Copeia, 2004(2), pp. 207–221
Small Fish in a Large Landscape: Diversification of Rhinichthys osculus (Cyprinidae) in Western North America DAVID D. OAKEY, MICHAEL E. DOUGLAS,
AND
MARLIS R. DOUGLAS
We mapped 112 restriction sites in the mitochondrial DNA genome of the Speckled Dace (Rhinichthys osculus), a small cyprinid fish broadly distributed in western North America. These data were used to derive a molecular phylogeny that was contrasted against the hydrographic evolution of the region. Although haplotypic variation was extensive among our 59 sampled populations and 104 individuals, their fidelity to current drainage basins was a hallmark of the study. Two large clades, representing the Colorado and Snake Rivers, were prominent in our results. The Colorado River clade was divided into four cohesive and well-defined subbasins that arose in profound isolation as an apparent response to regional aridity and tectonism. The Lower and Little Colorado River subbasins are sister to one another and (with the Upper Colorado River) form a large clade of higher-elevation populations that seemingly reflect postglacial recolonization from refugia in the Middle Colorado River. The latter subbasin is sister to the Los Angeles Basin and, thus, supports the hypothesis of an ancient connection between the two. A haplotype from the Northern Bonneville was sister to the entire Colorado River clade. The Snake River clade revealed a strongly supported Lahontan group that did not share haplotypes with surrounding basins. It contained instead scattered sites from former Pluvial Lake Lahontan, as well as from eastern California. It was, in turn, sister to the Owens River, whereas Rhinichthys falcatus was sister to this larger clade. The hypothesis of a southerly, ‘‘fishhook’’-configured tributary associated with a westward-draining Pliocene Snake River was manifested by the relationship of this Lahontan clade to upper Snake and northern Bonneville localities. The Klamath/Pit and Columbia Rivers were sisters in a clade basal to all the above, which in turn supported the hypothesis of a pre-Pliocene western passage of the Snake River. Our data also suggested at least three separate ichthyofaunal invasions of California, as well as a Bonneville Basin fragmented by a north-south connection between southeastern Idaho and the Colorado River. The dual western and southern movements of R. osculus from southern Idaho argued for a northern origin, possibly associated with Tertiary Lake Idaho.
T
HE freshwater fishes of western North America comprise an isolated and endemic fauna. As such, they are evolutionarily unique (Minckley and Douglas, 1991). Few species evolved among basins; most instead originated within basins that became transient over geological time (Miller, 1958; Minckley et al., 1986). During these periods of reconstructive tectonism, fishes and their water sources either endured in situ or were instead diverted into newer amalgams of older systems. The only constant for these fishes through time was their relative seclusion. Through the Tertiary, this isolation was exacerbated by an ever-increasing aridity that eventually culminated by Late Cenozoic in numerous extinctions (Smith, 1978). These events effectively abbreviated and molded an already isolated fauna to the extent that it became not only depauperate in overall diversity but also concomitantly rich in distinctive morphologies (Douglas, 1993). Our biological under-
standing of this fauna has been impeded for more than a century by the imposing topography of the region and by the relative inaccessibility of its rivers (Minckley and Douglas, 1991). The Escalante River, a tributary of the Colorado River in southeastern Utah, was the last river in the continental United States to be discovered and named (Dellenbaugh, 1873; Stegner, 1954: 142). Thus, our knowledge base for this fauna is at best rather fragmentary, and hence interrelationships among populations and species are incompletely known. The fishes of western North America are also ancient. The integration of modern drainage basins began during early Miocene and was antecedent to the evolution of most Western fish genera. The latter differentiated instead during the tortuous 20-million-year passage into Pleistocene (Smith, 1981; Minckley et al., 1986). Many of these species are now characterized by widespread distributions and extensive morpho-
q 2004 by the American Society of Ichthyologists and Herpetologists
208
COPEIA, 2004, NO. 2
logical variation (Smith, 1966; Behnke, 1992; Stearley and Smith, 1993). In the cyprinid genus Siphateles, for example, geographic variation occurred exclusively within basins yet was largely unaffected by progressive Pleistocene desiccation (Hubbs et al., 1974). The endemic catostomids (Pantosteus and Catostomus) also seemingly differentiated within basins yet were directly impacted by orogeny rather than aridity. Other cyprinids and salmonids (i.e., Gila, Rhinichthys, Oncorhynchus) demonstrated even more complex patterns that apparently stemmed from an amalgam of both processes (above), as well as from dispersal (Miller, 1946a; Hubbs et al., 1974; Allendorf and Leary, 1988). And finally, morphological variability and speciation in western fishes was affected not only by age and isolation but also by occasional interspecific hybridization (DeMarais et al., 1992; Minckley and DeMarais, 2000). Those Western North American freshwater fish genera most amenable to large-scale phylogeographic analyses are Oncorhynchus (Behnke, 1992; Stearley and Smith, 1993; McCusker et al., 2000), Siphateles (Hubbs et al., 1974), and Rhinichthys (Minckley et al., 1986). The latter contains a widespread western species (Rhinichthys osculus, the Speckled Dace, a ‘‘mountaincreek type’’ sensu Miller, 1958) that has attained its current broad distribution in part by headwater capture and stream transfer across low divides (Minckley, 1973; for review of processes, see Bishop, 1995). This species exhibits extensive morphological variation and, as such, has suffered a long and tortuous taxonomic history (La Rivers, 1962; Miller, 1984). For example, when first recognized as the western genus (now subgenus) Apocope, it was thought to comprise some 12 species ( Jordan et al., 1930). However, Miller (in Miller and Miller, 1948) stated: ‘‘The forms of Rhinichthys (subgenus Apocope) in the West exhibit so much overlap in their characters that most of the nominal species are now regarded as comprising a single, wide-ranging species, R. osculus (Girard).’’ Morphological variation in this species is now recognized through application of geographic trinomials (sensu La Rivers, 1962; Smith et al., 2002). The evolutionary relationships within this species, and how these juxtapose onto the harsh landscape of the American West, form the backdrop of our study. The numerous physical and biological developments described above have had a clear and extensive impact on the evolution of freshwater fishes in the water-poor American West. Yet testing these for strength of signal has proven to be a difficult and nontrivial task. Morphology has
often yielded little explanatory power (but see Douglas et al., 1999), and there has instead been a growing emphasis on molecular methods as a means to ascertain and weigh the veracity of evolutionary hypotheses. Here, mitochondrial (mt) DNA, because of its rapid evolution, neutrality, and matrilineal inheritance, has often proven advantageous in recovering shared histories of closely related taxa (Avise et al., 1987). Studies employing mtDNA data often seem to alternate between two extrema: not enough variation (i.e., presence of minimal polymorphism), or too much (stochastic lineage sorting at polymorphic sites; Moritz et al., 1987). Moreover, as a single gene tree, mtDNA represents only a small fraction of the total history within a sexual pedigree (Schneider et al., 1998). In this sense, molecular studies have not always proven to be the window to deep history that many had expected or would desire (see, for example, Fu, 2000). In this study, we employed restriction enzymes to cut mtDNA at specific sites and mapping techniques (as per Dowling et al., 1996) to identify and quantify these cleavage sites within the mtDNA genome. The resulting data are independent variables that infer mutually exclusive character states and can be rendered into binary data that correspond to presence/absence of restriction sites. As such, they are readily amendable to a large variety of analytical programs (Swofford et al., 1996). Mapped mtDNA restriction sites were used herein to derive an intraspecific phylogeny of R. osculus that was juxtaposed with the hydrologic evolution of Western North America. MATERIALS
AND
METHODS
Sampling.—From three to seven individuals were sampled in each of 59 populations throughout the range of R. osculus (Fig. 1; Appendix 1). Many (i.e., 52%) were collected by regional fish biologists, whereas the remainder (48%) were collected by DDO and Arizona State University personnel. At least two populations were sampled per subbasin so as to more fully represent potentially significant regions such as the Great Basin. In addition, three other species were also examined as outgroups: Rhinichthys atratulus (Rouge River, MI); Rhinichthys cataractae (Wind River, WY; South Platte River, CO; Snake River, ID); and Rhinichthys falcatus (Columbia River, OR). Rhinichthys atratulus was identified as a sister species to R. osculus by Coburn and Cavender (1992) and Woodman (1992).
