SPECIATION IN FISHES OF THE GENERA CYPRINODON AND EMPETRICHTHYS, INHABITING THE DEATH VALLEY REGION 1
ROBERT RUSH MILLER Museum of Zoology, University of Michigan
Received February 6, 1950
INTRODUCTION The present account is limited to a dis During the Pleistocene history of the cussion of speciation in the six species desert basins of western North America, represented by the cyprinodontid genera particularly in the Great Basin, block Cyprinodon. and Empetrichthys, which faulting produced dozens of independent were recently treated in detail (Miller, valleys separated by precipitous mountain 1948). ranges. In Pluvial 2 periods many of these valleys held lakes and large streams, DIFFERENTIATION but the desiccation of postglacial times The genus Cyprinodon, which com disrupted the drainages between, as well prises small fishes that tolerate a wide as within, the once continuous basins. variety of physical and chemical condi Isolated populations of animals and plants tions, is represented by four distinct spe resulted, and the Great Basin has become cies: C. radiosus in Owens Valley, along a land of relicts. the eastern base of the Sierra Nevada The Death Valley region, lying in the Range; C. salinus in Salt Creek, on the southwestern extension of the Great floor of Death Valley; C. nevadensis in Basin, provides an excellent example of the Amargosa River drainage system, such disconnected drainages and faunas. from the south end of Death Valley to During late Pleistocene time, an integrated just across the California-Nevada border; river system spread over this desert re and C. diabolis of Devil's Hole, an iso gion to unite the four now isolated basins lated spring in Ash Meadows, Nevada of what may be called the Death Valley (Map 2). Each of these species is sharply system (Map 1). differentated, and their characters run Although only a limited fish fauna in nearly the entire gamut of known varia habits this area, it is rich in material for tion in Cyprinodon, a genus found also in the student of evolution and is highly eastern and southern United States and endemic. The living fishes comprise 3 southward to northern South America. families, 5 genera, 10 species and about All four are tied together, however, by a 24 kinds including subspecies. One basic pattern of scale structure that dis genus, Emoetrichthys, is endemic, and tinguishes the species of the Death Valley nine species are confined to the system. system from those elsewhere (Miller, High endemism on the species level sug 1948, PI. 2). This harmony in funda gests a Wisconsin or pre-Wisconsin ori mental scale pattern is a striking con gin of the fauna. firmation of the interrelationship of these 1 This paper is summarized from the lengthy species, and is convincing zoological evi original (Miller, 1948) with modifications and additions. dence to support the geological data that 2 The term Pluvial refers to the two or more the waters of the Death Valley system stages, commonly regarded as of late Pleisto were once continuous (Map 1). Differ cene (Wisconsin) age, when desert basins in entiation in C. nevadensis, of the Amar western North America held lakes that have left definite records of their existence in the gosa River drainage, has been sufficient form of shore features. for the naming of 6 subspecies and the EVOLUTION 4: 155--163. June, 1950. 155 156 ROBERT RUSH :MILLER
..... _..-.,. c"-"j ~ __.i""""",c·"l'·' I \: I. \. "
MAP 1. Maximum extent of late Pleistocene waters in the southern Great Basin and ad joining regions. The wider boundary line marks the separation between coastal and interior or Colorado River drainages. The narrower line indicates the outline of the Death Valley system and enclosed basins (see inset). The authorities for the Pluvial lakes are cited by ;'diller (1946) and by Hubbs and Miller (1948). 1. Lake Tulare 8. Lake Panamint 16. Lake Manix 2. Walker Lake arm of Lake 9. Lake Manly 17. Lake Harper Lahontan 10. Lake Tecopa 18. Lake Kane 3. Lake Russell! 11. Lake Pahrump 19. Lake Thompson 4. Lake Long Valley 12. Lake Mesquite 20. Lake Amboy 5. Lake Adobe 13. Lake Ivanpah 21. Lake Ward 6. Lake Owens 14. Lake Mohave 22. Lake LeConte 7. Lake Searles 15. Little Lake Mohave
1 Named by Putnam (1949) THE GENERA CYPRINODON AND EMPETRICHTHYS 157
N
LAS VEGAS.
