Fragmented landscapes, habitat specificity, and conservation genetics of three lizards in Florida scrub

Conservation Genetics 4: 199 212, 2003. 2003 Kluwer Academic Publishers. Printed in the Netherlands. 199 Fragmented landscapes, habitat specificity, and conservation genetics of three lizards in Florida scrub Lyn C. Branch 1,,A.-M.Clark 2,P.E.Moler 3 &B.W.Bowen 4 1 Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, Florida 32611, USA; 2 BEECS Genetic Analysis Core, 421 Carr Hall, University of Florida, Gainesville, Florida 32611, USA; 3 Florida Fish and Wildlife Conservation Commission, 4005 South Main Street, Gainesville, Florida 32601, USA; 4 Hawaii Institute of Marine Biology, University of Hawaii, P.O. Box 1346, Kaneohe HI 96744, USA ( Author for correspondence: Phone: 352-846-0564; Fax: 352-392-6984; E-mail: BranchL@wec.ufl.edu) Received 9 January 2002; accepted 14 May 2002 Key words: Eumeces, Florida scrub, habitat fragmentation, mitochondrial DNA, Neoseps, phylogeography, Sceloporus, skinks Abstract Dry, sandy scrub habitats of the Florida peninsula represent naturally fragmented remnants of xeric ecosystems that were widespread during the Pliocene and early Pleistocene. This habitat is characterized by high endemism, and distribution of genetic and evolutionary diversity among scrub islands is of compelling interest because Florida scrub is rapidly disappearing under human development. We compare range-wide diversity in mitochondrial cytochrome b sequences for three scrub-associated lizards with contrasting levels of habitat specificity. All species show strong geographic partitioning of genetic diversity, supporting the hypothesis that scrub fauna is highly restricted by vicariant separations. The mole skink (Eumeces egregius), the least habitat specific, has the lowest phylogeographic structure among the lizards (φ st = 0.631). The mtdna geneology for E. egregius is not entirely concordant with the five recognized subspecies and supports a link between populations in central Florida (E. e. lividus) and the Florida Keys (E.e. egregius) rather than a previously proposed affiliation between northern and southern populations. The Florida scrub lizard (Sceloporus woodi) is the most habitat specific of the lizards and has the strongest phylogeographic structure (φ st = 0.876). The sand skink (Neoseps reynoldsi) falls between the mole skink and scrub lizard in terms of habitat specificity and phylogeographic structure (φ st = 0.667). For all three species, networks of mtdna haplotypes coalesce on two central ridges that contain the oldest scrub. The geographic structure and deep evolutionary lineages observed in these species have strong implications for conservation, including strategies for translocation, reserve design, and management of landscape connectivity. Introduction Since the inception of biogeography as a scientific discipline, researchers have searched for concordant patterns in species distributions to illuminate the history of regional biotas. In the last few decades, this search has employed patterns of genetic diversity, or concordance in gene genealogies, to resolve elements of shared evolutionary history. The southeast corner of North America is perhaps the most thoroughly surveyed biogeographic province in this regard, yielding a rich body of genetic and historical information. For example, freshwater vertebrates can generally be partitioned into evolutionary lineages corresponding to drainages of the Atlantic coast, Gulf of Mexico, and peninsular Florida (Bermingham and Avise 1986; Walker and Avise 1998), reinforcing biogeographic boundaries originally postulated from species distributions. Broad patterns of congruence have been revealed for coastal marine organisms (Bowen and Avise 1990; Avise 1992) and, to a lesser extent, terrestrial vertebrates (Avise 2000). However,

200 phylogeographic processes may operate on still smaller scales in terrestrial systems, and few studies have explicitly searched for these patterns. In Florida, patches of dry sandy habitat (known as scrub) are distributed along ridges that extend down the peninsula (Figure 1a). These ridges represent ancient shorelines that formed as the Florida peninsula contracted and expanded with changes in sea level. At least six shorelines were established between the late Pliocene and the most recent interglacial period (Webb 1990). The Mt. Dora and Lake Wales ridges in the center of the Florida peninsula are believed to be the oldest scrub habitats (Figure 1a), followed by progressively younger scrubs towards the contemporary coastline. During high sea levels associated with interglacial periods, the central ridges were isolated for long periods of time, resulting in a plethora of endemic scrub taxa, including plants, insects, lizards, and a bird (Deyrup 1989, 1996; Huck et al. 1989; McCoy and Mushinsky 1992; McDonald and Hamrick 1996). Xeric conditions, a key feature of scrub habitat, were widespread in Florida during the late Pliocene. As mesic habitat expanded during the Pleistocene (Watts and Hansen 1988), xeric habitats were fragmented and shrank into an archipelago of scrub patches. Hence vicariant events associated with the geology, oceanography, and climate of the