EVOLUTIOKARY GENETICS OF IKSULAR ADRIATIC LIZARDS

GEORGEC. GORMAN Department of Biology, University of California, Los Angeles, Los Angeles, California 90024

MICHAELSOUL^ AND SUHYUNGYANG' Department of Biology, University of California, Sun Diego, La Jolla, California 92037

AND EVIATARNEVO Biology Department, University of Haifa, Haifa, Israel

Received March 16, 1974

The use of electrophoresis to study sicula is found on virtually every island in allozyme variation allows the evolutionary the Adriatic. The pattern of distribution in biologist to compare various genetic param- places is completely interdigitating, but eters of populations. In this paper we sympatry is completely absent on islands employ this technique in an analysis of the smaller than several sauare kilometers. On ev61utionary genetics of the genus Lacerta some large islands both species can be in the Adriatic region. We compare genetic found, but there is usually parapatry or similarity among populations and species only marginal sympatry. Second, both L. to obtain a quantitative estimate of related- melisellensis and L. sicula have undergoneu ness. We compare genetic variability on considerable evolution in size, color, and islands and on the mainland to examine pattern. Thus Mertens and \Yermuth genetic consequences of insularity. \Ye (1960) list 18 subspecies for Lacerta examine the relative contributions of island melisellensis. and no fewer than 39 for L. size, habitat complexity, and various sicula. The subspecies controversy aside, measures of isolation to the determination there is no doubt that some of the insular of the observed levels of heterozygosity. isolates are very distinctive. For example, Finally, we attempt to estimate the relative there are populations of L. 7nelisellensis importance of selection and genetic drift, comprised of large (adult males > 10 and apply our findings to the question of grams) jet-black lizards, others of small allelic neutrality. brown lizards (adult males -- 5 grams) with The Adriatic waters of Yugoslavia are a solid green middorsal area, and still others dotted with more than 1000 islands of that are highly patterned. In addition to diverse size and habitat complexity, ranging being polytypic, many populations are from rocks barely emergent from the sea to highly polymorphic for dorsal pattern and substantial islands measuring hundreds of for ventral coloration (salmon, cream, yel- square kilometers. Lacertid lizards are low, and pale blue may exist in a single abundant on most of these islands and the population). adjacent coastal strip. They have interested Lacerta melisellensis and L. sicula appear evolutionists for two reasons: First, there to be of broad ecological tolerance occur- is a classic case of competitive exclusion ring in a variety of terrestrial situations. (RadovanoviC, 1956, 1959 cited in hlayr, They are most often collected in shrub and 1963). Either Lacerta melisellensis or L. grassy areas, and also around small stones.

A third species enters marginally- into the Present address: Museum of Vertebrate Zoology, University of California, Berkeley, Calif. present study. L. oxycephala is restricted to 94720. large boulders and rock faces on the main- EVOLUTION29: 52-71. March 1975 52 E\'OLUTIONARY GENETICS 53 land and on the islands. Unlike the other Populations Sampled two species, it does not show inter-island 1) Lacerta sicula. variation, nor is it polymorphic within a) L. s. campestris.-This is one of the populations. Xo subspecies have been more widely distributed races of L. sicula. designated (Mertens and n'ermuth, 1960). Population samples were collected at Zadar L. oxycephala is broadly sympatric with L. (n =30) and Trogir (n = 30), Yugo- melisellensis in the southern portion of the slavia; and Pescara (n= 30)) Italy (all latter's range, and has been found on at mainland). All collections were made within least one island in sympatry with L. sicula an area < 1 km sq. (Xevo et al., 1972). What we now call L. b) L. s. pe1agosae.-This form is endemic melisellensis and L. sicula have historically to Palagruia (n = 16). been considered closely related in con- c) L. s. cazzae.-Three island popula- ventional taxonomic terms (even con- tions were sampled: SuSac (n = 34), specific), and L, oxycephala is considered KopiSte (n = 19)) and Pod KopiSte (n = distant from them (Boulenger, 1920). 30). Recently, Arnold (1973) split the genus Lacerta into several genera, leaving oxy- 2) Lacerta melisel1ensis.-(islands only, cephala in Lacerta and placing melisellensis we were unable to collect a mainland and sicula in Podarcis. For convenience series). only, we retain the older nomenclature. a) L. m. curzo1ensis.-Collected at grnovo, island of KorEula (n = 14) ; within an area < 1 km sq. b) L. m. 1issana.-Collected at Pasadur, Species Studied island of Lastovo (n = 24) and Pod AlrEaru 1) Lacerta sicu1a.-This species occurs (n = 34)) a tiny island near Lastovo; the in Italy and along the Adriatic coast of town of Komiia, island of Vis (n = 17), Yugoslavia. Its present distribution in and Greben (n = 30), a small island adja- northern Yugoslavia suggests that it has cent to Vis; and BiSevo (n = 28), a mod- been a relatively recent invader, having erately large island west of Vis. come from north-eastern Italy around the c) L. m. gigas.-This subspecies is Adriatic. Its occurrence on some south- endemic to the tiny island of Mali Parianj, central Adriatic islands, and its absence adjacent to the east coast of Vis (n = 11). from the Yugoslav mainland at these lati- d) L. m. kammereri.-This subspecies is L2 tudes suggests direct over-water coloniza- endemic to the tiny island of Mali Barjak, tion of these islands (Palagruia, SuSac, adjacent to the west coast of Vis (n = 7). KopiSte, Pod KopiSte) from Italy. L. sicula e) L. m. digenea.-This form is endemic is also present in southern Yugoslavia, its to Svetac, west of Vis and BiSevo (n =32). presence there is presumably the result of f) L. m. melisel1ensis.-The nominate another over-water colonization. About race is endemic to Brusnik near Svetac one-third of the described subspecies are (n = 28). on islands in the eastern Adriatic. g) L. m. ga1vagnii.-A race endemic to Kamik near Svetac (n = 20). 2) Lacerta melisel1ensis.-This is an h) L. m. pornoensis.-Endemic to the autochthonous species to the coastal regions tiny isolated island of Jabuka west of of Yugoslavia. Of the 18 named subspecies, Svetac (n = 18). one is on mainland Yugoslavia, the remain- der are all insular isolates. 3) Lacerta oxycepha1a.-Only 7 indi- viduals were studied, three from Lastovo 3) Lacerta oxycepha1a.-This species is and two each from Greben and Vis. also restricted to Yugoslavia, occurring on As we are not concerned with nomen- the Dalmatian coast and islands. clature at the subspecific level, the popu- 54 G. C. GORMAN ET AL.

lations hereinafter are primarily referred to scorable for all populations of all species by locality only. Figures 14show collect- and Sn calculations are based only on these ing sites, and Table 1 presents data on the 19 loci. localities.

