American Fisheries Society Symposium 17:350-359, 1995 © Copyright by the American Fisheries Society 1995 Desert Aquatic Ecosystems and the Genetic and Morphological Diversity of Death Valley System Speckled Dace DONALD W. SADA Environmental Studies Program, University of Nevada-Las Vegas 2689 Highland Drive, Bishop, California 93514, USA HUGH B. BRITTEN AND PETER F. BRUSSARD Biodiversity Research Center, Department of Biology, University of Nevada Reno, Nevada 89557-0015, USA Abstract.—The morphological and genetic diversities of fishes in North American deserts have been examined to estimate evolutionary rates, to create models of interbasin pluvial connectivity, and to justify protection of aquatic ecosystems throughout the region. Morphological and genetic studies comparing 13 populations of speckled dace Rhinichthys osculus from the Death Valley system, Lahontan basin, and lower Colorado River were conducted to quantify differences among populations. Differences in meristic and mensural characteristics among populations were highly significant, but differences in body shape were slight and best explained as representing two forms, one deep-bodied and short, the other elongate and slender. Starch gel electrophoretic assays of 23 loci showed isolated populations to be genetically unique. Fifty-nine taxa are identified as endemic to wetland and aquatic habitats in the Death Valley system: 16 forms of fish, 1 amphibian, 22 mollusks, 7 aquatic insects, 3 mammals, and 10 forms of flowering plants. Genetic and morphological differentiation of isolated speckled dace populations and the diversity and number of endemic forms associated with wetlands and aquatic habitats in the Death Valley system suggest that each desert wetland community functions as an evolutionarily significant unit. Taxonomic descriptions of relict faunas occupy- Pleistocene pluvial periods (Miller 1946; Hubbs and ing isolated aquatic habitats in the endorheic desert Miller 1948). The ecology, systematics, biogeogra- basins of western North America began with recog- phy, and status of most fish species in the system nition of unique fishes during land surveys before have been actively studied (Miller 1948; Soltz and the West was settled (see Minckley and Douglas Naiman 1978; Echelle and Echelle 1993) with 1991) and attention expanded quickly to the inver- the exception of speckled dace Rhinichthys oscu- tebrate fauna (Brues 1932). This fauna is depauper- lus, which occupies the Owens and Amargosa ate compared to those in more mesic regions, but basins. many studies over the last 50 years have shown that Concern for the declining status of DVS speck- a wide diversity of endemic plants and animals is led dace caused us to conduct morphological, associated with regional wetlands that range in size genetic, and status studies to determine charac- from less than 0.1 ha to several thousand hectares. teristics of population variation and assist in the The small size of these habitats and the vulnerabil- design of conservation programs to protect and ity of isolated populations to habitat degradation enhance extant populations. In this paper we and invasion of nonnative species have been factors summarize results of these genetic and morpho- in the decline of many fish populations (Minckley logical studies and examine the utility of this and Deacon 1968; Williams et al. 1989; Rinne and analysis for identifying evolutionarily significant Minckley 1991). Approximately 60% of the North units. Then we consider other endemic organisms American fish species currently listed as threatened associated with DVS aquatic habitats and wet- or endangered by the U.S. Fish and Wildlife Service lands to determine if considering communities of occupy the desert southwestern region of the these organisms provides stronger justification United States (Williams and Sada 1985; Williams for regional wetland conservation than protection et. al 1989). strategies based on either populations or single The Death Valley system (DVS) is an endorheic species. We believe that a wide diversity of en- basin in southwestern Nevada and eastern Califor- demic species associated with these habitats nia. It comprises the Owens River, Amargosa River, would suggest that conservation programs pro- and Mojave River basins, which were tributary to tecting endemic species could also protect desert Lake Manly (present day Death Valley) during aquatic and wetland ecosystems. 350 DIVERSITY OF DESERT POPULATIONS 351 I. LAHONTAN BASIN 1. Smoke Creek 2 2. Reese River 2 3. Huntington Creek 1,2 1,2 4. East Fork Walker River II. DEATH VALLEY SYSTEM 5. Spring at Little Alkali Lake 2 14 6. Whitmore Hot Springs 7. Hot Creek' 8. Marble Creek 1.2 1,2 9. Pine/Rock Creeks 2 10. A-1 Drain near Bishop Canal south of Bishop 12. Little Lake 13. Amargosa River at Tecopa 1,2 14. Tubbs Spring Ranch' 15. Bradford Spring 1.2 16. Amargosa River at Beatty 1,2 III. LOWER COLORADO RIVER DRAINAGE 17. Beaver Dam Wash 1.2 FIGURE 1.