Florida peninsula may have been major influences on the diversity and present distribution of the region s flora and fauna. If the endemism of Florida scrub taxa is the result of such vicariant separations, this historical process should be reflected in concordant phylogeographic patterns among species. These scrub archipelagos are the setting for a phylogeographic survey of scrub lizards, including the sand skink (Neoseps reynoldsi, family Scincidae), mole skink (Eumeces egregius, family Scincidae), and Florida scrub lizard (Sceloporus woodi, family Phrynosomatidae). The sand skink has a restricted distribution on the central Florida ridges, with populations limited to scrub patches and occasionally sandhill or other habitats with sandy soils (Figure 1b, Christman 1992a; McCoy et al. 1999; Sutton et al. 1999). In contrast, the widespread mole skink inhabits Florida, southern Alabama, and southern Georgia (Figure 1c, Mount 1968). As this distribution implies, the mole skink is less dependent on scrub habitat and occurs on a variety of dry, sandy substrates. Five subspecies of mole skink have been described based on coloration and morphology (including scale counts, tail pigmentation, and position and width of dorsal stripes) but the relationships among these subspecies are uncertain (Mount 1965, 1968). The most widespread subspecies, E. egregius similis, occurs in northern Florida, Alabama, and Georgia. The peninsula mole skink (E. e. onocrepis) occurs throughout much of Florida, off the Lake Wales Ridge (Mount 1968). The blue-tailed mole skink (E. e. lividus) occurs only on the central and southern Lake Wales Ridge in central Florida (Figure 1c). Two subspecies are restricted to islands: E. e. egregius on the Florida Keys and Dry Tortugas, and E. e. insularis on Cedar Key and neighboring islands of the Gulf of Mexico (Figure 1c). These insular populations and those on the central ridges have been impacted heavily by habitat destruction (Christman 1992b, 1992c). The blue-tailed mole skink and the sand skink are listed as Threatened under the U.S. Endangered Species Act (United States Fish and Wildlife Service 1987). The third lizard species, the Florida scrub lizard (Sceloporus woodi), is a scrub endemic that was previously surveyed with mitochondrial DNA (mtdna) cytochrome b sequences (Clark et al. 1999). This species occurs on the central Florida ridges and along the Atlantic Coast (Figure 1d). Populations also previously occurred on the lower Gulf Coast, but these may have been extirpated. Clark et al. (1999) reported deep genetic separations (d max = 9.9%) among disjunct ridges, and strong population structure among scrub patches within ridges. These findings bolster the conclusion, initially formulated from landscape studies (Hokit et al. 1999), that populations of Florida scrub lizards are effectively isolated by broad zones of unsuitable habitat and, more recently, by anthropogenic factors. The observation of ancient separations within S. woodi has clear implications for the management of evolutionary potential and genetic diversity (Bowen 1998; Clark et al. 1999). A key question is whether this pattern extends to other scrub faunas. Are disjunct scrub faunas, in general, characterized by ancient genetic separations? Based on habitat restrictions (irrespective of taxonomy), the scrub lizard and, to a lesser degree, the sand skink would be predicted to have the strongest genetic structure, and the widespread mole skink would be predicted to have the least structure. A ranking system based on taxonomy would predict the opposite, as the mole skink is a polytypic species encompassing five recognized subspecies. The conservation context of this study is compelling because of the precarious status of Florida scrub habitats. Many of the plant species endemic to

Figure 1. (a) Major scrub ridges in Florida (redrawn from Clark et al. 1999) (b) Distribution of the sand skink (redrawn from Christman 1992a, courtesy of the University Press of Florida), (c) Distribution of the five subspecies of the mole skink shown by different shading as follows: E=E. e. egregius, O=E. e. onocrepis, S=E. e. similis, L=E. e. lividus, I=E. e. insularis (redrawn from Mount 1968), (d) Distribution of the Florida scrub lizard (redrawn from DeMarco 1992, courtesy of the University Press of Florida). 201

202 scrub are restricted to the Lake Wales Ridge, and corresponding conservation efforts focus on this area (United States Fish and Wildlife Service 1991). Examination of the geographic patterns of genetic diversity in endemic lizards allows us to address relevant questions such as: 1) Would most of the genetic diversity in these lizards be captured by conservation initiatives focused on the Lake Wales Ridge?, and 2) In order to preserve genetic diversity in these species, is it necessary to preserve a large number of populations, or would a few large populations capture most of the diversity (McCoy and Mushinsky 1994)? Methods Tissue samples (tail segments in most cases) were obtained from throughout the range of the three lizard species (Table 1). Samples were preserved in a saturated salt buffer [saturated NaCl; 25mM EDTA ph 7.5; 20% DMSO] using a protocol modified from Amos and Hoelzel (1991). DNA