Electrophoresis Allele frequency data for 20 populations Specimens were collected and transported (oxycephala treated as one population) are alive to the laboratory where they were presented in Table 2, and levels of poly- placed in a -68C freezer. Before processing morphism and heterozygosity are presented for electrophoresis, the head, feet, and skin in Table 1. Data on genetic similarity (SR) were removed for morphological studies. appear in Table 3. After removal of the gut, the body was homogenized in two volumes of grinding 1) Taxonomic Comparisons buffer (0.1 M Tris, 0.001 M EDTA, and 5 X M NADP, pH adjusted to 7.0 with The similarity coefficients in Table 3 con- concentrated HCl) and centrifuged with firm the morphologically based taxonomy 0.5 ml of toluene at 18,000 r.p.m. for 30 for the species level. Within-species simi- minutes. The supernatant was removed and larities range from .84-1.0 among L. frozen at -68C. melisellensis populations and .88-.99 for L. Starch gel electrophoresis was carried out sicula. Between species similarities for this with the same proteins and procedures pair show no overlap with the above, rang- described by AlcKinney et al. (1972) for ing from .61-.74. L. oxycephala is less 22 loci with the following exceptions: closely related to L. melisellensis and L. Albumin was not electrophoresed; 6-phos- sicula than the latter are to each other. The phogluconate dehydrogenase (6 Pgd) was mean similarity value is .42 with sicula and demonstrated with a Tris maleate EDTA .44 with melisellensis. This confirms the buffer, pH 7.4, and isocitrate dehydrogenase results of a preliminary study (Gorman, (Idh) with a Tris citrate buffer, pH 8.0. 1972). Fumarase (Fum) was demonstrated with These data are relevant to the general a Poulik buffer system, pH 8.1, for the problem of how similar congeners can be tray buffer, and pH 8.6 for the gel buffer. genetically, and are relevant to the question Heterozygosity estimates are based on of whether there need be a "genetic revolu- actual counts of presumed heterozygotic tion" following speciation. Two races of genotypes. Mean heterozygosity (H) is Danish house mice (Mus ~nusculus) have defined as the number of presumed hetero- average SE values between them of about zygotic genotypes recorded in a sample 0.50 (Rogers, 1972) and disjunct popula- divided by the product of the number of tions of the lizard Anolis carolinensis from individuals and the number of loci sur- Texas and the Bahamas have a value of veyed. Polymorphism estimates are based 0.68 (Webster et al., 1972). Within the on the number of loci with more than a relatively tightly-knit roquet species group single allele divided by the total number of of Anolis, between species SR values are as loci. In this study a locus is considered high as 0.70 for sympatric congeners and as polymorphic if any variation was found. low as 0.37 for the least similar pair in the In our analyses of genetic variation we group (Soul6 and Gorman, unpubl. data). primarily use H as our dependent variable Thus Lacerta melisellensis and L. sicula are because it is less sample size dependent more similar to one another than are so- than percent polymorphism. called subspecies of certain mammalian and Genetic similarity between populations reptilian species, and even the "distantly was estimated with the Rogers coefficient related" L. oxycephala seems genetically (SR) (Rogers, 1972). Only 19 loci were closer to its congeners than are members TABLE1. Biogeogvaphic and electvophovetic data fov the Lacerta populations.

- .------~ ------l'erccnt:' Areal Elevation1 Channel' RIean Standard loci (sq. km) (m.1 Tlistanre' in km squared (source) dcplli heterozygosity rrror polymorphic ------L. melisellensis Kortula 2 .89 (mainland) 36 Lastovo 900 (mainland) 66 Pod Mrearu 0.64 (MrEara) ; 5.76 (Lastovo) (66)66 Vis 256 (Hvar) 89 Greben 1.0 (Vis) (89)38 Mali Parfanj 0.81 (Vis) (89129 ~MaliBarjak 0.36 (Vis) (89) ? BiSevo 16.81 (Vis) 97 Svetac 361 (BiSevo) - 100 Brusnik 256 (BiSevo) ; 12.25: (Svetac) - 100 Kamik 1.69 (Svetac) - 100 Jabuka 529 (Svetac) 144

L. sicula 13) Zadar ...... 8.41 (1.14) 14) Trogir ...... 12.85 (1.05) 15) Pescara ...... 5 .88 (0.75) 16) Palagrufa 0.3 91 2,601.0 (mainland) - 110 5.65 (0.89) 17) SuSac 4.6 243 9,801 (mainland) ; 1,764 (Palagrufa) - 160 2.94 (0.49) 18) KopiSte 1.O 93 196 (SuSac) - 130 4.76 (0.89) 19) Pod KopiSte 0.05 30 0.36 (KopiSte) (- 130)20 4.40 (0.46)

L. oxycephala 20) Pooled sample ...... 5.26 (1.55) 2 1 ------. ------I Data on island area, elevation, and distance obtained from The Yugoslav Coast, Guide Book and Atlas, Yugoslav Lexicographical Institute, Zagreb, 1966; and the following nautical charts: Jabuka-Vis, 1:100,000, # 100-22 ; Sibenik-0. Lastovo, 1:200,000, # 153; both published by Hidrografski Institut Jugoslavenske ratne mornarice, Split; and Promontorio del Gargano to Ortona, 1:167,520, #H.O. 3952, U.S. Navy Hydrographic Office. Elevations of Mali Parianj and Mali Barjak are estimates, as they do not appear on the charts. 'Channel depths in meters based upon hydrographic charts listed above. The depths in parentheses are those of the parent island. "Estimates of polymorphism rounded to nearest whole number. 56 G. C. GORMAN ET AL.

'0 '"I .h '0 . .0 .0 ' . . .q .q ...., .q .m :? : : :zG%., . ., ., . . I

MN"" "?"0"' 'NM '0 '0 'MNO ' ' ' 0‘0 ....., . .* ....0q .; ., .cad., ... EVOLUTIONARY GENETICS 5 7

*0""' ...... ,'0 '. .*'0, ..' ' .a,?'mh .r;'0 ..' ' :63 :q : :xx

0""'..... '0., .' .*'0 ..' ' .?N'mh .,q'MN .' .,q'hm :z: :XX

0 " '00 ' '0, ' '"h * , . :w :ex :q :z :z : : : : : .,* . .- . .la, 5 8 G. C. GORMAS ET AL.