-, SHOSHONE! ...... \ "'NYO '. COUNTY "ftTECOPA <._. 5AN-~BERNA-ROIN~;COU • ~ 'Pc(j/A,A:l~~:~; ... J(-;'.•, ..~~:
G, .". . \ i
M o II. BAKERA' \ ... ..
E RT
~LOW.
. PARKE " ue
MAP 2. Sketch map of southeastern California and adjacent areas, showing the shrunken remnants of Pluvial waters and localities mentioned in the text. recogrntion of numerous races." Isolated resented by two species: E. inerriami, in springs and a few disconnected surface habiting Ash Meadows of the Amargosa flows are all that remain of this river, but River drainage, and E. latos, known only during periods of heavy rain in the head from Pahrump Valley (Map 2). The waters, Amargosa River comes to life as latter species has differentiated into 3 it pours its flood waters onto the great subspecies. These localities, both in Death Valley salt flat-the remnant of southern Nye County, Nevada, are iso Pluvial Lake Manly. Even during such lated by a low aluvial divide. Pahrump flood stages, however, there is no evidence Valley is a wholly enclosed basin, whereas that the many fish populations inter during high water stages the outflows of mingle. the springs in Ash Meadows join Amar The genus E1'1lpetrichtlz}'s, restricted to gosa River, which also contains water at warm springs and their outflows, is rep- such times. Until recently, Enipetrichthys 3 Following rather general usage in ichthyol was not known to have any close relative, ogy, the term race is applied to the next sys and at one time it was thought to be re tematic category below the subspecies; it does not ordinarily receive nomenclatorial recog lated to Orestias, a genus peculiar to Lake nition. Titicaca and other elevated waters of 158 ROBERT RUSH MILLER
the Andes. Representatives of this re total population are possible. The num markable genus are very rare in Ash ber of individuals of this species fluctuates Meadows, for only 22 specimens have between about 50 and 400. At times the been secured over a six-year period, but minimum effective breeding population they are rather common in Pahrump Val may thus fall below 50 individuals. No ley. In Ash Meadows Empetrichthys is doubt the Sewall Wright effect has been associated with Cyprinodon nevadensis a very significant factor in the evolution and the cyprinid, Rhinichthys osculus, but of this remarkable species, which is one in Pahrump Valley it constitutes the sole of the most distinctive in the genus. native fish life. The comparative abun dance of E. latos there is perhaps due DURATION OF ISOLATION largely to lack of competition. There may The postpluvial disruption of rivers, be some local differentiation in the popula lakes, and springs in the arid West re tions of E. merriami inhabiting the various ulted in the separation of fish populations springs in Ash Meadows, but if so the into isolated stocks, some of which have material collected to date is insufficient subsequently differentiated into distinct to determine whether the observed varia species and subspecies. Others have at tions are significant. tained only the racial level of differentia tion. It may be stated as a general rule, FACTORS AFFECTING SPECIATION RATES for desert fishes, that a positive correla There is evidence that at least 4 factors tion exists between duration of isolation play an important role in modifying rates and extent of differentiation (Hubbs and of evolution in these desert fishes. Miller, 1948). In the Death Valley re gion, those populations of Cyprinodon POPULATION SIZE estimated to have been isolated since In very small, closely inbreeding popu early Wisconsin time are now distinct lations, the probability of accidental sur species, sharply differentiated from their vival of particular genes or gene combina nearest relatives. If placed in the hands tions is greatly increased. Sewall Wright of a systematist unfamiliar with the group, has ably demonstrated the theoretical they might readily be regarded as com basis for this chance fixation of characters prising two or three genera. On the other which, in the absence of interspecific com hand, the close relationship among the petition, may be more important than six subspecies and the many races of selection in the evolution of such isolates. Cyprinodon nevadensis in the basin of The chances for recessive mutations to be Amargosa River indicates recency of iso come homozygous are better than they lation. Geologists believe that the Amar are in large populations of more wide gosa may have been a continuous river as spread distribution. The determining fac recently as a few centuries ago, at most tor is the size of the minimum effective a few thousand years ago. Nevertheless, breeding population. In some of the we cannot be sure that postpluvial disrup spring-inhibiting fish in the Death Valley tion of drainages accounts for all such region, the breeding population frequently speciation, since permament rivers today drops to a minimum of several hundred in other parts of the West also contain individuals or considerably less. The endemic forms. most striking example is afforded by The duration of time since Pahrump Cyprinodon diabolis, a dwarf species con Valley and Ash Meadows (of the Amar fined to a single hot-spring hole where the gosa River basin) have been isolated entire population occupies a habitat about from each other cannot be reliably esti 20 feet long, 8 feet wide, and a few inches mated in the absence of detailed studies of to a few feet deep. Since every fish is in the past hydrography of these two regions. view at one time, reliable estimates of the The isolation has been of sufficient dura- THE GENERA CYPRINODON AND EMPETRICHTHYS 159
