Genetic Differentiation in Speciose Versus Depauperate Phylads: Evidence from the California Author(s): John C. Avise and Francisco J. Ayala Source: Evolution, Vol. 30, No. 1 (Mar., 1976), pp. 46-58 Published by: Society for the Study of Evolution Stable URL: http://www.jstor.org/stable/2407671 . Accessed: 14/07/2011 14:09

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=ssevol. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to Evolution.

http://www.jstor.org GENETIC DIFFERENTIATION IN SPECIOSE VERSUS DEPAUPERATE PHYLADS: EVIDENCE FROM THE CALIFORNIA MINNOWS

JOHN C. AVISE' AND FRANCISCO J. AYALA Department of Genetics, University of California, Davis, California 95616

Received February 3, 1975

The systematics of cyprinid native habiting California, using techniques of gel to California has had a chaotic history. electrophoresis. These techniques provide Considerable controversy exists about the considerable information to elucidate evo- number of species, their placement in lutionary relationships among closely re- genera, and the evolutionary affinities lated species (Avise, 1974). Our results among the genera (Uyeno, 1961; Hopkirk, suggest that at least four, and possibly 1973). About 16 species and 10 genera are five of the genera of California minnows recognized at present (Moyle, 1974). Five are very similar in genic content and are of these genera (Orthodon, Mylopharodon, probably of recent monophyletic origin. Hesperoleucus,, and ) However, the other California minnows are are monotypic and restricted in distribu- less closely related, and the biochemical tion to a single drainage basin, the Sacra- differences observed between these species mento-San Joaquin. may be typical of mean levels of divergence All cyprinids in North America are between other North American cyprinids. thought to belong to the subfamily Leucis- Models are considered which describe cinae with a single possible exception; amounts of genetic differentiation expected several subfamilies are recognized in the between species within a group. One model Old World (Miller, 1959). It is commonly assumes that genetic distance between accepted that relatively few cyprinid fishes species is proportional to the time since migrated from their center of origin in they shared a common ancestor; the other Eurasia to North America by way of a model assumes that genetic distance is Bering land bridge during the Miocene. proportional to the number of cladogenetic Therefore, most cyprinid species in North events (speciations) in the evolutionary America, including the California minnows, history of the group. Those models lead may share a relatively recent common an- to distinct predictions of mean amounts of cestry. "Not only the lack of basic mor- genetic distance between species in species- phological diversity but also the readiness diverse versus species-depauperatephylads with which most American minnows hy- of equal evolutionary age. Our data on the bridize (Hubbs, 1955) supports the evi- California minnows suggest that mean time dence that the group has not been here since divergence from a common ancestor long enough to develop strongly divergent is more important than mean number of lines" (Miller, 1959). It has, nevertheless, cladogenetic events in the evolutionary been suggested that close affinities exist history of a group of species as a predictor between certain Western cyprinid genera of levels of biochemical differentiation. (Mylopharodon, ) and those in China MATERIALS AND METHODS and Japan (see Miller, 1965). We have analyzed patterns of genetic were collected by seine and im- variability at 24 gene loci in species be- mediately placed on dry ice. The nine longing to nine genera of cyprinids in- species and the collection sites are listed in Table 1. The data for two species, Hes- ' Present address: Department of Zoology, peroleucus symmetricus and Lavinia exili- Univ. of Georgia, Athens. Georgia. cauda, are taken from Avise et al. (1975).

EVOLUTION 30:46-58. March 1976 46 GENETIC DIFFERENTIATION IN MINNOWS 47

TABLE 1. Samples of California minnows and estimates of genetic variability in a study of 24 gene loci coding for soluble proteins and enzymes.

H

t1o ) 0 E

Common Collection c Species name site M ,c

Hesperoleucus symmetricus roach Russian River, 6.67 ? 2.54 25 1.46 32 Mendocino Co. hitch Bass Lake, Madera Co. 4.80 ? 2.38 17 1.21 48 Mylopharodon conocephalus hardhead Chowchilla River, 0.62 ? 0.62 4 1.04 60 Madera Co. grandis Sacramento Russian River, 1.13 ? 1.13 4 1.13 28 squawfish Mendocino Co. Orthodon microlepidotus Sacramento Clear Lake, 1.50 ? 1.13 4 1.08 23 blackfish Lake Co. Pogonichthys macrolepidotus splittail San Joaquin River, 3.57 ? 1.07 8 1.13 7 Contra Costa Co. egregius Lahontan Sagehen Creek, 3.00 + 2.39 8 1.13 22 redside Placer Co. Gila bicolor tui chub Lake Crowley, 5.92 + 2.77 21 1.25 24 Mono Co. Notemigonus crysoleucus golden Bass Lake, Madera Co. 6.80 + 2.92 21 1.29 15 shiner

* Frequency most common allele < 0.95.

The electrophoretic techniques used substitutions per locus which have accu- have been previously described (Avise et mulated since any two populations diverged al., 1975). Basically these procedurescon- from a common ancestor. sist of placing samples of muscle or liver extract from individual fish into starch GENETIC VARIATION WITHIN SPECIES gels, and applying electric currents to the Little or no allelic variation is found at gels. Proteins migrate through the gels most of the 24 loci in any species. If a according to their net charge; their posi- locus is considered polymorphic when the tion is visualized with specific staining most common allele has a frequency no mixtures. A total of 24 loci have been greater than 0.95, the number of poly- studied in every species; 12 enzymes are morphic loci ranges from one (4.2% in encoded by 19 gene loci and five non- Mylopharodon conocephalus, Orthodon enzymatic proteins are encoded by five microlepidotus and Ptychocheilus grandis) 48 J. C. AVISE AND F. J. AYALA

TABLE 2. Allele frequencies at variable loci in nine species of California minnows. Enzyme abbrevia- tions given by Avise et al. (1975). Alleles are designated according to differences in electrophoretic mobility, with the common allele in Hesperoleucus usually designated 100.

