Perognathus amplus and P. longimembris are morphologically very similar (Hoffmeister 1986). They are, for the most part, allopatrically distributed (Hall 1981), but are known to occur sympatrically at one locality: 2.5 miles north of Wenden, La Paz County, Arizona (Hoffmeister 1986; this study). Here, as over much of both species' respective ranges, P. longimembris is noticeably smaller in all morphological features except relative tail length (Hoffmeister 1986; cf. Hall 1981). In northern Arizona, however, where the ranges of the two species are separated by the the Colorado River and its canyon, these two mice are so similar in size as to be nearly indistinguishable (Hoffmeister 1986). This observation led Hoffmeister to hypothesize that P. amplus and P. longimembris may not be separate species; he proposed that P. amplus may represent one divergent end of a ring species demonstrating circular overlap. According to this hypothesis, the relatively large P. longimembris in northern Arizona or adjacent Nevada crossed the Colorado River and spread through Arizona, becoming progressively larger. These large-bodied "P. amplus" can coexist in sympatry with the relatively small P. longimembris that have more recently crossed the Colorado River into west-central Arizona.

If Hoffmeister's (1986) hypothesis is correct, two different phylogenies depicting the hypothetical relationships among the operational taxonomic units (OTUs) in this study would be possible: (1) The OTUs representing Perognathus amplus would form a clade within a paraphyletic Perognathus longimembris (fig. 1a). This result would be similar to the relationships between Peromyscus polionotus and Peromyscus maniculatus found by Avise et al. (1983), and between Thomomys townsendii and T. bottae found by Patton and Smith (1989, 1994), and will be called the "paraphyletic hypothesis." (2) The OTUs of the two species would form two monophyletic groups; the OTUs nearest the root within each of the two clades would be those from the two sides of the Colorado River in northern Arizona, and the OTUs representing Perognathus amplus and Perognathus longimembris from the zone of sympatry in west-central Arizona would be separated by large patristic distances (fig. 1b). This will be called the "Northern Origin hypothesis." Several alternative hypotheses to that of Hoffmeister (1986) can be proposed, but the phylogenies they suggest are divisible into two groups. (3) The OTUs of the two species are not consistently joined into clades, thus leaving the status of the two taxa unresolved (fig. 1c). This would occur if Perognathus amplus arose from multiple colonizations across the Colorado River and will be called the "Polyphyletic" hypothesis. (4) The OTUs are resolved into two monophyletic groups, but with OTUs other than those from northern Arizona near the root (fig. 1d). As an alternative to the Northern Origin hypothesis, this will be called the "Southern Origin" hypothesis.

In this paper, I use the restriction-fragment patterns of the entire mitochondrial DNA (mtDNA) molecule of Perognathus amplus and P. longimembris from Arizona, California, Utah, and Nevada (fig. 2, Appendix) to identify distinct haplotypes, and to analyze the within- and between-population component of haplotype variation. Because the homology of similar restriction fragments is often difficult to judge, and because restriction fragments violate the assumption of independence of characters, they are unsuitable for phylogeny reconstruction (Swofford and Olsen 1990). I therefore directly sequenced parts of the mitochondrial cytochrome-b gene of single representatives from each OTU and used these sequence data for phylogeny reconstruction. In addition to testing Hoffmeister's ring hypothesis, I will describe the way in which mitochondrial variation is distributed through two closely related polytypic rodent species. This is of particular interest because P. amplus and P. longimembris have similarly restricted habitat preferences and are exclusively found in desert scrub of the Mojave, Sonoran, and parts of the Great Basin deserts (Hoffmeister 1986). Neither species ever ventures far into wooded habitats; it is possible to capture these mice at the ecotone between desert scrub and juniper woodland, but they are not to be found further up in the juniper woodland (pers. obs.). These small mice therefore can potentially demonstrate extreme levels of local differentiation.


Collection of Specimens

In the summers of 1986, 1987, and 1988, with the help of my field assistants, I collected pocket mice with Sherman live traps baited with a commercial bird seed mixture. We placed 100-250 traps at 10-m intervals through the habitat, either in straight lines or in clover-leaf patterns, each night of trapping. Many of the populations were sampled on only one night and at only one place, but some (e.g., Wupatki Kingman, Wenden, and Wellton) were sampled at more than one place or in more than 1 yr (see Appendix). Though this form of sampling has the potential to accentuate the similarity within and the differences between populations, this will be seen to have had no great effect on the patterns revealed below. Specimens were returned to and housed in the University of Illinois, Urbana, School of Life Sciences animal care facility until processed.