tion to allow for the differentiation of a determined, at least in part, is indicated species of Enipetriclithys in each basin. by preliminary breeding experiments and In Pahrump Valley, the evolution of 3 by the more carefully controlled work on subspecies of E. latos in isolated springs Fundulus by Gabriel (1944). These suggests that the lake which formerly cov show that F 1 generations retain essen ered the valley floor was of late Pleistocene tially the same meristic characters found age, and the physiographical evidence sup in the natural stocks. A few exceptions. ports this view. Two of the springs are however, indicate that altered environ less than 1 mile apart and the third is ment, as well as heredity, is a factor de only 7 miles distant. termining the number of parts. Observations and experiments have ECOLOGICAL CONDITIONS shown that there is a very general cor Ecological conditions in desert springs relation between temperature and certain are believed to play an important role in meristic characters in fishes. Thus an hastening evolution. These springs are increase in temperature usually is asso very diverse, physically and chemically, ciated with a decrease in number of fin and their contained fish life presumably rays, scales or vertebrae, and, vice versa, has become highly adapted to this great in cooler waters the number of segments variety of conditions (Sumner and is typically increased. This picture is Lanham, 1942). It may be assumed that oversimplified, however and citing tem the conditions in each spring and creek in perature as the only, or major, causal some way influence the development of factor is treading on uncertain ground. the remnant populations. What effect Many other factors undoubtedly operate these chemical and physical characteristics in conjunction with accelerated tempera may have in modifying the characters of ture, but what these factors may be and the fishes is known only for two striking how they affect the organism remains to features of springs and creeks, namely, be determined. high temperature and salinity. Some of Among certain subspecies and races of the springs in the Death Valley region are Cyprinodon. nevadensis, a striking lack of remarkably high in boron content, but no correlation between temperature and me investigation has been undertaken to de ristic characters may be demonstrated termine what effect, if any, this element (table 1). For example, in comparing may have on the fishes. pectoralis with the race of mionecies in Much evidence has been accumulated to habiting the source pool of Point of Rocks demonstrate that high temperature plays Spring (columns 2 and 4, table 1), it is an important role in accelerating the rate seen that the meristic characters are con of speciation (Plough. 1942. and other sistently and significantly higher for pec authors in same symposium). The repeti toralis even though it inhabits the warmer tion of certain striking trends in warm of the two springs. This suggests that spring fishes is a significant expression of high temperature may have accelerated this relationship: (1) the number of me the production of mutations and that the ristic segments is generally reduced; (2) random inheritance of these changes may the position of the dorsal fin is often more have produced the variant structure of posterior; (3) the' head. eye and other such populations. Plough (1942: 14-15) anterior parts .of the body are enlarged; has shown that high temperature. as well (4) the size is very frequently reduced; as temperature shocks, greatly increases (5) the pelvic fins and rays of Cyprinodon the mutation rate in Drosophila. tend to be reduced in size and number. or Although no clear-cut correlation be the fins may be entirely lacking; and other tween temperature and meristic charac modifications have been observed. That ters can be demonstrated by comparing some or all of these trends are genetically distinct subspecies or races within one 160 ROBERT RUSH MILLER area or between two regions, a general proximated temperatures, the average correlation of this nature is evident. This number of body scales is often not the is epecially true when populations con same. For example, the temperature of fined to the limits of a single spring and Hidden Spring is nearly identical with its outlet are considered. Here one finds that of Eagle or of Deep Spring, but the a significant positive correlation between average number of body scales differs temperature and certain meristic char significantly from the samples from either acters. This correlation is particularly of those springs (fig. 1). well illustrated by the number of scales Increasing salinity seems to have about around the body (fig. 1). In the samples the same effect on speciation as decreas from the spring outlets, which are con ing temperature, namely an increase in sistently cooler than the source springs, certain meristic parts. In C}Iprinodon the body scales are significantly increased salinus, the species that inhabits the briny when compared with the fish from the waters of Salt Creek on the floor of Death springs. In each complex of spring Valley, the scales are smaller than in source and outlet there appears to be a any other known species of the genus. distinct genetic stock, the members of In the Death Valley race of Cyprinodon which respond to temperature differences nevadensis amarqosae, found in the waters in the same way. Although the spring of Amargosa River, which are less saline sources may have identical or closely ap- than those of Salt Creek, the scale counts
TABLE 1. Correlation between certain meristic characters and temperature infive warm-spring populations of Cyprinodon nevadensis Means are given above, and number of specimens and standard error of the mean below in parentheses.