Species Allele (frequency) Hesperoleucus Ldh-l Ldh-2 Gpd-1 Pgi-1 Ipo-1 Est-2 symmetricus 100(0.70) 100(0.86) 300(0.95) 110(0.03) 100(0.70) 100(0.91) 85(0.30) 0(0.14) 100(0.05) 100(0.95) 50(0.30) 97(0.09) 10(0.02) Tpi-1 Mdh-3 Pgm-1 Got-2 100(0.98) 100(0.99) 110(0.02) -100(0.97) 50(0.02) 0(0.01) 100(0.98) -20(0.03) Lavinia Ldh-l Pgi-2 Pgm-1 Ipo-1 Est-2 exilicauda 100(0.80) 200(0.13) 100(0.68) 150(0.18) 103 (0.01) 85 (0.20) 100(0.87) 90(0.32) 100(0.82) 100(0.99) Mylopharodon Gpd-1 conocephalus 100(0.92) 75(0.08) Ptychocheilus Adh-i Pgm-1 Ipo-1 grandis 100(0.90) 100(0.04) 150(0.02) 90(0.10) 90(0.96) 100(0.98) Orthodon Pgi-2 Gpd-1 microlepidotus 100(0.80) 300(0.02) 10(0.20) 200(0.98) Pogonichthys Mdh-1 Pgi-2 macrolepidotus 85 (0.92) 98(0.43) 70(0.08) 0(0.50) -10(0.07) Richardsonius Pgi-1 Est-2 Pgm-1 egregius 120(0.39) 97(0.91) 110(0.98) 110(0.61) 95(0.09) 00(0.02) Gila bicolor Mdh-2 Idh-2 Ipo-1 Est-2 Pt-3 Gpd-1 100(0.83) 100(0.94) 50(0.94) 100(0.27) 100(0.52) 100(0.99) 75(0.17) 90(0.06) 10(0.06) 97(0.73) 80(0.48) 75(0.01) Notemigonus Pgd-i Pgi-1 Ipo-1 Est-1 Pt-3 Pgi-2 crysoleucus 175(0.74) 120(0.20) 175(0.87) 100(0.95) 120(0.50) 100(0.04) 160(0.23) 110(0.80) 125(0.13) 98(0.05) 100(0.50) 10(0.96) 140(0.03) genetically variable loci are shown in The confidence of the estimates is low since Table 2. so' few loci contribute to the heterozygosity The levels of genetic variation found in estimates, and the numbers of individuals the California minnows are within the low sampled are small. part of the range characteristic of verte- Little genetic differentiation exists be- brate species (Selander and Kaufman, tween populations of a given species. Avise 1973). Three species (Mylopharodon et al. (1975) have studied geographic vari- conocephalus, Ptychocheilus grandis, and ation within the species Hesperoleucus Orthodonmicrolepidotus) have particularly symmetricus and Lavinia exilicauda, and low heterozygosities. But little significance have found little differentiation among should be attached to the different levels conspecific populations even when sepa- of genetic variation among the species. rated by considerabledistances. Moreover, GENETIC DIFFERENTIATION IN MINNOWS 49

TABLE 3. Matrix of genetic distances (above diagonal) and genetic similarities (below diagonal) between species of California minnows, based on 24 loci. Distances and similarities calculated using the method of Nei (1972).

1 2 3 4 5 6 7 8 9 1) Hesperoleucus 0.055 0.095 0.194 0.518 0.705 0.432 0.251 0.901 2) Lavinia 0.948 0.147 0.216 0.616 0.746 0.519 0.354 0.919 3) Mylopharodon 0.909 0.863 - 0.131 0.546 0.600 0.453 0.174 0.790 4) Ptychocheilus 0.824 0.806 0.877 - 0.541 0.600 0.526 0.333 0.989 5) Orthodon 0.596 0.540 0.579 0.582 - 1.079 0.776 0.518 1.094 6) Pogonichthys 0.494 0.474 0.549 0.549 0.340 0.519 0.679 1.118 7) Richardsonius 0.649 0.595 0.636 0.591 0.460 0.595 0.443 0.976 8) Gila 0.778 0.702 0.840 0.717 0.596 0.507 0.642 0.884 9) Notemigonus 0.406 0.399 0.454 0.372 0.335 0.327 0.377 0.413

we have sampled a second population of and completely different (I < 0.05) at Ptychockeilus grandis, from the Chow- about 30% of the loci. Few loci have ge- chilla River, Madera County. The two netic similarities in the broad range be- populations of this species share common tween 0.05 and 0.95. This U-shaped alleles at all loci studied; their genetic distribution of genetic similarities is often similarity is I = 0.99. This situation agrees found in comparisons between closely re- with what has been generally found in studies of allozyme variation in many 60- vertebrate and invertebrate species: geo- graphic populations of a given species are genetically very similar, yielding similarity coefficients usually greater than 0.95 (re- 50 view in Avise, 1974). Therefore, for pur- poses of interspecific comparisons, a single 9 GENERA OF population provides an adequate represen- CA L I FO R N I A 440 MINW tation of a species. 4 O- ~~~MI NN O W S