TABLE3. Rogers' (1972) coefficients of genetic similavity for all the Lacerta populations. The population number key is given in Table 1. Decimals are omitted from the values, which should be multiplied by lo-'.

within a tightly-knit species group in Vis are placed in the same subspecies (L. m. another genus of lizards. lissana) and the Kortula form is placed in The within-species similarities suggest another subspecies (L. ?n. curzolensis) . some relationships that depart from present However, the SR data indicate that the taxonomy. The populations of Lastovo and Lastovo population is slightly more similar to KorEula than it is to Vis, whereas the populations most similar to Vis are on its fringing islands, and on all the islands west of Vis extending to Jabuka (Figs. 2, 3). Thus at least six designated subspecies are more similar electrophoretically to L. m.

MIL' BAPJAK ...' GPESEN 0 ,

ITALY a'.,

FIG. 2. Vis and its fringing islands. Black areas of pie diagrams indicate percent hetero- FIG. 1. The Adriatic coast. Mainland collecting zygosity for populations of L. melisellensis; sites and some of the major islands are indicated. complete black would represent 10%. EVOLUTIONARY GENETICS

I 0- 2 4 6 8Km

FIG. 3. Vis to Jabuka transect. For proper scaling, several islands had to be displaced, crosses indicate true latitudes and longitudes. Island sizes are drawn to scale. True minimum distances between islands are indicated where displacement would be misleading. Black areas of pie diagrams represent percent heterozygosity for populations of L. melisellensis; complete black would represent 10%.