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST
209
Fig. 1. (A) Major drainage basins of western North America: Colorado River (AZ, UT, WY, CO); Columbia River (WA, OR, ID); Snake River (WA, ID, UT, WY); Sacramento River (CA, OR); Klamath River (CA, OR); Bonneville Basin (UT, NV, ID); Lahontan Basin (NV, CA, OR); Death Valley (CA, NV). (B) Collection localities for 61 Rhinichthys osculus samples in western North America. This paper reports analyses of 59 localities: Frenchman’s Lake and Last Chance Creek are both represented by Squaw Queen Creek. Locality information is in Appendix 1.
Mitochondrial DNA extraction.—MtDNA was extracted from mt-rich tissues (i.e., eggs, heart, liver, spleen, kidney) and isolated by CsCl-gradient centrifugation (Dowling et al., 1996). Aliquots of purified mtDNA were digested separately by 15 hexameric restriction endonucleases (i.e., BamHI, BclI, BglII, BstEII, EcoRI, HindIII, MluI, NcoI, NdeI, NheII, PvuII, SacI, SacII, XbaI, XhoI). Success depends upon obtaining complete digestion of the DNA with each enzyme. Cleavage fragments were end-labeled with all 4 a-32p dNTPs, separated by electrophoresis through 1% agarose and 4% acrylamide gels, and visualized by autoradiography. Size standards (phage lambda DNA digested with HindIII, and phage FX 174RF DNA digested with HaeIII) were included on each gel, providing estimates of fragment size. Differences in fragment profiles could be readily attributed to simple losses and gains of restriction sites, and letters were assigned to unique profiles in order of appearance. Each individual was assigned an alphanumeric code denoting composite mtDNA haplotype compiled across 15 enzymes. A single preparation of end-labeled DNA can
be used to map recognition sites for several different enzymes. The generation of comprehensive mtDNA restriction-site maps becomes a relatively efficient process, and these were constructed for 104 haplotypes. Restriction sites in a given enzyme profile were mapped relative to cleavage sites generated in pairwise double digests by other enzymes. Five endonucleases (i.e., BamHI, BstEII, EcoRI, HindIII, and PvuII) formed the foundation for every restriction map, and three to four pairwise digestions were performed for each of the remaining 10 restriction enzymes. Although it is labor-intensive to use multiple gel mediums and to hand-construct restriction maps, these dramatically reduce the uncertainty regarding differential migration of very small fragments caused by secondary conformational differences (Dowling et al., 1996). When the placement of a restriction site was ambiguous, it was removed from the analysis. In all, 14 such sites (8.6%) were deleted. Phylogenetic analyses.—The most widely used strategy for finding optimal trees under a par-
210
COPEIA, 2004, NO. 2
simony approach is to employ random addition sequences in conjunction with tree bisection-reconnection branch swapping (Nixon, 1999). This strategy works well when numbers of taxa are reduced, but larger datasets (i.e., those . 40–50 taxa) have proven problematic (Goloboff, 1999). This is because the latter contain numerous composite optima (or tree islands) that, in turn, make it difficult to identify a globally optimum tree. A large number of suboptimal trees are instead produced, and although these reflect identical tree lengths, they differ among themselves with regard to minor rearrangements. Their accumulation often fills system memory to capacity and overly taxes the patience of researchers. Thus, larger datasets often require search strategies that specifically deal with the problem of composite optima. One (the parsimony ratchet) was demonstrated in Nixon (1999), whereas a second (i.e., TNT [Tree analysis using New Technology, vers. 0.2 g; P. A. Goloboff, J. S. Farris, and K.C. Nixon; www. zmuc.dk/public/phylogeny]) was described in Goloboff (1999). We employed the latter to derive minimum length trees from an initial data matrix of 104 haplotypes and 154 binary characters by selecting the following parameters: Random Sectorial Searches (RSS) 5 15/35/3/5; Consensus-Based Sectorial Searches (CSS) 5 same as above; Tree-Drifting (DFT) 5 30/4/0/20/0; and Tree-Fusing (TF) 5 5 (see Goloboff, 1999, for explanation of parameters). A series of arguments have been presented to support (Goloboff, 1995; Allard and Carpenter, 1996; Nixon and Carpenter, 1996) or critique (Turner and Zandee, 1995) a posteriori character weighting as an analytical strategy. The intent of the process is to increase efficiency of phylogenetic analysis by differentially weighting characters according to cladistic reliability (the latter defined by Farris [1969] as the fit between character and phylogeny). This procedure removes heterogeneity from data while improving congruence among informative and usually more conservative characters (Allard and Carpenter, 1996). Here, successive approximation weighting (Farris, 1969) has been most frequently used (see, for example, Wills et al., 1998; Anderson, 2000; Platnick, 2000). In this study, we used PAUP* (vers. 4.0b8; D. L. Swofford, unpubl.) to produce a strict consensus of the 395 fused trees from TNT. A heuristic search was then performed in PAUP*, with trees rooted at R. atratulus and with characters weighted by the consistency index (CI; Farris, 1969), as determined from the consensus tree. The CI, rather than the rescaled consistency index (RCI), was used following recommenda-
tions of Archie (1996:158), who noted that the CI was already properly scaled between (0,1) and thus did not require rescaling. Constant (n 5 10) and uninformative (n 5 32) characters were given weight 5 0. Thus, the final matrix consisted of 104 individuals and 112 parsimony-informative characters. The heuristic search employed tree bisection-reconnection (TBR), saved multiple trees (MULTREES), kept only best trees, and used the input tree as the starting tree. The strict consensus from this analysis then served as input for a second pass through the data, with parameters set as above and with the CI recalculated from the new input tree. A majority-rule consensus tree (identical in topology to the strict consensus) was then derived from the resulting 477 most parsimonius trees. RESULTS Restriction site variation.—Our hexameric restriction sites had a relatively uniform distribution around the mtDNA molecule, with 888 corresponding base-pairs representing approximately 5.3% of the estimated 16.78 kb mtDNA genome. The total number of restriction sites generated per haplotype map ranged from 137 to 144. Evidence of length variation (i.e., 1 100bp) was evident in one individual from the Lahontan Basin (Oakey, 2001). A majority of the 59 populations was represented by unique haplotypes, ranging from one (of seven individuals, Amargosa River) to five (of six individuals, Reese River), and these were generally differentiated by one or two restriction sites (Oakey, 2001). Populations from adjacent localities in close proximity often shared the same haplotypes. Other haplotypes were broadly distributed in larger watersheds. Several localities in the middle Colorado, Lahontan, and Bonneville Basins (Fig. 1) revealed strongly divergent haplotypes that exhibited phylogenetic affinity with other basins. Phylogenetic analysis.—Results of the weighted MP analysis revealed that R. osculus was paraphyletic, with a single R. falcatus from the Columbia River buried within it (Fig. 2). The tree was rooted at R. atratulus (Michigan) and was followed immediately by haplotypes of R. cataractae from eastern and western Rocky Mountain drainages. The smaller Columbia River clade was sister to a clade composed of the Clearwater River 1 two divergent haplotypes from Southern Bonneville (i.e., SEV4 and BSC2). These were sister to the Klamath 1 Pit clade, and this larger clade was, in turn, sister and basal to the
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST
211
western Great Basin (i.e., Lahontan) and the Colorado River physiographic regions. The latter two formed the largest and most extensive clades in the tree. Each exhibited geographic subbasins within which the fidelity of mtDNA haplotyes was quite high. The strongly supported clade of Colorado River dace (top, Fig. 2) was partitioned into four geographically defined subbasins, and also contained haplotypes from the Southern Bonneville and Los Angeles Basins. It consists of a Lower Basin 1 Little Colorado River (LCR) subclade is sister to these, the Upper Basin subclade. The Middle Colorado subbasin is sister to the Los Angeles River and this clade was sister to all the above. Finally, a Northern Bonneville haplotype (i.e., BOX1) is basal to the entire Colorado Basin. A second, strongly supported Lahontan Clade (middle, Fig. 2) was composed of individuals from the Humboldt River 1 eastern California, and its sister, the Owens River. Basal to the Lahontan clade was R. falcatus. Northern Bonneville and Upper Snake River haplotypes were each, in turn, sister to the Lahontan 1 R. falcatus clade. Overall, haplotypes clustered within monophyletic drainage basins that were relatively resolved. DISCUSSION
Fig. 2. Majority-rule consensus of 477 trees produced in PAUP* and based on 104 Rhinichthys osculus haplotypes and 112 restriction sites. Characters were weighted by consistency index with constant and unreliable characters given weight 5 0. Original tree was consensus of 377 most parsimonious trees produced in TNT. Location names at tips of branches are arranged in couplets with the first name representing the upper branch and the trailing name the second branch. Location data are provided in Appendix 1.