isolations were conducted with standard phenol/chloroform methodology (Hillis et al. 1996). The cytochrome b portion of the mtdna was amplified using primers Cyb 8: 5 - CGA AGC TTG ATA TGA AAA ACC ATC GTT G- 3", and Cyb 2: 5 -AAA CTG CAG CCC CTC AGA ATG ATA TTT GTC CTC A- 3 (Kessing et al. 1989; Kocher et al. 1989). Polymerase chain reactions (PCR) were accomplished in a 25-ul solution containing 200 um of each dntp, 3.0 mm MgCl 2, and 0.26 um of each primer. The following thermocycling parameters were used: 1 cycle at 94 C (3 min) followed by 35 cycles at 94 C(1min),52 C(1min),and72 C (1 min). For all three species, standard precautions, including negative controls (template-free PCR reactions), were used to test for contamination and to assure the fidelity of PCR reactions. For the Florida scrub lizard, a single-stranded PCR product was generated by denaturing the doublestranded DNA with fresh 0.2 M NaOH and using the non-biotinylated strand as a template for sequencing reactions. Samples from the two skinks were processed as double-stranded PCR products and purified with 30,000 MW filters (Millipore Corp.). Singlestranded and double-stranded PCR products were analyzed with an automated DNA sequencer (Applied Biosystems model 373A, ABI, Foster City, CA) in the DNA Sequencing Core at the University of Florida. Raw data from the sequencer were edited and aligned using Sequencher software (Gene Codes Corp., Ann Arbor, MI). Those mtdna sequences that matched known haplotypes were collated for analysis, whereas new haplotypes were resequenced to assure the accuracy of nucleotide sequence designations. Haplotype diversities (h) and nucleotide diversities (π) were estimated by the method described by Nei (1987: equations 8.4 and 10.5) as implemented in Arlequin version 2.0 (Schneider et al. 2000). Genetic distances (d values) between haplotypes were determined with the Kimura two-parameter model (K2P, Kimura 1980) and empirically derived ti:tv ratios of 4.7:1 for the mole skink, 12.5:1 for the sand skink, and 2.4:1 for the scrub lizard. To determine evolutionary relationships among haplotypes, dendrograms were generated with the parsimony and neighborjoining algorithms of PAUP version 4.0b6 (Saitou and Nei 1987; Swofford 2000). Support for nodes in the mtdna trees was assessed with 100 bootstrap permutations of the dataset (Felsenstein 1985). In addition, mtdna lineages were linked in an unrooted parsimony network and imposed on a map of Florida to resolve phylogeographic patterns. We examined genetic partitioning among scrub ridges for S. woodi and N. reynoldsi using analyses of molecular variance (AMOVA program, Excoffier et al. 1992) implemented in Arlequin version 2.0 (Schneider et al. 2000). Similar analyses were conducted for E. egregius, with geographic partitions corresponding to the ranges of recognized subspecies except for E. e. onocrepis, which was divided into peninsula and Atlantic Coast populations. In addition, we calculated pairwise φ st values and corresponding P values for the subspecies of E. egregius (Schneider et al. 2000). Results Mole skink Forty-one haplotypes were observed in 78 specimens of the mole skink (Table 1). Based on a cytochrome b fragment of 416 bp, we identified 95 variable sites, including 86 transitions and 18 transversions. Sequences from haplotypes A-P have been deposited under Genbank accession numbers AF470631- AF470646. Haplotype diversity was highest in the subspecies endemic to scrub (E. e. lividus) andlowest for the two insular subspecies (Table 1), particularly the Cedar Key mole skink, which also has the most restricted distribution. Sequence divergence between haplotypes of the mole skink averaged 6.3%

Table 1. Estimates of genetic diversity (± SD) with samples partitioned among subspecies a for the mole skink (E. egregius) and among scrub ridges for the sand skink (N. reynoldsi) and the Florida scrub lizard (S. woodi). Species Location Subspecies a N Nucleotide Haplotype Haplotypes diversity (π) diversity (h) Eumeces egregius Florida Panhandle similis 8 0.006 ± 0.005 0.904 ± 0.103 L, M, N, P, X, Z, hh Cedar Key area insularis 7 0.010 ± 0.006 0.250 ± 0.180 A, B Northern Peninsular, FL, onocrepis 39 0.033 ± 0.017 0.884 ± 0.035 B, I, J, K, Q, R, S, T, U, including the Mt. Dora Ridge V, W, ii, jj, kk, nn, oo Atlantic Coast onocrepis 9 0.026 ± 0.015 0.833 ± 0.127 C, D, F, Y, ll, mm Central Florida lividus 10 0.025 ± 0.013 0.956 ± 0.059 G, H, O, aa, bb, cc, dd, ee (Lake Wales Ridge) Florida Keys egregius 5 0.022 ± 0.014 0.700 ± 0.219 E, ff, gg All locations 78 0.057 ± 0.028 0.959 ± 0.011 Neoseps reynoldsi Mt. Dora Ridge 6 0.000 0.000 G North Lakes Wales Ridge 13 0.005 ± 0.004 0.782 ± 0.079 A, B, C, D, W Central Lake Wales Ridge 25 0.012 ± 0.007 0.933 ± 0.036 E, F, H, I, J, K, L, M, N, O, P, Q, R, S, T, U South Lake Wales Ridge 10 0.024 ± 0.013 0.978 ± 0.054 F, V, X, Y, Z, aa, bb, cc, dd All locations 54 0.028 ± 0.014 0.959 ± 0.013 Sceloporus woodi Mt. Dora Ridge 36 0.005 ± 0.004 0.714 ± 0.054 A, B, C, G, I, H, dd Central Lake Wales Ridge 20 0.002 ± 0.002 0.353 ± 0.123 D, L, T South Lake Wales Ridge 19 0.003 ± 0.002 0.386 ± 0.139 J, K, Q, R, S Bombing Range Ridge 30 0.004 ± 0.003 0.756 ± 0.070 D, E, F, M, N, O, U, V Gulf