6' 40'E I Pod yr;oru

0 8 C \,. -. i,. _ --,-- ~ .~ " . . , 4Is53'kJ ITALY ,/ -. \*,,Plc 5 irnnceico

FIG. 4. A portion of the Italian mainland and islands inhabited by L. sicula. In addition, two islands that have L. melisellensis are shown (Lastovo, Pod Mrraru). For proper scaling, some islands had to be displaced. Crosses indicate true latitude and longitude. True minimum distances are indicated where displacement would be misleading. Land mass sizes are drawn to scale. 60 G. C. GORMAN ET AL. lissana from Vis than is the latter to the L. melisellensis and L. sicula are same so-called subspecies from Lastovo. ecologically similar. If there is occasional The L. sicda comparisons generally con- immigration, then hybridization and intro- firm our expectations while illuminating gression might also be encountered occa- some problems. First, the SuSac, KopiSte, sionally. It is less likely that rnelisellensis and Pod RopiSte populations are very and oxycephala would hybridize, for they similar. The island of Palagruia, a possible are broadly sympatric, ecologically distinct, stepping stone, is situated between these and presumably have well established iso- islands and the Italian mainland (Fig. 4), lating mechanisms. It is, of course, not yet the SuSac-KopiSte group seems more unlikely that the "shared" allele in this case closely related to Pescara in Italy and Zadar is not the same protein, but merely has the in northern Yugoslavia than to the Pala- same electrophoretic mobility. gruia or the nearby Trogir population. Little is known about the relationship This strengthens the supposition that these between genetic distance and successful populations were founded by Italian sicula. hybridization in reptiles. The best docu- It also suggests that the Palagruia coloniza- mented case is the hybridization of Anolis tion may have been independent of the trinitatis and A. aeneus on Trinidad. SuSac-KopiSte colonization and not a step- Although the F1 is predominantly sterile ping stone for the latter. (Gorman, et al., 1971), we now have sev- eral examples of backcrossed individuals, 2) Interspecific Intro,tyession? indicating that introgression is possible (Gorman and Yang, 1975). Their simi- The small island of Pod MrEaru lies very larity coefficient is 0.72, which is about close to Lastovo (Fig. 4). There are two the same as the most similar Lacerta striking peculiarities about the genetics of melisellensis-L. sicula populations (Gorman the Pod MrEaru melisellensis population. and SoulC, unpublished data). First is the relative lack of similarity to Lastovo (Table 3). The S, value is the 3) Amounts of Genetic Variation lowest among the fringing island-parent island comparisons (see below). Second, A) Mainlands vs. islands.-Our collect- from Table 2 we see that the Pod RIrEaru ing efforts for mainland Lacerta were population has three presumed alleles that successful only with L, sicula, which were are unique among rnelisellensis populations extremely abundant at mainland sites. The (and unique alleles are uncommon among mainland populations of L. sicula in Yugo- the populations studied). slavia and Italy are highly polymorphic and Two of these alleles are electrophoret- heterozygous for the systems we surveyed. ically identical with alleles found in L. Half the loci at Trogir and about a third at sicula (Est-4band Gp-la), and one (Gp-IC) Pescara were polymorphic. The lowest is found in oxycephala which is sympatric percent polymorphism on the mainland with rnelisellensis on Pod MrEaru. At the (Zadar) was greater than the maximum for latitude of Lastovo, Pod MrEaru is at the an island population (Palagruia) . Hetero- species border. All islands to the west have zygosity was greatest at the three main- L. sicula, islands to the east have L. land localities, and lowest on the four melisellensis. The evidence is circum- islands. Mean heterozygosity for the three stantial, but hybridization and introgression mainland populations (giving equal weight are strongly implicated. It is perhaps more to each population) is 9.05%, while mean than coincidence that the S, values between heterozygosity for the four island popula- ~nelisellensisand sicula are highest for Pod tions is 4.33%. Palagruia was only slightly MrCaru rnelisellensis compared with sicula lower than Zadar in polymorphism and just from Pod KopiSte, the nearest sicula island below Pescara in heterozygosity (Table 1 ) . (Table 3 ) . Highest variability was found in sicula EVOLUTIONARY GENETICS 61 from Trogir. The distribution of sicula km) only 0.6 km from KopiSte. As shown along the Yugoslav coast has a gap that in Table 1, the heterozygosity estimates are begins just south of Trogir and extends virtually the same: the KopiSte population south for several hunderd kilometers. A is polymorphic at an additional locus. separate invasion from Italy probably KopiSte has two unique alleles (6-Pgd-b, accounts for the southern populations; thus Gp-3-C) not found in any of the other the gap is not believed to be the result of Lacerta populations. Pod KopiSte had a extinction, and the Trogir population is rare allele 6-Pgd-" that was absent on presumably near the moving front of an KopiSte, but present in all other L. sicula invasios. One might be surprised to find populations, and might be assumed to be that this peripheral mainland population present, but rare on KopiSte. Otherwise, is the most variable of all populations sam- they were fixed for the same alleles, or they pled. Power (1971), however, studied shared polymorphisms. morphological variation in Brewer's black- We studied three fringing populations in bird and found that a recently invaded the vicinity of the large island of Vis: Mali region had birds with as much variation as Barjak, Mali Parianj and Greben. All are birds from the original range. What little within one kilometer of Vis. The two Malis evidence exists, therefore, suggests that are truly tiny ("mali" means small in successful overland range expansions need Serbo-Croatian), each less than 0.01 sq. not be accompanied by genetic depauper- km, and they are relatively flat. Greben ization at or near the periphery. is not only larger (0.75 sq. km), but much Although we lack mainland melisellensis, more diverse topographically, rising mod- the largest melisellensis islands (Korcula, erately steeply on the west side, and drop- Lastovo, and Vis) might be considered ping precipitously on the east.. Mali Barjak approximations of the mainland ecological was the only population sampled with zero condition. This is almost certainly true for variability, and Mali Parianj was very low Korrula which is less than 2 km from the compared to Vis (although not as low as mainland and is separated by a shallow some of the most remote islands). Our water channel. If this is so, it appears that sample sizes were too small on these islands there is slightly less variability in "main- to account for many of the possible rare land" melisellensis than mainland sicula. polymorphisms, but there can be little itre also note that variability is highest in doubt that the low heterozygosity is real these three largest melisellensis islands and significantly different from Vis. (plus Greben, see Fringing Islands, below) Greben has less than 1% the land area and considerably lower on the smaller of Vis, is slightly farther from Vis than the islands. Thus the pattern is similar for both two tiny islands, yet had the same vari- melisellensis and sicula. ability as Vis. Some rare alleles not in our B) Fringing islands and the "s~nall \'is sample were found on Greben (Pgm-lC, island effect."-Our data allow us to com- a GP~-~),but one of these appears in the pare the genetic variation in fringing island Lastovo sample, the other in BiSevo, and it and "parent" island populations. We define is not unlikely that they would be found in a fringing island as a small island that Vis with additional sampling. probably has been disconnected by eustatic The two small islands of Brusnik and sea level rise from the adjacent larger island Kamik might be considered fringing islands within the recent geological past, as evi- of Svetac. However, neither of their lizard denced by shallow channel depths. In all, populations show particularly strong affin- seven fringing island populations associated ity with Svetac (Table 3). The Brusnik with four parent islands were sampled. The population is equal to Svetac in poly- only L. sicula fringing population studied morphism and higher in heterozygosity. It was Pod KopiSte, a small island (0 05 sq. is considerably larger than Kamik (0.05 62 G. C. GORMAN ET AL. vs. 0.01 sq. km) and is shown on the hydro- We dislike the second explanation graphic charts to be surrounded by because we have calculated elsewhere extremely shallow water, so that in rela- (SoulC et al., in prep.) that colonies estab- tively recent times Brusnik was perhaps lished by a single, fertilized female should, twice as large as at present. The Kamik on the average, retain about 34% of the population shows reduction in polymor- polymorphisms in the source population, phism compared to Svetac and at loci and these small island populations do not. (Pgm-2 and Got-1), it is fixed or predomi- Unfortunately, the data do not permit us nant for alleles that are relatively rare on to distinguish between the drift and Svetac. selection hypotheses. Jabuka is not a fringing island. It is C) Retained polymorphisms: Are the isolated and. surrounded by deep water and allozymes neutral?-Although genetic drift almost certainly has not had recent con- might account for the possible loss of alleles nection with the other islands. Thus the in the smallest fringing populations, this in low variability on Jabuka is in accord with no way implies that the alleles are behaving the results for the other very small island neutrally. In this section we try to estimate populations, but other factors, such as the proportion of polymorphic loci for founder effect, might be the real cause of which alleles could be effectively neutral. low variation. The major difficulty in such analyses The fringing island of Pod MrEaru has was succinctly stated by Lewontin and lower variability than Lastovo, but more Krakauer ( 1973) : "Any observed distri- than we might expect from its size. How- bution of gene frequencies over space or ever, introgression with sicula, as stated time, if considered to be in a steady state above, might well account for the elevated can be explained by a suitable.choice of N, levels of heterozygosity and polymorphism. m and p, with s being made arbitrarily From these results it appears that we small. . . ." For example, if many popula- can define a "small island effect" in which tions are sampled and the gene frequencies variability is markedly reduced from the are everywhere very similar this can be parent island. This effect is limited to tiny attributed to strong, uniform selection islands indeed (0.01 sq. km and less) and is throughout the range. On the other hand, a not seen on islands as small as 0.05 sq. km. neutral model can be applied by assuming A similar phenomenon has been found for that the effective migration between neigh- Anolis cristatellus on the fringing islands of boring populations is of the order of one Puerto Rico and the Virgin Islands (SoulC individual per generation (Wright, 195 1 ) . and Gorman, unpubl. data). We have no Lewontin and Krakauer resolve this conun- rigorous estimates of population size on such drum by pointing out that there is one tiny islands, but based on collecting experi- feature of the drift-migration-neutrality ence our estimates range from 100 to about hypothesis that produces a different result 1000. than does a selection model: "While natural We have considered three explanations selection will operate differently for each for the severe reduction in genetic variation locus and each allele at a locus, the effect of on tiny islands: (1) Genetic drift is likely breeding structure is uniform over all loci and its impact would be even greater if in and all alleles." That is, if for some loci the past the population went through a there is spatial or temporal gene frequency bottleneck, such as would occur during heterogeneity, while for others there is uni- prolonged drought; (2) extinction of the formity in gene frequencies, and therefore, original relict population followed by over- all loci together form a heterogeneous or bi- water colonization by one or a few indi- modal distribution, then it is safe to assume viduals from the large island; (3) strong that the neutral model is not a sufficient directional selection on very small islands. explanation. In other words, if drift, migra- EVOLUTIONARY GEXETICS 63