DNA Extraction and Restriction-Fragment Analysis

I removed the liver, heart, and one kidney from each of 35 Perognathus longimembris and 45 P. amplus individuals, as well as those from one P. apache and one P. flavus, and isolated pure mtDNA from these pooled tissues by standard techniques (Spolsky and Uzzell 1984, 1986; Mack et al. 1986). I used 16 type II restriction endonucleases (ApaLI, AvaI, BamHI, BclI, DraI, EcoRI, EcoRV, HaeII, HindIII, KpnI, NdeI, PstI, Pvu II, SstI, SstII, and XbaI) to digest 5-10 ng of mtDNA from each P. longimembris and P. amplus, and followed the reaction conditions recommended by the supplier (Bethesda Research Labs, New England BioLabs). I end-labeled the resulting DNA fragments with mixed (G, A, T, C) alpha-32P-deoxynucleoside triphosphates using the large (Klenow) fragment of DNA polymerase I, and separated them electrophoretically by length on horizontal 1% agarose gels. The separated fragments were localized by autoradiography of dried gels. I estimated fragment sizes by comparison to known length markers (lambda phage DNA digested with HindlII, and a 1-kb ladder).

I assigned the mtDNA of individual pocket mice to haplotypes based on their fragment patterns for all informative enzymes (see below). I calculated the proportion of shared fragments (F) between haplotypes using equation (21) of Nei and Li (1979) and an estimate of the number of base pair substitutions per nucleotide (%p) by Upholt's (1977) method. To examine the variation in haplotypes at all localities with more than one individual, I calculated F and %p in pairwise comparisons for each individual from these populations, then calculated the average percent nucleotide divergence between individuals within populations, between populations within species, and between species.

Cytochrome-b Sequences

I amplified portions of the mitochondrial cytochrome-b gene using the primer pair gludgL-cytb2H of Palumbi et al. (1991). I performed symmetric and asymmetric Taq DNA polymerase mediated amplification reactions in 50-[mu]l volumes. Reaction mixes, exclusive of primers, were 50.0 mM KC1, 10.0 mM Tris-HC1 (pH 9.0), 0.1% Triton X-100, 2.0 mM [MgC1.sub.2], 12.5 [mu]M of each deoxynucleotide (dATP, dCTP, dGTP, dTTP), 1.25 units of Taq DNA polymerase, and 1.02.0 [mu]g template DNA. Symmetric amplifications were 0.5 [mu]M for each oligonucleotide primer (except for the degenerate primer gludgL, which was used at twice the normal concentration in all amplifications), and were subjected to 30 PCR cycles (1 min, 94 [degrees]C; 45 s, 50 [degrees]C; 2 min + 2 e/cycle 72 [degrees]C) in a Perkin-Elmer Cetus thermal cycler; asymmetric amplifications were 1.0 [mu]M for the excess primer and 0.02 [mu]M for the limiting primer, and were subjected to 35 cycles as above. I purified the double-stranded products by electrophoresis of 10%-20% of the reaction on 1% high-melting-point agarose gels, visualized the single-product bands with ethidium bromide and long wavelength UV, sliced out the bands, and melted them each in 0.5 ml of distilled water in microcentrifuge tubes at 94 [degrees]C. I used 2 [mu]l from each of these melted gel slices as templates in asymmetric reactions to make single-stranded DNA products for sequencing, and sequenced both single DNA strands of the amplified region using a commercial kit (Sequenase, United States Biochemical). DNA sequences were recorded, inverted, and tested for conformity using the GeneJockey sequence processor (Bio-soft, Cambridge, U.K.), then aligned by eye. I used PAUP 3.1.1 (Swofford 1993) to find the most parsimonious phylogeny and to calculate the percent sequence divergence based on the sequence data, and MacClade 3.0 (Maddison and Maddison 1992) to investigate character evolution and to examine the support for alternate hypotheses of relationship. I also used a neighbor-joining bootstrap program (NJBOOT2, written and supplied by K. Tamura) to provide bootstrap values for the various nodes of the phylogeny.