Phylogenetic analyses and basin fidelity.—Our analyses of mapped mtDNA restriction sites revealed a nonmonophyletic R. (Apocope) osculus. The inclusion of R. falcatus in the osculus-clade, and the possible presence of additional but undescribed forms in the Columbia River Basin (Peden and Hughes, 1981, 1988; Hughes and Peden, 1989), emphasize the need for large-scale studies of Rhinichthys in this region. For example, R. falcatus and R. osculus coexist across much of the former’s distribution (Peden and Hughes, 1988), and R. falcatus was initially included as a member of the R. osculus group (Hubbs et al., 1974). Thus, a more thorough assessment of genetic variability is now required for R. falcatus before its position in this topology can be properly interpreted. Perhaps R. falcatus is simply a geographic form of R. osculus, as suggested by Hubbs et al. (1974) and above. A prominent feature of our data is the remarkably high fidelity by which R. osculus clusters within designated subbasins (as synopsized in Fig. 3). Indeed, our tree is relatively consistent with the idea that western fishes differentiated within basins, with each of the latter now characterized by high endemism and few species in common (i.e., Miller’s [1958] ‘‘centers
212
COPEIA, 2004, NO. 2 cause it is an ecological generalist and an extinction-resistant dispersalist (as per Smith, 1981). It has endured in a fractured landscape by using numerous but intermittent geographic connections that existed during the Cenozoic and by surviving in favorable habitats with greater frequency than other taxa (Smith, 1978). Thus, large gaps in haplotypic diversity created by long-range dispersal and subsequent largearea extinctions do not characterize R. osculus. Instead, its ubiquitous distribution and extremely large populations (i.e., ‘‘millions,’’ per Jordan, 1891) would be influential in maintaining this diversity in the American West and in retarding the time necessary for phylogenetic diversification. We discuss these and other aspects in greater detail below, as we dissect each of the well-resolved clades in our tree.
Fig. 3. Relationships among drainage basins in western North America as inferred from mtDNA restriction site data for Rhinichthys osculus.
of endemism’’ concept). Yet, R. osculus was recognized by Jordan and Evermann (1896) and Jordan et al. (1930) as a distinct species largely because of its ‘‘morphological and geographic cohesiveness.’’ More extensive (i.e., dense taxon) sampling will be required to determine whether morphologically described subspecies are congruent with the molecular results depicted herein. Most R. osculus populations exhibited extensive restriction site variation. That is, nearly every population was represented by one or more unique haplotypes distinguished by three restriction sites. This Type-III phylogeographic pattern (Avise et al., 1987), suggests that longterm zoogeographical barriers have not limited gene flow. A more typical phylogeographic pattern would instead reveal a few widespread genotypes, with others but a few steps from the common types. This clearly does not typify R. osculus in western North America. Possibly, this species maintains considerable variation be-
The Colorado Basin clade.—This well-supported clade includes Upper, Middle, Lower, and Little Colorado River subbasins, plus the Los Angeles River and a basal Northern Bonneville haplotype. The high fidelity of haplotypes within subbasins typifies an endemism largely attributable to the prolonged development of these reaches as isolated segments ( Jordan, 1891; Uyeno and Miller, 1963; Hunt, 1969). In this sense, the Colorado River may be far older than previously imagined (Hershler et al., 1999; Howard, 1996, 2000). During upper Paleocene to mid-Miocene, the Los Angeles and Ventura Basins received drainage from several major eastern precursors (i.e., a lower Colorado that drained the Sonoran provinces, and an Amargosa/Colorado that drained the Mojave provinces; Howard, 1996, 2000). In addition, Gila (Cyprinidae) from the Lower Colorado River Basin seemingly diverged morphologically as a result of early to mid-Pliocene vicariant events (Douglas et al., 1999). These data juxtapose with the hypothesis that the modern western ichthyofauna is indeed ancient (i.e., Oligocene-Miocene) (Minckley et al., 1986). We suggest that R. osculus was present within these early and interior western drainages and that our data reflect these ancient connections. Los Angeles Basin.—Haplotypes from the Santa Ana and San Gabriel Rivers in the Los Angeles Basin formed a well-supported monophyletic assemblage that was sister to the Middle Colorado River clade. Similarly, Cornelius (1969) found that Rhinichthys osculus carringtoni (the geographic form found in the Los Angeles Basin) was meristically and morphologically most similar to Rhinichthys osculus yarrowi from the Middle Colorado River than it was to R. o. carringtoni
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST from north-coastal and northeast California. At least two other Los Angeles Basin fishes (i.e., Catostomus [Pantosteus] santaanae and Gila orcutti) also have hypothesized nearest relatives in the Lower Colorado River (Smith, 1966). Catostomus (Pantosteus) santaanae probably arrived in southern California coastal drainages as the result of an early (e.g., Pliocene) and westwarddraining Colorado River (Smith, 1966). In our analyses, the basal nature exhibited by the Los Angeles Basin suggests an old connection with the Middle Colorado River, followed by a long period of isolation. It also suggests that the Colorado River and the Pacific coastal drainages were linked by an ancient fluvial connection. However, additional studies (both genetic and geologic) are needed to more fully develop such a biogeographic synthesis (see Minckley et al., 1986). Lower elevation Colorado River drainages.—The Middle Colorado Basin (i.e., Pluvial White, Moapa, and Lower Virgin Rivers) was sister to the Los Angeles Basin, with both sub-clades basal and sister to the remaining Colorado River haplotypes. The Middle Colorado historically drained the southwestern margin of the Colorado Plateau, and it is characterized by elevated endemism (Miller and Hubbs, 1960). Portions of the Middle Colorado represent the lowest elevations in the watershed, and the high numbers of haplotypes found there (Oakey, 2001) suggest that effective population sizes were previously quite large. In addition, the Middle Colorado may have provided refugia for the eventual recolonization of higher elevation sites in the Upper Colorado River Basin and the LCR (W. L. Minckley, pers. comm.). The close relationship between Middle and Upper Colorado basins is supported by the shared presence of a morphological attribute (i.e., a frenum) in both Upper Basin R. o. yarrowi, and Middle Basin (i.e., Pluvial White River) Rhinichthys osculus velifer (Miller, 1984). In addition, haplotypes of Pahranagat Valley R. o. velifer were virtually identical in our analyses with those from the Moapa River (Oakey, 2001). These results conflict with the interpretation of Williams (1978), who regarded the Moapa form as meristically intermediate between R. o. velifer and R. o. yarrowi. The close genetic relationship between R. o. velifer and the ‘‘Moapa River’’ form is not surprising, particularly since both occur within drainages that are subject to intermittent flooding (Hubbs and Miller, 1948; Miller and Hubbs, 1960). Monophyly of Lower Colorado River R. osculus is not surprising, particularly given the com-
213
plex basin and range faulting that occurred within this region during much of the Tertiary (Nations et al., 1985). This upheaval contributed in large part to the long history of isolation this region has experienced. The Lower Colorado also revealed a high number of haplotypes (Oakey, 2001) that, in turn, suggested the former presence of large population sizes. These haplotypes segregated into two lineages, Verde River/San Pedro-Santa Cruz rivers versus upper Gila River. In a study similar to ours, Lower Colorado River populations of a second small cyprinid (Agosia chrysogaster) were also characterized by relatively close affinities, a poor population structure, and a separation of Verde River populations from those in the upper Gila River (Tibbets and Dowling, 1996). A founder-flush scenario, coupled with frequent dispersal may have enabled R. osculus to move freely throughout this subbasin and to maintain high effective population sizes over time, thus retarding phylogenetic resolution. Upper elevation Colorado River drainages.—Streams on the Colorado Plateau may only interconnect during storm events and then in an unpredictable and ephemeral manner. Given this, it was not unusual to find that the LCR drainage was represented by six unique haplotypes from four widely scattered localities. The lack of structure in this clade may again reflect the isolation of populations and the stochastic loss of lineages in those that are reduced in numbers (as per Minckley et al., 1986). Minckley (1973) also suggested the possibility that R. o. osculus and R. o. yarrowi may have intergraded chaotically across the Mogollon Rim (the southern edge of the Colorado Plateau in north-central Arizona) as this region gradually eroded northward. A hint of this is reflected in the sister relationship between the LCR with the Lower Colorado, and particularly by the basal position of Nutrioso Creek (NUC) in the LCR drainage, which lies adjacent to headwater populations in the Lower Basin. The Upper Colorado River clade is sister to the Lower-LCR clade and is composed of the Green, San Juan, and upper Colorado rivers, as well as haplotypes from the Southern Bonneville. Although it contains haplotypes separated by great geographic distances, their close genetic affinities and shallow clade depth suggest relatively young populations. Freshwater fishes from nonglaciated areas of the northern latitudes (Bodaly et al., 1992; Hansen et al., 1999; Wilson and Hebert, 1998) demonstrate on average deeper topologies and greater genetic diversities than those seen above. At least 20 sep-
214
COPEIA, 2004, NO. 2