tion and neutrality explain the constancy of We analyzed seven polymorphic loci. gene frequencies at the uniform loci, then The Vis population was scored as mono- selection, acting differently in different morphic at one of these loci (N-Gpd), but places or times, must account for those loci we expect that the rare allele is actually which have heterogeneous gene frequencies. present at a low frequency because of its Alternatively, if drift is assumed to account occurrence on both Greben and BiSevo. for the loci which have heterogeneous gene The same is probably true for the rare freqeuncies, then uniform selection must alleles in many of the populations. The explain the loci which are geographically presence or absence of rare alleles is sam- or temporally uniform. Lewontin and pling error and should not seriously affect Krakauer tested for heterogeneity in the the results. gene frequency distributions for a given The results are given in Table $. The locus by estimating the parameter F,, the ratio of the observed variance of F to the "effective inbreeding coefficient." The expected variance is .06981/.01177 = 5.93 estimate is: (P < .001). (In calculating u2 we set k = 2, a conservative test. Degrees of freedom were determined as per Lewontin and A when F, = estimate of effective inbreeding, Krakauer.) Some loci, at least, apparently s2, = variance of one of the two alternate are selected. Aside from the data in Table alleles from population to population, and 4, chi-square analyses (Snedecor, 1956, p. $ = mean frequency of the allele over the 227) suggest that only two or three loci ensemble of populations. Next, the ratio of are significantly heterogeneous geograph- the observed variance over all loci to the ically, the two esterases and possibly Got-1. theoretical variance (see Lewontin and The null hypothesis is accepted in five Krakauer, 1973), s2p/u2, is calculated. A of the seven cases (P for rejection 0.01). high and statistically significant ratio is The results are as follows: Pgi, x2= 3.70, taken as evidence for heterogeneity of the P > 0.40; Got-1, x2= 10.15, 0.05 >P > 0.02 ; Got-2, = 1.16, P 0.80; N Gpd, values and for selection at some of the x2 > fl = 1.87, P 0.70; Pgm-2, = 4.1 7, loci. x2 > x2 P > 0.30; Es-1, = 94.65, P < 0.01; Es-4, We chose the most homogeneous set of x2 = 44.53, P 0.01. The alleles at the populations-those on Vis, its fringing x2 < island, Greben, Mali Parianj and Mali other loci comprise a "conservative" set. Barjak, and the population on Biievo, Three interpretations of this result are which, though more than 4 km from Vis, possible. First, neither set of alleles is was connected to it about 10,000 to 15,000 neutral; the geographically unstable loci years ago. During the last glaciation all of manifest the effects of different selection these islands were part of a single, emergent regimes on different islands, whereas the land mass. Sea level reached its present conservative loci are everywhere being height about 5000 years ago. The best stabilized by the same balancing selection guess, then, about the ages of these popu- forces. Second, alleles at the geographically lations is that they have been isolated for unstable loci are relatively neutral and are at least 1000 to 2000 generations. To drifting, whereas the conservative loci are include populations such as those on the responding to selection as in the first more distant islands of Svetac, Brusnik, Kamik and Jabuka, might weaken the interpretation. The third possibility is that analysis because there is a distinct like- there is sufficient gene flow between islands lihood that they have not been so recently to swamp the effects of selection differen- connected to Vis, hence their lizards might tials for most loci; the other loci, the two not be samples of the same original gene esterases and possibly Got-1, are diverging pool. in the face of the gene flow because they are 64 G. C. GORMAN ET AL.

TABLE4. Estimates of effective inbreeding coefficients for seven polymorphic loci.

Locus and allele frequency Population Vis Greben Mali Parianj Mali Barjak BiSevo s2n P responding to very strong and very different estimates for these loci are artificially low, selection pressures. the rare allele being present in only two of The last interpretation is the least the five populations for the first four alleles probable. It is difficult for us to accept in Table 4. Nevertheless, for three of these that lizards are migrating across the chan- four cases, the frequencies at the rare alleles nels at the rate of one per generation (the in the two populations where they were generation time is about one per year), the detected are virtually identical. In contrast, rate required to maintain effective panmixis if, for esterases, one chooses at random in semi-isolated colonies (Wright, 1951; pairs of populations where the allele is Maruyama, 1970). The three islets are neither fixed nor absent and examines the uninhabited and are rarely visited by man. frequencies, the gene frequency differences The actual rates of immigration are more will average about 0.3 1. likely to be two orders of magnitude less D) Biogeographic correlates of variation. than that required for panmixis. -Here we address one of the central prob- Both the first two interpretations require lems of contemporary population genetics- non-neutrality at most of the loci. Circum- the determinants of genetic diversity in stantial evidence, therefore, weighs in favor populations. A perusal of Figure 3 and of a selectionist interpretation of the pat- Table 1will show that the populations of L. terns, at the majority, but definitely not all melise22ensis on the more remote islands of the loci. The esterase alleles, especially, have lower values. This simple pattern could well be effectively neutral. is, however, confounded by a negative One source of bias in the above analysis is correlation between distance and island uncorrected. Many of the loci were fixed area; in general, the more remote an island, in the smaller island populations (for t$e the smaller it is. Hence, the smaller an common allele in all cases). Hence, the F, island the lower the H value. Which is more important, distance or size? Multiple TABLE5. Correlation coefficients from m?~lti$le regression is a statistical tool for deter- regression analyses. The coefficients above the mining the importance of individual vari- diagonal are from analyses 111, IV and V, those ables in such multivariate problems. The below from analysis 11. independent variables (IV's) we chose to test were:

A - .66 -.42 -.83 - .63 A = area of the island: A' .66 - -.25 -.91 - .91 A' = log of area of the island; D2 -.07 .37 - .2 1 - .08 M -.77 -.48 .13 - - -,92 E = maximum elevation of the island; E- .78 .93 .29 -.60 - D2 = the distance squared between the H .61 .67 .24 -.78 .72 - island and the mainland or, in the EVOLUTIOK.4RY GENETICS 65

TABLE6. Coefficients of multiple correlation, R2, f~omfour stepwise multiple ~egressiofzaizalyses of the biogeographical correlates of genetic variation of Lacerta melisellensis. Boldface IV's indicate a negative partial vegression coefficient.