Mitochondrial DNA Restriction Fragments

All of the 16 restriction enzymes digested the mtDNA of at least one mouse from the entire study, and produced a total of 225 different-sized fragments. The restriction-fragment patterns from six of the enzymes were either uninformative (i.e., they produced identical patterns for all specimens: ApaLI, SstII), or produced single restriction fragments (i.e., cut the circular mtDNA molecule at only one restriction site: BamHI, HindIII, KpnI, and PstI) for more than one individual. The central assumption of the restriction-fragment method is that fragments of the same length (inferred by co-migration through an electrophoretic gel) shared by two or more different individuals are products of the same restriction sites. This assumption is potentially violated in the case of single restriction sites and restriction fragments because the active site could be any one of the more numerous sites found in other individuals. I chose not to pursue further the nature of these single restriction fragments; instead I eliminated these enzymes from the analysis. This led to only a small loss of information as these enzymes were largely invariant from locality to locality within the two species.

Based on the patterns of restriction fragments, I identified 38 distinct mitochondrial DNA genotypes (haplotypes) among the specimens of Perognathus longimembris and P. amplus (table 1). Average sequence divergence estimates within populations are generally low (tables 2, 3); nearly all are less than 1%, and most are less than 0.5%. Considering these low values and the distribution of divergence estimates in the two largest samples (fig. 3), it appears that at the collecting-locality level, one can expect to find a small number (e.g., 3-6) of slightly different haplotypes that are probably the result of mutation in situ. More divergent haplotypes sometimes occurred at collecting localities (e.g., at Stanfield, fig. 3b), and may have been brought in by a more divergent female immigrant. All of the high divergence estimates in figure 3b are due to the presence of one individual pocket mouse (table 1, haplotype 15); this individual's haplotype is very similar (%p = 0.21) to haplotype 20 from Buckeye, differing only by the possession of an additional DraI restriction site in haplotype 15. One other noteworthy pattern of local haplotype variation is no variation, as seen in the four mice sampled from Marble Canyon and Page, and the seven mice from Wupatki, Cameron, and Navajo Spring (tables 1, 2).


The letter designations of the haplotypes (table 1), and the percent sequence-divergence estimates (table 2) within and between collecting localities show that the haplotypes sampled are descendants of eight separate female lineages based in part on the presence of identical haplotypes at two or more localities (e.g., Wupatki/Cameron/Navajo Spring, Stanfield/ Gila Bend/Picacho, Marble Canyon/Page, Ft. Irwin/Laughlin, Wellton/Imperial), and in part on the low (<2%) between-locality sequence divergences (table 3). The three lineages within P. amplus are the Wupatki-Cameron-Navajo Spring group (haplotype number 1, table 1), the Wickenburg-King-man-Wenden-Aguila group (2-11), and the Stanfield-Gila Bend-Picacho-Buckeye group (12-20). The five lineages of P. longimembris are Beaver Dam (21), Mesquite (22-26), the Marble Canyon-Page group (27), the Ft. Irwin-Laughlin group (28-33), and the Wellton-Imperial-Wenden group (3438). Though the specimen of P. longimembris from Wenden (haplotype 34) has a lower divergence from the other members of its haplotype group than the 2.0% cutoff and is a member of the Wellton-Imperial-Wenden group, it will be kept separate because Wenden is the one place in this study where the two species coexist. With the exception of P. longimembris from Wenden, in the rest of this paper I will refer to the multilocality haplotype groups by the first-named locality in the groups above (i.e., Wupatki, Wickenburg, Stanfield, Ft. Irwin, and Wellton).


Percent sequence-divergence estimates from restriction fragments between the major haplotype groups within species range from a little less than 4% to just under 8% (table 2). These values, when compared to the low within-locality divergences, substantiate the treatment of the major haplotype groups as separate evolutionary lineages, and justify the use of single representatives from each group for sequencing. In general, the between-species divergences are greater still; most are larger than 9.5%. These between-species values should be used with caution. Although restriction fragments are very useful and reliable for the identification of the major haplotype groups and assessing within-population variability, one begins to lose confidence in the homology of similar-sized fragments in comparisons between more distant (both geographically and genetically) populations within species. At the between-species level, for these pocket mice, the use of restriction fragments becomes unreliable; there are very few co-migrating fragments shared between the taxa, and many of these may have had independent origins. It is probably safest to consider only the average of all between-species divergences (10.4%) as a rough estimate of the distance between the two species.