Subspecies...... shoshone mionectes mionectes pectoralis calidae Localitvt ...... _...... A B C 0 E Temperature...... 30° C. 31° C. 31° C. 33° C. 37° C.
Anal rays 10.07 9.62 9.88 10.03 10.08 (133, ±.04) (103, ±.05) (67, ±.05) (100, ±.04) (100, ±.04) Pectoral rays 15.99 15.68 15.54 17.10 15.88 (214, ±.04) (193, ±.05) (233, ±.04) (200, ±.04) (200, ±.05) Pelvic rays' 4.98 2.99 5.01 5.12 5.33 (258, -) (206, -) (234, -) (200, -) (200, -) Lateral scales 25.72 24.59 25.40 25.42 25.58 (116, ±.05) (95, ±.06) (67, ±.07) (90, ±.06) (100, ±.05) Dorsal to pelvic scales 9.39 8.75 9.14 9.79 9.35 (48, ±.08) (36, ±.07) (49, ±.08) (49, ±.08) (94, ±.06) Dorsal to anal scales 8.88 8.50 8.86 8.94 8.94 (50, ±.06) (50, ±.07) (50, ±.07) (50, ±.05) (100, ±.05) Predorsal scales 18.50 17.96 17.82 17.91 17.71 (111, ±.11) (50, ±.13) (67, ±.12) (90, ±.12) (99, ±.10) Caudal peduncle scales' 15.64 13.00 14.34 15.44 15.71 (84, -) (95, -) (67, -) (89, -) (100, -) Body scales 23.25 20.44 23.10 24.75 23.06 (84, ±.17) (91, ±.12) (67, ±.22) (89, ±.19) (100, ±.10)
1 A = Outlet, Shoshone Spring B = Source, Point of Rocks Spring C = Big Spring D = Lovell's Spring E = Outlet, South Tecopa Hot Spring A and E lie about 8 miles apart in the Middle Amargosa River basin, California; B, C, and Dare 172 to 9 miles apart in Ash Meadows, Nye County, Nevada (Map 2). 2 The standard error of the mean was not calculated for these data as they present a markedly skewed curve and the derivation of this statistic assumes a normal frequency curve (Miller, 19,48: 15). THE GENERA CYPRINODON AND EMPETRICHTHYS 161
28 -
(/) 27 i-- W .J 18 - 27' 23° 28° 23° 28° 25' 32° 28° 34° 30° SOURCE OUTLET SOURCE OUTLET SOURCE. OUTLET SOURCE: OUTLET SOURCE OUTLET DEEP EAGLE HIDDEN POINT SHOSHONE SPRING OF ROCKS SPRING SPR ING SPRING SPRING TEMPERATURE IN DEGREES CENTIGRADE FIG. 1. Correlation between body circumference scales and temperature in certain populations of Cyprillodoll neuadensis from the Amargosa River basin. average higher than in any other form of The numerous warm springs of the C. nevadensis. In the spring-fed lakes at Death Valley region have temperatures Saratoga Springs, Death Valley, where that vary from about 22° to 43° C. (72° C. nevadensis neuadensis spawns, the to 109° F.), although fish do not tolerate waters are only moderately salty and the temperatures higher than 40° C. (l04° average number of scales around the body F.). Individually the temperature of each is lower than it is at either of the preced spring fluctuates very narrowly over the ing localities. A direct, environmental in years (Miller, 1948: 130). There is evi fluence is indicated, but it is thought that dence that Cyprinodon (and presumably the observed differences are due in part Empetrichthys also) spawns every month to genetic adaptations that parallel those of the year in the warmer springs (those due to the direct effect of the surround 28° C. or higher), whereas the limits of mgs. the breeding period for the stream popula tions are approximately from April to NUMBER OF GENERATrONS PER October. In contrast to the 2 (or perhaps YEAR 3) generations produced per year by the Speciation is evidently a particularly stream stocks, as many as 8 or 10 genera rapid and extensive process in warm tions may be the annual production for springs. The role that high temperature the warm-spring fish. Experimental may play in hastening this process already studies have shown that Cyprinodon has been discussed. A secondary and neuadensis may reach maturity and pro highly significant effect of raised tem duce young when only 2 months old. peratures is the increase in number of Since these fish were born and developed generations that are produced each year in much cooler water than that of their by the warm-spring fishes. native warm springs, it seems safe to as- 162 ROBERT RUSH MILLER sume that the warm-spring fish are capa