arate glacial epochs were recorded during the Pleistocene (Martinson et al., 1987; Dawson, 1992), and higher elevation populations of R. osculus in the upper basin may in fact represent recent recolonization from lower basin refugia (W. L. Minckley, pers. comm.). That is, upper elevation fishes may have been driven to lower elevations at onset of colder glacial periods, only to recolonize the higher elevation sites again during warmer interglacials. Low haplotype diversity may result from serial bottlenecks (Dowling et al., 1996) as Speckled Dace progressively recolonized further upstream following glacial retreat. This hypothesis is supported by the relatively low number of haplotypes found in the Upper Colorado River clade and its sister relationship with the lower-LCR Basin. Southern Bonneville localities were buried within the Upper Colorado River clade. In general, Bonneville species share closest relatives with those found in the Upper Basin (Miller, 1958). This was also suggested by the fact that Gila cypha (a Colorado River endemic) was more closely related to Sevier River (i.e., Bonneville Basin) Gila atraria than to its Colorado River sister species, Gila elegans or Gila robusta (Dowling and DeMarais, 1993). Montane species are the link between the Bonneville and Upper Colorado Basins, and these in turn suggest interbasin transfers between the two, largely from headwater stream captures and drainage reversals over low divides (Hubbs and Miller, 1948; Miller, 1958; Smith, 1978). Such a scenario could explain why upper Virgin River haplotypes were found in the Upper but not the Middle Colorado River clade. The Virgin River lies along the boundary between the Great Basin and the Colorado Plateau (Minckley et al., 1986). Its lower portion drains basin and range topography, whereas the upper drains the Plateau. The Upper Colorado River Basin formerly drained portions of the upper Virgin, but this flow was reversed by the continued uplift in this region, effectively allowing these headwater reaches to be captured by the lower-in-elevation Virgin River. Lahontan (or western Great Basin) clade.—Haplotypes from northern Great Basin (i.e., Northern Bonneville) and those basins further west are all basal to the Lahontan clade. These basins were once allied to the premodern Snake River drainage, prior to its union with the Columbia River in late Pliocene (Malde, 1991; Smith et al., 2000). This early Snake River was a western outlet for Lake Idaho (now southern Idaho), a system of heterogeneous lacustrine habitats that formed Miocene and Pliocene. The faunas dur-
ing these two periods were spatially homogeneous, with each a major focus of ichthyofaunal diversity (Smith, 1975, 1987; Middleton et al. 1985). Lake Idaho has been the subject of numerous studies, at least three of which have considerable bearing on the present investigation. Taylor (1966, 1985), Miller and Smith (1967), and Smith (1975) found that Pliocene bivalves and fishes from Lake Idaho had closest relatives in the Sacramento-San Joaquin and Klamath basins to the west. These disjunctions resulted from several hypothesized drainages. One such (i.e., the ‘‘fishhook:’’ Taylor, 1966) is believed to have run westward from southeastern Idaho, through southeastern Oregon and western Great Basin, then southward along the eastern Sierra Nevada to the Death Valley system. Numerous fossil bivalves and fishes (Taylor, 1966; 1985:289, fig. 18; Taylor and Smith, 1981; Miller and Smith, 1981) documented the existence of this drainage. However, the timing of such an early Snake River connection is rather uncertain in part because of vast temporal and spatial intervals coupled with imprecise dating of fossils (Smith et al., 2000). The sister-relationship of the Columbia 1 Upper Snake to the Lahontan and other western basins suggests that northern R. osculus has been strongly influenced by earlier basin alignments and by several connections to the early Snake River. The interior of the Lahontan clade consists of fluvial localities in the Humboldt River drainage, as well as high-desert isolates scattered widely in the former basin. This region was isolated as the Sierra Nevada Mountains uplifted during late Plio-Pleistocene. It is characterized not only by high levels of endemism but also by extensive variation within and among its widespread forms (Hubbs and Miller, 1948; Hubbs et al., 1974). The high number of mtDNA haplotypes within the interior of the clade suggested the former presence of a metapopulation, with influences extending as far west as the closed basins of eastern California. As demonstration, 45 unique haplotypes were recovered from 60 individuals (e.g., Reese River); none was shared with the Bonneville, Columbia, or Colorado Basins (Oakey, 2001). The interior of the Lahontan clade was, however, poorly resolved and exhibited a confusing geographic structure. This is likely attributable to a long history of intermittently connected habitats that displayed numerous secondary contacts, coupled with incomplete lineage sorting of ancestral variation. A similar stochastic structure was also evident in Cyprinodon from Death Valley (Duvernell and Turner, 1998). The Lahontan clade consisted of a well-sup-
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST ported interior, plus an apical-to-basal progression that runs westward to eastern California, and southwest to the Owens/Amargosa system. The eastern California and Owens River localities, on the eastern front of the Sierra Nevada Mountains, represented the southern extension of the fishhook distribution (discussed above). Our eastern California locations consisted of Lake Almanor, Eagle, and Honey Lakes and sites in the upper Feather River that lie on a plateau between the Sierra Nevada and its eastern escarpment. Honey and Eagle Lake Basins are now closed systems, and several impassable canyons resulting from the developing escarpment on the western slope of the Sierra Nevada now protect the upper Feather River (Moyle, 1976). We speculate that these older Snake River haplotypes have persisted by simply occupying protected, hanging tributaries and, thus, represent, in part, the original Lahontan form of R. osculus. Owens River clade.—A strongly supported (100%) Owens River clade is basal to the Lahontan and represents two mainstream localities in addition to Amargosa River and Whitmore Hot Springs (WHS). The latter is isolated behind a 0.7-million-year-old caldera (Hill et al., 1985) and may retain R. osculus haplotypes from an earlier period. Our results corroborated Miller (1946b) who suggested that Owens River R. osculus spilled across Mono Basin and into Owens Valley, but physical evidence for this event was obscured by volcanic ash (Hubbs and Miller, 1948). Again, the lack of a close relationship between Owens River and the Los Angeles Basin corroborates the earlier morphological studies of Cornelius (1969). The close affinity between Amargosa and Owens Rivers is likely caused by their occasional fusion as Lake Manley enlarged during years of extreme precipitation (Miller, 1946b). However, Miller (1946b, 1984) also argued that R. osculus in the Amargosa River were closely allied to those in the Colorado River, as the former was briefly connected to the latter by Pluvial Las Vegas Wash, a flood-tributary. Similarly, Howard (1996, 2000) suggested that both the Amargosa and Gila paleorivers drained the Los Angeles Basin and were later joined by the Colorado River after it diverted southward in Late Miocene. However, our data fail to link the Amargosa and Colorado Rivers. The Amargosa Basin is instead a geological as well as biological composite with relationships both to the north and south (Hershler et al., 1999). Our Amargosa River samples, positioned intermediate between the Colorado
215
and Owens Rivers, may reflect but a single aspect of this complex hydrology. Columbia River Basin.—The Columbia River clade, although highly supported by our data, was not a monophyletic assemblage. That is, individuals from some localities showed greater affinities to basins outside the present watershed. For example, the Columbia River clade included samples from the middle and upper Columbia and lower Snake Rivers but not the upper Snake River. The latter sample was instead sister to the Bear River of the Northern Bonneville, and these were sister to other northern Bonneville localities. The basal position of Columbia 1 Klamath/Pit, and the close affinity of Clearwater River (CLR) with Southern Bonneville (SEV4, BSC2), suggested that the developing premodern Snake River exerted an early influence upon the Columbia River. The current separation of the upper Snake and Columbia faunas is attributable to the long intervals of isolation, as well as to the extinction of Columbia River forms caused by vulcanism on the Snake River Plain (McPhail and Lindsey, 1986; Malde, 1991). A close relationship between the Upper Snake and the Northern Bonneville is not surprising, given a scenario of repeated exchanges between the two, including a spectacular overtopping of Lake Bonneville in late Pleistocene (Gilbert, 1890). The hypothesis that upper Snake River populations resulted from Bonneville Basin immigrants (Miller and Miller, 1948) was supported in our study by the presence of a common haplotype in both upper Snake and Bear Rivers (Oakey, 2001). However, the Upper Snake reflects low haplotype diversity, which is likely a result of local bottlenecks, extinctions, and rapid loss of mtDNA lineages, perhaps associated with regional vulcanism or glacial periods (Smith, 1966; Malde, 1991). One lower Columbia River locality (i.e., Deschutes River; DSC) did not cluster with upper basin haplotypes but was instead sister to the Northern Bonneville. This can, in part, be explained by the complex history of the premodern Snake River and the Oregon Lakes region. During Pliocene, a suggested outlet for the premodern Snake River was through Harney and Malheur Lakes, as the Snake passed from southern Idaho to the Pacific Ocean (Smith, 1975; Taylor, 1985:fig. 5). A subsequent connection between the Deschutes River and the Oregon Lakes district was of brief duration and was outlined by Behnke (1979) and Taylor (1985). Apparently, the upper Deschutes River flowed into the Oregon Lakes when the western flow of the Snake River was reversed as a result of uplift.