Dependent Variable Populations ff B P I1 All excluding M 0,61*a Pod MrEaru MD' 0,72*" MD2A' 0.76* MD2A'A 0.78* MD'A'AE 0.79

111' Iv1,2 V1 Non-fringing only M 0.8 ja* A' 0.83" A' 0.68" MD2 0.92" A'D' 0.93" A'D2 0.82 MD2A' 0.97* A'D2A 0.95 A'D2A 0.86 A'D2AM 0.87 E not included in analysis. 'M not included in analysis. * P < 0.05. ** P < 0.01.

case of a small island, distance were deleted because they are probably squared to the likeliest source island recent isolates of the source population, and (see Table 1 ) ; are therefore an additional source of hetero- M = channel depth: the sea level drop geneity. In analysis IV the variable M was in meters that would create a land deleted. The product-moment correlation land bridge between the island and coefficients among the variables are given the mainland. in Table 5. L. sicula populations could not be treated in a similar manner because the The first three IV's are estimates of island sample size of islands is too small (fewer size and, by correlation, habitat diversity than the number of "independent" vari- (see below), while the latter two IV's are ables). estimates of isolation. We use D2 rather The results excepting analysis I are given than distance itself because if colonization in Table 6. The first and second analyses plays any role, the rate is probably a give very similar results, although removing decreasing exponential function of distance Pod hlIrdaru gives slightly higher coeffi- (MacArthur and \Vilson, 1967). We use cients of multiple correlation, R2 values. M because it may be correlated with the An impressive improvement in R2 values time since the population was isolated by occurs upon removal of the fringing popu- rising eustatic sea levels following the last lations (analyses 111, IV and V) . Virtually glaciation. all of the variance of H values is explained Five analyses were performed. The first by the three variables M, D2, and A', with four used H as the dependent variable; the M alone accounting for 85% of the variance. fifth used P. Analysis I employed all of the In this stepwise design, M is the best L. melisellensis populations. Analysis I1 predictor of heterozygosity. A', however, was like the first except that the Pod is virtually as good, and the combination blrdaru population was deleted because its A'D2 gives a slightly higher R2 than MD2. H estimate is probably a result of inter- When P is the dependent variable, A' is the specific hybridization. In analyses 111, IV best predictor and A'D2 again is the best and V, all of the fringing island populations two-variable combination. 6 6 G. C. GORMAN ET AL.

Interpretation is hazardous, especially tion at most of large island and mainland because of the small sample size. We are lizards. encouraged, nevertheless, by the similarity The radically different ecology of small between these results and those from an islands eventually results in predictable analogous study of side-blotched lizards evolutionary changes in behavior, size, (Uta) on islands in the Gulf of California, color, scales and other characters (Mertens, Mexico (Soul6 and Yang, 1973). In the 1934; Soul6, 1966). Evolutionary rates former study we found that the same two- seem to be particularly high for the small variable combination, A'D2, gave the island Lacerta. This is more easily demon- highest R2 values. Equally good was the strable for color and size than for allozymes. combination MD2 where M was the number Thus Jabuka, Kamik, and Brusnik may of sympatric, confamilial lizard species. In have achieved melanism and relatively both the Uta and Lacerta studies A' large body size independently, but even if accounted for about 75 to 85% of the they form a single stock, Mali Barjak is variance in heterozygosity or polymorphism, tending in the same direction with a popu- and DGccounted for about 10%. lation that is distinctly larger, darker and Our hypothesis for interpreting the Uta less patterned than on the parent island of results (which we have called the "time- Vis. divergence theory of variation") was that What we postulate, then, for these (1) genetic variation is lost as a conse- Adriatic Lacerta is differential rates of loss quence of directional selection at rates of alleles, the rate being proportional to the proportional to average evolutionary rates; ecological distinctness of the island com- (2) evolutionary rates of island reptiles pared to the mainland. The populations are everywhere seem to be inversely propor- apparently not old enough tr, have accumu- tional to island size, and the reason for this lated new electrophoretic variation through is that (3) the relative ecological distinct- mutation. ness of an island in comparison to the main- The reasons we do not invoke genetic land is much greater on small islands than drift to explain the correlation of H and A' on larger ones. Specifically, we argued that are (1) the allele frequency data do not the ecological dissimilarity, or relative support a drift effect except on very tiny simplicity, of an island compared to the islands, none of which are included in mainland can be rather crudely estimated analyses 111, IV and V, and (2) genetic by the log of island area, or by the number drift is not a satisfactory explanation for of sympatric confamilial species. The the intermediate values on intermediate former estimates distinctness, we argued, sized islands such as BiSevo and Svetac; because the log of habitat diversity might these islands have lizard populations, per- be expected to increase linearly with the haps 10"o lo5, too large to be subject to much drift. log of area in an analogous fashion to the A confounding problem in this study is species-area relationship. In qualitative the correlation between the two variables terms, the small islands in both the Gulf A' and M (r = 0.91). This is a geographic of California and the Adriatic are exceed- coincidence, a consequence of the absence ingly arid and rocky with a sparse, unique of large, distant islands in the Adriatic. vegetation and a maritime climate. Other Hence, the importance of M in analysis I11 important features are the virtual absence could easily be an artifact. of predators, a highly modified insect Another possible explanation for the fauna, and the occasional overwhelming importance of M is that it is an estimate of presence of sea bird nesting colonies (not island age; i.e., the deeper the channel, the in the Adriatic, however). This distinct longer the island in question has been cut set of conditions affects only a tiny frac- off from a larger island or the mainland. EVOLUTIOK.4RY GEKETICS 67