—Delineation of the Lahontan basin (I), Death Valley system (II), and Colorado River drainage (III), and locations (numbered 1-17) of Rhinichthys osculus populations examined in morphological and genetic studies. Super- script 1 on place names denotes museum collections examined in morphological studies; superscript 2 denotes populations sampled for genetic analysis. Methods them from Pine and Rock creeks and the other Morphological and allozyme analyses were con- from A-1 Drain, were combined), and 14 collections ducted on several speckled dace populations from from 13 localities were used for analysis of meristic the Lahontan basin, one population from the lower and mensural characteristics. Colorado River drainage, and all known popula- Morphometrics.—Eighteen traditional truss mea- tions in the DVS (Figure 1). Of the 17 collections surements, fin lengths, and six meristic variables examined, 13 were used for the allozyme analysis (lateral line scales and pores and rays of the dorsal, (two proximate collections near Bishop, one of anal, pelvic, and pectoral fins, all counted by the 352 SADA ET AL. FIGURE 2.—Morphometric measurements taken on speckled dace from the Death Valley system: (1) head width; (2) least bony interorbital width; (3) snout tip to supratemporal canal; (4) snout tip to branchiostegal junction; (5) branchiostegal junction to supratemporal canal; (6) supratemporal canal to pelvic fin origin; (7) supratemporal canal to dorsal fin origin; (8) branchiostegal junction to dorsal fin origin; (9) branchiostegal junction to pelvic fin origin; (10) dorsal fin origin to pelvic fin origin; (11) dorsal fin origin to anal fin origin; (12) pelvic fin origin to base last dorsal ray; (13) dorsal fin base length; (14) pelvic fin origin to anal fin origin; (15) base last dorsal ray to anal fin origin; (16) base last dorsal ray to upper caudal peduncle; (17) base last dorsal ray to lower caudal peduncle; (18) anal fin origin to upper caudal peduncle; (19) anal fin origin to lower caudal peduncle; (20) vertical through junction of caudal vertebrae 5 and 6 anterior of hypural plate; (21) depressed dorsal fin length; (22) depressed anal fin length; (23) pectoral fin length; (24) pelvic fin length; (25) standard length. Measurements 1-20 were used in sheared principle components analysis. methods of Hubbs et al. 1974) were recorded from Isozyme data analysis. —Estimates of polymor- 484 fish (Figure 2). Individuals were segregated by phism and heterozygosity and tests for conformance sex for analysis of mensural characters, and only fish to Hardy-Weinberg expectations were made with longer than 25 mm standard length (SL) were ex- the BIOSYS-1 program (Swofford and Selander amined. Truss measurements and two measures of 1981). A X2 test for heterogeneity was used to test body width were analyzed by principal components the significance of allele frequency differences be- analysis with shearing to reduce the effects of size tween populations, and fixation (F) statistics were (Rohlf and Bookstein 1987). Components were calculated for all sampled populations. Differences sheared by locality and calculated from the covari- among populations were assessed by principal com- ance matrix of data transformed to basee loga- ponents analysis of arcsine-transformed allele fre- rithms. Proportional truss measurements (fractions quencies for polymorphic loci. of SL multiplied by 1,000), fin length measure- Distribution of endemic wetland species.—En- ments, and meristic counts were log-transformed demic wetland species associated with DVS valley (basee) and tested by one-way analysis of variance floor wetlands were identified from the literature (ANOVA) for differences among populations. and by consultation with regional scientists and land Protein electrophoresis.—The variability of iso- managers. zymes coded by 23 gene loci (Table 1) was assayed for 311 speckled dace from 13 localities (Figure 1). Whole fish were stored at -80°C until processed for Results electrophoresis. Approximately 1 g of muscle tissue Morphological Analysis was dissected from the caudal end of each fish. This was minced in an iced spot plate and combined with Meristic and proportional mensural characters 0.75 mL of chilled 0.05 M tris-HC1 extraction buffer were all within ranges documented for speckled (pH 7.1) in a centrifuge tube for further macera- dace (Hubbs et al. 1974). Highly significant differ- tion. Electrophoretic and histochemical staining ences among all populations for all meristic and procedures followed those of May (1992). mensural characters