DIFFERENTIATION BETWEEN SPECIES =0.589?0.029 Table 3 gives the estimates of genetic 3 0- similarity, I, and genetic distance, D, (cal- culated according to the method of Nei, 1972) between all species pairs. They spread over a broad spectrum ranging from 2 0- D = 0.055 (Hesperoleucus symmetricus versus Lavinia exilicauda) to D = 1.118 (Pogonickthys macrolepidotus versus No- temigonus crysoleucus). The average value 10 of D for all pairwise comparisonsis 0.57 ? 0.05; i.e., on the average about 0.57 elec- trophoretically detectable allelic substitu- tions per locus have occurred in the 0 .1 .2 .3 4 5 .6 .7 .8 .9 1 separate evolutions of any two species. SIMILARITY Figure 1 shows the distribution of ge- netic similarities among loci. Pairs of spe- FIG. 1. Percentageof loci within a given range of genetic similarity values in the comparisons cies are, on the average, essentially identi- among species belonging to nine genera of Cali- cal (I > 0.95) at nearly 60% of the loci, fornia minnows. 50 J. C. AVISE AND F. J. AYALA lated species, whether they are relatively The UPGMA and WPGMA are simple monomorphiclike the California minnows, agglomerative clustering procedures, cy- or very polymorphic like the species of the cling repeatedly through the data matrix Drosophila willistoni group (Ayala et al., and admitting for membershipto previously 1974a,b). This bimodal distribution of formed clusters the single entity (a species loci with respect to genetic similarity has in our case) exhibiting the highest level of bearing on the question of the sampling similarity to a cluster. In UPGMA, the strategies to be followed when estimating average similarity of a candidate species to genetic differentiation between closely re- an extant cluster is determined by weight- lated species. At a given locus any two ing equally each species in the extant species are either essentially identical in cluster. WPGMA differs from UPGMA allelic configuration, or else completely by weighting the species most recently distinct. Therefore, it is not essential to joined to the cluster equal to all previous sample many individuals at each locus, species in the cluster. UPGMA and since a few individuals provide nearly as WPGMA imply that the resulting dendro- much information as large numbers. On grams represent estimates of evolutionary the other hand, the bimodal distribution trees only if overall rates of evolutionary of genetic similarities makes it important divergence are homogeneous in different that as many loci as possible be sampled. phyletic lines. Although there is no a To put this matter differently, with respect priori reason for preferring one method to genetic similarities the variance about over another, the relative fits of the dendro- individuals is small relative to the variance grams to the original matrix (Table 3) may about loci; therefore, the precision of the be computed as "cophenetic" correlation estimates is much more dependent in the coefficients. From our data, the cophenetic number of loci than on the numbers of correlationsfor UPGMA and WPGMA are individuals sampled. In our study, al- 0.96 and 0.93, respectively, indicating that though several species are repre- both methods distort little the information sented by few individuals, the number of in the similarity matrix. loci sampled (24) is fairly large. The Wagner tree method relaxes the To a first approximation, phylogenetic assumption of homogeneous evolutionary relationships can be estimated from ge- rates, and, indeed, has been used to test netic or phenotypic information. Methods whether evolutionary rates are homoge- exist for clustering groups of organisms neous. This method results in the forma- when a set of measurements have been tion of most parsimonious trees, that is, made in each group, and there is no reason trees with minimum length (defined as the to weight any measurement more than minimum sum of differences between all others. The genetic information provided nodes forming endpoints of branches). by electrophoretic studies is of this kind. Divergence between species estimated by We have evaluated the information con- the Wagner procedure are greater than or tained in Table 3 using three such equal to observed differences between spe- methods: (1) unweighted pair group cies. The cophenetic correlation of the method with arithmetic means or UPGMA Wagner tree with our data matrix is 0.99. (Sneath and Sokal, 1973); (2) weighted Figure 2 shows the dendrogramobtained pair group method with arithmetic means using the UPGMA method of clustering. or WPGMA (Sneath and Sokal, 1973); The dendrogramobtained by the WPGMA and (3) a modification of Wagner's tree method is virtually identical to that, ex- method (Farris, 1972). We shall briefly cept for the positions of Pogonickthys and discuss these clustering techniques since Orthodonwhich are reversed. The dendro- they give somewhat different dendrograms gram obtained by the modified Wagner when applied to our data. tree method is shown in Figure 3. All three GENETIC DIFFERENTIATION IN MINNOWS 51

.002 HIESPEROLEUCUS .053 iHESPEROLEUCUS .054 .108 IAVINIA LAVINIA .016 . 024 OYLOPP1ARODON MYLOPHARODON .038 3 PTYCH1OCHEI.LUS PTYCHOCHEILUS