Coast 11 0.000 0.000 P Atlantic Coast 20 0.019 ± 0.011 0.718 ± 0.073 W, X, Y, Z, aa, bb, cc All locations 136 0.039 ± 0.020 0.909 ± 0.013 a Subspecies designations are based on geographic location (Mount 1968), as morphology was not examined in all specimens sampled in the wild. 203 (range, 0.2 11.9%). An analysis of molecular variance revealed significant genetic structure among samples from geographic regions corresponding to current subspecies designations (φ st = 0.631, P < 0.0001; Table 2). Similarly, pairwise comparisons of the fixation indices among subspecies indicated significant differentiation for all pairs of subspecies, but the φ st for Atlantic coast populations and populations on the Lake Wales Ridge was much lower than those for other subspecies pairs (Table 3). A neighbor-joining analysis of the mole skink haplotypes revealed a deep partition between northern and southern ridges (Figure 2a). The subspecies of mole skinks are not monophyletic for mtdna lineages. Haplotype B was observed in both E. e. insularis on Scale Key in the Cedar Key area and E. e. onocrepis on the mainland (Levy County, Table 1). In the phylogenetic tree, haplotype hh from north Florida (E. e. similis, Suwannee County) clustered with haplotypes Y and ll from the Atlantic Coast (E. e. onocrepis, Figure 2a). Other haplotypes from the Atlantic Coast and haplotypes from the Florida Keys (E. e. egregius) clustered with haplotypes from the Lake Wales Ridge (E. e. lividus). Samples from the Mt. Dora Ridge (part of the E. e. onocrepis range) formed a monophyletic group, distinguished by 8.4% divergence from haplotypes on the Lake Wales Ridge. In the parsimony network (Figure 2b), northern lineages, Atlantic coast lineages, and the Florida Keys lineages radiate out from the Lake Wales Ridge. Sand skink For the sand skink, a 425 bp fragment of cytochrome b revealed 30 haplotypes in 54 specimens (Table 1). We documented 53 variable sites with 50 transitions and 4 transversions. Sequences from haplotypes A, B, E, F, M, N, Y, and Z have been deposited under Genbank accession numbers AF470647-AF47054. Overall haplotype diversity in this species was high (h

204 Table 2. Analysis of molecular variance (AMOVA) of mitochondrial DNA haplotype variation with samples partitioned among subspecies a for the mole skink (E. egregius) and among scrub ridges for the sand skink (N. reynoldsi) and the Florida scrub lizard (S. woodi). Source of variation d.f. Variance % of P Fixation components variation index (φ st ) E. egregius Among subspecies 5 9.12 63.1 <0.0001 0.631 Within subspecies 72 5.33 36.9 N. reynoldsi Among ridges 3 5.04 66.7 <0.0001 0.667 Within ridges 50 2.51 33.3 S. woodi Among ridges 4 5.60 87.6 <0.0001 0.876 Within ridges 130 0.79 12.4 a Subspecies designations are based on geographic location (Mount 1968) of sample sites. E. e. onocrepis was divided into peninsula and Atlantic Coast populations for this analysis. Table 3. Pairwise fixation indices (φ st ) for subspecies a of the mole skink. similis insularis onocrepis onocrepis lividus (peninsula) (Atlantic coast) E. e. similis E. e. insularis 0.876 E. e. onocrepis (peninsula) 0.572 0.434 E. e. onocrepis (Atlantic Coast) 0.784 0.759 0.633 E. e. lividus 0.801 0.766 0.621 0.076 E. e. egregius 0.835 0.777 0.592 0.566 0.570 P = 0.03; P = 0.009; all other pairwise fixation indices, P < 0.0001. a Subspecies designations are based on geographic location (Mount 1968) of sample sites. = 0.959). Haplotypes differed by an average sequence divergence of 2.9% (range, 0.2 5.7%). Populations of the sand skink were highly structured (φ st = 0.667, P < 0.0001; Table 2), with most of the genetic variance partitioned among four lineages (Figure 3a). These lineages were distinguished by 10 or more base pair substitutions and corresponded to: 1) the Mt. Dora Ridge, 2) the northern Lake Wales Ridge (Orange, Lake, and northern Polk counties), 3) central Lake Wales Ridge, including Polk County and northern Highlands County, and 4) southern Lake Wales Ridge (south of Josephine Creek in Highlands County). The haplotype from the Mt. Dora Ridge exhibited an average sequence divergence of 3.2% from haplotypes in the northern Lake Wales Ridge and 4.5% divergence from all other haplotypes. Sequence divergence between the most northern population of sand skinks (Mt. Dora Ridge) and the most southern population (Lake Wales Ridge, south of Josephine Creek) averaged 4.8%. Haplotypes central to the parsimony network are all located in the central Lake Wales Ridge (Figure 3b). As in the mole skink, it appears that the oldest lineages radiate out from this ridge. Florida scrub lizard Sequence comparisons (272 bp fragment) among specimens of the Florida scrub lizard revealed 42 variable sites containing 33 transitions and 14 transversions. A total of 30 haplotypes was observed in 136 individuals (Clark et al. 1999). Representative haplotypes have been submitted to Genbank under accession numbers AF144631-AF144635. A large number of haplotypes occurred in scrub lizard populations on all the major scrub ridges except the Gulf Coast, where lizards were found only in one small scrub patch (Table 1). Haplotypes differed by an average sequence divergence of 4.8% (range, 0.37 9.9%).