This, of course, assumes the validity of the time, respectively, are so highly correlated commonly accepted figure of 100 to 150 with each other, we cannot distinguish meters for the eustatic sea level rise follow- between these two factors or allocate to each ing the last Pleistocene glacial period. As the fraction of the variance that it might stated above, Jabuka is the only island explain. What we can say is that these are included in this study that probably has the two variables that would be expected to not been connected to the Yugoslav coast be important given the general validity of during the last glaciation. Only Jabuka, the time-divergence hypothesis. and perhaps Svetac and its fringing islands, Some technical points of comparison could have populations isolated for more between the Uta and Lacerta studies remain than about 15,000 years. to be noted. In the Uta study we con- It is, of course, possible that all of the sidered that the variables that best repre- island populations stem from post-Pleis- sented time were (1) phenetic distance tocene, over-water colonizations, since from a mainland population (Q), and, persistence of the island during glaciation perhaps (2) D2, a measure of isolation. does not mean persistence of these helio- Finally, M in this study and F or F' in the thermic lizards. There are other reasons, Uta study, although they are all channel however, for rejecting this hypothesis. depth variables, are not strictly com- First, island populations such as Vis and parable. All of the islands in the Lacerta BiSevo have higher levels of electrophoretic study are strictly continental (connected variation than would be expected if these to the mainland in the recent past), populations were founded by rafting prop- whereas many of the islands in the Uta agules. Second, the virtual electrophoretic study are oceanic (never with a mainland identity of alleles and allele frequencies of connection). There is no a priori reason for these two populations (as well as others) supposing a correlation between channel strongly implies a pre-existing connection in depth and age for oceanic islands. the Pleistocene; the most recent time such E) Biogeographical cor~elatesoJ hetero- a connection could have existed was during zygosity: Lacerta sicu1a.-As stated above, the last glaciation. we cannot apply the same type of analysis If, in fact, M values are fair estimates of to L. sicula because there are too few population age (time since isolation), why populations. But we can discuss our should there have been a steady attrition of observations in the context of our working variation with time? It is explicit in the hypotheses relative to L. melisellensis. hypothesis discussed above, that the level It was our original belief based on of variation in an island population is a biogeographic data that most L. sicula function of the length of time selection has populations, in contrast to the L. nzelisel- been affecting a diminution of allelic lensis populations, were founded by over- diversity. The longer this time, the less water colonizations. If this is correct, it variation is to be expected. Indeed, M is follows that the sicula populations never negatively correlated with H. had very much electrophoretic variation. In conclusion, we have proposed two The results support our hypothesis. Instead explanations for the major result of the of a positive correlation between A' and multiple regression analyses. The first is H as in melisellensis, we observe an inverse that the loss of allelic diversity is a function relationship between area and 2; thus of the ecological distinctness of the island. SuSac is the largest, but the least hetero- The second is that the loss of variation is a zygous of the three island populations and function of the age of the population. Palagruia is the smallest and most hetero- Because the two relevant independent vari- zygous. ables, A' and M, which we believe to be Next we note an inverse relationship correlated with ecological distinctness and between distance and H. Palagruia is 68 G. C. GORMAN ET AL. closest to the Italian mainland and is most previous study of island lizards, are con- heterozygous; it is approximately on the 100 sistent with the predictions of the time- meter contour and may have been con- divergence theory of variation, but do not nected to the Italian mainland. SuSac and allow firm conclusions. The theory predicts KopiSte are virtually equidistant from that the smaller (more ecologically distinct) Palagruia (or the mainland) and are less and the older an island, the lower the heterozygous. Finally we note that SuSac heterozygosity in its population. It is and KopiSte are surrounded by deep water argued that directional selection is the main (>130 m) and were probably not con- force eroding genetic variation : the smaller nected to each other, nor to Palagruia nor an island, the more directional selection will the mainland. prevail over stabilizing selection; the older an island, the longer such selection has operated. Very small islands, where genetic drift appears to reduce variability, were Twenty populations of Lacerta sicula, L. excluded from most of the regression nzelisellensis, and L. oxycephala from the studies. coast and islands of the Adriatic were studied electrophoretically. Genetic simi- larity measured by the Rogers coefficient Field work in Yugoslavia was supported (Rogers, 1972) ranges from 0.84-1.0 by a grant from the Office of International among populations of melisellensis, and Activities of the Smithsonian Institution from 0.88-0.99 among populations of sicula. from Public Law 480 funds. A supple- Interspecific values range from 0.61-0.74 in mentary grant from the American Philo- this pair, and both have values of about sophical Society also aided us, especially 0.43 compared to oxycephala. There is in Italy. Laboratory efforts were sup- circumstantial evidence for hybridization ported by NSF Grant GB 27358 to SoulC, and introgression in one population. and preparation of the manuscript was Among Lacerta, amounts of genetic vari- supported by NSF Grant GB 40070 to ation are greater in (1) mainland popu- Gorman. C. Wills, V. JovanoviC, and lations than in putatively relict island D. Lilge kindly commented on the manu- populations, and (2) on large islands than script. T. Papenfuss, A. Keller, R. Clover, on small islands. Some fringing islands that D. Goldsmith, and A. Gorman actively are separated by a short linear distance and gathered lizards. Special thanks go to Dr. shallow channel depth from the parent V. JovanoviC of Belgrade not only for direct island show a "small island effect," i.e., a help in the field but also for logistical precipitous drop in variation compared with advice, arranging collecting permits, and the mother island but only if these fringing providing needed linguistic help. Mr. and islands are tiny (" < 0.01 sq. km). Mrs. D. Meredith, formerly of KorEula, Genetic drift is implicated. Larger fringing provided advice on housing, and most islands do not show an area effect and are important, arranged for our boat. This genetically virtually identical to the parent study could never have been accomplished island. without the help of Captain Niko Sambrailo An analysis of allele frequencies in popu- who, though he hated lizards, and doubted lations on different islands suggests that our sanity, took us to 35 islands in as many most of the polymorphisms are being days. selected, but that some (e.g., esterases) have a pattern consistent with the neutral- LITERATURECITED drift hypothesis. ARNOLD,E. IT. 1973. Relationships of the Pale- Stepwise multiple regression was used to arctic lizards assigned to the genera Lacerta, Algyroides, and Psammodromus (Reptilia: analyze the geographic correlates of mean Lacertidae). Bull. Brit. Mus. (Nat. Hist.) 25: heterozygosity. The results, like those of a 291-366. EVOLUTIONARY GENETICS 69

BOULENGER,G. A. 1920. Monograph of the Sou~i,M. 1966. Trends in the insular radiation Lacertidae. Vol. 1. Brit. Mus. Nat. Hist., of a lizard. Amer. Natur. 100:47-64. London. Sou~i,M. ASD S. Y. YANG. 1973. Genetic vari- GORXAN,G. C. 1972. Evolutionary genetics of ation in side-blotched lizards on islands in island lizard populations. Yearb. Am. Phil. the Gulf of California. Evolution 27:593-600. SOC.1971:318-319. WEBSTER,T. P., R. K. SELANDER,AND S. Y. YANG. GORX~N,G. C., P. LICHT, H. C. DESSAUER,ASD 1972. Genetic variability and similarity in the 0. Boos. 1971. Reproductive failure among Anolis lizards of Bimini. Evolution 20:5?3-535. the hybridizing Anolis lizards of Trinidad. WRIGHT, S. 1951. The genetical structure of Syst. Zool. 20: 1-18. populations. Ann. Eugen. 153323-354. GORLIAS, G. C. AND S. Y. YANG. 1975. A lorn level of backcrossing between the hybridizing APPENDIX Anolis lizards of Trinidad. Herpetologica (in Genetic Distance of the Lacertid Lizards press) . LEWOSTIN,R. C., ASD J. KRAKAUER.1973. D:s- Several estimates have been proposed to examine tribution of gene frequency as a test of the genetic similarity (or distance) between popu- theory of selective neutrality of polymorphisms. lations. The Rogers Coefficient (Rogers, 1972) Genetics 74: 175-195. that we used in the body of our paper was chosen, MAC~~RTHUR,R. AND E. 0. WILSON. 1967. The in part, out of convenience (the program was theory of island biogeography. Princeton Univ. readily available to us) and, in part, to provide Press, Princeton, N.J. p, 203. direct comparison with other works that have used MARUY.~~~A,T. 1970. On the rate of decrease of this measure. Nei (1972) provided another esti- heterozygosity in circular stepping stone mate of genetic distance which is probably models of populations. Theoret. Pop. Biol. superior, because it estimates codon differences per 1 :101-119. locus, and may well be linearly related to diver- MAPR, E. 1963. Animal species and evolution. gence time. Dr. Nei's laboratory, in the person of The Belknap Press of Harvard Univ. Press, Dr. A. Roychoudhury, kindly consented to com- Cambridge, Mass. p. 797. pute genetic distance from our gene frequency data. MCKINNEY, C. O., R. K. SELASDER,W. E. Their program can only calculate pairmise com- JOHSSON, ASD S. Y. YANG. 1972. Genetic parisons for 15 or fewer populations, so me variation in the side-blotched lizard (Uta cannot compare all 20 populations in the same stansbzwiana). Studies in Genetics VII. Univ. analysis. Therefore, with the consent of the Texas Publ. 72 13:307-318. editors, we present the Nei estimates of genetic MERTENS, R. 1934. Die Insel-reptilien, ihre distance as an appendix. We compare distances Ausbreitung, Variation, und Artbildung. Zoo- between populations within species, then pool the logica 32 :1-209. within species gene frequency data to make between species comparisons. We emphasize that MERTENS, R., AND H. WERXIUTH. 1960. Die none of the conclusions (or speculations) from Amphibien und Reptilien Europas. Waldemar the body of the paper are altered, merely refined. Kramer Verlag, Frankfurt am Main. p. 264. Appendix Tables la and lb present standard NEVO,E., G. GORM-\N,M. SOULI?,S. Y. YASG, R. genetic distance (and standard error of standard CLOVERAND V. JOVANOVI~.1972. Competi- genetic distance) between populations of Lacerta tive exclusion between insular Lacerta species melisellensis and L. sicula, respectively. Table 2 (Sauria, Lacertidae) . Notes on experimental presents the standard genetic distance between the introductions. Oecologia 10: 183-190. three species of Lacerta. POWER,D. M. 1971. Range expansion of Brew- Note that the greatest within species genetic er's blackbird-phenetics of a new population. distance is 0.11 for L. melisellensis (KorEula and Can. J. Zool. 49: 175-183. Kamik) and only 0.06 for L. sicula (Trogir and R.~DOVANOVI~,M. 1956. Rassenbildung bei den Pod KopiSte), but the between species values are Eidechsen auf Adriatischen Inseln. 0sterr. much higher (0.4 between melisellensis and sicula) . Akad. Wiss. 110: 1-82. Establishing divergence time is very much in the 1959. Zum Problem der Speziation Lei realm of speculation. First, we have to make the Inseleidechsen. 2001. Jahr. Abt. System. 0kol. major assumption that genetic distance is linearly U. Geogr. 86:395-436. related to time, And, based upon that premise, ROGERS,J. S. 1972. Measures of genetic simi- which is admittedly questionable, we have to larity and genetic distance. Studies in Genetics. calibrate our clock. Elsewhere (Yang et al., 1974) VII. Univ. Texas Publ. 72 13 : 145-153. we have attempted such a "ballpark" estimate, SNEDECOR,G. W. 1956. Statistical methods ap- and me shall do so again, eniphasizing that me are plied to experiments in aqriculture and groping for a realistic estimate, and not pretending biology. The Iowa State College Press, Amcs, that me can provide the exact point in time for Iowa. p. 534. divergence. Our estimate was based upon the 7 0 G. C. GORMAN ET AL.

APPENDIXTABLEla. Standard genetic distance (and standard error)l between populations of Lacerta melisellensis. -- --- Kor Las Pdm 1 Gre 3Ilp hllb Bis Sle Bru Kam Jab 1 2 3 4 5 6 7 8 9 10 11 12

TABLElb. Standard genetic distance (and standard error) between popzdations of Lacerta sicula

Zad Tro Pes Pala SuF Kop Pdk 13 14 15 16 17 18 19

Standard error appears immediately below standard genetic distance. EVOLUTIONARY GENETICS 7 1

concordance between Nei's Genetic Distance and APPENDIX TABLE 2. Standard genetic distance Sarich and \lrilson's Immunological Distance (? standard error) of three species of Lacerta. (where I. D. 30 = G. D. 1.0) and where I. D. 30 = 18 million years of divergence (see Yang et al., 1974). This estimate gives a predicted divergence time L. sicula 0.415 -C 0.169 of about 7 million years between L. melisellensis L. oxycephala 0.894 k 0.285 0.9478 -C 0.300 and L. siczda, and about 16 million years between L. oxycephala and the other species. Since the genus Lacerta (sensu lato) is known as far back as the Miocene (Romer, 1966), and since Arnold (1973) thinks that oxycephala is generically distinct from melisellensis and sicula, our estimate of 16 Genetic Distance was computed in the laboratory million years might not be too high. of Dr. M. Nei with support from PHs grant GM At the low end, we have estimated that the 20293. populations isolated on islands should probably be dated in the tens of thousands of years, based upon changes in eustatic sea level. When we examine the NEI, M. 1972. Genetic Distance between popu- relationships between Vis and the islands to the lations. Amer. Natur. 106:283-292. west (see Fig. 3), the genetic distances generally RORIER, A. S. 1966. Vertebrate Paleontology, lie between 0.01 (180,000 years) and 0.003 (60,000 Third Edition. Chicago, Univ, of Chicago years). These latter estimates are a bit high, but Press. VIII + 468 p. the genetic distances are so small, and depend so YANG,S. Y., M SOULI?AND G. C. GORMAS. 1974. heavily on differences in allele frequencies rather Anolis lizards of the eastern Caribbean: A case than gene substitutions, that we are not displeased study in evolution. I. Genetic relationships, with the order of magnitude provided by the phylogeny and colonization sequence of the estimates. roqzdet group. Syst. Zool. 23:387-399.

View publication stats