.047 .0 GILA _ ~~~~~~~~GILA .099 .2 RICHARDSONIUS ORTIODON

ORTHODON .127 POGONICHTHYS .140 POGONICHTHYS RICHARDSONIUS

NOTEMICO N-US .610 NOTESMIGONUS

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 FIG. 3. Phylogenetic tree of California min- GENETIC IDENTITY nows based on the modified Wagner tree method. The tree 'is rotated so the root occupies the mid- FIG. 2. Dendrogram of California minnows point of the longest path a of based on the unweighted pair-group method with connecting pair species on the arithmetic means (UPGMA). Scale is in units of tree, that connecting Notemigonus to amounts genetic identity, I. The cophenetic correlation Orthodon. Numbers indicate presumed of in and in units equals 0.96. The dendrogram based on the evolution various branches are of genetic weighted pair-group method (WPGMA) is vir- distance, D. tually identical, except that the positions of Pogonichthys and Orthodon are reversed. a more complete discussion of these species, see Avise et al., 1975. methods of clustering give similar results Mylopharodon conocephalus (hardhead) for the five species genetically most similar. and Ptychocheilus grandis (Sacramento Therefore our phylogenetic interpretations squawfish).-These two species are very are most reliable where they involve the similar in external morphology, but can be closely related upper clusters in the dendro- separated on the basis of head shape and grams. We now discuss the phylogenetic presence of a frenum connecting the middle affinities between the minnow species based of the premaxilla to the head in Mylo- on our estimates of biochemical similarity, pharodon. Hardhead and Sacramento in relation to previous interpretations squawfish are endemic to the Sacramento- based on morphology, geographic distribu- San Joaquin drainage and other minor tion, and the fossil record. basins formerly connected to it. Ptycho- Hesperoleucus symmetricus (roach) and cheilus contains three described species Lavinia exilicauda (kitch).-These two besides grandis; oregonensis primarily in species are endemic to the Sacramento-San the Columbia River drainage, umpquae in Joaquin drainage and have no known fossil the Sinslaw and Umpqua drainages in record. Adults differ in a number of ap- Oregon, and lucius in the Colorado River. parently independent morphological fea- Fossils assigned to Ptychocheilus are tures, including body size and shape, num- known from the early Pliocene in Oregon ber of anal and dorsal rays, and pharyngeal (Shotwell, 1963) and Middle Pliocene in tooth formula (Avise et al., 1975). None- Arizona (Uyeno and Miller, 1965). Mylo- theless, this morphological differentiation pharodon is presently a monotypic . has occurred within the context of very Fossils assigned to "Leucus" from Middle little overall genetic divergence (D = Pliocene deposits in Idaho are similar to 0.055). The close biochemical similarity living hardhead, and may be referable to between Hesperoleucus and Lavinia most Mylopharodon (Uyeno and Miller, 1963; likely evinces a relatively recent speciation, Miller, 1965). which probably occurred in the Sacra- These two genera are quite similar in mento-San Joaquin basin. These two min- genic constitution (I = 0.88). This high nows should be considered congeners. For level of similarity seems surprising in view 52 J. C. AVISE AND F. J. AYALA

of the possible age of these genera as Richardsoniusegregius (redside) .-Rich- evinced by their widespread geographic ardsonius now contains four species in the distributions and by the fossil record. western United States, and is thought to be Comparablelevels of biochemicalsimilarity closely allied to the eastern genus Clinos- between widespread sympatric species are tomus. As pointed out above, our confi- unusual, but have been described in the dence in determining the phylogenetic genus Peromyscus (Avise et al., 1974). relationships among the minnows decreases Nonetheless, the close morphological sim- as similarity values become lower. Thus, ilarity between Mylopharodon and Ptycho- Richardsonius is placed in different posi- cheilus led Hopkirk (1973) to argue that tions by the PGMA and the Wagner tree the two are similar enough to be combined methods (Figs. 2 and 3). In any event, into one genus. The biochemical data sup- Richardsonius appears slightly outside the port this conclusion. The hardhead and range of similarities found between the five the Sacramento squawfish are genically previously discussed genera (Table 2), al- also closely similar to Hesperoleucus and though it is somewhat more similar to Gila Lavinia (see Table 3 and Figs. 2 and 3). and its derivatives than are Orthodon and Gila bicolor (tui chub).-This species is Pogonichthys. Our data do not support found in Californiain the Klamath, Lahon- (nor do they strongly deny) Uyeno's tan, Owens Valley, and Sacramento-San (1961) conclusion that Richardsonius di- Joaquin drainages. It belongs to a rather verged from the Gila line after Ptycho- large genus with at least 17 described cheilus. species distributed throughout the western Orthodon microlepidotus (blackfish) United States and northern Mexico. Fish and Pogonichthys macrolepidotus (split- resembling Gila are reported from Mid- tail) .-These two minnows are biochem- Miocene (Miller, 1965), making it the ically quite distinct from each other and oldest known cyprinid genus in North from the previously discussed species. America. Their position in the dendrogramsrelative The systematic relationshipsof the genus to each other and to Richardsoniuschanges, Gila are controversial. Members of Gila depending on the method of clustering are known to hybridize with Lavinia (Mil- used. Both species are endemic to the ler, 1963) and Richardsonius (Hopkirk Sacramento-SanJoaquin drainage and have and Benke, 1966). Hopkirk (1973) be- no fossil record. Hopkirk (1973) believes lieves that Hesperoleucus is a creek- that the original cyprinid invader of North adapted form derived from the Gila phyletic America was similar to Pogonickthys in line. Uyeno and Miller (1965) feel that general morphology,and that Pogonichthys Pogonichthys was also derived from a was not derived from a Gila-like ancestor. Gila-like ancestor, but Hopkirk (1973) dis- Jordon and Evermann (1896) feel that putes this claim. Uyeno (1961) examined species with straight teeth in one series, osteological relationships among three such as Orthodon, are more primitive. Our genera and concluded that Ptychocheilus findings are consistent with other authors' diverged earlier from a line leading to Gila interpretations that Pogonichthys and and Richardsonius. Orthodon are not closely similar to other On the basis of biochemical information, California minnows. Gila appears most similar to the Lavinia- Notemigonus crysoleucus (golden Hesperoleucus- Ptychockeilus - Mylopharo- shiner).-This minnow is not native to don group. It seems possible that these California, but has been introduced from four genera were derived from a Gila-like the eastern United States. Notemigonus ancestor, but the study of other species of may be the only North American member Gila appears necessary to confirm these )f the Abramidinae, primarily an Old inferences. World group (Miller, 1959). It is also the GENETIC DIFFERENTIATION IN MINNOWS 53