Figure 2. (a) Neighbor-joining tree depicting the relationship among mtdna haplotypes in the mole skink. Sample localities and subspecies (based on locality, not morphotype) are shown. Numbers at the nodes indicate bootstrap support >60%. = haplotype also found in the Cedar Key area. (b) Parsimony network for the mole skink illustrating geographic linkages among clusters of mtdna haplotypes. For simplicity, representative haplotypes (indicated within circles) are used to illustrate the overall structure of the network. Subspecies (based on locality) for each haplotype are identified as E = E. e. egregius, O=E. e. onocrepis, S=E. e. similis, L=E. e. lividus, I=E. e. insularis. 205

206 Populations of the Florida scrub lizard exhibited strong geographic structure, with the largest fraction of genetic variance found among ridges (φ st = 0.876, P < 0.0001; Table 2). Samples from the five major scrub ridges were characterized by fixed differences in mtdna sequences, except for one haplotype shared between samples from the central Lake Wales Ridge and the Bombing Range Ridge to the east (Figure 4a). In addition, fixed differences occurred between: 1) populations on the central Lake Wales Ridge and the southern part of this ridge, and 2) populations on the north and south segments of the Mt. Dora Ridge. The neighbor-joining analysis demonstrated three primary branches within the Florida scrub lizard, distinguished by 11 21 base substitutions (Figure 4a). The three lineages correspond to 1) the Mt. Dora Ridge, 2) the Bombing Range, Lake Wales, and Gulf Coast ridges, and 3) the Atlantic Coast ridge. The lineage of the Mt. Dora Ridge is distinguished by 4.4% mean sequence divergence from the lineage found on the Bombing Range and Lake Wales ridges. Two sample locations on the Atlantic coast of Florida exhibit 3.4% mean sequence divergence from each other and 7.3% from all other localities (Clark et al. 1999). A parsimony network of haplotypes from the Florida scrub lizard (Figure 4b) reaffirms the relationships observed in the neighbor-joining tree. The Bombing Range Ridge and the central Lake Wales Ridge form one haplotype cluster and the southern Lake Wales Ridge is represented by another cluster. The single haplotype observed on the Gulf Coast of Florida is linked to haplotypes observed on the southern end of the Lake Wales Ridge. Atlantic coast sites are affiliated with the Mt. Dora Ridge. Discussion Figure 3. (a) Neighbor-joining tree depicting the relationship among mtdna haplotypes in the sand skink. Samples were collected on scrub ridges as shown. Numbers at the nodes indicate bootstrap support >60%. = haplotype also found on the south Lake Wales Ridge. (b) Parsimony network for the sand skink illustrating geographic linkages among clusters of mtdna haplotypes. For simplicity, representative haplotypes are used to illustrate the overall structure of the network. The three lizard species surveyed here are endemic to Florida scrub and, in the case of the two skinks, similar habitats with well-drained, sandy soils. All three species show strong population structure and deep intraspecific partitions in mtdna phylogenies, clearly reflecting the earth history (geography, oceanography, and climate) that shaped the distribution of scrub habitats.

207 Population structure and evolutionary partitions Figure 4. (a) Neighbor-joining tree depicting the relationship among mtdna haplotypes in the Florida scrub lizard (redrawn from Clark et al. 1999). Samples were collected on scrub ridges as shown. Numbers at the nodes indicate bootstrap support >60%. = haplotype also found on the Bombing Range Ridge. (b) Parsimony network for the Florida scrub lizard illustrating geographic linkages among clusters of mtdna haplotypes and the probable sequence of colonization from north to south on the Florida peninsula (redrawn from Clark et al. 1999). For simplicity, representative haplotypes are used to illustrate the overall structure of the network. Mole skink Mole skinks show considerable variation in morphology and coloration, and the relationships among subspecies are not well resolved. In the most recent and thorough taxonomic revision of the mole skink, Mount (1965) recognized three previously-described subspecies (E. e. egregius, E. e. onocrepis, and E. e. similis) and described two additional subspecies, E. e. lividus and E. e. insularis. He suggested that the overall geographic variation in this species strongly reflected the Pleistocene history of the southeast, with E. e. lividus (on the central and southern Lake Wales Ridge) probably representing the ancestral stock and E. e. onocrepis (peninsular form) representing an intergrade between the northern populations (E. e. similis) and the Lake Wales Ridge populations. Consistent with this picture, mtdna lineages radiate out from the Lake Wales Ridge (Figure 2b), and some mtdna haplotypes of the proposed intergrade (E. e. onocrepis) clustered with E. e. similis, whereas others clustered with E. e. lividus. TheAtlantic Coast populations clustered most closely with the Lake Wales Ridge populations, rather than with populations from the Mt. Dora Ridge as in S. woodi. The strong genetic divergence (d = 8.4%) between the two oldest ridges Lake Wales Ridge (E. e. lividus) and Mt. Dora Ridge (E. e. onocrepis) suggests a separation of approximately 4 MY based on conventional estimates of molecular evolutionary rates for cytochrome b (Irwin et al. 1991), a time frame that overlaps the proposed Pliocene origin of scrub habitats (Webb 1990). Overall population structure in the mole skink was high (φ st = 0.631), indicating limited dispersal among sandy habitats. However, pairwise comparisons among subspecies revealed less structure for populations on the Atlantic Coast and the Lake Wales Ridge than for other subspecies pairs. The intermingling of haplotypes from the Atlantic Coast and the Lake Wales ridges in the neighbor-joining tree suggests that multiple exchanges may have occurred between these areas, which were separated by water when sea levels were only slightly higher than current levels. The mole skink commonly occurs in coastal habitats, including beaches and tidal wrack, and is found on numerous islands (Mount 1965; Smith et al. 1993). Rafting could be an important dispersal mechanism in this species. Alternatively, the patterns observed for these subspecies could represent incomplete lineage sorting.