Mitochondrial Cytochrome-b Sequences

I obtained 375-396 bp of sequence data from the cytochrome-b region of representatives of the major haplotype groups and from two outgroup taxa (fig. 4). The sequences were all confirmed by sequencing both the "light" and the "heavy" strands. The topology of the most parsimonious cladogram from PAUP (276 steps, CI = 0.66, RI = 0.57, resealed CI = 0.38, fig. 5) was identical to that of the phylogeny produced by the neighbor-joining method (fig. 6). In this phylogeny, the two species (P. amplus and P. longimembris) are clearly and unambiguously separated into two monophyletic groups, allowing for rejection of both the Polyphyletic and the Paraphyletic hypotheses. This arrangement is robust as shown by the high bootstrap values at the bases of each monophyletic group under both of the tree-building strategies. This phylogeny is consistent with the Northern Origin Hypothesis since the taxa nearest the root within each clade are from both sides of the Colorado River in northern Arizona. Within the P. amplus clade, Wickenburg is basal; the lineage represented by Wickenburg is geographically the most widely distributed lineage in P. amplus, and members of this lineage are found at least as far north as 12.75 miles northeast of Kingman, Mohave County, Arizona. Within the P. longimembris clade, the most parsimonious arrangement is to place Mesquite nearest the root as shown in figure 5; however, placement of either Beaver Dam or Marble Canyon nearest the root, or making all three northern populations sister taxa to the remaining P. longimembris, adds only a few additional steps to the length of the cladogram. This is also reflected by the low bootstrap values for these nodes in the trees (figs. 5, 6). Regardless of which of these three taxa is truly the most closely related to P. amplus, they are all found near the north bank of the Colorado River in northern Arizona, and thus lend support to the northern origin hypothesis. The Southern Origin hypothesis is not supported by these data. A cladogram constructed with the southernmost taxa of P. amplus and P. longimembris (Stanfield and Wenden/Wellton, respectively) nearest the root was much less parsimonious (winning sites test, G = 10.82, P < 0.005; Prager and Wilson 1988; as modified by Edwards et al. 1991) than the topology in figure 5; it was 13 steps longer with 289 steps (CI = 0.63, RI = 0.52, resealed CI = 0.32).


Are Perognathus amplus and Perognathus longimembris a Single Ring Species?

As a result of his study of morphological variation within and between Perognathus amplus and P. longimembris, Hoffmeister (1986, pp. 254-255) proposed that these two forms may be extremes of a single ring-species demonstrating circular overlap. Though he continues to regard these two mice as separate species in his taxonomy, his hypothesis, if verified, would necessitate the inclusion of the two forms within a single species. Hoffmeister's hypothesis would have gained the strongest support from a paraphyletic relationship between P. amplus and P. longimembris, indicating that P. amplus is a recent descendant lineage within P. longimembris, and that continuing gene flow across the Colorado River is preventing the complete sorting of genetic variants. None of the data presented here indicate paraphyly of either taxon. All the data indicate that P. amplus and P. longimembris are monophyletic groups that diverged long ago. Based on the commonly used rates of nucleotide substitution for the whole mammalian mtDNA molecule (2% per my; Brown et al. 1979) and for mammalian cytochrome-b (2.5% per my; Meyer et al. 1990; Irwin et al. 1991), and the average sequence divergences between the two species from the restriction fragments and the sequence data, P. amplus and P. longimembris diverged approximately 5-6 mya.

Even though P. amplus and P. longimembris are old, molecularly well-differentiated species, the phylogenetic relationship of the lineages within both species is consistent with the Northern Origin hypothesis. This is made clear by superimposing the phylogeny produced from the cytochrome-b sequence data on a map of the collecting localities (fig. 7). The phylogeny on this map would be outgroup rooted along the branch that crosses the Colorado River in northwestern Arizona. Similar to the result of Moritz et al. (1992) with Ensatina, the longest terminal branches (i.e., the oldest living lineages) on the neighbor joining phylogeny (fig. 6) are the basal taxa Wickenburg and Mesquite. The poor resolution of the branching order among the three most basal lineages within P. longimembris is also similar to the paraphyly of the basal taxa of Ensatina: the retention of old, divergent lineages and the extinction of intermediate forms that could have resolved the phylogeny.

Though I agree with Hoffmeister (1986) that P. amplus and P. longimembris were separated by a vicariant or colonization event that occurred in northern Arizona, and that their evolution in and around Arizona has many of the attributes of a ring species, I disagree that this calls their species status into question. For P. amplus and P. longimembris to be members of a true ring species, they must be connected by freely intergrading populations around the entire circumference of the ring; the upper Colorado River currently prevents this from happening. Similar to the cases of circular overlaps studied by Mayr (1970), the ring connecting populations of P. amplus and P. longimembris is incomplete, thus making this a case of geographic speciation. Given that the two species are allopatric everywhere except 2.5 miles north of Wenden, one can test their biological species status only at this place. One bit of evidence in support of a geographic speciation model as opposed to speciation around a ring is the fact that all populations of P. amplus studied by McKnight and Lee (1992) possessed two pairs of chromosomes with secondary constrictions, whereas all populations of P. longimembris had only one pair with secondary constrictions. This uniform similarity within both species and difference between them would be more easily explained by a single vicariant or colonization event, not by accumulated difference around a ring.