216
COPEIA, 2004, NO. 2
Klamath-Pit clade.—The Klamath 1 Pit clade is composed not only of samples from widely scattered localities in the Klamath and Pit Rivers but also those from south-coastal (i.e., San Benito River) and eastern California (i.e., Eagle Lake). Two general scenarios may explain the close relationship between the Pit River and the south-coastal San Bonito River (Pajaro Basin). Rhinichthys osculus may have entered the Pajaro Basin by way of a headwater transfer with the San Joaquin River (Snyder, 1905; Murphy, 1941). Prior to 1.5 mya, the Sacramento River also flowed through the San Francisco Trough to Monterey Bay, whereas the San Benito River flowed north into San Francisco Bay (Taylor, 1985:312, fig. 38). The general extension of the Sacramento River ichthyofauna into southcoastal drainages was clearly demonstrated by these patterns (Snyder, 1905), as represented an extension of the R. osculus form in the Snake River ( Jordan and Evermann, 1896; Cornelius, 1969). The Pit River is centrally positioned in this region and was the center of intense orogeny and volcanism during Pliocene and Pleistocene. The Klamath-Cascade region acted as a single, coherent block when its western end was displaced 340 km to the south about 20 mya (Magill and Cox, 1981). Thus, a close relationship between Klamath and Pit Rivers may in part be the result of these drainage realignments (Minckley et al., 1986). Fossils from the early Pliocene connected both Pit and Klamath Basins with the premodern Snake River as the latter drained to the Pacific Ocean (Miller and Smith, 1967; Smith, 1975; Taylor, 1985). The position of this clade in Figure 2 points to its ancient connections, but the timing is clearly uncertain, in part because of a lack of physical evidence coupled with the uncertainty in aging fossil materials (Smith et al., 2000). Bonneville Basin and the origin of Rhinichthys osculus.—The numerous examples of differentiated fauna in the Bonneville Basin were recognized by Cope and Yarrow (1875) as stemming from long intervals of piecemeal isolation. Their conclusions are strongly supported by the patterns we uncovered in R. osculus. For example, a majority of haplotypes from the Northern Bonneville clustered with the Deschutes River (DSC), whereas one (i.e., Box Elder County, UT; BOX1) was consistently sister to the Colorado River. Southern Bonneville haplotypes were also sister to the Upper Colorado, whereas two highly divergent haplotypes were sister to the Lower Snake (Clearwater River; CLR). Taylor (1983; 1985:296, fig. 25) suggested a Late-
Miocene drainage connection between southeastern Idaho and the lower Colorado River Basin, a route supported by living and fossil molluscs in western Bonneville Basin. Hubbs and Miller (1948) identified this drainage as a structural trough leading to Pluvial White and Carpenter Lakes. This north-south connection between the Bonneville Basin and the Colorado River is also reflected in the distribution of Gila (now Snyderichthys) copei ( Johnson and Jordan, 2000). Haplotypes of this species are separated into northern (e.g., Bear and upper Snake Rivers) and southern (e.g., Utah Lake and Sevier River) clades, with the northern clade more genetically similar to the outgroup taxon (Lepidomeda mollispinis mollispinis) from Virgin River. The fragmented history of the Bonneville Basin is clearly evident, and these studies provide compelling evidence of its role as a north-south conduit between southern Idaho and the Colorado River. Taylor’s (1985) western Bonneville route crossed the drainage of Big Spring Creek, at the upper end of Snake Valley (Tooele County, UT), where Smith (1978) noted a distinct, undescribed dace that shared many ‘‘spring isolate’’ characters with the extinct R. deaconi (Miller, 1984:table 2). Our two divergent southern Bonneville haplotypes (i.e., Big Springs Creek [BSC2] and Sevier River [SEV4]) differed from conspecifics by greater than eight restriction sites. Finding two highly divergent haplotypes at the same locality represents an uncommon Type-II phylogeographic situation (Avise et al., 1987), generally attributed to secondary contact among allopatric populations. Our study suggests these divergent, southern Bonneville haplotypes may represent a widespread and undescribed form related to R. osculus but with an earlier connection to the north. This is supported by the position of these Southern Bonneville haplotypes in our trees, and by the basal location of Northern Bonneville, Los Angeles and Middle Colorado Basins within their respective clades. The close affinity of Colorado River R. osculus with those in the Los Angeles Basin, but not with Death Valley, suggests two different invasions in that region. The Northern Bonneville-to-Colorado-to-Los Angeles connection was likely earlier than the Lahontan-toOwens connection. It is tempting to infer from these data the origin of R. osculus in Western North America. Patterns of haplotype distribution suggest that the premodern Snake River and Lake Idaho had major roles in the distribution and subsequent evolution of R. osculus in surrounding basins. The basal position of those basins allied to
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST the early Snake River (e.g., Upper Snake, Northern Bonneville, Klamath-Pit, and Columbia) could represent the earliest appearance of modern R. osculus in the west. Bonneville haplotypes join the tree at separate yet earlier positions, and may represent an early R. osculus form in the western paleodrainages of the Mohave and Sonoran desert provinces (sensu Minckley et al., 1986). The long isolation of the Los Angeles Basin, and its sister relationships with the Northern Bonneville and Colorado clades suggest the possibility of an earlier R. osculus-like form in this region. The presence of undescribed Rhinichthys (Peden and Hughes, 1988) in the Columbia River Basin also indicates that evolution of the group may have been northerly, possibly associated with retreating ice (McPhail and Lindsey, 1986; Bodaly et al., 1992). However, the early Bonneville-to-Colorado distribution, coupled with the accompanying fishhook drainage, suggest that R. osculus may, in fact, have originated much earlier, possibly associated with Tertiary Lake Idaho. ACKNOWLEDGMENTS DDO received enormous assistance collecting fish samples from around the West (a full list of collectors is provided in Oakey, 2001:appendix 5). However, two friends stand foremost in this endeavor: J. Dunham and M. Andersen. Collecting permits were provided by the states of AZ, CA, ID, MT, NV, and UT, whereas the Arizona State University IACUC (Institutional Animal Care and Use Committee) approved collecting protocols. Numerous friends assisted between campus and field: J. Bann, B. Bartram, R. Broughton, P. Brunner, J. Chesser, B. DeMarais, E. Goldstein, A. Dauberman, D. McElroy, G. Naylor, S. Norris, R. Olson, C. Secor, A. Tibbets, R. Timmons, B. Trapido-Lurie, P. Unmack, T. Velasco, and M. Wurzburger. This manuscript represents part of a dissertation submitted by DDO in partial requirement for the Ph.D. degree at Arizona State University. Support from a National Science Foundation Dissertation Improvement Grant is greatly acknowledged. This manuscript is dedicated to the memory of Professor W. L. Minckley of Arizona State University, who for 35 years studied the origin and evolution of native fishes in western North America. He (and colleagues) struggled mightily to conserve this fauna against rampant water diversion and the unbridled urban sprawl that now characterize the ‘‘New West.’’ He was a motivating force for this study, and his death on 22 June 2001 left an inestimable void in both our knowledge of indigenous, western North Amer-