208 A number of authors have suggested that the Florida Keys mole skink (E. e. egregius) is most closely related to the northern subspecies (E. e. similis; Taylor 1935; McConkey 1957; Christman 1992c). However, our mtdna analyses demonstrate links between the Florida Keys mole skink and populations in the center of the peninsula, with mtdna haplotypes of the Florida Keys subspecies most closely related to haplotypes of E. e. lividus on the Lake Wales Ridge rather than to haplotypes of the northern subspecies or the geographically nearest populations (Atlantic Coast, E. e. onocrepis). This link between the Florida Keys and central Florida is consistent with the pattern reported by Christman (1980) based on morphological analyses of several snake species. He suggested that part of the Florida Keys, as well as the central ridges, may have remained as islands when the majority of the peninsula was submerged. Thus, these populations may be affiliated due to concordant geographic history. Sand skink Populations of the sand skink exhibit strong genetic structure that corresponds to predictions based on geologic history. As in the mole skink, lineages of this species coalesce on the central Lake Wales Ridge, supporting the view of this ridge as an ancient refugium. Populations of the sand skink on the Mt. Dora Ridge are very distinct from populations on the Lake Wales Ridge, though not as divergent as comparable populations of the mole skink. Under the conventional molecular clock, the 4.5% divergence in sand skinks between these two ridges would correspond to about a 2 MY separation. Florida scrub lizard The parsimony network for the Florida scrub lizards (Figure 4b) shows a strong north-south orientation of mtdna lineages, indicating colonization down the Florida peninsula during the Pliocene (see Clark et al. 1999 for details). Populations on the Mt. Dora Ridge and the Lake Wales Ridge are divergent at alevel(d = 4.4%) that closely corresponds to the parallel estimate for sand skinks (d =4.5%).However, branching patterns of the three primary lineages of scrub lizards indicate dispersal events from the Mt. Dora Ridge, rather than radiation from the Lake Wales Ridge as in the two skinks (Figure 4b). Although the Atlantic Coast populations are highly divergent from populations on the central ridges (d = 7.8 8.0%), the west coast sample indicates a more recent derivation from the Lake Wales Ridge. The phylogeographic data are roughly concordant with the resolution of two morphotypes for the southern (Lake Wales) and northern (Mt. Dora) ridges (Jackson 1973), but raise the possibility of a third evolutionary entity on the Atlantic Coast. Population structure is highly significant in this species with either fixed differences or significant frequency shifts in mtdna haplotypes between most localities separated by more than a few hundred meters of non-scrub habitat (Clark et al. 1999). Considering the low dispersal documented for this species, all lines of evidence indicate that scrub patches within ridges are distinct demographic units (Hokit et al. 1999). Superimposed on this population structure are the separations of 2.0 8.0% between major scrub archipelagos. These deeper divergences likely qualify as evolutionarily significant units (ESUs) under the criteria of Waples (1991, 1995) and Moritz (1994). Comparative phylogeography The mole skink, sand skink, and scrub lizard have a number of notable similarities. Haplotype diversity was relatively high for all three lizard species. However, exceptions to this pattern are apparent; low mtdna diversity was observed in small scrub patches (e.g., in S. woodi on the Atlantic and Gulf coasts; Clark et al. 1999), insular populations (e.g., in E. e. insularis on Cedar Key and adjacent keys), and in N. reynoldsi on the Mt. Dora Ridge. This low diversity may result from loss of haplotypes through genetic drift in small populations, or from founder effects in populations that were colonized by only a few individuals. The strong phylogenetic structure apparent in all three lizard species, particularly the Florida scrub lizard, likely reflects both the historical isolation of lizard populations during the Pliocene and Pleistocene and the low vagility of these species. Hokit et al. (1999) documented very low dispersal rates among adjacent scrub habitats for S. woodi. No dispersal data are available for the two skink species, but the limited distribution of the sand skink and the genetic partitioning in this species are consistent with low vagility and high habitat specificity. These skinks move by swimming through loose sand (Christman 1992a); thus, habitat features such as dense vegetation may restrict dispersal. Mole skinks inhabit a wider range of habitats than either the Florida scrub lizard or

209 the sand skink (Mount 1963). Corresponding genetic structuring was weaker in the mole skink and mtdna lineages did not always correspond to geographic partitions (Figure 2a). These patterns may reflect both higher vagility in this species and the ability to persist in a broader array of habitats. In terms of geography, all three species have a north-south split, associated with the two major Pliocene islands of terrestrial habitat in central peninsular Florida (Mt. Dora and Lake Wales ridges). Networks of mtdna haplotypes coalesce on these ridges (Figures 2, 3, and 4). Hence the genetic data strongly support the hypothesis that scrub habitats on the two central ridges are the cradle of subsequent Pleistocene and Holocene diversity. It is notable that the divergences between northern and southern ridges are 4.5% for the sand skink and 4.4% for the scrub lizard. This striking similarity in evolutionary depth (and geographic orientation) indicates a concordant biogeographic history. Both species demonstrate a vicariant separation that dates approximately to the onset of modern glacial cycles in the late Pliocene (Webb 1990). The 8.4% divergence between ridges in the mole skink indicates an earlier separation in this species. Dispersal, recolonization, and translocation The observed genetic partitions for scrub-associated lizards have several implications for conservation. First, the low dispersal rates indicate