One final point where I disagree with Hoffmeister's (1986) view of the evolution of these pocket mice is his assumption that P. amplus is descended from P. longimembris, and that morphological changes within P. amplus as it spread into Arizona allowed for the sympatry of the two forms. None of the evidence presented in this paper is consistent with an interpretation of P. amplus as a descendant of P. longimembris. That the most derived form of P. longimembris studied here is the form that occurs in sympatry with P. amplus contributes to the ringlike nature of their relationship. Whether P. longimembris from Mesquite or any of the other collecting localities would be able to coexist with P. amplus is an important question that will probably never be answered. Contrary to Hoffmeister's (1986) scenario, however, the population of P. amplus that is currently sympatric with P. longimembris, Wenden, is a member of the same mitochondrial lineage that my analysis identifies as the most basal lineage, Wickenburg. The genetic changes that preclude hybridization between the two species probably did not occur within the Wickenburg haplotype group. It is likely they occurred after the vicariant event that split the two species, or accumulated during the radiation and spread of P. longimembris.

The agreement of this study with Hoffmeister's hypothesis of a northern separation of the two species is in conflict with earlier opinions as to the evolution of these pocket mice. McKnight and Lee (1992) found that the most parsimonious reconstruction of karyotypic evolution in these pocket mice required a southern connection between them. In light of the new cytochrome-b sequence data presented here, a reinterpretation of their data seems warranted. Both P. amplus and P. longimembris have 56 chromosomes (2n = 56) and similar sex chromosomes; the only differences seen in their karyotypes are in the number of autosomal arms (fundamental number, FN), and, as noted above, the number of autosomal pairs with secondary constrictions (Patton 1967, 1970; Williams 1978; McKnight and Lee 1992). Perognathus amplus shows the greatest number of different karyotypes across its range with FN variants of 84, 86, 88, 90, 92, and 94; P. longimembris has only two documented FN variants, 86 and 88. McKnight and Lee (1992) considered the FN = 88 karyo-type ancestral, with three avenues of evolution away from it: (1) to lower FN in P. longimembris, (2) to lower FN in P. amplus, and (3) to higher FN in P. amplus. If the cytochromeb sequence cladogram (fig. 5) truly reflects the evolutionary history of these pocket mice, then the ancestral karyotype of the two species was 2n = 56, FN = 86. This new reconstruction of the karyotype data requires only one additional origination of the FN = 88 karyotype. This new view of karyotypic evolution in these mice, in conjuction with the cytochrome-b cladogram, suggests an intriguing explanation for the extreme variation in FN seen in P. amplus in central and southern Arizona. During the isolation allowing the differentiation of the Stanfield haplotype group, pericentric inversions produced the FN = 94 karyotype found in P. a. taylori from the vicinity of Tucson (Patton 1967, 1970). When this group of populations (or subspecies) came back in contact with the Wickenburg haplotype group (FN = 86), hybridization took place, and through [F.sub.1] and backcross hybridizations all the intermediate karyotypes were formed.

Populations Lacking Mitochondrial DNA Haplotype Diversity

Two separate haplotype groups, one in each species, were found to be completely lacking in mtDNA haplotype diversity. All of the P. amplus sampled from Wupatki National Monument, Cameron, and Navajo Spring had the same mtDNA haplotype (table 1, number 1) as revealed by their restriction fragments. Similarly, all of the P. longimembris from Page and Marble Canyon had a single mtDNA haplotype (table 1, number 27). Reduced levels of genetic diversity such as these could result from the colonization events that gave rise to these populations (Wade et al. 1994), but this is not likely because of the high sequence divergences separating these populations and their conspecifics (table 2). The amount of time necessary to achieve these levels of sequence divergence (2 4 my) is certainly enough that divergent haplotypes would have arisen. Alternatively, more recent population bottlenecks, possibly because of habitat restriction during Pleistocene glaciations, could reduce genetic diversity to these levels. Both of these haplotype groups occur in high elevation Great Basin scrub habitats that would have been reduced severely in extent during glacial periods, as forest habitat expanded down-slope from the surrounding plateaus.