217
ican fishes and the landscape within which they evolved. LITERATURE CITED ALLARD, M. W., AND J. M. CARPENTER. 1996. On weighting and congruence. Cladistics 12:183–198. ALLENDORF, F. W., AND R. F. LEARY. 1988. Conservation and distribution of genetic variation in a polytypic species, the Cutthroat Trout. Conserv. Biol. 2:170– 183. ANDERSON, F. E. 2000. Phylogenetic relationships among loliginid squids (Cephalopoda: Myopsidae) based upon analysis of multiple data sets. Zool. J. Linn. Soc. 130:603–633. ARCHIE, J. W. 1996. Measures of homoplasy, p. 153– 188. In: Homoplasy: the recurrence of similarity in evolution. M. J. Sanderson and L. Hufford (eds.). Academic Press, New York. AVISE, J. C., ARNOLD, J., BALL, R. M., BERMINGHAM, E., LAMB, T., NEIGEL, J. E., REEB, C. A., AND N. C. SAUNDERS. 1987. Intraspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annu. Rev. Ecol. Syst. 18: 489–522. BEHNKE, R. J. 1979. Monograph of the native trouts of the genus Salmo of western North America. U.S. Fish and Wildlife Service, Denver, CO. ———. 1992. Native trout of western North America. Am. Fish. Soc., Spec. Monogr. 6:1–275. BISHOP, P. 1995. Drainage rearrangement by river capture, beheading and diversion. Prog. Phys. Geogr. 19:449–473. BODALY, R. A., J. W. CLAYTON, C. C. LINDSEY, AND J. VUORINEN. 1992. Evolution of Lake Whitefish (Coregonus clupeaformis) in North America during the Pleistocene: genetic differentiation between sympatric populations. Can. J. Fish. Aquat. Sci. 49:769– 779. COBURN, M. M., AND T. M. CAVENDER. 1992. Interrelationships of North American cyprinid fishes, p. 328–373. In: Systematics, historical ecology, and North American freshwater fishes. R. L. Mayden (ed.). Stanford Univ. Press, Stanford, CA. COPE, E. D., AND H. C. YARROW. 1875. Report upon the collections of fishes made in portions of Nevada, Utah, California, Colorado, New Mexico, and Arizona, during 1871, 1872, 1873, and 1874, p. 635–703. In: United States Army engineer department report on the geography and geology of the explorations and surveys west of the 100th Meridian, in Charge of George M. Wheeler. Vol. 5 (Zoology). Government Printing Office, Washington, DC. CORNELIUS, R. H. 1969. The systematics and zoogeography of Rhinichthys osculus (Girard) in southern California. Unpubl. master’s thesis, California State Univ., Fullerton. DAWSON, A. G. 1992. Ice Age Earth. Routledge Press, London. DELLENBAUGH, R. H. 1873. A canyon voyage. Academic Press, New York. DEMARAIS, B. D., T. E. DOWLING, M. E. DOUGLAS, W.
218
COPEIA, 2004, NO. 2
L. MINCKLEY, AND P. C. MARSH. 1992. Origin of Gila seminuda (Teleostei: Cyprinidae) through introgressive hybridization: implications for evolution and conservation. Proc. Natl. Acad. Sci. USA 89:2747– 2751. DOUGLAS, M. E. 1993. An analysis of sexual dimorphism in an endangered cyprinid fish (Gila cypha Miller) using video image technology. Copeia 1993: 334–343. ———, W. L. MINCKLEY, AND B. D. DEMARAIS. 1999. Did vicariance mold phenotypes of western North American fishes? Evidence from Gila River cyprinids. Evolution 53:238–246. DOWLING, T. E., AND B. D. DEMARAIS. 1993. Evolutionary significance of introgressive hybridization in cyprinid fishes. Nature 362:444–446. ———, C. MORITZ, J. D. PALMER, AND L. H. RIESBERG. 1996. Nucleic acids III: analysis of fragments and restriction sites, p. 249–320. In: Molecular systematics. 2d ed. D. M. Hillis, C. Moritz, and B. K. Mable (eds.). Sinauer Associates, New York. DUVERNELL, D. D., AND B. J. TURNER. 1998. Evolutionary genetics of Death Valley pupfish populations: mitochondrial DNA sequence variation and population structure. Mol. Ecol. 7:279–288. FARRIS, J. S. 1969. A successive approximations approach to character weighting. Syst. Zool. 18:374– 385. FU, J. 2000. Toward the phylogeny of the family Lacertidae—why 4,708 base pairs of mtDNA sequences cannot draw the picture. Biol. J. Linn. Soc. 71:203– 217. GILBERT, G. K. 1890. Lake Bonneville. U.S. Geol. Surv. Monogr. 1:1–438. GOLOBOFF, P. A. 1995. Parsimony and weighting: a reply to Turner and Zandee. Cladistics 11:91–104. ———. 1999. Analyzing large data sets in reasonable times: solutions for composite optima. Ibid. 15:415– 428. HANSEN, M. M., K.-L. D. MENSBERG, AND S. BERG. 1999. Postglacial recolonization patterns and genetic relationships among whitefish (Coregonus sp.) populations in Denmark, inferred from mitochondrial DNA and microsatellite characters. Mol. Ecol. 8: 239–252. HERSHLER, R., L. HSIU-PING, AND M. MULVEY. 1999. Phylogenetic relationships within the aquatic snail genus Tyronia: Implications for biogeography of the North American southwest. Mol. Phylogenet. Evol. 13:377–391. HILL, D. P., R. A. BAILEY, AND A. S. RYALL. 1985. Active tectonic and magmatic processes beneath Long Valley caldera, eastern California: an overview. J. Geophys. Res. 90(B13):11111–11120. HOWARD, J. L. 1996. Paleocene to Holocene paleodeltas of ancestral Colorado River offset by the San Andreas fault system, southern California. Geology 24:783–786. ———. 2000. Provenance of quartzite clasts in the Eocene-Oligocene sespe formation: paleogeographic implications for southern California and the ancestral Colorado River. Bull. Geol. Sci. Am. 112:1635– 1649. HUBBS, C. L., AND R. R. MILLER. 1948. The zoological
evidence: correlation between fish distributions and hydrographic history in the desert basins of western United States, p. 17–144. In: The Great Basin, with emphasis on glacial and post-glacial times. Bulletin of the Univ. of Utah 38, Biol. Ser. 10, Salt Lake City. ———, ———, AND L. C. HUBBS. 1974. Hydrographic history and relict fishes of the north-central Great Basin. Calif. Acad. Sci. Mem. 7:1–259. HUGHES, G. W., AND A. E. PEDEN. 1989. Status of the Umatilla Dace, Rhinichthys umatilla, in Canada. Can. Field Nat. 103:193–200. HUNT, C. B. 1969. Geologic history of the Colorado River, p. 59–130. In: The Colorado River region and John Wesley Powell. U.S. Geological Survey Professional Papers 669. Government Printing Office, Washington, DC. JOHNSON, J. B., AND S. JORDAN. 2000. Phylogenetic divergence in Leatherside Chub (Gila copei) inferred from mitochondrial cytochrome b sequences. Mol. Ecol. 9:1029–1035. JORDAN, D. S. 1891. Report of explorations in Colorado and Utah during the summer of 1889, with an account of the fishes found in each of the river basins examined. Bull. U.S. Fish Comm. 9:1–40. ———, AND B. W. EVERMANN. 1896–1900. The fishes of North and Middle America. Bull. U.S. Nat. Mus. 47(4 parts):i–ix,1–3313. ———, ———, AND H. W. CLARK. 