that loss of scrub patches represents a critical loss of landscape connectivity for scrub-associated lizards. Demographic rescue and recolonization are likely to occur only among patches that are in close proximity or are linked by suitable habitat (Tiebout and Anderson 1997; Hokit et al. 1999). Secondly, low vagility coupled with the historical isolation of scrub habitats in Florida has led to high levels of genetic diversity and deep intraspecific divergence. This finding prompts concerns about translocations, which have become a prominent strategy for conservation of threatened and endangered species (Griffith et al. 1989; Dodd and Seigel 1991). Translocations have been implemented as mitigation procedures for sand skinks in central Florida (Mushinsky, pers. com.), and reintroductions of mole skinks and Florida scrub lizards are potential conservation strategies in restored habitat (Humphrey et al. 1985; Enge et al. 1986). In the case of scrub-associated species, translocations could compromise the integrity of genetic differences that have accumulated over millions of years. Movement of animals between patches that have been isolated historically may result in loss of unique mtdna haplotypes through genetic swamping or disruption of co-adapted gene complexes and loss of local adaptations (Vrijenhoek 1989; Storfer 1999). Although the importance of these processes in determining fitness of translocated organisms is rarely known, the probability of success in translocation generally is expected to increase with geographic proximity of source of propagules. However, if translocation is used as a management strategy for lizards in Florida scrub, it is important to consider not only the geographic distance between the two areas (which often is correlated with genetic distance between populations), but also the phylogeographic history. Lizard populations located close together but on different ridges may be much more divergent than populations separated by a greater distance but located on the same ridge. Management decisions about translocation should incorporate recognition of strong population structure and evolutionary partitions within these species. Preservation of diversity in endemic scrub lizards: Multiple ESUs in each species The results presented here demonstrate the role of comparative phylogeography in guiding conservation policy (Avise 1992; Moritz and Faith 1998; Rocha et al., 2002). Fragmentation of landscapes through geologic time, and subsequent isolation and genetic divergence of populations, are key processes in the generation of biological diversity (Mayr 1970). In contrast, human-induced habitat fragmentation, on a vastly shorter temporal scale, threatens to eradicate much of this diversity. The conservation challenges are enormous in taxonomic groups, such as scrub lizards, with patchy distributions and high levels of genetic divergence associated with habitat fragments. Most of the variance in lizard mtdna diversity occurs between regions, rather than being contained within local populations. Given the depth of mtdna divergences, indicating separations of millions of years, we believe these results almost certainly reflect molecular evolutionary separations in the nuclear genome as well. Under this assumption, the backbone of programs for conserving genetic diversity in these species should include protection of populations on isolated ridges. Current strategies designed to protect scrub plants, which focus on the Lake

210 Wales Ridge (United States Fish And Wildlife Service 1991), would preserve only a fraction of the genetic diversity in scrub-associated lizards. Fortunately, large populations of the Florida scrub lizard occur on state and federal lands in several areas (e.g., Ocala National Forest, Mt. Dora Ridge; Avon Park Air Force Range, Bombing Range Ridge; Jonathan Dickinson State Park, Atlantic Coast Ridge). In contrast, Gulf Coastal scrub has largely disappeared. The distribution of the blue-tailed mole skink, as currently recognized (Mount 1965), encompasses only part of the Lake Wales Ridge. Areas proposed as reserves for scrub plants (United States Fish And Wildlife Service 1991) may be important in protecting diversity within this subspecies, if the patches are large enough to sustain lizard populations. Likewise, these reserves and Ocala National Forest may be important for preserving genetic diversity in the sand skink. Notably, for all three species of lizards, samples from the Mt. Dora Ridge are genetically distinct from other areas. Clearly, Ocala National Forest, the largest public landholding on the Mt. Dora Ridge, is an important refuge for these lizards and other scrub organisms (Deyrup 1996). A strategy of targeting populations on historically isolated ridges would be effective in capturing the most diversity in the smallest number of protected areas (Pressey et al. 1993). However, in all three species, unique haplotypes were found at all localities where more than five individuals were sampled. Thus, the large-scale development of central and coastal Florida, and corresponding disappearance of scrub habitat, has probably erased some of the genetic diversity in these species. If the relatively deep evolutionary separations observed in endemic lizards are indicative of a general pattern in scrub species, a comprehensive plan will be required to conserve genetic diversity in the endemic flora and fauna of Florida scrub. To the extent that conservation plans strive to protect genetic diversity, individual populations of scrub-associated lizards merit high conservation priority, and populations on major scrub ridges could qualify for protection as distinct evolutionary entities under the U.S. Endangered Species Act. Acknowledgements We thank G. Hokit, B. Stith, and D. Cook for facilitating this project in many ways. For field assistance and tissue collections, we thank A. Abercrombie, F. Antonio, J. Arnett, W. Ballentine, R. Bartlett, J. Berish, S. Berish, A. Coley, K. Dyer, K. Enge, P. Frank, S. Godley, C. Greenberg, J. Jensen, S. Johnson, B. Kaiser, W. Kellner, B. Kemker, B. Mansell, J. Marais, C. May, J. Matter, M. Robson, D. Stevenson, S. Telford, W. Van Devender, J. Vicente, B. Willis, K. Wood, and