biochemically most distinct minnow ex- An important question in evolutionary ge- amined. It is well differentiated from the netics is whether a high degree of specia- native California minnows, exhibiting an tion (splitting or cladogenesis) is the result average of about one allelic substitution per of rapid genetic change (phyletic evolution locus relative to the other species (Table or anagenesis). To formulate the question 2). if Notemigonus does indeed belong to differently: are rates of cladogenesis or the Abramidinae,its genetic differentiation splitting closely correlated with rates of from the North American anagenesis or phyletic evolution? (such as the California minnows) should Avise and Smith (1974) have studied be representative of the degree of differ- genetic differentiation at 14 loci among entiation occurring between the most di- species of another group of fishes, the vergent New World minnows. centrarchid genus Lepomis. This genus consists of 11 extant species, all endemic PHYLETIC EVOLUTIONVERSUS to North America. Lepomis has never con- CLADOGENESIS tained many species, and appears in the As discussed earlier, most of the 250 fossil record at the Miocene-Pliocene species of North American minnows belong border (Miller, 1965). The genus Lepomis, to the Leuciscinae, a subfamily which first therefore, represents a species-depauperate appears in the fossil record of North group relative to the species-rich North America in the Miocene. The California American minnows. Although the phyletic minnows share geographic range but are evolution of both groups extends for sim- not necessarily a monophyletic assemblage. ilar lengths of time, fewer cladogenetic It seems likely that at least some of the events have taken place in Lepomis than species may have their closest relatives in the minnows. among the eastern United States fauna Avise and Ayala (1975) have formu- (i.e., Richardsonius with ). lated theoretical models predicting relative Furthermore, Notemigonus is possibly a levels of genetic differentiation in speciose member of a different subfamily, the versus depauperategroups, under two com- Abramidinae; comparedto other California peting assumptions: (1) genetic differen- minnows Notemigonus may be nearly as tiation is a function of time, unrelated to distant phylogenetically as the most sep- the number of cladogenetic events (Model arated North American minnows. As a 1) and (2) genetic differentiation is pro- group the California minnows may be a portional to the number of cladogenetic fairly representative sample of levels of events in the group (Model 2). These divergence among the North American models provide qualitatively distinct pre- minnows. dictions about levels of genetic divergence The North American minnows are a depending upon the relationship between highly species-diverse group, whose evolu- rate of speciation and amount of genetic Mio- tionary origin has been traced to the change. When genetic distance is a func- cene (see above). Most of the California tion of time, mean minnows examined occur sympatrically in genetic distances be- at least some localities in California and tween species in speciose and depauperate still retain their identities. The species phylads of comparable evolutionary age status of all species in the study is well are very similar. On the contrary, when established even for the species most closely genetic distance is a function of the number related in every aspect, Hesperoleucus of speciations in the history of a phylad, symmetricus and Lavinia exilicauda (Avise members of speciose phylads are much et al., 1975). How much genetic differen- more distinct than members of depauperate tiation has occurred among these minnows, phylads, on the average, and the ratio of comparedwith species-depauperategroups? mean distances increases as the frequency 54 1. C. AVISE AND F. T. AYALA of speciation events in one group relative to are very similar. Maximum D = 1.12 and the other becomes greater. 1.02, for the minnows and Lepomis re- The two models are, of course, over- spectively. simplifications. We do not expect genetic Our results do not agree with the pre- distances between species to be solely a dictions derived from model 2, i.e., that function of time since divergence, or of the rate of genetic evolution is proportional number of cladogenetic events in their to the amount of cladogenesis. The mean history. Genetic distances almost certainly as well as the maximum genetic distances reflect both of these factors, as well as the between species are very similar in both particular selective regimes faced by the groups, the speciose minnows, and the populations, the population sizes main- depauperate Lepomis. These results are in tained, and other parameters (Ayala, 1974; much better agreementwith the predictions Ayala and Gilpin, 1975). Nonetheless, if of model 1, namely that genetic divergence greater numbers of speciation events reflect is approximately constant through time a higher average rate of genetic evolution, independently of the amount of clado- we would expect to find greater mean dis- genesis. tances between species in diversified than The reliability of this test of the models in depauperatephylads of comparable evo- rests heavily upon the assumption that lutionary age. For example, we should levels of genetic divergence between the find mean genetic distances to be greater California minnows are representative of among the speciose Leuciscinae than among mean levels of divergence between North the depauperateLepomis. American minnows. We have earlier given Using the data of Avise and Smith arguments showing why this may be a (1974), we have calculated genetic sim- reasonable assumption. However, we know ilarity (I) and genetic distance (D) be- very little about the specific evolutionary tween 10 of 11 extant species of Lepomis histories of the California minnows, nor according to the method of Nei (1972). can we be certain about the true number For all pairwise comparisons, the average of cladogenetic events in the histories of genetic similarity is I = 0.54, with a range Lepomis or the minnows since the Miocene, of 0.36 to 0.85; the average genetic dis- although this number is probably greater tance is D = 0.63, with a range of 0.16 to in the minnows. Although minnows com- 1.02. For the Californiaminnows examined prise the largest number of known fossil in the present study, I = 0.59 (range 0.33 species of any group of fishes in North to 0.95), and D = 0.57 (range 0.05 to America, the record is far from complete. 1.12); if we exclude comparisons among The evolutionary origin of the North the four most closely related minnows, American minnows presently traces back these statistics become I = 0.54, and D = to the mid-Miocene; thus they are as old 0.65. It is, of course, risky to compare or older than the genus Lepomis whose amounts of genetic differentiationin groups evolutionary origin is dated from the as different as sunfish and minnows, par- Miocene-Pliocene boundary. It appears ticularly when different sets of loci are that the minnows are not evolving genet- assayed. Nonetheless, our data certainly ically at a faster rate than Lepomis. Never- give no indication that representatives of theless, the predictions of the models the speciose minnows are genically more should definitely be tested in other or- distinct than are membersof the depauper- ganisms as well. Among fishes, a particu- ate genus Lepomis. If anything, mean sim- larly promising group for comparison to ilarity appears slightly higher among the the depauperate Lepomis is the highly minnows than among Lepomis species. species-diverseminnow genus Notropis, ap- Furthermore, the maximum D observed parently dating from mid-Pliocene (Miller, between any two species in the two groups 1965). GENETIC DIFFERENTIATION IN MINNOWS 55