1930. Checklist of the fishes and fishlike vertebrates of North and Middle America north of the northern boundary of Venezuela and Columbia. Rept. U.S. Comm. Fish. 1928:1–670. LA RIVERS, I. 1962. Fishes and fisheries of Nevada. Nevada Fish and Game Commission, Reno. MAGILL, J., AND A. COX. 1981. Post-Oligocene tectonic rotation of the Oregon Western Cascade Range and the Klamath Mountains. Geology 9:127–131. MALDE, H. E. 1991. Quaternary geology and structural history of the Snake River plain, Idaho and Oregon, p. 251–281. In: Quaternary nonglacial geology: conterminous U.S. R. B. Morrison (ed.). Geological Society of America, Boulder, CO. MARTINSON, D. G., N. G. PISIAS, J. D. HAYS, J. IMBRIE, T. C. MOORE JR., AND N. J. SHACKLETON. 1987. Age, dating, and orbital theory of the Ice Ages: development of a high-resolution 0–300,000 year chronostratigraphy. Quart. Res. 27:1–29. MCCUSKER, M. R., E. PARKINSON, AND E. R. LINDSEY. 2000. Mitochondrial DNA variation in Rainbow Trout (Oncorhynchus mykiss) across its native range: testing biogeographical hypotheses and their relevance to conservation. Mol. Ecol. 9:2089–2108. MCPHAIL, J. D., AND C. C. LINDSEY. 1986. Zoogeography of the freshwater fishes of Cascadia (the Columbia system and rivers north to the Sikine), p. 615–638. In: The zoogeography of North American freshwater fishes. C. H. Hocutt and E. O. Wiley (eds.). John Wiley and Sons, New York. MIDDLETON, L. T., M. L. PORTER, AND P. G. KIMMEL. 1985. Depositional settings of the Chalk Hills and Glenns Ferry formations west of Bruneau, Idaho, p. 37–53. In: Cenozoic paleogeography of west-central United States. R. M. Flores and S. S. Kaplan
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST (eds.). Rocky Mountain Section—Soc. Econ. Paleont. Mineral., Denver, CO. MILLER, R. R. 1946a. Gila cypha, a remarkable new species of fish from the Colorado River in Grand Canyon, Arizona. J. Wash. Acad. Sci. 36:403–415. ———. 1946b. Correlation between fish distribution and Pleistocene hydrography in eastern California and southwestern Nevada, with a map of the Pleistocene waters. J. Geol. 54:43–53. ———. 1958. Origin and affinities of the freshwater fish fauna of western North America, p. 187–222. In: Zoogeography. C. L. Hubbs (ed.). American Association for the Advancement of Science, Publ. 51, Washington, DC. ———. 1984. Rhinichthys deaconi, a new species of dace (Pisces: Cyprinidae) from southern Nevada. Occ. Pap. Mus. Zool. Univ. Mich. 707:1–21. ———, AND C. L. HUBBS. 1960. The spiny-rayed cyprinid fishes (Plagopterini) of the Colorado River System. Misc. Publ. Mus. Zool. Univ. Mich. 115:1– 39. ———, AND R. G. MILLER. 1948. The contribution of the Columbia River system to the fish fauna of Nevada: five species unrecorded from the state. Copeia 1948:174–187. ———, AND G. R. SMITH. 1967. New fossil fishes from Plio-Pleistocene Lake Idaho. Occ. Pap. Mus. Zool. Univ. Mich. 654:1–24. ———, AND ———. 1981. Distribution and evolution of Chasmistes (Pisces: Catostomidae) in western North America. Ibid. 696:1–46. MINCKLEY, W. L. 1973. Fishes of Arizona. Arizona Game and Fish Department, Phoenix. ———, AND B. D. DEMARAIS. 2000. Taxonomy of chubs (Teleostei, Cyprinidae, genus Gila) in the American southwest with comments on conservation. Copeia 2000:251–256. ———, AND M. E. DOUGLAS. 1991. Discovery and extinction of western fishes: a blink of the eye in geologic time, p. 7–17. In: Battle against extinction: native fish management in the American west. W. L. Minckley and J. E. Deacon (eds.). Univ. of Arizona Press, Tucson. ———, D. A. HENDRICKSON, AND C. E. BOND. 1986. Geography of western North American freshwater fishes: description and relationships to intracontinental tectonism, p. 519–614. In: The zoogeography of North American freshwater fishes. C. H. Hocutt and E. O. Wiley (eds.). John Wiley and Sons, New York. MORITZ, C., T. E. DOWLING, AND W. M. BROWN. 1987. Evolution of animal mitochondrial DNA: relevance for population biology and systematics. Annu Rev. Syst. Ecol. 18:269–292. MOYLE, P. B. 1976. Inland fishes of California. Univ. of Calif. Press, Berkeley. MURPHY, G. 1941. A key to the fishes of the Sacramento–San Joaquin Basin. Calif. Fish Game 1941: 165–171. NATIONS, D., J. C. WILT, AND R. H. HELVY. 1985. Cenozoic paleogeography of Arizona, p. 335–355. In: Cenozoic paleogeography of west-central United States. R. M. Flores and S. S. Kaplan (eds.). Rocky
219
Mountain Section—Soc. Econ. Paleont. Mineral., Denver, CO. NIXON, K. C. 1999. The parsimony ratchet, a new method for rapid parsimony analysis. Cladistics 15: 407–414. ———, AND J. M. CARPENTER. 1996. On consensus, collapsibility, and clade concordance. Ibid. 12:305– 312. OAKEY, D. D. 2001. Genetic and geographic variation in Rhinichthys osculus (Teleostei: Cyprinidae) from western North America. Unpubl. Ph.D. diss., Arizona State Univ., Tempe. PEDEN, A. E., AND G. W. HUGHES. 1981. Life-history notes relevant to the Canadian status of the Speckled Dace (Rhinichthys osculus). Syesis 14:21–31. ———, AND ———. 1988. Sympatry in four species of Rhinichthys (Pisces), including the first documented occurrences of R. umatilla in the Canadian drainages of the Columbia River. Can. J. Zool. 66: 1846–1856. PLATNICK, N. I. 2000. A re-limitation and revision of the Australasian ground spider family Lamponidae (Araneae: Gnaphosoidea). Bull. Am. Mus. Nat. Hist. 245:1–328. SCHNEIDER, C. J., M. CUNNINGHAM, AND C. MORITZ. 1998. Comparative phylogeography and the history of endemic vertebrates in the wet tropic rainforests of Australia. Mol. Ecol. 7:487–498. SMITH, G. R. 1966. Distribution and evolution of the catostomid fishes of the subgenus Pantosteus, genus Catostomus. Misc. Pub. Mus. Zool. Univ. Mich. 129: 1–132. ———. 1975. Fishes of the Pliocene Glenns Ferry Formation, Southwest Idaho. Univ. Mich. Mus. Paleont. Pap. Paleont. 14:1–68. ———. 1978. Biogeography of intermountain fishes, p. 17–42. In: Intermountain biogeography, a symposium. K. T. Harper and J. L. Reveal (eds.). Great Basin Nat. Mem. 2, Bricham Young Univ., Provo, UT. ———. 1981. Late Cenozoic freshwater fishes of North America. Annu. Rev. Syst. Ecol. 12:163–193. ———. 1987. Fish speciation in a western North American Pliocene rift lake. Palaios 2:436–445. ———, MORGAN, N., AND E. GUSTAFSON. 2000. Fishes of the Mio-Pliocene Ringold Formation, Washington: Pliocene capture of the Snake River by the Columbia River. Univ. Mich. Mus. Paleont. Pap. 32:1– 47. ———, T. E. DOWLING, K. W. GOBALET, T. LUGASKI, D. K. SHIOZAWA, AND R. P. EVANS. 2002. Biogeography and timing of evolutionary events among Great Basin fishes, p. 175–234. In: Great Basin aquatic systems history. R. Hershler, D. B. Madsen and D. R. Currey (eds). Smiths. Contrib. Earth Sci., No. 33. Smithsonian Institution Press, Washington, DC. SNYDER, J. O. 1905. Notes on the fishes of the streams flowing into San Francisco Bay, California. U.S. Fish Comm. Rept. 1904:327–338. STEARLEY, R. F., AND G. R. SMITH. 1993. Phylogeny of Pacific trouts and salmon (Oncorhynchus) and genera of family Salmonidae. Trans. Am. Fish. Soc. 122: 1–33.