amargosae, only moderate success was ble of breeding in nature when only 1 obtained, but when males of amarqosae month or 6 weeks old. This production were hybridized with females of macu of numerous generations each year may larius, very good results ensued. Some greatly accelerate the time rate of evolu of the variations in productivity were tion of such populations. probably influenced by differences such The isolation of small, breeding popula as size, temperature, and exposure, in the tions that are subjected to an accelerated various pools used in the experiments. mutation rate and an increase in number In the hybridization between subspecies of generations, due to high temperature, of C. neuadcnsis, best results were obtained are believed to result in the striking dif when the two spring forms tncuadcusis ferentiation exhibited by such species as and shoshone) were crossed. The two C:yprinodon diabolis. other combinations, tunarqosae X neva densis and amarqosae X shoshone, in EXPERIMENTAL \iVORK volved one parent form (a11largosae) that The rearing of three subspecies of develops in nature in saline water. . The Cypril1odon ncuadensis in concrete pools, experimental pools contained fresh water. under physical and chemical conditions Whether this factor had an important unlike those in their natural habitats, has bearing on the poor results when a111a1' led to some interesting results that need gosae was used as one parent, or whether further checking. With only few excep neuadcusis and shoshone are more closely tions, analysis of meristic characters of the related to each other than either one is to F] has demonstrated that these genera amargosae, is problematical. tions retain the characters of their respec tive subspecies in nature. In general, the PALEONTOLOGICAL EVIDENCE average values for the first generations of Although the paleontological data on the three subspecies lie in the same direc fishes from western United States are tion as do the values for the natural stocks woefully incomplete. a study of the few of these subspecies. If the subspecies in available Miocene fossils reveals a fauna nature do not differ significantly in a quite unlike that now inhabiting the re particular count, the values for the F] gion. It is therefore assumed that the stock likewise are not significantly dif recent fish fauna is not much older than ferent. Pliocene. Pleistocene fish remains should Interspecific, as weIl as intraspecific, be of great importance in working out the hybridization was carried out. When spe more recent history of the fish fauna. but cies of Cyprinodon were crossed, lowered the few specimens coIlected thus far have fertility generaIly resulted, particularly in been largely fragmentary. certain reciprocal crosses. Attempts to The occurrence of 3 species of Fundulus mate males of Cyprinodon salinus with in Death VaIley and the Mohave Desert females of Cyprinodon inacularius (a spe in late Pliocene or early Pleistocene time, cies confined to the lower Colorado River and of a fossil Cyp1'inodon in Death Val basin) failed. Mating females of salinas ley during the same period (MiIIer. 1945). with males of neuadensis resulted in sev is of significance in attempting to elucidate eral generations of hybrids, but compara the origin of the cyprinodontid fauna. Fun tively few fish were produced. The most dulus and Cyprinodon are widespread successful interspecific matings took place today in regions of low altitude and are between neuodensis and macularius, but particularly abundant near and along the success of some of these combinatinons coastal areas. In particular. the presence varied with the reciprocal crossings. For of Fundulus suggests a lowland connec example. when males of C. macularius tion to the sea from the southwest, as the were crossed with females of C. neuadensis only living members of this genus in the