The finding of lack of correlation be- Selander (1973; see also Selander et al., tween amount of genetic differentiation 1969) found that two subspecies or semi- and number of speciation events is, of species of Mus musculus differ genetically course, hardly surprising. There are many as much as the Drosophila subspecies. well known instances of high rates of A different situation obtains during the anagenetic change with little cladogenesis, second stage of geographicspeciation, when and vice versa. In the evolution of the reproductive isolation is being completed genus Homo, a great deal of anagenesis under the influence of natural selection act- with respect to morphological, behavioral ing upon genetically different populations and other traits has taken place, while no which have regained geographic contact. cladogenetic events seem to have occurred In the D. willistoni group, Ayala et al. after that genus evolved from Australo- (1 974b) found that populations in the pithecus. Groups of sibling species, which second stage of speciation do not differ are common in insects as well as in other significantly more than populations in the organisms, are examples of the opposite first stage. It appears that sexual isolation situation: speciation with little or no or other pre-zygotic isolating mechanisms morphologicaldifferentiation. may develop without changing a sub- stantial proportion of structural genes. DISCUSSION The second survey strategy employed to An important problem currently dis- answer the question, "How much genetic cussed by evolutionists is "How much ge- differentiation accompanies speciation?" netic differentiation accompanies the pro- involves sampling species which by other cess of speciation?" Two general survey criteria appear especially closely related. strategies have been employed to answer Since species exhibit genetic differences this question. The first method is most which have accumulated subsequent to, as direct; it involves sampling populations at well as during the speciation process, the stages when reproductive isolation, and most telling results will be those at the hence speciation, is being completed. The low end of the distribution of genetic dif- most thorough study of populations at ferences. The rationale is that if enough these stages of divergence is that by Ayala cases are examined, examples may be found and coworkers on the Drosophila willistoni of species which differ little in genic con- group (review in Ayala et al., 1974b). The tent, and hence, in which speciation has generally accepted model of geographic involved little genic change. Results of speciation applies to the group. Ayala et al. such surveys have generally indicated that (1974b) find a substantial degree of ge- species which appear closely related, such netic differentiation during the first stage as sibling (morphologically nearly indis- of speciation, when isolated allopatric pop- tinguishable) species and hybridizing spe- ulations show evidence of partial reproduc- cies, are completely distinct in allelic com- tive isolation; some 0.23 electrophoretically position at about one third to one half of detectable allelic substitutions per locus their loci (review by Avise, 1975). Such had taken place between populations repre- results appear compatible with the hy- sentative of this stage. Similar amounts of pothesis that significant genetic divergence genetic differentiation have been found in accompanies the development of reproduc- other groups of Drosophila flies; for ex- tive isolation. ample, about 0.19 electrophoretically de- We have found that mean genetic dif- tectable substitutions per locus exist be- ferentiation between species is no greater tween the subspecies D. pseudoobscura in a species-rich group, the North Ameri- pseudoobscuraand D. p. bogotana (Ayala can minnows, than in a depauperategroup, and Dobzhansky, 1974; for other Drosoph- the genus Lepomis. This finding is in- ila species see Zouros, 1973). Hunt and compatible with two hypotheses: (1) that J. C. AVISE AND F. T. AVALA increased degrees of cladogenesis are the likely to involve changes in both structural result of higher rates of genetic evolution; genes and regulatory genes. Wallace and (2) that the process of speciation per (1963), Stebbins (1969), Britten and se (i.e., development of reproductive iso- Davidson (1969; Davidson and Britten, lating mechanisms) involves allelic changes 1973), and others have pointed out that at a substantial proportion of structural regulatory genes may play a crucial role in gene loci. If speciation per se were to in- evolution. Most recently, Wilson et al. volve much genetic change, we would (1974a,b) have suggested that adaptive expect to find on the average greater ge- evolution and speciation may depend more netic differentiation in speciose groups on changes in gene regulation than on (where the number of speciation events is amino acid substitutions in protein se- greater) than in depauperategroups. quences. Nevertheless, a wealth of direct Moreover we have found that the genetic and inferential evidence exists showing that distance between the two most similar spe- alleles of structural genes are subject to cies in our study, Hesperoleucus symmetri- selection, and that they become organized cus and Lavinia exilicauda, is fairly small, in coadapted genomes (see, e.g., Dobzhan- D = 0.055; i.e., these two species differ by sky, 1970; Prakash and Lewontin, 1968; about 5.5 allelic substitutions for every 100 Clegg et al., 1972; Ayala, 1972, 1974). loci. We have shown elsewhere that these Moreover, we should keep in mind that the two species are, indeed, "good" species genes studied by electrophoresisare for the (Avise et al., 1975). Although 5.5% of the most part involved in basic cell metabolism. genome represents a large number of loci Other kinds of structural genes may be (1,650 if we assume a genome size of equally or more important than electro- 30,000 gene loci), this value is much lower phoretic loci in producing morphological than the average of 20-25 allelic substitu- and ecological differentiation and reproduc- tions per 100 loci observed