The Utility of Studying Ring Species

Why is it worthwhile to study ring species? Fully connected, freely intergrading ring species, if they exist, demonstrate speciation by distance. Such species would have low vagility (i.e., low, but positive, gene flow), and distributions encircling some absolutely insurmountable geographic or habitat barrier. Species with low vagility, however, have concordantly low propensities to cross seemingly trivial dispersal barriers. Thus, the species that are most likely to demonstrate speciation by distance around a ring are also the species that are most susceptable to geographic or allopatric speciation, leading in most cases to uncertainty about the ultimate cause of reproductive isolation. Because of this I do not believe that ring species are valuable as exemplars of a particular mode of speciation. Ring species, either completely connected ones or broken ones, are valuable because they provide clear records of evolutionarily important colonization events as they expand their distributions around the central barriers. The pocket mice studied here are a good example.

Pocket mice have an ancient history; the earliest identifiable Perognathus fossils are from the early Miocene, Hemingfordian fauna, circa 16-20 mya (Wahlert 1993), and, by Barstovian time (12-16 mya), four different kinds of Perognathus are distinguishable (Savage and Russel 1983). During the late Miocene (i.e., 5-10 mya), I envision the ancestor of P. amplus and P. longimembris inhabiting the Miocene semidesert of Axelrod (1979, p. 28) from southern and western Arizona, around the head of the proto-Gulf of California into eastern California and southern Nevada and Utah. The split (ca. 5 mya) of the amplus-longimembris lineage into two major components most likely was caused by the barrier that separates the two species today, the Colorado River. Regrettably, the geological history of the development of the Grand Canyon and the modern course of the Colorado River is still uncertain. There is evidence, however, that the lower Colorado River (from the mouth of the Grand Canyon, near present-day Lake Mead, to the vicinity of Yuma, Ariz.) began to develop after the river changed course and river sediments were deposited into the head of the proto-Gulf of California between 5 and 8 mya (Buising 1990; Lucchitta 1990); where the Colorado River flowed before this time is unknown. This event, whatever its cause, agrees well with the estimated time of divergence of P. amplus and P. longimembris inferred from the mtDNA sequence data. With the rapid uplift of the Sierra Nevada, Transverse, and Peninsular mountain ranges of California during the Pliocene, a severe rain shadow was cast over the lands to the east, and the dry Mojave and Colorado (i.e., the Sonoran Desert in southeastern California) deserts attained nearly their present form (Axelrod 1983). As P. longimembris spread into these harsh deserts, it was selected for smaller body size.

As Grinnell (1914) pointed out, the Colorado River is not a potent barrier to dispersal over its entire length. The upper river, within the confines of the Grand Canyon, and further down stream near Needles, California, seldom changes its course, and is effective in separating closely related taxa on either side. The lower river tends to meander, and these course changes and oxbows have the potential to "transport" taxa from one side of the river to the other. This has happened for the P. longimembris in this study from the localities Wellton, Imperial, and Wenden. Some of the mice from Wellton and Imperial shared the same mtDNA haplotype, and those from Wenden were similar. On the basis of this similarity, I conclude that the presence of P. longimembris in southwestern Arizona is the result of a relatively recent colonization event. The selection for small size imposed by the harsh Colorado Desert no doubt contributed to the eventual sympatry of P. longimembris with the larger P. amplus.



This paper is part of a dissertation presented in partial fulfillment of the Ph.D. in Ecology, Ethology and Evolution at the University of Illinois, Urbana-Champaign. For assistance in collecting pocket mice, I thank T. W. Jesse, D. K. McKnight, and B. S. Seward. Financial and logistical support was provided by The American Museum of Natural History, Theodore Roosevelt Memorial Fund; U.S. Army, Construction Engineering Research Laboratory; University of Illinois Research Board and Dissertation Research Fund; D. K. and B. McKnight; C. Spolsky; and National Science Foundation grant BSR 90- 1 8686 to H. B. Shaffer. For granting permission to collect pocket mice on lands under their control, I thank the Navajo Nation and the U. S. Department of Interior: National Parks Service (Glen Canyon National Recreation Area); Fish and Wildlife Service (Kofa National Wildlife Refuge); and the Departments of Fish and Game of Arizona, California and Nevada. I also thank P. Marko, S. M. Rollmann, H. B. Shaffer, and two anonymous reviewers for helpful criticism of an earlier draft of the manuscript.


Avise, J. C., J. F. Shapira, S. W. Daniel, C. F. Aquadro, and R. A. Lansman. 1983. Mitochondrial DNA differentiation during the speciation process in Peromysous. Molecular Biology and Evolution 1:38-56.