220
COPEIA, 2004, NO. 2
STEGNER, W. 1954. Beyond the hundreth meridian. Explorations of the intermontane west by J. M. Powell. Penguin Books, New York. SWOFFORD, D. L. OLSEN, G. J., WADDELL, P. J., AND D. M. HILLIS. 1996. Phylogenetic inference, p. 407– 514. In: Molecular systematics. 2d ed. D. M. Hillis, C. Moritz, and B. K. Mable (eds.). Sinauer Associates, New York. TAYLOR, D. W. 1966. Summary of North American Blancan nonmarine mollusks. Malacologia 4:1–172. ———. 1983. Late Tertiary mollusks from the lower Colorado River Valley. Univ. Mich. Mus. Paleont. Contrib. 26:289–298. ———. 1985. Evolution of freshwater drainages and molluscs in western North America, p. 265–321. In: Late Cenozoic history of the Pacific Northwest. C. J. Hocutt and A. B. Leviton (eds.). American Association for the Advancement of Science Symposium, San Francisco CA. ———, AND G. R. SMITH. 1981. Pliocene molluscs and fishes from northeastern California and northwestern Nevada. Univ. Mich. Mus. Paleont. Contrib. 25: 339–413. TIBBETS, C. A., AND T. E. DOWLING. 1996. Effects of intrinsic and extrinsic factors on population fragmentation in three species of North American minnows (Teleostei: Cyprinidae). Evolution 50:1280– 1292. TURNER, H., AND R. ZANDEE. 1995. The behavior of Goloboff’s tree fitness measure, F. Cladistics 11:57– 72.
UYENO, T., AND R. R. MILLER. 1963. Summary of late Cenozoic freshwater fish records for North America. Occ. Pap. Mus. Zool. Univ. Mich. 631:1–34. WILLIAMS, J. E. 1978. Taxonomic status of Rhinichthys osculus (Cyprinidae) in the Moapa River, Nevada. Southwest. Nat. 23:511–518. WILLS, M. A., D. E. G. BRIGGS, R. A. FORTNEY, M. WILKINSON, AND P. H. A. SNEATH. 1998. An arthropod phylogeny based on fossil and recent taxa, p. 33– 105. In: Arthropod fossils and phylogeny. G. D. Edgecomb (ed.). Columbia Univ. Press, New York. WILSON, C. C., AND P. D. N. HEBERT. 1998. Phylogeography and postglacial dispersal of lake trout (Salvelinus namaykush) in North America. Can. J. Aquat. Sci. 55:1010–1024. WOODMAN, D. A. 1992. Systematic relationships within the cyprinid genus Rhinichthys, p. 374–391. In: Systematics, historical ecology, and North American freshwater fishes. R. L. Mayden (ed.). Stanford Univ. Press, Stanford, CA.
(DDO) DEPARTMENT OF BIOLOGY AND MUSEUM, ARIZONA STATE UNIVERSITY, TEMPE, ARIZONA 85287-1501; AND (MED, MRD) DEPARTMENT OF FISHERY AND WILDLIFE BIOLOGY, COLORADO STATE UNIVERSITY, FT. COLLINS, COLORADO 80523-1474. PRESENT ADDRESS: (DDO) 323 SOUTH HOBSON, MESA, ARIZONA 85204. Email: (MED)
[email protected]. Submitted: 26 Nov. 2002. Accepted: 22 Jan. 2004. Section editor: R. M. Wood.
OAKEY ET AL.—SPECKLED DACE IN THE AMERICAN WEST APPENDIX 1. COLLECTION ABBREVIATIONS
AND
221
LOCALITIES FOR 61 Rhinichthys osculus POPULATIONS USED STUDY (SEE FIG. 1).
OTU
Locality
County
State
1-AMR 2-ANR 3-APC 4-BCC 5-BEN 6-BLU 7-BOC 8-BOX 9-BRR 10-BSC 11-CBC 12-TUR 13-CHV 14-CLR 15-CON 16-DOL 17-DPB 18-DSC 19-ECC 20-FRE 21-FRN 22-GAN 23-GLN 24-GRJ 25-HAR 26-HIT 27-LCH 28-LAC 29-LIT 30-LVA 31-MAY 32-MOA 33-NFH 34-NUC 35-ORB 36-PAR 37-PIN 38-PIT 39-RES 40-RFC 41-SAN 42-SBR 43-SEV 44-SFK 45-SFR 46-SGR 47-SMO 48-SOC 49-SQQ 50-TCN 51-TCT 52-TET 53-THS 54-VEL 55-WCC 56-WHI 57-WHS 58-WYO 59-YAK 60-YMP 61-YRC
Amargosa R. Animas R. Apache Cr. Black Canyon Cr. Benner Cr. Blue R. Bonita Cr. Rabbit Spr. Bear R. Big Springs Cr. Campbell Blue Cr. Turkey Cr. Chevelon Cr. Clearwater R. Condor Canyon Dove Cr. Diana’s Punch Bowl Deschutes R. East Clear Cr. Frenchman’s Cr. Francis Cr. Gance Cr. South Canyon Cr. Colarado R. Marble Cr. Coulee Cr. Last Chance Cr. LaVerkin Cr. Virgin R. Honey Lk. Pahranagat R. Moapa R. Little Humboldt R. Nutrioso Cr. Owens R. Paria R. Eagle Lk. Pit R. Reese R. Redfield Canyon Santa Ana R. San Benito R. Sevier R. So. Fk. Humboldt R San Francisco R. San Gabriel R. Smoke Cr. Sonoita Cr. Squaw Queen Cr. Tucannon R. Touchet R. Teton R. Thousand Spr. Cr. White R. West Clear Cr. White R. Whitmore Hot Sps. Gros Ventre R. Yakima R. Yampa R. Yreka Cr.
Nye San Juan Yavapai Apache Plumas Greenlee Graham Box Elder Rich White Pine Greenlee Cochise Apache Clearwater Lincoln Dolores Nye Sherman Coconino Plumas La Paz Elko Garfield Mesa Mono Stevens Plumas Washington Washington Lassen Lincoln Clark Humboldt Greenlee Inyo Coconino Lassen Lake Nye Graham San Bernardino San Benito Sanpete Elko Greenlee San Bernardino Washoe Santa Cruz Lassen Columbia Walla Walla Teton Elko Lincoln Yavapai Rio Blanco Mono Sublette Kittitas Moffat Siskiyou
NV NM AZ AZ CA AZ AZ UT UT NV AZ AZ AZ ID NV CO NV OR AZ CA AZ NV CO CO CA WA CA UT UT CA NV NV NV AZ CA AZ CA OR NV AZ CA CA UT NV NM CA NV AZ CA WA WA ID NV NV AZ CO CA WY WA CO CA
IN
THIS
Lat.–Long.
368529N 368449N 348539N 358449N 408209N 338209N 338259N 418249N 418479N 388459N 338459N 318459N 348459N 468089N 378509N 378459N 398029N 458379N 348329N 398539N 348409N 418159N 398339N 398039N 378469N 478449N 408259N 378169N 368539N 398449N 378129N 368409N 418469N 348049N 378209N 368539N 408379N 428179N 408279N 328309N 348109N 368309N 398249N 408389N 338239N 348219N 408359N 318329N 408029N 468309N 468039N 438339N 418299N 378299N 348329N 408009N 378369N 448109N 468579N 408319N 418409N
1668459W 1088139W 1128559W 1098059W 1218139W 1098109W 1098359W 1138529W 1118049W 1148039W 1098079W 1098059W 1108409W 1158479W 1148229W 1088559W 1168409W 1208549W 1118109W 1208169W 1138259W 1158489W 1078259W 1088319W 1188259W 1178439W 1208229W 1138159W 1138559W 1208029W 1158029W 1148409W 1178209W 1098139W 1188309W 1118369W 1208599W 1208239W 1178039W 1108209W 1178109W 1218109W 1128039W 1158449W 1088549W 1178519W 1198589W 1108469W 1208309W 1188029W 1188419W 1118029W 1148149W 1158109W 1118309W 1078389W 1188469W 1108449W 1208459W 1088599W 1228339W