during specia- tive isolation. Instances of speciation with- tion in species of Drosophila and in Mus. out much change in genes studied by Other cases of speciation involving little electrophoresis do not necessarily imply change in structural genes have been re- that few changes have taken place in struc- ported. Speciation in gophers of the genus tural genes. Nevertheless, gene regulation Thomomys has been accomplished appar- must play an important role in evolutionary ently through extensive remodeling of change in general, and speciation in par- karyotypes by Robertsonian fusions and ticular. fissions, as well as other chromosomal re- arrangements with little structural gene SUMMARY divergence (Nevo et al., 1974). Among We have examined electrophoreticvaria- diploid plants, an instance of speciation tion in proteins encoded by 24 gene loci involving a change in reproduction from in natural populations of nine genera of obligate outcrossing to self-pollinating has minnows (family ) endemic to taken place with very little structural gene waters of California. The mean proportion change (Gottlieb, 1973). of polymorphic loci per population is Speciation, or cladogenetic evolution, 12.5%, and the mean frequency of hetero- occurs not only according to the general zygotes per locus is 3.78 ? 0.80%. These model of geographic speciation, but also by levels of genetic variation are within the a variety of other processes, including low part of the range characteristic of chromosomalreorganization, changes in re- vertebrate species. productive system, polyploidy and others. Average genetic distance, D, for all pair- Some of these modes of speciation may wise comparisons among species is 0.57, involve little change in structural genes. i.e., about 57 allelic substitutions, on the The process of geographic speciation is average, are estimated to have occurred for GENETIC DIFFERENTIATION IN MINNOWS 57 every 100 loci in the separate evolution of appreciation to Professor T. Dobzhansky any two species. At least four genera and to Drs. Hiram Li and Peter Moyle for (Hesperoleucus, Lavinia, Mytopharodon, critically reviewing the manuscript, al- and Ptychockeilus) are genetically very though this does not imply that they agree similar, and have probably evolved from with all our conclusions. Work was sup- a relatively recent common ancestor. The ported by an NIH training grant in ge- other genera are less similar; levels of ge- netics and by AEC contract AT(04-3)34. netic differentiation among them may be fairly representative for the very species- LITERATURE CITED diverse North American minnows. We have AVISE,J. C. 1974. Systematic value of electro- also calculated the mean genetic distance phoretic data. Syst. Zool. 23:465-481. among 10 of the 11 known species of the AVISE, J. C., AND F. J. AYALA. 1975. Genetic (in genus Lepomis; this is D = 0.63. The change and rates of cladogenesis. Genetics press),. North American minnows and Lepomis are AVISE, J. C., AND M. H. SMITH. 1974. Bio- of approximately equal evolutionary age, chemical genetics of sunfish. II. Genic sim- although the minnows are highly speciose ilarity between hybridizing species. Amer. (about 250 species), while Lepomis is rela- Natur. 108:458-472. AVISE, J. C., J. J. SMITH, AND F. J. AYALA. 1975. tively depauperate (11 species). Adaptive differentiation with little genic To compare the amount of genetic dif- change between two native California min- ferentiation in a speciose group and a nows. Evolution (in press). depauperategroup, we have considered two AVISE, J. C., M. H. SMITH, R. K. SELANDER, T. E. LAWLOR, AND P. R. RAMSEY. 1974. Bio- alternative models: (1) genetic differentia- chemical polymorphism and systematics in the tion is a function of time, unrelated to the genus Peromyscus. V. Insular and mainland numberof cladogenetic events; (2) genetic species of the subgenus Haplomylomys. Syst. differentiation is proportional to the num- Zool. 23:226-238. ber of cladogenetic events in the group. AYALA, F. J. 1972. Darwinian versus non- Darwinian evolution in natural populations of According to model 1, the values of D are Drosophila. Proc. Sixth Berkeley Symp. Math. approximately equal in speciose and de- Stat. Prob. 5:211-236. pauperate phylads of comparable age. 1974. Biological evolution: natural se- However, according to model 2, the value lection or random walk. Amer. Sci. 62:692- of D is substantially greater in a speciose 701. AYALA, F. J., AND T. DOBZHANSKY. 1974. Dro- than in a depauperate phylad. Our find- sophila pseudoobscura bogotana, a new sub- ings of about equal average amount of ge- species. Pan Pacific Entomol. 50:211-219. netic differentiation in the speciose min- AYALA, F. J., AND M. E. GILPIN. 1975. Gene nows and in the depauperate Lepomis, frequency comparisons between taxa: support support the notion that time since di- for the natural selection of protein polymor- Acad. Sci. 71:4847-4849. vergence from a common ancestor is more phisms. Proc. Nat. AYALA, F. J., M. L. TRACEY, L. G. BARR, J. F. important than the number of intermediate McDoNALD, AND S. PEIREZ-SALAS. 1974a. Ge- cladogenetic events in determiningthe level netic variation in natural populations of five of genetic divergence between species. Ap- 'Drosophila species and the hypothesis of parently, the development of reproductive the selective neutrality of protein polymor- phism. Genetics 77:343-384. in- isolating mechanisms per se does not AYALA, F. J., M. L. TRACEY, D. HEDGECOCK, AND volve change at a substantial proportion R. C. RiciCMOND. 1974b. Genetic differen- of structural genes. tiation during the speciation process in Dro- sophila. Evolution 28:576-592. BRITTEN, J., AND E. H. DAVIDSON. 1969. Gene ACKNOWLEDGMENTS R. regulation for higher cells: a theory. Science A number of people assisted in collect- 165:349-357. ing. We especially want to thank Dr. CLEGG, M. T., R. W. ALLARD, AND A. L. KAHLER. 1972. Is the gene the unit of selection ? Martin Tracey, Dr. Peter Moyle, and Ms. Evidence from two experimental plant popula- Lori Barr for their help. We extend our tions. Proc. Nat. Acad. Sci. 69:2474-2478. 58 J. C. AVISE AND F. J. AYALA