Axelrod, D. I. 1979. Age and origin of Sonoran desert vegetation. California Academy of Science Occasional Papers 132.

--. 1983. Paleobotanical history of the western deserts. Pp. 113-129 in S. G. Wells and D. R. Haragan, eds. Origin and evolution of deserts. University of New Mexico Press, Albuquerque.

Brown, W. M., M. George, Jr., and A. C. Wilson. 1979. Rapid evolution of animal mitochondrial DNA. Proceedings of the National Academy of Sciences, USA 76:1967-1971.

Buising, A. V. 1990. The Bouse formation and bracketing units southeastern California and western Arizona: implications for the evolution of the Proto-Gulf of California and the lower Colorado River. Journal of Geophysical Research 95 (B12):20111-20132.

Edwards, S. V., P. Arctander, and A. C. Wilson. 1991. Mitochrondrial resolution of a deep branch in the genealogical tree for perching birds. Proceedings of the Royal Society of London, B. Biological Sciences 243:99-107.

Grinnell, J. 1914. An account of the mammals and birds of the Lower Colorado Valley with especial reference to the distributional problems presented. University of California Publications in Zoology 12:51-294.

Hall, E. R. 1981. The mammals of North America. Wiley, New York.

Hoffmeister, D. F. 1986. Mammals of Arizona. University of Arizona Press and Arizona Game and Fish Department, Tucson.

Irwin, D. M., T. D. Kocher, and A. C. Wilson. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32:128-144.

Lucchitta, 1. 1990. History of the Grand Canyon and of the Colorado River in Arizona. Pp. 311-332 in S. S. Beus and M. Morales, eds. Grand Canyon geology. Oxford University Press, New York.

Mack, A. L., F. B. Gill, R. Colburn, and C. Spolsky. 1986. Mitochondrial DNA: a source of genetic markers for studies of similar passerine bird species. Auk 103:676 681.

Maddison, W. P., and D. R. Maddison. 1992. MacClade: analysis of phylogeny and character evolution, Version 3. Sinauer, Sunderland, Mass.

Mayr, E. 1970. Populations, species, and evolution. Belknap Press, Cambridge, Mass.

--. 1982. Processes of speciation in animals in C. Barigozzi, ed. Mechanisms of speciation. Alan R. Liss, New York.

McKnight, M. L., and M. R. Lee. 1992. Karyotypic variation in the pocket mice Perognathus amplus and P. Iongimembris. Journal of Mammalogy 73:625-629.

Meyer, A., T. D. Kocher, P. Basasibwaki, and A. C. Wilson. 1990. Monophyletic origin of Lake Victoria cichlid fishes suggested by mitochondrial DNA sequences. Nature 347:550-553.

Moritz, C., C. J. Schneider, and D. B. Wake. 1992. Evolutionary relationships within the Ensatina eschscholtzii complex confirm the ring species interpretation. Systematic Biology 41:273-291.

Nei, M., and W. H. Li. 1979. Mathematical model for studying genetic variation in terms of restriction endonucleases. Proceedings of the National Academy of Sciences, USA 76:5269-5273.

Palumbi, S., A. Martin, S. Romano, W. O. McMillan, L. Stice, and G. Grabowski. 1991. The simple fool's guide to PCR. Version 2.0. University of Hawaii, Honolulu. Distributed by the authors.

Patton, J. L. 1967. Chromosomes and evolutionary trends in the pocket mouse subgenus Perognathus (Rodentia: Heteromyidae). Southwestern Naturalist 12:429-438.

--. 1970. Karyotypes of five species of pocket mice, genus Perognathus (Rodentia: Heteromyidae), and a summary of chromosome data for the genus. Mammalian Chromosomes Newsletter 11: 3-8.

Patton, J. L., and M. F. Smith. 1989. Population structure and the genetic and morphologic divergence among pocket gopher species (genus Thomomys). Pp. 284-304 in D. Otte and J. A. Endler, eds. Speciation and its consequences. Sinauer, Sunderland, Mass.

--1994. Paraphyly, polyphyly, and the nature of species boundaries in pocket gophers (genus Thomomys). Systematic Biology 43:11-26.

Prager, E. M., and A. C. Wilson. 1988. Ancient origin of lactalbumin from Iysozyme: analysis of DNA and amino acid sequences. Journal of Molecular Evolution 27:326-335.