THE GENERA CYPRINODON AND EMPETRICHTHYS 163 Pacific drainage are now found from Mon which may be closely approximated. One terey Bay, California, nearly to the tip of unique species, Cyprinodon diabolis, is Lower California. Cyprinodon clearly confined to a single small spring in which came in from the southeast, in the direc its total numbers fluctuate from about 50 tion of the Colorado River (Miller, 1946). to 400 individuals. Experimental studies The endemic genus Enipetrichtliys, have demonstrated that F, generations of whose nearest living relative is Crenich pure stocks of three subspecies of Cy tliys of eastern Nevada, is evidently an old prinodon ueuadcusis, reared under en (pre-Pleistocene?) relict. Both genera vironmental conditions greatly altered were presumably derived from Fundulus, from those of their natural habitat, main and Empetrichth-ys may have descended tained essentially the same meristic char from a species similar to F. curryi, one of acters as are found in the natural popula the fossils from Death Valley (Miller, tions. A few exceptions, however, indi 1945: fig. 1). cate that altered environment. as well as heredity, is a factor in determining the SUMMARY number of segments. Speciation in the cyprinodontid fishes inhabiting the Death Valley region is cor LITERATURE CITED related with the disruption of waters since GABRIEL, M. L. 1944. Factors affecting the late Pleistocene times, with population number and form of vertebrae in Fundulus size, with the number of generations pro heteroclitus. Jour. Exp. Zool., 95: 105-147. duced each year, and with physical and HUBBS, C. L., AND R. R. MILLER. 1948. Cor chemical conditions of the remnant springs relation between fish distribution and hydro graphic history in the desert basins of west and creeks. There are two living genera, ern United States. Ill: The Great Basin, Cyprinodon with 4 species, one of which with emphasis on glacial and postglacial has 6 subspecies, and Empetrichtltys with times. Bull. Univ. Utah, 38, No. 20 (BioI. 2 species of which one has 3 subspecies. Ser., 10, No.7) : 18-166, Figs. 10-29, 1 map. All of the species and subspecies are MILLER, R. R. 1945. Four new species of fos sil cyprinodont fishes from eastern Cali endemic, as is Einpetrichtliys which is evi fornia. Jour. Wash. Acad. Sci., 35 (10): dently an ancient relict. Three fossil rep 315-321, Figs. 1-4. resentatives of Fundulus, probably of late MILLER, R. R. 1946. Correlation between fish Pliocene or early Pleistocene age, indicate distribution and Pleistocene hydrography a lowland connection to the southwest. in eastern California and southwestern Ne One of the species may have been near vada, with a map of the Pleistocene waters. Jour. Geol., 54: 43-53, Figs. 1-2. the ancestral stock from which Empetri MILLER, R. R. 1948. The Cyprinodont fishes c!tthys was derived. The Cyprinodon of the Death Valley system of eastern Cali reached Death Valley from the southeast, fornia and southwestern Nevada. Misc. by way of the Colorado River basin or its Publ. Mus. Zool. Univ. Mich., No. 68: 1 antecedent. Most of the subspecies of ISS, Figs. 1-5, PIs. I-XV, Maps 1-3. PLOUGH, H. H. 1942..Temperature and spon Cyprinodon nevadensis show racial dif taneous mutation', BioI: Symposia, 6: 9-20, ferentiation, particularly C. n. mionectes Figs. 1-2. of Ash Meadows, Nevada. High tem PUTNAM, W. C. 1949. Quaternary geology of perature is believed to have accelerated the June Lake district, California, Bull. the time rate of evolution in warm-spring Geol. Soc. Amer., 60: 1281-1302, Figs. 1-6, populations by increasing the mutation PIs. 1-7. rate and the number of generations pro SUMNER, F. B., AND U. N. LANHAM. 1942. Studies of the respiratory metabolism of duced each year. Many of the local sub warm and cool spring fishes. BioI. Bull., species and races have very small ranges, 82 (2): 313-327, Figs. 1-4.