DAVIDSON,E. H., AND R. J. BRITTEN. 1973. Or- PRAKASH, S. AND R. C. LEWONTIN. 1968. A ganization, transcription, and regulation in molecular approach to the study of genic the genome. Quart. Rev. Biol. 48: heterozygosity in natural populations. III. 565-613. Direct evidence of coadaptation in gene ar- DOBZHANSKY, T. 1970. Genetics of the evolu- rangements of Drosophila. Proc. Nat. Acad. tionary process. Columbia Univ. Press, New Sci. 59:398-405. York. SELANDER, R. K., W. G. HUNT, AND S. Y. YANG. FARRIS, J. S. 1972. Estimating phylogenetic 1969. Protein polymorphism and genic hetero- trees from distance matrices. Amer. Natur. zygosity in two European subspecies of the 106:645-668. house mouse. Evolution 23:379-390. GOTTLIEB, L. D. 1973. Genetic differentiation, SELANDER, R. K., AND D. W. KAUFMAN. 1973. sympatric speciation, and the origin of a Genic variability and strategies of adaptation diploid species of Stephanomeria. Amer. J. in . Proc. Nat. Acad. Sci. 70:1875- Bot. 60:545-55. 1877. HopiiRI, J. D. 1973. Endemism of fishes of SHOTWELL, J. A. (ed.) 1963. The Juntura basin: the Clear Lake region of central California. studies in earth history and paleoecology. Univ. Calif. Pub. Zool. 96:1-135. Amer. Phil. Soc. Trans. 53:1-77. HopRIRK, J. D., AND R. J. BEHNKE. 1966. Ad- SNEATH, P. H. A., AND R. R. SORAL. 1973. ditions to the known native fish fauna of Numerical . W. H. Freeman, San Nevada. Copeia 1966:134-136. Francisco. fish HUBBS, C. L. 1955. Hybridization between STEBBINS, G. L. 1969. The basis of progressive species in nature. Syst. Zool. 4:1-20. evolution. Univ. North Carolina Press, Chapel HUNT, W. G., AND R. K. SOELANDER.1973. Bio- Hill. chemical genetics of hybridization in European UYENO, T. 1961. Osteology and phylogeny of house mice. Heredity 31:11-33. the American cyprinid fishes allied to the AND EVERMANN. 1896. The JORDAN,D. S., B. W. genus Gila. Ph.D. Dissertation. Univ. of fishes of North and Middle America: a de- Michigan. scriptive catalogue of the species of fish-like UYENO, T., AND R. R. MITLER. 1963. Summary vertebrates in the waters of North America, of late Cenezoic freshwater fish records of north of the Isthmus of Panama. Bull. U.S. North America: Univ. Michigan Mus. Zool. Nat. Mus. 47:1-1240. Occ. Pap. 631, 34 p. MILLER, R. R. 1959. Origin and affinities of the freshwater fish fauna of western North 1965. Middle Pliocene cyprinid fishes America. Zoogeography. Amer. Ass. Adv. Sci. from the Bidahochi formation, Arizona. Pub. 51:187-222. Copeia 1965:28-41. . 1963. Synonomy, characters, and varia- WALLACE, B. 1963. Genetic diversity, genetic tion of Gila crassicauda, a rare Californian uniformity, and heterosis. Canad. J. Genet. minnow, with an account of its hybridization Cytol. 5:239-253. with Lavinia exilicauda. Calif. Fish and Game WILSON, A. C., L. R. MAXSON, AND V. M. SARICH. 49:20-29. 1974a. Two types of molecular evolution. 1965. Quaternary freshwater fishes of Evidence from studies of interspecific hybrid- North America. p. 569-581. In The Quater- ization. Proc. Nat. Acad. Sci. 71:2843-2847. nary of the United States, Princeton Univ. WILSON, A. C., V. M. SARICH, AND L. R. MAXSON. Press, Princeton, New Jersey. 1974b. The importance of gene rearrangement MOYLE, P. B. 1974. Inland fishes of California. in evolution: evidence from studies on rates Univ. Calif. Press (in press). of chromosomal, protein, and anatomical NEI, M. 1972. Genetic distance between popu- evolution. Proc. Nat. Acad. Sci. 71:3028- lations. Amer. Natur. 106:283-292. 3030. NEVO, E., Y. J. Kim, C. R. SHAW, AND C. S. ZOUROS,E. 1973. Genic differentiation associ- THAELER. 1974. Genetic variation, selec- ated with the early stages of speciation in tion, and speciation in Thomomys talpoides the mulleri subgroup of Drosophila. Evolu- pocket gophers. Evolution 28:1-23. tion 27:601-621.