Savage, D. E., and D. E. Russell 1983. Mammalian paleofaunas of the world. Addison-Wesley, Reading, Mass.

Spolsky, C., and T. Uzzell. 1984. Natural interspecies transfer of mitochondrial DNA in amphibians. Proceedings of the National Academy of Science, USA 81:5802-5805.

--. 1986. Evolutionary history of the hybridogenetic hybrid frog Rana esculenta as deduced from mtDNA analysis. Molecular Biology and Evolution 3:44-56.

Stebbins, R. C. 1949. Speciation in salamanders of the plethodontid genus Ensatina. University of California Publications in Zoology 48:377-526.

Swofford, D. L. 1993. PAUP: phylogenetic analysis using parsimony. Version 3.1.1. Illinois Natural History Survey, Champaign, III.

Swofford, D. L., and G. J. Olsen. 1990. Phylogeny reconstruction. Pp. 411-501 in D. M. Hillis and C. Moritz, eds. Molecular systematics. Sinauer, Sunderland, Mass.

Upholt, W. B. 1977. Estimation of DNA sequence divergence from comparison of restriction endonuclease digests. Nucleic Acids Research 4:1257-1265.

Wade, M. J., M. L. McKnight, and H. B. Shaffer. 1994. The effects of kin-structured colonization on nuclear and cytoplasmic genetic diversity. Evolution 48:1114-1120.

Wahlert, J. H. 1993. The fossil record. Pp. 1-37 in H. H. Genoways and J. H. Brown, eds. Biology of the Heteromyidae. Special Publication no. 10. American Society of Mammalogists.

White, M. J. D. 1978. Modes of speciation. W. H. Freeman, San Francisco.

Williams, D. F. 1978. Karyological affinities of the species groups of silky pocket mice (Rodentia: Heteromyidae). Journal of Mammalogy 59:599-612.


Voucher specimens of all pocket mice used in this study are housed in the University of Illinois Museum of Natural History, Urbana. The names of the collecting localities are in capital letters and are arranged below by state and county, from North and West to South and East. The number of specimens is in parentheses.

Perognathus amplus: Arizona, Coconino Co.: NAVAJO SPRING, 2 mi SE Marble Canyon (1); CAMERON, 4 mi S, 0.25 mi W Cameron (1); WUPATKI, 2.5 mi N, 3 mi W Wupatki Ruin (5); Mohave Co.: KINGMAN, 8.5 and 12.75 mi NE Kingman (2); La Paz Co.: WENDEN, 2.5 and 3.5 mi N Wenden (3); Maricopa Co.: AGUILA, 5 mi WSW Aguila P.O. (4); WICKENBURG, 2.5 mi W, 2.75 mi S intersection AZ 93 and US 60, Wickenburg (2); 20 mi SSW Wickenburg (2); 3.5 mi E junction AZ 74 and US 60 (1); BUCKEYE, 13 mi S, 1.6 mi E Buckeye (5); GILA BEND, 10 mi E Gila Bend (1), Pinal Co.: STANFIELD, 6 mi W Stanfield (16); PICACHO, 2.8 mi S Picacho P.O. (2).

P. longimembris: California, San Bernardino Co.: FT. IRWIN, 2 mi W Echo Site, Goldstone, Ft. Irwin (5); Imperial Co.: IMPERIAL, 2 mi N 1-8 on county hwy. S 34 (2); Nevada, Clark Co.: LAUGHLIN, 19 mi S, 3.5 mi E Searchlight (3); Utah, Kane Co.: PAGE, 0.8 mi NNE of hwy. 89, Lone Rock Rd., Glen Canyon NRA (3); Arizona, Mohave Co.: BEAVER DAM, 5 mi N Beaver Dam (1); MESQUITE, 4 mi ESE Mesquite, Clark Co., NV (16); Coconino Co.: MARBLE CANYON, 0.3 mi SSW Marble Canyon (1); La Paz Co.: WENDEN, 2.5 mi N Wenden (1); Yuma Co.: WELLTON, 9.5 mi W Wellton (3).

P. apache: Arizona, Coconino Co.: Elk Rd. West, near cemetery, Page (1). P. flavus: Arizona, Coconino Co.: 2.5 mi N, 3 mi W Wupatki Ruin (1).
COPYRIGHT 1995 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:McKnight, Mark L.
Date:Oct 1, 1995
Previous Article:Local adaptation and agents of selection in a mobile insect.
Next Article:Evolutionary origin of a parthenoform, the Amazon molly Poecilia formosa, on the basis of a molecular genealogy.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters