A Phylogenetic Analysis of Body Size Evolution and Biogeography in (Sauromalus) and Other Iguanines Author(s): Kenneth Petren and Ted J. Case Source: Evolution, Vol. 51, No. 1 (Feb., 1997), pp. 206-219 Published by: Society for the Study of Evolution Stable URL: http://www.jstor.org/stable/2410974 Accessed: 14/09/2010 18:44

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 Evolution, 51(1), 1997, pp. 206-219

A PHYLOGENETICANALYSIS OF BODY SIZE EVOLUTIONAND BIOGEOGRAPHYIN CHUCKWALLAS(SAUROMALUS) AND OTHER IGUANINES

KENNETH PETREN1 AND TED J. CASE Department of Biology, University of California at San Diego, La Jolla, California 92093-0116

Abstract.-The evolution of body size was reconstructedin chuckwallas ( Sauromalus),large herbivorous lizards of southwest North America, using a phylogeny derived from sequence variation in the mitochondrial cytochrome b gene. The body mass of two endemic island (S. hispidus and S. varius) is typically fivefold larger than mainland species. We tested the hypothesis that large body size has evolved on these islands in response to local ecological conditions against the alternative hypothesis that large size is simply retained from large iguanine ancestors. The most parsimonious tree topology depicts the insular gigantic Sauromalusas monophyletic, having diverged from a common ancestor on the Baja California peninsula after the radiation of smaller bodied clades. In a robustness analysis of this topology, we found general supportfor this tree over alternativetopologies representingminimum evolution hypotheses that imply large body size is retained from large iguanine ancestors. The most parsimonious reconstruction of body size evolution implies a change from large to small size after the Sauromalus ancestor diverged from , and one reversal back to large size within Sauromalus.The large size increase in the gigantic clade contrasts with evolutionary stasis of small body size (for an iguanine) in mainland populations. The gigantic species show 3-4% total sequence divergence from S. obesus populations on the nearby Baja California peninsula, and mainland populations of S. obesus obesus show similar levels of divergence from each other.An analysis of charactertransitions and comparativebehavior implicates predation, and its relaxation on isolated islands, as a strong selective force in Sauromalus. Patterns of genetic differentiation in Sauromalus and biogeographic implications are discussed. Key words.-Body size, charactertransitions, cytochrome b, Gulf of California, insular gigantism, lizards, predation, Sea of Cort6z.

Received March 1, 1996. Accepted September 17, 1996.

Island organisms are well known for their unusual traits differ not only in size, but in morphology, coloration, life that often obfuscate their origin and history. Body size is history, and behavior (Shaw 1945; Case 1978, 1982; Iverson arguably the most fundamental characteristic defining dif- 1980; Ryan 1982; Carothers 1984; Smits 1985). ferences between closely related species (Darwin 1859; Elton Competing hypotheses exist regarding the origin of gi- 1949; Hutchinson 1959; Peters 1983; Gittleman 1985; Brown gantism in the insular Sauromalus.Shaw (1945) hypothesized 1995; Pianka 1995) and is a trait that seems to change most that the gigantic species diverged earliest, which implies (if readily on islands (Mertens 1934; Foster 1964; Case 1978; it is assumed that the genera most closely related to Saurom- Lister 1989). Islands often harbor gigantic or dwarfed forms alus were large-bodied) that large size in S. hispidus and S. compared to mainland relatives, yet closely related species varius was simply retained from iguanine ancestors (e.g., may show different responses to biotically depauperateisland Iguana, , Conolophus, ). The recent evo- environments, suggesting that multiple processes interact to lution of small body size in mainland forms would be con- determine size shifts (Foster 1964; Case 1978). For example, sistent with a pattern of Holocene dwarfism associated with black tiger snakes (Notechis ater nigra) occur on islands off recent human impacts (Pregill 1986; Case et al. 1992). Al- South Australia, which were connected to one another or to ternatively, Case (1978, 1982) suggested that large size may the mainland as recently as 6000 years ago (Robinson et al. have evolved in Sauromalus on these islands from smaller 1986), yet today snakes on these islands show up to five-fold mainland ancestors as a response to ecological variation in body size (Schwaner 1985; Schwaner and Sarre conditions on these islands. This implies a different sequence 1990). In the Sea of Cortez, the rattlesnakeCrotalus mitchellii of evolutionary events: small size evolved early in mainland has evolved a gigantic form on the island of Angel de la Sauromalus, followed by the reappearance of large size in Guarda, but the closely related endemic form on the island insular S. hispidus and S. varius. The appearance of large of El Muerto are dwarfs (Klauber 1956). The maximum body body size as a recently derived character on islands is not lengths of these two forms differ by a factor of two and span without in other insular lizards. A phylogenetic nearly the entire range of body size for the genus Crotalus. precedent We reconstructedphylogenetic relationships in the herbiv- reconstructionof the xantusiid lizards using mtDNA (Hedges orous iguanine lizard genus Sauromalusto determine the or- et al. 1991) suggests that the large insular species of the igin of the endemic gigantic species living on islands in the California Channel Islands, Xantusia riversiana, is sister tax- Sea of Cortez. The island endemics Sauromalushispidus and on to the smaller mainland X. vigilis and does not represent S. varius are up to five-fold larger in body mass compared an endemic genus as previously believed (Savage 1963). to mainland species and other Sauromalus on islands in the Sauromalushispidus inhabits Angel de la Guarda,an island Sea of Cortez (Shaw 1945; Case 1982). The gigantic species estimated to have separated from the Baja California pen- insula approximately one million years ago (Moore 1973; Lonsdale 1989). Sauromalusvarius is endemic to San Esteban 1 Present address: Department of Ecology and Evolutionary Bi- ology, Princeton University, Princeton, New Jersey 08544-1003; Island, which, conversely, shares geological and biogeo- E-mail: [email protected]. graphical affinities with mainland Mexico (Felger and Lowe 206 C) 1997 The Society for the Study of Evolution. All rights reserved. BODY SIZE EVOLUTION IN CHUCKWALLAS 207

1976; Case 1983; Cody et al. 1983; Murphy 1983a). It is not Sequencing of asymmetric templates was proceeded by known whether the endemic Sauromaluson these two islands purification with centrifugal filtration (30 pumMillipore car- owe their origin to vicariant events associated with their for- tridge filters) to remove unincorporatedprimer. The appro- mation, or whether they colonized at a later time. priate PCR primers were used for external or internal se- We used cytochrome b sequence divergence to reconstruct quencing in a standardreaction with 35S dATP (T7 Sequenase phylogenetic relationships in Sauromalus. Other iguanine v 2.0, U.S. Biochemical). Sequencing reactions were elec- genera were included to improve phylogenetic and character trophoresed on 6% denaturing polyacrylamide gels, fixed reconstructions.Historically, there has been considerable un- with acetic acid and methanol, dried, and exposed for 2-4 d certainty regarding the relationships among iguanine genera (Maniatis et al. 1982). Both strandswere sequenced (average (Avery and Tanner 1971; de Queiroz 1987; Norell and de overlap was 75%) transcribed using MacVector (1991), Queiroz 1991). The robustness of the best phylogenetic tree aligned by eye in PAUP (Swofford 1991) and transferredto was evaluated against alternative topologies that lead to dif- other programs for analysis as indicated below. ferent conclusions of body size evolution based on the cri- terion of maximum parsimony. Character State Reconstructions Uncovering the sequence of evolutionary events helps to distinguish adaptationsfrom exaptations, and strengthensin- We used snout-vent length (SVL) as our measure of body ferences of selective forces based on current environmental size, and treated this variable as continuous (Table 2). Ig- conditions (Gould and Vrba 1982; Baum and Larson 1991). uanine genera show considerable body size variation, so an Body size may be influenced by many diverse selection pres- approximate value reflecting the general body size within sures, so it is expected that the direction of body size evo- each genus was used. Within populations of Sauromalus, lution will depend on a number of other traits and environ- measures of body size vary extensively over time, and choos- mental circumstances. We employ the method of inferring ing a continuous metric for comparison is problematic. transitions of characters from ancestral state reconstructions Chuckwallas are known to forgo breeding during drought to interpret body size evolution (Farris 1970; Brooks and years when resources are low (Berry 1974; Case 1982), and McLennan 1991; Miles and Dunham 1993), and discuss ev- droughts may last several years, causing great fluctuationsin idence for the currentutility of body size differences in Sau- the age and size structureof populations. Upper decile snout- romalus. vent length (UDSVL) was used because this metric is less prone to such fluctuations than the arithmetic mean SVL MATERIALSAND METHODS (Case 1976). Size variation among populations on the main- land and Baja California peninsula are nearly an order of Multiple Sauromaluspopulations were sampled throughout magnitude less than the difference between these populations the range of the genus to detect ambiguous species and sub- and the gigantic species (Case 1976). species boundaries (Fig. 1, Table 1). Representative samples Of the eight currently recognized iguanine genera, only from all iguanine genera (except Amblyrhynchus)were in- , Dipsosaurus, and Sauromalus are mostly small cluded to improve phylogenetic and body size reconstruc- bodied (approximately less than 1 kg and < 220 mm SVL). tions. Whole blood (0.25-0.5 mL) was taken from the caudal Many others (Cyclura, Iguana, Conolophus, Amblyrhynchus, vein (Esra et al. 1975) in the presence of EDTA and stored and Ctenosaura) are considerably larger than the largest in an equal volume of phosphate buffered saline and two chuckwalla, but some Ctenosaura species are similar in size volumes of 95% ethanol. For some individuals, toe clips or to the smaller-sized Sauromalus.Upper decile SVL are avail- tail clips in 95% ethanol were used. able for all Sauromalustaxa, but could only be approximated Total DNA was isolated by salt precipitation (Miller et al. for other iguanine genera from maximum and mean SVL data. 1988) followed by one extraction with phenol-chloroform/ The body size at ancestral nodes was inferred using the isoamyl alcohol (24:23:1), and one with chloroform/isoamyl squared-changeparsimony method (Maddison 1991), as well alcohol (23:1; Maniatis et al. 1982). Total genomic DNA (0.1 as the linear change parsimony method (Swofford and Mad- p,g) was used in a PCR reaction (Mullis et al. 1986; White dison 1987; Maddison and Maddison 1992), which assumes with combinations of universal and specifically et al. 1989) a model allowing more rapid change along branches. designed primers (numbers refer to human mtDNA loca- tions): L14703-24, L15149-73, H15577-57, and H15173-49 (Palumbi et al. 1991); L14817-41, H15775-52 (Kocher et al. RESULTS 1989); and H15341-65 (Radtkey et al. 1995). Reaction con- Genetic Divergence and Phylogenetic Reconstruction ditions varied, but most yielded a single clean product by denaturing at 93?C for 1 min., annealing for 1 min at 46?C Total Jukes-Cantorcorrected sequence divergence in cy- to 48?C, extending for 2 min at 72?C, and repeating these tochrome b within Sauromalus was less than 7% (Table 3). steps for 35 cycles. We included 3.0 mM MgCl2, 0.2 mM The gigantic species S. hispidus and S. varius differed by 2- dNTP, 0.4 p,M of each primer,and 1-2 units Taq polymerase 3% from each other, and by 3-4% from the nearest mainland (Promega) in a total volume of 50 p,L. Asymmetric PCR populations, whereas some S. o. obesus populations showed products were generated for direct sequencing by adding 5 similar levels of divergence. Sequences of individuals con- p,L of the first round reaction to a similar mixture of com- sidered to be S. o. obesus from Baja California differed from ponents in 100 p,Ltotal volume, but with the addition of only other S. o. obesus populations by 5-6%. The relatively small one primer. divergence within Sauromalus contrasted with the much 208 K. PETRENAND T. J. CASE

-\ 14 \\\\\\\ - 1 S. obesus ~~~~~~~~~~~ =Nv x ___ E_ multiforamina tus

= 1J

-16e D1 t E ~~~S.obesus tumidus

S. hispidus S.~~~~~~~S obesus townsendiS.hsiu

000 ~~Angelde laGuarda (1- 2) Smith australis 2 0 (3) \ S.aterkLaVentana (4)

San Lorenzo Sur obesus ~~~~~~~~~~~~~~~~~(5)

50 km

9 ~~~~~~~~~SanEsteban

S. australis ~ae hw

S. ater slevini ( 1 1- 13)

S. ater klauberi (10)

S. ater ater

FIG. 1. The geographic distribution of Sauromalus.The classical nomenclature and distributions are based on Shaw (1945), Tannerand Avery (1964), Soule and Sloan (1966), Robinson (1972), Stebbins (1985), and Grismer et al. (1994). Numbers indicate locations of populations included in this study. The borders between S. australis and S. obesus obesus and between subspecies of S. obesus are approximate. Stippling represents land area exposed during the most recent glacial maximum (1100-1500 YBP, Gastil et al. 1983). greater sequence divergence (> 12%) observed between There was a single most parsimonious topology as deter- and C. hemilopha (Table 4). mined by an heuristic search using PAUP (Swofford 1991), Of the 32 individual Sauromalus sequenced, 27 unique under a global maximum parsimony criterion with Oplurus haplotypes were found. Terminal taxa for phylogenetic re- quadrimaculatusas the functional outgroup (Fig. 2). We ini- construction were determined by grouping sequences differ- tially employed a weighting scheme of 4:1 transversions to ing by 1.5% or less (haplotype groups, Table 1). This thresh- transitions, however we explored the effect of different old was chosen because of the naturalbreak in the data: there weighting schemes on the topology (see below). Most phy- were no pairwise distances between 1.5% and 2.3% (the dis- logenetic information within Sauromalus resides in transi- tance between the two gigantic species). Haplotype groups tional changes (130 inferred transitions compared to 15 in- were coded as polymorphic for tree reconstructions when ferred transversions). Transitional differences between ig- maximum parsimony was used. A subset of 1000 randomly uanine genera are closer to saturation(183 inferredtransitions generatedtrees yielded a G1 statistic of -0.61, suggesting that compared to 141 inferred transversions). Within Sauromalus there is strong phylogenetic signal in the data (P < 0.01; most substitutions inferred by parsimony reconstructions Hillis and Hulsenbeck 1992). were at the third codon position (189/242 or 78%), however BODY SIZE EVOLUTION IN CHUCKWALLAS 209

TABLE 1. Source localities for Sauromalusand iguaninel populations. Haplotype groups (defined as haplotypes showing less than 1.5% Jukes-Cantorcorrected sequence divergence) are terminal nodes for parsimony tree reconstructions. Locations of populations are shown in Figure 1. Nomenclature follows that of Shaw (1945) and Soule and Sloan (1966).

Unique Unique haplo- Max. Popu- haplo- Haplotype group N types difference2 lation Location N types S. hispidus 7 5 1.1% 1 Angel de la Guarda Is. (east) 2 2 2 Angel de la Guarda Is. (north) 2 1 3 Smith Is. 1 1 4 La Ventana Is. 1 1 5 San Lorenzo Sur Is. 1 1 S. varius 4 4 0.4% 6 San Esteban Is. 4 4 S. o. obesus (mid-north Baja) 4 4 0.6% 7 20 km w. of Bajia Los Angeles, Mex. 3 3 S. australis 8 4 km w. of Bajia Los Angeles, Mex. 1 1 S. ater klauberi 1 1 9 8 km s. of Bajia Concepcion 1 1 S. ater slevini 1 1 10 Santa Catalina Is. 1 1 3 3 0.6% 11 Carmen Is. 1 1 12 Danzante Is. 1 1 13 Monserrate Is. 1 1 S. o. obesus (UT/CA) 2 2 1.5% 14 8 km w. of Virgin, UT 1 1 15 17 km e. of Panamint Springs, CA 1 1 S. o. obesus (CA) 4 1 16 Amboy, CA 4 1 S. o. obesus (s. CA) 1 1 17 Canbrake, CA 1 1 S. o. tumidus 1 1 18 8 km w. of Dateland, AZ 1 1 S. obesus townsendi 4 4 1.2% 19 80 km w. of Sonoyta, Mexico 1 1 20 40 km s. of Sonoyta, Mex. 1 1 21 8 km sw. of Caborca, Mex. 1 1 22 50 km sw. of Caborca, Mex. 1 1 Iguana iguana 1 1 23 South America (San Diego Zoo) 1 1 Ctenosaura hemilopha 2 2 0.1% 24 San Esteban Is. 2 2 Ctenosaura similis 1 1 25 Guanacaste, Costa Rica 1 1 Conolophus subcristatus 1 1 26 Galapagos Is. 1 1 1 1 27 Cuba 1 1 Dipsosaurus dorsalis 1 1 28 California 1 1 1 1 29 Fiji 1 1 Oplurus quadrimaculatus 1 1 30 Madagascar 1 1 I Sources of iguanine tissue: Iguana iguana (A. Alberts, San Diego Zoo); Ctenosaurasimilis (T. Wright, T. Langen); Conolophussubcristatus (J. W. Sites Jr., loan from U.N.M); Cyclura nubila (A. Alberts); Dipsosaurus dorsalis (B. Mautz); Brachylophusfasciatus and Oplurusquadrimaculatus (R. N. Fisher). 2 Maximum Jukes-Cantorsequence divergence among haplotypes included in group.

average levels of divergence at these sites (13%) falls well Sauromalus using maximum parsimony. Two other tree- below the saturationlevel (Brown et al. 1982). searching algorithms, the neighbor-joining distance method (Saitou and Nei 1987) and the maximum likelihood method Robustness of the Topology (1:1 and 4:1 weighting; Hasegawa and Yano 1984; Felsen- stein 1989; Kishino and Hasegawa 1989), also arrived at We evaluated the robustness of the most parsimoniousphy- topologies congruentwith the maximumparsimony topology. logeny in a variety of ways. First, distance and maximum- likelihood tree-construction algorithms were used to search Using Iguana iguana, the closest relative of Sauromalus, as for the best tree. We tested different weighting schemes, and the functional outgroup, and excluding other iguanines did bootstrap support for key nodes was also evaluated. We then not result in different topologies for the major Sauromalus directly tested a priori hypotheses and the most parsimonious clades. a posteriori hypothesis that would lead to different interpre- Some combinations of terminal taxa and weighting tations of body size evolution. The robustness analysis was schemes resulted in best tree topologies that differed in the focused on the part of the topology that affects our conclu- order of branching of the major Sauromalusclades, such that sions about body size evolution: the order of branching of the order of branchingof the mid-northBaja California clade the four major Sauromalus clades (Fig. 2: A, B, C, and D). and southernclade was reversed (orderA, C, B, D). However, Finally, tree reconstructions, bootstrap analyses and tests of monophyly of the gigantics (clade D) was always supported, alternative hypotheses were repeated with different subsets and this clade was always most derived relative to the other of Sauromalus haplotypes. This was done for two reasons. clades. Also, the main S. obesus clade of the U.S. and Sonora, First, some methods do not allow for polymorphic coding of Mexico was always the sister to the rest of Sauromalus.This terminal taxa, and second, we wished to assess the level of pattern was reflected in bootstrap replicates (Felsenstein inaccuracy that could be caused by incomplete sampling 1985). Using different terminal taxa, values for the node de- within the geographic range of recognized taxonomic groups. fining clade D ranged from 89-100%. However, bootstrap Weighting schemes of 4:1 and 7:1 yielded the same basic values for each of the two nodes defining the order of branch- topology for iguanine genera as well as the major clades of ing of A, B, and C showed wide variation (50% to 85%). 210 K. PETREN AND T. J. CASE

TABLE2. Body sizes of Sauromaluspopulations and iguanine genera. For Sauromalus,letters in brackets represent the broad categories of large (1), small (s), or medium (m) body size. Chuckwalla populations classified as "small" are smaller than most iguanines, but are still considerably larger than most other North American lizards. The values for Little Lake and Amboy approximatethe range of upper decile snout vent lengths (UDSVL) and maximum body mass for mainland and peninsular populations. Snout-vent lengths for iguanine genera are approximate, as it is not possible to accurately characterize each genus with a single value in the absence of complete phylogenetic information. Species in the genus Ctenosaura show body size variation comparable to Sauromalus.

Sauromalus1 species Location n UDSVL (mm) Max. mass (g) S. obesus obesus Amboy, CA 110 175 180(s) S. obesus obesus Little Lake, CA 20 218 380(s) S. obesus obesus Southwestern Utah 38 178 (s) S. obesus tumidus Arizona 40 201 (s) S. obesus townsendi Sonora, Mex. 41 176 (s) S. obesus obesus Northern Baja, Mexico 48 189. (s) S. australis Southern Baja, Mexico 16 172 (s) S. ater ater Southern Gulf islands 111 163 (s) S. ater sleveni Carmen Island 22 181 (s) S. ater klauberi Santa Catalina Island 16 171 (s) S. ater sleveni Monserrate Island 46 210 550(m) S. varius San Esteban Island 90 301 1800(1) S. hispidus Angel de la Guarda Island 213 219 1400(1) S. hispidus San Lorenzo Island 32 295 (1) Iguanines2 genus Distribution No. species SVL (mm) Max. mass (g) Iguana Central and South Amreica 2 375+ 3500+ Ctenosaura Mexico-Central America 12 150-350+ C. hemilopha Mexico 300+ 2800+ C. similis Costa Rica 350+ 2000+ Conolophus Galapagos Islands 2 400+ 10,000+ Amblyrhynchus Galapagos Islands 1 400+ 6000+ Cyclura Caribbean Islands 7-8 400+ 5400+ Brachylophus Fiji Island 2 190 404 Dipsosaurus Southwestern North America 1 140 100 1 Sources for Sauromalus:Berry 1974; Case 1976, 1982; specimens in the California Academy of Science and the San Diego Natural History Museum. 2 Sources for iguanines: Burghardtand Rand 1982; Dugan and Wiewandt 1982; Etheridge 1982; Gibbons and Watkins 1982; Van Devender 1982; Weiwandt 1982; Boersma 1983; Eibl-Eibesfeldt 1984; de Queiroz 1995.

These values bracket the value of 70% proposed by Hillis potential topologies constructed from biogeographic hypoth- and Bull (1993), which represents 95% confidence that the eses put forth in Shaw (1945) and Murphy (1983a,b); it takes node is significantly supported by the data in phylogenies into account the assumption that the gigantic clades evolved with < 20% divergence between terminal taxa; although it in situ on their respective islands as vicariant relicts, and has been suggested that this criterion may be too conservative which also supports minimal size evolution in Sauromalus. (Felsenstein and Kishino 1993). Topologies that rooted the The remainderof the topology is reconstructedfrom the rel- gigantic clade basal with respect to the other Sauromalus ative distances of current populations from the point where clades (e.g., order D, A, B, C) were not observed in any the island of Angel de la Guardasplit from the Baja California bootstrap replicates under a variety of conditions (< 0.1%). peninsula. The a posteriori tree (Fig. 3C) was derived by Within Sauromalus clades, the order of branching within preserving the iguanine topology, constraining the gigantic clade B ([S. australis, S. ater klauberi] and S. ater slevini) clade to be basal in Sauromalus, and conducting a heuristic was consistent, while the topology within clade A (S. o. obe- search for the best topology for the remaining Sauromalus sus, S. o. tumidus, and S. o. townsendi) varied greatly. This taxa using the sequence data (PAUP; Swofford 1991). suggests that the order of branching of the three S. obesus We tested these topologies against the best topology (Fig. subspecies is not resolvable, and there is little support for 2) using maximum parsimony and maximum-likelihood these subspecies designations based on cytochrome b se- methods (Felsenstein 1989), which compare the overlap of quence divergence. the standarddeviation of the respective robustness estimates of the trees being compared. Terminal taxa were limited to Testing Alternative Tree Topologies previously recognized species and subspecies (S. ater ater, Two alternativea priori.topologies, and the best alternative S. ater shawi, and S. obesus multiforaminatus were not in- a posteriori topology that rooted the gigantic species basal cluded in the analysis). with respect to other Sauromaluswere evaluated against the The a priori hypotheses were rejected as being significantly most parsimonious topology (Fig. 2). These alternative to- less parsimonious and significantly less likely than the best pologies imply minimum evolution of body size. Shaw's topology. The a posteriori hypothesis is significantly less (1945) hypothesized phylogeny was based primarilyon mor- parsimonious, but not significantly less likely than the best phological features including scutellation, coloration and size tree according to these tests (Table 5). In testing the sensi- (Fig. 3A). The second topology (Fig. 3B) is one of many tivity of these tests to substitution of taxa in the main S. BODY SIZE EVOLUTIONIN CHUCKWALLAS 211

TABLE3. Pairwise genetic distances between Sauromaluspopulations. Distances are the percentage of nucleotide differences (corrected for multiple substitutions using the method of Jukes and Cantor 1969) for 902 bp of the cytochrome b gene. Numbers below diagonal are all changes, numbers above the diagonal are transversions only. (See Table 1 for locations and within-group haplotype divergence).

S.o.o. S. o. o. S. o. o. S. o. o. Species/haplotype group S.h. S.v. n. Baja S. au. S. a. k. S. a. s. CA UT/CA s. CA S. o. tu. S. o. to. S. hispidus 0.2 0.3 0.5 0.6 0.5 0.5 0.6 1.0 0.7 0.8 S. varius 2.3 0.3 0.4 0.6 0.4 0.4 0.5 1.0 0.7 0.7 S. o. obesus (mid-north Baja, Mex.) 3.8 3.9 0.1 0.2 0.1 0.1 0.2 0.7 0.3 0.4 S. australis 4.9 5.0 4.7 0.3 0.2 0.2 0.3 0.8 0.4 0.5 S. ater klauberi 5.1 5.6 5.1 2.8 - 0.3 0.3 0.5 0.9 0.6 0.6 S. ater slevini 4.7 5.0 4.2 3.5 4.0 0.2 0.3 0.8 0.4 0.5 S. o. obesus (CA) 4.9 4.9 5.3 5.6 5.8 5.7 0.1 0.6 0.2 0.2 S. o. obesus (UT/CA) 5.4 6.1 6.2 5.8 6.3 6.1 3.6 0.7 0.2 0.4 S. o. obesus (s. CA) 5.0 5.9 5.6 6.2 6.4 6.1 3.1 3.0 0.6 0.8 S. o. tumidus 5.3 5.9 5.7 5.9 6.1 5.4 2.9 3.4 3.2 0.4 S. o. townsendi 5.3 6.0 5.7 6.3 6.5 6.6 2.9 3.1 2.8 3.2 obesus clade (Fig. 2: clade A), sometimes the best tree was and Hasegawa 1989), all branch length estimates for the best not significantly different from the a posteriori best alter- tree were always significantly different from zero (P < 0.01), native tree according to both tests, but with different rep- while alternative trees always had one or more branches not resentative taxa both tests showed significant differences. Us- significantly different from zero (P > 0.05); In-likelihood ing ln-likelihood variance tests (Felsenstein 1989; Kishino tests should be interpreted cautiously, because they do not

1 S. hispidus (1-5) Gigantic 7 1 6 1 8 S. varius (6) 1S. 14 obesus obesus (7-8) Mid-north 11 ~~~~~~~~~~~~~~~BajaCalif. 6 S. australis (9) .21 ater klauberi(10) Baja Calif. 8 3 7 1 6 S. ater slevini (1 1 -13) Mex.

C 6 - S. obesus obesus (14,15)

10 _-S. obesus obesus (16) Main 2 'S. obesus obesus (17) |obesus 5 S. obesus tumidus (18) Mex.; U.S. 8 S. obesus townsendi (19-22) Iguana iguana (23) 115 Ctenosaura hemilopha (24) 2 61 Ctenosaura similis (25)

1 5S 9 3 -Conolophus subcristatus (26) 21 1 Cycluranubila (27) 247 Brachylophus fasciatus (28) . 227 Dipsosaurus dorsalis (29) 254 Oplurusquadrimaculatus (30) FIG.2. The most parsimonious phylogenetic tree for iguanine and Sauromalusrelationships based on 902 bp of cytochrome b sequence (corresponding to positions 14797-15698 of human sequence; Anderson et al. 1981). Numbers represent inferred branch lengths using a weighting scheme of 4:1 transversions to transitions. Taxon numbers indicate source locations (Fig. 1, Table 1). Multiple sequences per terminus were coded as polymorphic. Letters define clades that are central to tests of alternative hypotheses (see text). Percentage representation in 500 bootstrap replicates are as follows: node A/BCD = 60%; B/CD = 55%; C/D = 99%; monophyly of Sauromalus = 100%. Within the S. obesus clade, all bootstraps are < 50%. 212 K. PETREN AND T. J. CASE

S. australis allow free rearrangement of other branches (Felsenstein 1989). Limitations are even more problematic for the branch L->S / S. atersievini length test because of possible inflated levels of significance (Type I error). In addition, there are at least 1800 trees with S. aterklauberi the same number or fewer steps as this alternative tree that do not support basal rooting of the gigantic clade relative to Large S. hispidus the other Sauromalusclades (detected by a parsimony search Ancestor v in PAUP; Swofford 1991). In sum, these results indicate that \ ~~~~~S. varius when compared to the best tree, the a posteriori alternative tree (Fig. 3C) should be considered marginally significantly S. obesus townsendi less likely to represent the true tree.

L- > S S. obesus tunidus Body Size Evolution S. obesus obesus The reconstruction of body size (Fig. 4) as a continuous A. A priorihypothesized phylogeny for Sauromnalus. character reveals that Sauromalus ancestors were generally large bodied (320-375 mm). Two major changes in body size are inferred after divergence of Sauromalusfrom its common S. obesus townsendi ancestor with Iguana. The first change from large to small S. obesus tutmidus size (300-375 mm to 181-189 mm) occurred before any diversification within the genus, after which size remained L->S S. obesusobesus small as speciation occurred.The second change back to large size (181-189 mm to 291 mm) occurred in the ancestor of S. ater- A-lauberi the gigantic species, and was retained upon divergence of S. Large S. ater slei,i hispidus and S. varius. Similar results are obtained if squared- Ancestor S a l change parsimony is used (Maddison 1991), or if size is S. australis treated as a discrete variable (large/small). The precision of these estimates must not be taken literally because body size varies among species in these genera, however substitution S. hispidus over the range of sizes produces the general result that the main chuckwalla lineage was small-bodied before the shift S. varius back to large size in the gigantic clade. An alternative evolutionary reconstruction would suggest B. A prioriminimum evolution /biogeographic hypothesis that the main chuckwalla lineage may have remained large bodied, while the three small-bodied clades each changed S. ater klauberi independently.This would imply that body size changed three times instead of twice, and is therefore less parsimonious. S. australis Unfortunately, we know of no way to statistically compare these alternative character reconstructions in a meaningful / S. ater slevini L- > S way. However, the alternative reconstruction would imply S. obesus townsendi that the gigantic species are relicts, and after they were se- quester,edon islands a major change occurred relatively re- Large S. obesus rumidus Ancestor cently over a vast area of mainland (sparsely inhabited by S. obesus obesus humans) that caused mainland forms to evolve smaller size. Characterstates of terminal taxa for extreme saxicoly, ar- boreality, and island status aid interpretation of selective S. hispidus pressures on body size (Fig. 4). The two major transitions implied by the body size reconstructions show parallel tran- S. varius sitions in these characters under the criterion of maximum parsimony. Considering charactercolumns 2-4 together (Fig. C. A posterioribest alternativetree showingminimum size or are in mainland evolution 4), arboreality saxicoly generally present FIG. 3. Alternativehypotheses of phylogenetic relationshipswithin forms, and generally absent in endemic forms on deep-water Sauromalusimply differentpatterns of size evolution. Dark lines rep- islands (islands that remained separated from the mainland resentlarge body size, thin lines representsmall size. (A) Shaw's(1945) during the last glacial regression and inundation approxi- hypothesized phylogeny suggests that the large size of S. varius and mately 15 thousand years ago; Gastil et al. 1983). Deep-water S. hispiduswas retainedfrom a large-bodiediguanine ancestor. (B) An a prioriphylogenetic hypothesis derived from biogeographicdata con- island forms generally attain large size without arboreality cerning the formationof the oceanic islands (Soule and Sloan 1966; or saxicoly (Amblyrhynchus, Conolophus, Cyclura, Saurom- Gastil et al. 1983; Murphy1983a,b), inferringminimal size evolution. alus varius, and S. hispidus). There are exceptions to this (C) The most parsimoniousalternative a posterioritopology supporting pattern. klauberi on the deep-water island minimal size evolution, generatedby an heuristic parsimony search using the sequence data (see text). of Santa Catalina remains saxicolous, and relatively small in BODY SIZE EVOLUTION IN CHUCKWALLAS 213

TABLE 4. Pairwise genetic distances between iguanine genera. Distances are the percentage of nucleotide differences (corrected for multiple substitutions using the method of Jukes and Cantor 1969) for 902 bp of the cytochrome b gene. Numbers below the diagonal are all changes, numbers above the diagonal are transversions only. Sauromalus distances are averaged across haplotype groups.

Genus/species Sauromalus L i. Ct. s. Ct. h. Co. s Cy. n D. d. B. f 0. q. Sauromalus 4.0 7.1 6.3 6.4 8.6 10.2 11.4 12.5 Iguana iguana 13.2 6.5 6.6 6.2 8.6 10.6 10.8 11.9 Ctenosaura similis 15.0 14.1 4.9 5.5 8.6 10.3 11.4 10.8 Ctenosaura hemilopha 15.0 16.4 13.4 5.2 9.4 9.7 10.2 12.5 Conolophus subcristatus 15.7 16.6 15.3 15.1 8.8 10.3 10.6 10.4 Cyclura nubila 18.9 19.8 18.7 20.0 20.5 11.2 13.8 11.7 Dipsosaurus dorsalis 21.9 22.6 20.7 22.2 23.4 24.6 12.5 11.5 Brachylophus fasciatus 25.2 24.7 24.9 24.0 23.2 28.2 26.4 12.8 Oplurus quadrimaculatus 24.3 25.7 23.4 25.5 24.0 26.0 24.7 30.0

body size. Dipsosaurus dorsalis remains small sized on some hemilopha and C. similis show little change over a much deep-water islands in the Sea of Cortez. Brachylophus fas- longer interval. Because most inferred cytochrome b se- ciatus of Fiji and Tonga is moderately sized (19.3 cm max- quence changes among Sauromalusare silent substitutions at imum SVL), whereas the larger B. vitiensis (22.3 cm maxi- the third codon position, phylogenetic reconstructions are mum SVL) is restricted to only a few small islands (Gibbons most likely independent of selective factors influencing body 1984; Zug 1991). Both are arboreal, but frequently move from size. tree to tree on the ground, and B. vitiensis inhabits dryer The best tree differs substantially from previous phylo- islands with shorter open forest (Gibbons 1984). A much genetic hypotheses of Sauromalus,which were based on scu- larger Brachylophus existed in Tonga in the Holocene (Pregill tellation, body size, and the color and banding patterns on 1993), but is now extinct. the dorsum (Shaw 1945). Previous allozyme and karyotype studies (Robinson 1972; Murphy 1983a) did not obtain DISCUSSION enough information for phylogenetic inference, yet these studies are in agreement with the cytochrome b phylogeny. Our results support the hypothesis that the gigantic en- The placement of genera of iguanines in our tree is consistent demic species S. hispidus and S. varius changed in size in with the most parsimonious trees based on morphological response to the island environment (Case 1978, 1982). From characters by de Queiroz (1987) and Norell and de Queiroz the phylogenetic hypothesis we infer one change from large (1991), except that they were unable to resolve the placement size to small size as Sauromalus split from Iguana stock, and of Ctenosaura and Sauromalus. The iguanine topology is one transition back to large size as the ancestor of the mono- identical to the most-parsimonioustrees determinedfrom se- phyletic gigantic species diverged from smaller peninsular quence analysis of the mitochondrial ND4-leucine/tRNA Sauromalus. Evidence from captive breeding and dentary ring genes (Sites et al. 1996). Mitochondrial DNA possesses nu- comparisons (Case 1976; Sylber 1985; C. R. Tracy, unpubl. merous advantagesover nuclear loci as tools for evolutionary data) suggest that size differences between mainland S. obe- reconstructions (Avise 1986, 1989; Moore 1995), however sus populations as well as mainland/island gigantic species mtDNA genes nevertheless representonly one locus. It would are not caused solely by different population age structures therefore be desirable to compare these results with phylog- or phenotypic plasticity (e.g., King 1989; Madsen and Shine enies reconstructedusing multiple independent nuclear loci. 1993), and therefore have a genetic component. Total sequence divergence in cytochrome b between these species and the closest small-bodied relative is between 3- Selective Factors on Body Size 4%, whereas body mass has increased by a factor of five. If Iguanines may be relatively large for lizards because unlike one assumes nucleotide substitution rates are similar in these most other lizards they are herbivores, and greater size en- lineages, many mainland S. obesus show little size change ables more efficient processing of plant material (Pough over comparable levels of evolutionary time, and Ctenosaura 1973; Case 1979a). In contrast with carnivorous lizards, the morphology of herbivores is not as constrained by the size TABLE 5. Results of tests of alternate hypotheses. The number of of food items taken (Schoener 1969, 1974; Case 1979b). In terminal taxa were limited to those previously recognized. The three herbivores, greatervariation in food resource availability may hypotheses represent the alternative trees shown in Figure 3. An select for larger body size (Case 1979a), and larger size may asterisk denotes that the tree is significantly different from the best be favored through sexual selection in territorial iguanines < tree (P 0.05). because larger individuals are better able to defend territories Maximum (Stamps 1977, 1983). Thus there are numerous reasons that No. parsimony Maximum may explain the general trend of large size in herbivorous Tree taxa (steps) SD likelihood (In) SD iguanines, but what factors cause body size differences Best tree 16 955 -5748.0 among Sauromalus? Hyp (A) 16 975* 4.9 -5800.6* 16.2 We consider, it turn, four potential island/mainland dif- Hyp (B) 16 971* 4.5 -5783.3* 11.8 ferences that could affect body size changes: sexual selection, Hyp (C) 16 959* 2.0 -5753.4 4.7 competition, resource availability, and predation. Generally, 214 K. PETREN AND T. J. CASE

-7

Small-> Large S. hispidus

S. varius @000 Small size: (1 81-189 mm) S. obesus obesus 00 0* S. aterklauberi 0000 Large -> Small S. australis 0000 S. aterslevini 0000 S. obesus obesus 0 000 S. obesus tumidus 0 000 Large size: S. obesus townsendl 0000

Iguana 000. Small->Large Ctenosaura @C00O Conolophus 0000 Amblyrhynchus 0 0 0 0 Cyclura 0000 Brachylophus 0000 Dipsosaurus 0000 Oplurus 0 1 2 3 4 FIG.4. Reconstructed evolution of body size in Sauromalus and other iguanines. Ancestral sizes were inferred using the linear change parsimony method (Swofford and Maddison 1987). Thick lines represent generally larger size, thin lines smaller size. Body sizes and sources for terminal taxa are shown in Table 2. Amblyrhynchuswas not included in the analysis, but is thought to be the sister clade to the other Galapagos genus Conolophus(Sites et al., in press). Ancestral reconstructedsizes did not differ qualitatively when the maximum ranges of body sizes for genera of iguanines or for mainland populations of S. obesus were used. Arboreal and saxicolous behavioral traits and island status of terminal taxa are included to evaluate the context of body size evolution. Filled circles represent presence of the trait, shaded circles indicates reduced or variable presence among species. islands harborfewer competitors than mainland habitats, but west North America, the transition toward increasing aridity historically mainland chuckwallas have had few competitors in the late Miocene and Pliocene (Axlerod 1979) may have for food resources. The island gigantic Sauromalus exhibit selected for alternative antipredatorstrategies as vegetation differences in territorial behavior (Case 1982; Carothers became more scarce. At some point after diverging from 1984), however the observed pattern that the larger species Iguana, ancestral Sauromalus evolved extremely saxicolous are less territorialis contrary to the predictions of the sexual behavior. selection hypothesis. Both mainland and island habitats have Mainland Sauromalus are found almost exclusively near highly variable food resources, as droughtscan last for years. rocks with deep, vertical crevasses. When threatened, they However as will become clear, differences in behavior pat- inflate their body cavity to wedge themselves in, often sus- terns between mainland and island gigantic chuckwallas can- pended above the ground, making extraction extremely dif- not be accounted for solely by differences in resource vari- ficult. Smaller chuckwallas can retreat into places that are ability (Case 1982). less accessible to larger predators than larger chuckwallas. Islands often have fewer large predators,and iguanines on Mainland mammals such as coyotes and foxes are known to islands lacking mammalian predators are generally large. prey on chuckwallas (Berry 1974; Case 1982), and the Pleis- Most large mainland iguanines occupy habitats where an ar- tocene presumably had an even richer predator fauna. This boreal escape or bounding into thickets is a feasible defense combination of smaller size and antipredatorbehavior may against larger predators. According to the phylogeny, the function as a predator defense in other herbivorous lizards. Iguana-Sauromalus ancestor was very likely semiarboreal, For example Ctenosaura defensor are small bodied relative much like extant mainland Iguana or Ctenosaura. In south- to other species in this genus, and have spiked tails and a BODY SIZE EVOLUTION IN CHUCKWALLAS 215 behavior whereby they retreat into a hole and curl their tail lective factors on this island are not similar to those on the to block the entrance. This complex of charactersis also seen islands with gigantics, and small size is favored in spite of in the old world agamid lizard genus Uromastyx.Dipsosaurus the absence of large predators; (2) in this population, (and dorsalis, which frequent more open terrain throughout the possibly other S. ater populations on deep-water islands not range of Sauromalus, are extremely wary, using flight and sampled here), genetic variation in the determinants of size taking refuge in mammalburrows. Interestingly, Dipsosaurus is lacking; and (3) this populationhas not had time to diverge, are small bodied and are also known to inflate their bodies in spite of not being on a landbridge island, because of more to hinder extraction (Smith 1946). recent overseas colonization or human introduction (see be- When pursued, the gigantic S. hispidus and S. varius often low). do not run into deep rock crevasses, but merely freeze or Because genetic divergences generally representmaximum hide under a bush or in a superficial burrow under a solitary estimates of the time an island population has been separated rock. They are very often found in the middle of arroyos far from a mainlandpopulation, it is possible that other mainland from rock outcrops: places where mainland individuals are populations may be more closely related to the island pop- rarely seen. The gigantic island species are clearly less lim- ulation than those sampled. This becomes important when ited spatially in foraging activity, and also show increased mainland populations show substantial variation over short temporal activity patternscompared to mainlandchuckwallas distances as is true for Sauromalus (Table 3; Avise et al. (Case 1982). The importance of large predators (and their 1989; Lamb et al. 1992), making complete sampling of all absence on islands) is furthersupported by data on tail break haplotypes difficult. This interpretationis supported by the frequencies, which show three- to tenfold increases in main- result that peninsular S. australis are more closely related to land populations relative to these island species (Case 1978). S. ater klauberi on Santa Catalina than they are to S. ater The only detectable sources of predation on the islands of slevini on landbridge islands adjacent to the peninsula (Fig. Angel de la Guardaand San Esteban are birds of prey, ravens, 2, Table 3). Also, S. ater slevini on Monserrate exhibit less and rattlesnakes (Case and Cody 1983). These predatorsmay than 0.6% sequence divergence from populations on the near- even select for larger size, as larger chuckwallas may escape by islands of Carmenand Danzante, suggesting that this pop- the prey size window for these predators. The endemic rat- ulation may have been recently established. The latterislands tlesnake Crotalus mitchellii angelensis on Angel de la Guarda had recent landbridge connections with the adjacent penin- is much larger than other C. mitchellii (Case 1978). They sula, but are not believed to have had historical landbridge too to prey on S. hispidus, however many are large pass connections with Monserrate,an island of questionable origin through the digestive tract as evidenced by regurgitated (Murphy 1983a). Therefore conclusions regarding the lack and dead rattlesnakes with very large adult S. chuckwallas, of evolution of large size in S. ater klauberi must await more hispidus part way through the digestive tract (Case 1982; extensive sampling of mainland and island populations. Petren and Case, pers. obs.). Dipsosaurus dorsalis presents another contrast to the evo- Thus there are three potential mechanisms where differ- lution of gigantism on islands in the Sea of Cortez, as they ences in predation in these island chuckwalla populations are found on numerous islands in the Sea of Cor- may select for increased size: (1) reduced predation from deep-water have not large predators, eliminating selection for extreme saxicoly, tez without large predators, yet evolved large size. for size stasis for ater forms and eliminating this constraint on body size; (2) reduced The possible reasons discussed S. predation leading to increased access to food resources both above also applies to Dipsosaurus, with the additional pos- temporally and spatially, and selection favoring larger body sibility that genetic, physiological, or ecological constraints sizes in general for herbivores; and (3) continued predation unique to these more distantly related iguanines may prevail. from smaller predators, favoring increased body size to es- cape the window of vulnerability. The relative contributions Genetic Differentiation in Sauromalus of these factors cannot be partitioned without manipulative experiments. The mid-northernBaja California region was thought to be a zone of overlap in the ranges of S. obesus obesus to the Lack of Large Size Evolution in north and S. australis to the south, however the morphological Sauromalus ater klauberi and Dipsosaurus evidence for this hypothesis is equivocal (Gates 1968; Welsh 1988). Cytochrome b haplotypes of chuckwallas from this The population of Sauromalus ater klauberi on Santa Cat- region appear to be distinct from other mainland species, alina Island appears to contradict the pattern of large size suggesting that furtherinvestigations and perhaps taxonomic evolution on deep-water islands. Santa Catalinawas certainly revisions are necessary. not connected to other land masses for tens of thousands of Sauromalus hispidus populations on the Holocene land- years, much like the islands of Angel de la Guarda and San bridge islands of La Ventana and Smith (Fig. 1) have nearly Esteban (Gastil et al. 1983). Accordingly, this island is char- identical cytochrome b sequence as some Angel de la Guarda acterized by high levels of endemism (Case and Cody 1983). S. hispidus. Because these islands have not had recent con- Our estimates of genetic divergence between this population nections to the deep-water island of Angel de la Guarda,the and S. australis are less than, but similar in magnitude to the most likely explanationis recent introductionsby Seri Indians levels of divergence between the gigantic species (S. hispidus and local fisherman for food (Aschmann 1959; Robinson and S. varius) and the adjacent Baja California population. 1972; Bahre 1980). The population on San Lorenzo Sur may There are three possible explanations for this result: (1) se- have also been recently introduced by humans, but there re- 216 K. PETREN AND T. J. CASE mains the possibly of a recent landbridgeconnection to Angel that female (and perhaps male) dispersal and gene flow is de la Guarda (Gastil et al. 1983; Murphy 1983a,b). limited. Avise et al. (1989) uncovered similar results based Monophyly of the gigantics was very well supported by on an analysis of whole-mtDNA sequence divergence, and bootstrap and maximum-likelihood analysis. Furthersupport concluded that chuckwallas harbored some of the highest for the close relationship of these species comes from ob- mtDNA diversity among . One reason for this pattern servations that S. hispidus and S. varius interbreedin captivity may be the extremely saxicolous behavior of mainland Sau- (Sylber 1985), and on a small islet in the eastern Sea of Cortez romalus. The disjunct distribution of rock outcrops suitable (Pelicano Island) where they have been recently introduced for habitation may lead to low levels of migration and gene (Robinson 1972; Bahre 1980; Petren and Case, pers. obs.). flow between populations, and an isolation-by-distance pat- This sister relationship is surprising in light of the biogeo- tern of genetic divergence. Because of this pattern,we cannot graphic affinities of their respective islands. Geological ev- attributelarger genetic divergences between Sauromaluspop- idence indicates that Angel de la Guarda and San Esteban ulations to past vicariant events, as intermediate haplotypes owe their origins to separate sides of the gulf (Moore 1973; may exist in unsampled geographic regions. These results Gastil et al. 1983; Murphy 1983a,b; Lonsdale 1989). The highlight the importanceof undertakingextensive geographic flora and fauna of Angel de la Guardamost closely resembles sampling before concluding that current genetic differences that of the Baja California peninsula, while species on San result from historical processes (Templeton et al. 1995). Esteban more closely resemble those of the Sonora, Mexico Phylogenetic reconstruction may be similarly affected by mainland (Soule and Sloan 1966; Case and Cody 1983; Cody highly fragmented patterns of genetic divergence. Substitut- et al. 1983; Lawlor 1983; Murphy 1983a,b). Perhaps further ing sequences from different populations of the main S. obe- molecular studies on other species in the region will uncover sus clade resulted in dramaticallydifferent bootstrapsupport patterns similar to those found in Sauromalus. for nodes elsewhere on the tree. Less complete sampling The phylogenetic topology suggests a peninsularorigin for within the ranges of recognized subspecies would have re- the common ancestor of S. varius and S. hispidus. However sulted in bootstrap values of 50% or 85% depending on we cannot distinguish between vicariance and overwater dis- chance. The result that bootstrap values may be profoundly persal modes of colonization of Angel de la Guarda. This affected by sequence variation within recognized taxonomic island is believed to have rifted from the peninsula approx- categories suggests that caution be exercised in placing con- imately one million years ago (Moore 1973). Because the fidence in bootstrap values before between-population vari- geologic evidence suggests San Esteban originated on the ation has been evaluated. The generality of this problem is opposite side of the gulf, and had no land connections to difficult to assess, as mitochondrial phylogenetic studies of- Angel de la Guarda, it seems likely that ancestral S. varius ten do not address genetic divergence within taxonomic cat- colonized San Esteban via overwater dispersal. Brachylophus egories. ancestors may have dispersed great distances over water to colonize their present locations (Gibbons 1981), and an ov- ACKNOWLEDGMENTS erwater dispersal event was most likely necessary for colo- nization of the Galapagos islands by Conolophus and Am- We would like to thank the Mexican government for per- blyrhynchusancestors, although in both cases vicariantevents mission to conduct this study (permitDAN 02318). We would cannot be ruled out (Sites et al., in press). Similarly, the sea like to thank J. W. Sites Jr. for access to specimens and floor in the midriff island region of the Sea of Cortez has comments on the manuscript.We gratefully acknowledge the been extremely dynamic over the last few million years assistance of A. Alberts, B. Epstein, R. N. Fisher, J. Heyman, (Lonsdale 1989), and we cannot discount the existence of a T. Langen, R. Murphy, G. Pregill, R. Radtkey, W. Mautz, C. temporary landbridge between these islands sometime in the Tracy,R. Vale, and T. Wright.We thank R. N. Fisher,J. Kohn, distant past. R. Radtkey, C. Wills, and especially two reviewers whose The tumultuous tectonic and climatic history of the Baja comments greatly improved the manuscript. This work was California peninsula has been implicated in the vicariant di- supported by National Science Foundation grant DEB- vergence of many endemics of this region (Savage 1960; 9318906 and National Institute of Health training grant 5T32 Axlerod 1979; Gastil et al. 1983; Murphy 1983a,b). Unfor- GM 07240 (KP). tunately, there are no known Sauromalus(nor any other squa- mate) fossils of pre-Pleistocene age from Baja Californiathat LITERATURE CITED might allow one to directly evaluate historical biogeographic ANDERSON, S., A. T. BANKIER, B. G. BARELL, M. L. H. DE BRUIJN, models (Gehlbach 1965; Estes 1983; Norell 1989). Other A. R. COULSON, J. DROUIN, I. C. EPERON, D. P. NIERLICH, B. A. show patterns superficially similar to Sauromalus, ROE, F SANGER, P. H. SCHREIER, A. J. H. SMITH, R. STADEN, with differentiated forms on opposite sides of the gulf, as AND I. G. YOUNG. 1981. Sequence and organization of the hu- well as northern and southern Baja California peninsular man mitochondrial genome. Nature 290:457-465. ASCHMANN, H. 1959. The central desert of Baja California: De- forms (Savage 1960; Murphy 1983a,b; Stebbins 1985; Sites mography and ecology. Ibero-Americana, No. 42. Univ. of Cal- et al. 1992). However evaluation of specific biogeographic ifornia, Berkeley. hypotheses must await more complete phylogenetic studies AVERY, D. F, AND W. W. TANNER. 1971. Evolution of the iguanine to determine if other species show congruent patterns of dif- lizards (Sauria, ) as determined by osteological and myological characters. Sci. Bull. Biol. Ser. Brigham Young Univ. ferentiation. 12:1-79. Cytochrome b haplotype divergence is high between geo- AVISE, J. C. 1986. Mitochondrial DNA and the evolution of higher graphically proximate populations of Sauromalus,suggesting animals. Phil. Trans. R. Soc. Lond. B 312:325-342. BODY SIZE EVOLUTION IN CHUCKWALLAS 217

. 1989. Gene trees and organismal histories: A phylogenetic ESRA, G. N., K. BENIRSCHKE, AND L. A. GRINER. 1975. Blood approach to population biology. Evolution 43:1192-1208. collecting technique in lizards. J. Am. Vet. Med. Assn. 167:555- AVISE, J. C., B. W. BOWEN, AND T. LAMB. 1989. DNA Fingerprints 556. from hypervariable mitochondrial genotypes. Mol. Biol. Evol. ESTES, R. 1983. Handbuchder Palioherpetologie, Sauria terrestria 6:258-269. Amphisbaenia. Teil IOA., Verlag, Stuttgart, Germany. AXELROD,D. I. 1979. Age and origin of SonoranDesert vegetation. ETHERIDGE, R. E. 1982. Checklist of the iguanine and malagasy Calif. Acad. Sci. Occ. Pap. 132:1-74. iguanid lizards. Pp. 7-37 in G. M. Burghardt and A. S. Rand, BAHRE,C. J. 1980. Historical Seri residence, range, and socio- eds. of the world. Noyes, Park Ridge, NJ. political structure. Kiva 45:197-209. FARRIS, J. S. 1970. Methods for computing Wagner trees. Syst. BAUM, D. A., AND A. LARSON. 1991. Adaptation reviewed: A phy- Zool. 19:83-92. logenetic methodology for studying character macroevolution. FELGER, R. S., AND C. H. LOWE. 1976. The island and coastal Syst. Zool. 40:1-18. vegetation and flora of the northernpart of the Gulf of California. BERRY, K. H. 1974. The ecology and social behavior of the chuck- Cont. Sci. Nat. Hist. Mus. Los Angeles County. 285:1-59. walla, Sauromalus obesus obesus Baird. Univ. Calif. Pub. Zool. FELSENSTEIN, J. 1985. Phylogenies and the comparative method. 101:1-60. Am. Nat. 125:1-15. BOERSMA,P. D. 1983. An ecological study of the Galapagos marine . 1989. PHYLIP (phylogeny inference package). Vers. 3.5c. iguana. Pp. 157-176 in R. I. Bowman, M. Berson, and A. E. Distributed by the author. Department of Genetics, Univ. of Leviton, eds. Patterns of evolution in Galapagos organisms. Washington, Seattle. American Association for the Advancement of Science, San FELSENSTEIN, J., AND H. KISHINO. 1993. Is there something wrong Francisco, CA. with the bootstrap on phylogenies? A reply to Hillis and Bull. BROOKS,D. R., AND D. A. McLENNAN.1991. Phylogeny, ecology Syst. Biol. 42:193-200. and behavior: A research programin comparativebiology. Univ. FOSTER, J. B. 1964. Evolution of mammals on islands. Nature 202: of Chicago Press, Chicago. 234-235. BROWN, J. H. 1995. Macroecology. Univ. of Chicago Press, Chi- GASTIL, G., J. MINCH AND R. P. PHILLIPS. 1983. The geology and cago. ages of the islands. Pp. 13-25 in T. J. Case, and M. L. Cody, BROWN, W. M., E. M. PRAGER, A. WANG, AND A. C. WILSON. 1982. eds. Island biogeography in the Sea of Cortez. Univ. of Cali- Mitochondrial DNA sequences of primates: Tempo and mode of fornia Press, Berkeley. evolution. J. Mol. Evol. 18:225-239. GATES, G. 0. 1968. Geographical distribution and character anal- BURGHARDT, G. M., AND A. S. RAND. 1982. Iguanas of the world. ysis of the iguanid lizard Sauromalusobesus in Baja California, Noyes, Park Ridge, NJ. Mexico. Herpetologica 24:285-288. CAROTHERS, J. H. 1984. Sexual selection and sexual dimorphism GEHLBACH, F R. 1965. Amphibians and reptiles from the Pliocene in some herbivorous lizards. Am. Nat. 124:244-254. and Pleistocene of North America: A chronological summary CASE, T. J. 1976. Body size differences between populations of and selected bibliography. Texas J. Sci. 17:56-70. the chuckwalla, Sauromalus obesus. Ecology 57:313-323. GIBBONS, J. R. H. 1981. The biogeography of the Brachylophus . 1978. A general explanation for insular body size trends (Iguaninae) including a description of a new species, B. vitiensis in terrestrial vertebrates. Ecology 59:1-18. from Fiji. J. Herp. 15:255-272. 1979a. Optimal body size and an 's diet. Acta Bioth- 1984. Iguanas of the South Pacific. Oryx 18:82-91. eor. 28:54-69. GIBBONS, J. R. H., AND I. F WATKINS. 1982. Behavior, ecology, . 1979b. Characterdisplacement and coevolution in some and conservation of south Pacific banded iguanas, Brachylophus, Cnemidophorouslizards. Fort. Zool. 25:235-282. including a newly discovered species. Pp. 418-441 in G. M. . 1982. Ecology and evolution of insular gigantic chuck- Burghardt and A. S. Rand, eds. Iguanas of the world. Noyes, wallas, Sauromalus hispidus and Sauromalus varius. Pp. 184- Park Ridge, NJ. 212 in G. M. Burghardt and A. S. Rand, eds. Iguanas of the GITTLEMAN, J. L. 1985. Carnivore body size: Ecological and tax- world. Noyes, Park Ridge, NJ. onomic correlates. Oecologia 67:540-554. . 1983. The reptiles: ecology. Pp. 159-209 in T. J. Case, GOULD, S. J., AND E. S. VRBA. 1982. Exaptation-A missing term and M. L. Cody, eds. Island biogeography in the Sea of Cort6z. in the science of form. Paleobiology 8:4-5. Univ. of California Press, Berkeley. GRISMER, L. L., J. A. McGUIRE, AND B. D. HOLLINGSWORTH. 1994. CASE, T. J. AND M. L. CODY. 1983. Island biogeography in the Sea A report on the herpetofauna of the Vizcaino Peninsula, Baja of Cortez. Univ. of California Press, Berkeley. California, Mexico, with a discussion of its biogeographic and CASE, T. J., D. T. BOLGER, AND A. RICHMAN. 1992. Reptilian ex- taxonomic implications. Bull. S. Calif. Acad. Sci. 93:45-80. tinctions: The last ten thousand years. Pp. 91-125 in P. Fiedler HASEGAWA, M., AND T. YANO. 1984. Maximum likelihood method and S. Jain, eds. Conservation biology. Chapmanand Hall, New of phylogenetic inference from DNA sequence data. Bull. Biom. York. Soc. Japan 5:1-7. CODY, M. L., R. MORAN, AND H. THOMPSON. 1983. The Plants. Pp. HEDGES, S. B., R. L. BEZY, AND L. R. MAXSON. 1991. Phylogenetic 49-97 in T. J. Case, and M. L. Cody, eds. Island biogeography relationships and biogeography of xantusiid lizards inferredfrom in the Sea of Cortez. Univ. of California Press, Berkeley. mitochondrial DNA sequences. Mol. Biol. Evol. 8:767-780. DARWIN, C. R. 1859. The origin of species by means of natural HILLIS, D. M., AND J. J. BULL. 1993. An empirical test of boot- selection. Murray, London. strapping as a method for assessing confidence in phylogenetic DE QUEIROZ, K. 1987. Phylogenetic systematics of the iguanine analysis. Syst. Biol. 42:182-192. lizards: A comparative osteological study. Univ. Calif. Pub. HILLIS, D. M., AND J. HULSENBECK. 1992. Signal, noise, and re- Zool. 118. liability in molecular phylogenetic analyses. J. Hered. 83:189- . 1995. Checklist and key to the extant species of Mexican 195. iguanas (Reptilia: Iguanidae). Spec. Publ. Mus. Zool. Univ. Na- HUTCHINSON, G. E. 1959. Homage to Santa Rosalia or why are cional Aut6noma de Mexico 9:1-48. there so many kinds of animals? Am. Nat. 93:145-159. DUGAN, B., AND T. A. WIEWANDT. 1982. Socio-ecological deter- IVERSON, J. B. 1980. Colic modifications in iguanine lizards. J. minants of mating strategies in iguanine lizards. Pp. 303-319 in Morphol. 163:79-93. G. M. Burghardt and A. S. Rand, eds. Iguanas of the world. JUKES, T. H., AND C. R. CANTOR. 1969. Evolution of protein mol- Noyes, Park Ridge, NJ. ecules. Pp. 21-132 in H. N. Munro, ed. Mammalian protein EIBL-EIBESFELDT, I. 1984. The large iguanas of the Galapagos Is- metabolism. Academic Press, New York. lands. Pp. 157-174 in R. Perry, ed. Key environments: Gala- KING, R. B. 1989. Body size variation among island and mainland pagos. I. U. C. N., Pergamon, Oxford. snake populations. Herpetologica 45:84-88. ELTON, C. S. 1949. Population interspersion: An essay on animal KISHINO,H., AND M. HASEGAWA. 1989. Evaluation of the maximum community patterns. J. Ecol. 37:1-23. likelihood estimate of evolutionary tree topologies from DNA 218 K. PETREN AND T. J. CASE

sequence data, and the branching order of Hominoidea. J. Mol. PETERS, R. H. 1983. The ecological implications of body size. Evol. 29:170-179. Cambridge Univ. Press, Cambridge. KLAUBER, L. M. 1956. Rattlesnakes. Their habits, life histories and PIANKA, E. R. 1995. Evolution of body size --varanid lizards as a influence on mankind. 2d ed. 2 vols. Univ. of California, Berke- model system. Am. Nat. 146:398-414. ley. POUGH, F. H. 1973. Lizard energetics and diet. Ecology 43:837- KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. EDWARDS, S. PAABO, 844. F X. VILLABLANCA, AND A. C. WILSON. 1989. Dynamics of PREGILL, G. K. 1986. Body size of insular lizards: A pattern of mitochondrial DNA evolution in higher animals: Amplification Holocene dwarfism. Evolution 40:997-1008. and sequencing using conserved primers. Proc. Nat. Acad. Sci. . 1993. Fossil lizards from the late Quaternary of 'Eua, 86:6196-6200. Tonga. Pac. Sci. 47:101-114. LAMB T., T. R. JONES, AND J. C. AvISE. 1992. Phylogeographic RADTKEY, R. R., S. DONNELLAN, R. N. FISHER, C. MORITZ, K. HAN- histories of representative herpetofauna of the southwestern U. LEY AND T. J. CASE. 1995. When species collide: The origin S.: Mitochondrial DNA variation in the (Dipso- and spread of an asexual species of gecko. Proc. R. Soc. Lond. saurus dorsalis) and the chuckwalla (Sauromalus obesus). J. B Biol. Sci. 258:145-152. Evol. Biol. 5:465-480. ROBINSON, A. C., P. D. CANTY, P. A. MOONEY, AND P. M. RUDDUCK. LAWLOR, T. E. 1983. The mammals. Pp. 265-289 in T. J. Case, 1986. South Australia's offshore islands. Australian National and M. L. Cody, eds. Island biogeography in the Sea of Cortez. Parks and Widlife Service Special Publication. Univ. of California Press, Berkeley. ROBINSON, M. D. 1972. Chromosomes, protein polymorphism, and LISTER, A. M. 1989. Rapid dwarfing of red deer on Jersey in the systematics of insular chuckwalla lizards (genus Sauromalus)in last interglacial. Nature 342:539-542. the gulf of California, Mexico. Ph.D. diss. Univ. of Arizona, LONSDALE, P. 1989. Geology and tectonic history of the Gulf of Tucson. California. Pp. 499-521 in E. L. Winterer,D. M. Hussong, and RYAN, M. J. 1982. Variation in iguanine social organization:Mat- R. W. Decker, eds. The geology of North America. Vol. N. The ing systems in chuckwallas (Sauromalus).Pp. 380-390 in G. M. Eastern Pacific and Hawaii. Geological Society of America, Burghardt and A. S. Rand, eds. Iguanas of the world. Noyes, Boulder, CO. Park Ridge, NJ. MACVECTOR. 1991. MacVector sequence analysis programs. Vers. SAITO, N., AND M. NEI. 1987. The neighbor-joiningmethod: A new 3.5. IBI Kodak, New Haven, CT. method for reconstructing phylogenetic trees. Mol. Biol. Evol. MADDISON, W. P. 1991. Squared-change parsimonyreconstructions 4:406-425. of ancestral states for continuous-valued characters on a phy- SAVAGE, J. M. 1960. Evolution of a peninsular herpetofauna.Syst. logenetic tree. Syst. Zool. 40:304-314. Zool. 9:184-212. . 1963. Studies on the lizard Xantusidae. IV. The MADDISON, W. P., AND D. R. MADDISON. 1992. MacClade: Analysis family of phylogeny and character evolution. Vers. 3.0. Sinauer, Sun- genera. Cont. Sci. Los Angeles County Mus. 71:1-38. derland MA. SCHOENER, T. W. 1969. Models of optimal size for solitary pred- ators. Am. Nat. 103:277-313. MADSEN, AND SHINE. in T., R. 1993. Phenotypic plasticity body in in . 1974. Resource partitioning ecological communities. sizes and sexual size dimorphism European grass snakes. Science 185:27-39. Evolution 47:321-325. SCHWANER, T. D. 1985. Population structureof black tiger snakes, MANIATIS, T., E. F. FRITSCH, AND J. SAMBROOK. 1982. Molecular Notechis ater niger, on offshore islands of South Australia. Pp. cloning: A laboratory manual. Cold Spring HarborLaboratory, 35-46 in G. Grigg, R. Shine, and H. Ehrmann, eds. Biology of Cold Spring Harbor,NY. Australian frogs and reptiles. Surrey Beatty, Sydney, Australia. MERTENS, R. 1934. Die Insel Reptilian. Zoologica 84:1-205. SCHWANER, T. D., AND S. D. SARRE. 1990. Body size and sexual MILES, D. B., AND A. E. DUNHAM. 1993. Historical perspectives dimorphism in mainland and island tiger snakes. J. Herp. 24: in ecology and evolutionary biology: The use of phylogenetic 320-323. comparative analyses. Ann. Rev. of Ecol. Syst. 24:587-619. SHAW, C. E. 1945. The chuckwallas, genus Sauromalus.Trans. San MILLER, S. A., D. D. DYKES, AND H. E POLESKY. 1988. A simple Diego Soc. Nat. Hist. 10:269-306. procedure for extracting DNA from human nucleated cells. Nuc. SITES, J. W., J. W. ARCHIE, C. J. COLE, AND 0. F VILLELA. 1992. Acids Res. 16:1215. A review of phylogenetic hypotheses for lizards of the genus MOORE, D. G. 1973. Plate-edge deformation and crustal growth, Sceloporus (Phrynosomatidae): Implications for ecological and Gulf of California structuralprovince. Geol. Soc. Am. Bull. 84: evolutionary studies. Bull. Am. Mus. Nat. Hist. 213:1-110. 1833-1906. SITES, J. W., JR., S. K. DAVIS, T. GEURRA, J. B. IVERSON, AND H. MOORE,W. S. 1995. Inferringphylogenies from mtDNA variation: L. SNELL. In press. Charactercongruence and phylogenetic sig- Mitochondrial-gene trees versus nuclear-gene trees. Evolution nal in molecular and morphological data sets: A case study in 49:718-726. the living iguanas (, Iguanidae). Mol. Biol. Evol. MULLIS, K. B., F. FALOONA, S. SCHARF, S. SAIKI, R. HORN, G. SMITH, H. M. 1946. Handbook of lizards: Lizards of the United EHRLICH, AND H. EHRLICH. 1986. Specific enzyme amplification States and Canada. Cornell Univ. Press, Ithaca, NY. if DNA in vitro: The polymerase chain reaction. Quant. Biol. SMITS, A. W. 1985. Behavior and dietary response to aridity in the 51:263-273. chuckwalla, Sauromalus hispidus. J. Herp. 19:441-449. MURPHY, R. W. 1983a. Paleobiography and genetic differentiation SOULE,M., AND A. J. SLOAN. 1966. Biogeography and distri- of the Baja California herpetofauna. Occ. Pap. Calif. Acad. Sci. bution of the reptiles and amphibians on islands in the Gulf 137:1-48. of California, Mexico. Trans. San Diego Soc. Nat. Hist. 14: . 1983b. The reptiles: Origins and evolution. Pp. 130-158 137-156. in T. J. Case and M. L. Cody, eds. Island biogeography in the STAMPS, J. A. 1977. Social behavior and spacing patternsin lizards. Sea of Cortez. Univ. of California Press, Berkeley. Pp. 265-334 in Biology of the Reptilia. C. Gans and D. W. NORELL, M. A. 1989. Late Cenozoic lizards of the Anza Borrego Tinkle, eds. Vol 7. Ecology A. Academic Press, New York. Desert, California. Cont. Sci. Nat. Hist. Mus. Los Angeles Coun- . 1983. Sexual selection, sexual dimorphism and territori- ty 414:1-31. ality in lizards. Pp. 169-204 in R. B. Huey, E. R. Pianka, and NORELL, M. A., AND K. DE QUEIROZ. 1991. The earliest iguanine T. W. Schoener, eds. Lizard ecology: Studies on a model organ- lizard (Reptilia: Squamata) and its bearing on iguanine phylog- ism. Harvard Univ. Press, Cambridge, MA. eny. Am. Mus. Nat. Hist. Nov. 2997:1-16. STEBBINS, R. C. 1985. Western reptiles and amphibians.Houghton PALUMBI, S., A. MARTIN, S. ROMANO, W. 0. MCMILLAN, L. STICE Mifflin, Boston, MA. AND G. GRABOWSKI. 1991. The simple fool's guide to PCR. SWOFFORD, D. L. 1991. Phylogenetic analysis using parsimony Vers. 2.0. Dept. of Zoology and Kewalo Marine Lab, Honolulu, (PAUP). Vers. 3.0s. Ill. Nat. Hist. Surv., Champaign. HI. SWOFFORD, D. L. AND W. P. MADDISON. 1987. Reconstructing an- BODY SIZE EVOLUTION IN CHUCKWALLAS 219

cestral character states under Wagner parsimony. Math. Biosci. of herbivory and large body size. Pp. 162-183 in G. M. Burg- 87:199-229. hardt and A. S. Rand, eds. Iguanas of the world. Noyes, Park SYLBER, S. K. 1985. Eggs and hatchlings of the yellow giant chuck- Ridge, NJ. walla and the black giant chuckwalla in captivity. Herp. Rev. WELSH, H. H., JR. 1988. An ecogeographic analysis of the her- 16:18-21. petofauna of the Sierra San Pedro Martirregion, Baja California, TANNER, W. W., AND D. F. AVERY. 1964. A new subspecies of with a contribution to the biogeography of the Baja California Sauromalusobesus from the upper Colorado Basin of Utah. Her- herpetofauna. Proc. Calif. Acad. Sci. 46:1-72. petologica 20:38-42. WHITE, T. J., N. ARNHEIM, AND H. A. EHRLICH. 1989. The poly- TEMPLETON, A.R., E. ROUTMAN, AND C. A. PHILLIPS. 1995. Sep- merase chain reaction. Trends Genet. 5:185-189. arating population structurefrom population history: A cladistic WIEWANDT, T. A. 1982. Evolution of nesting patterns in iguanine analysis of the geographical distribution of mitochondrialDNA lizards. Pp. 119-141 in G. M. Burghardt and A. S. Rand, eds. haplotypes in the tiger salamander,Ambystoma tigrinum. Ge- Iguanas of the world. Noyes, Park Ridge, NJ. netics 140:767-782. ZUG, G. R. 1991. The lizards of Fiji: Natural history and system- atics. Bishop Mus. Bull. Zool. 2:1-136. VAN DEVENDER, R. W. 1982. Growth and ecology of spiny-tailed and green iguanas in Costa Rica, with comments on the evolution CorrespondingEditor: J. Sites Jr.