K. Wray. For logistic support we are indebted to P. Ebersbach, J. Owens, R. Progulske, and P. Walsh. Field studies were facilitated by staff at Avon Park Air Force Range, Archbold Biological Station, Lake Wales State Forest, and Florida Division of Forestry. For assistance with the genetic analysis, we thank E. Almira, S. Shanker, S. Encalada, A. Garcia, A. Bass, and the DNA Seqencing Core at University of Florida. We thank M. Deyrup, R. Franz, S. Edwards, H. Mushinsky, E. McCoy, D. Webb, and many other Florida naturalists who contributed ideas and expertise. R. Franz reviewed the manuscript. J. Ernst provided invaluable assistance with data analysis. This project was funded primarily by the Florida Fish and Wildlife Conservation Commission and the Department of Defense, with additional support from the Biological Resources Division of the USGS, the Species at Risk Initiative of the National BiologicalService, and the University of Florida. This is Florida Agricultural Experiment Station Journal Series No. R-09058. References Amos B, Hoelzel AR (1991) Long-term preservation of whale skin for DNA analysis. Reports of the International Whaling Commission, Special Issue, 13, 99 103. Avise JC (1992) Molecular population structure and the biogeographic history of a regional fauna: A case history with lessons for conservation biology. Oikos, 63, 62 76. Avise JC (2000) Phylogeography: The History and Formation of Species. Harvard University Press, Cambridge, Massachusetts. Bermingham E, Avise JC (1986) Molecular zoogeography of freshwater fishes in the southeastern United States. Genetics, 113, 939 965. Bowen BW (1998) What is wrong with ESUs? The gap between evolutionary theory and conservation principles. Journal of Shellfish Research, 17, 1355 1358. Bowen BW, Avise JC (1990) Genetic structure of Atlantic and Gulf of Mexico populations of sea bass, menhaden, and sturgeon: Influence of zoogeographic factors and life-history patterns. Marine Biology, 107, 371 381. Christman SP (1980) Patterns of geographic variation in Florida snakes. Bulletin of the Florida State Museum, Biological Sciences, 25, 157 256. Christman SP (1992a) Sand skink. In: Rare and Endangered Biota of Florida. Vol. III Amphibians and Reptiles (ed. Moler PE), pp. 135 140. University of Florida Press, Gainesville, Florida. Christman SP (1992b) Bluetail mole skink. In: Rare and Endangered Biota of Florida. Vol. III Amphibians and Reptiles

211 (ed. Moler PE), pp. 117 122. University of Florida Press, Gainesville, Florida. Christman SP (1992c) Florida Keys mole skink. In: Rare and Endangered Biota of Florida. Vol. III Amphibians and Reptiles (ed. Moler PE), pp. 178 180. University of Florida Press, Gainesville, Florida. Clark AM, Bowen BW, Branch LC (1999) Effects of natural habitat fragmentation on an endemic scrub lizard (Sceloporus woodi): An historical perspective based on a mitochondrial DNA gene genealogy. Molecular Ecology, 8, 1093 1104. Demarco V (1992) Florida scrub lizard. In: Rare and Endangered Biota of Florida. Vol. III Amphibians and Reptiles (ed. Moler PE), pp. 141 145. University of Florida Press, Gainesville, Florida. Deyrup M (1989) Arthropods endemic to Florida scrub. Florida Scientist, 52, 254 270. Deyrup M (1996) Two new grasshoppers from relict uplands of Florida (Orthoptera: Acrididae). Transactions of the American Entomological Society, 122, 199 211. Dodd Jr CK, Seigel RA (1991) Relocation, repatriation, and translocation of amphibians and reptiles: Are they conservation strategies that work? Herpetologica, 47, 336 350. Enge KM, Bentzien MM, Percival HF (1986) Florida Scrub Lizard Status Survey. Florida Cooperative Fish and Wildlife Research Unit, Technical Report No. 26, University of Florida, Gainesville, Florida. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: Application to human mitochondrial restriction data. Genetics, 131, 479 491. Felsenstein J (1985) Confidence limits on phylogenies: An approach using the bootstrap. Evolution, 39, 783 791. Griffith B, Scott JM, Carpenter JW, Reed C (1989) Translocation as a species conservation tool: Status and strategy. Science, 245, 477 480. Hillis DM, Mable BK, Larson A, Davis SK, Zimmer EA (1996) Nucleic Acids IV: sequencing and cloning. In: Molecular Systematics (eds. Hillis DM, Moritz C, Mable BK), pp. 321 381. Sinauer Associates, Sunderland, Massachusetts. Hokit DG, Stith BM, Branch LC (1999) The effect of landscape structure in Florida scrub: A population perspective. Ecological Applications, 9, 124 134. Huck RB, Judd WS, Whitten WM, Skean Jr JD, Wunderlin RP, Delaney KR (1989) A new Dicerandra (Labiatae) from the Lake Wales Ridge of Florida, with a cladistic analysis and discussion of endemism. Systematic Botany, 14, 197 213. Humphrey SR, Eisenberg JF, Franz R (1985) Possibilities for restoring wildlife of a longleaf pine savanna in an abandoned citrus grove. Wildlife Society Bulletin, 13, 487 496. Irwin DM, Kocher TD, Alison AC (1991) Evolution of the cytochrome b gene of animals. Journal of Molecular Evolution, 32, 128 144. Jackson JF (1973) Distribution and population phenetics of the Florida scrub lizard, Sceloporus woodi. Copeia, 1973, 746 761. Kessing B, Croom H, Martin A, McIntosh C, McMillan WO, Palumbi S (1989) The Simple Fool s Guide to PCR. Department of Zoology, University of Hawaii, Honolulu, Hawaii. Kimura M (1980) A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution, 16, 111 129. Kocher TD, Thomas WK, Meyer A, Edwards SV, Pääbo S, Villablanca FX, Wilson AC (1989) Dynamics of mitochondrial DNA evolution in animals: Amplification and sequencing with conserved primers. Proceedings of the National Academy of Sciences, 86, 6196 6200. Mayr E (1970) Populations, Species, and Evolution. Harvard University Press, Cambridge, Massachusetts. McConkey EH (1957) The subspecies of Eumeces egregius, a lizard of the southeastern United States. Bulletin of the Florida State Museum, 2, 13 23. McCoy ED, Mushinsky HR (1992) Rarity of organisms in the sand pine scrub habitat of Florida. Conservation Biology, 6, 537 548. McCoy ED, Mushinsky HR (1994) Effects of fragmentation on the richness of vertebrates in the Florida scrub habitat. Ecology, 75, 446 457. McCoy ED, Sutton PE, Mushinksy HR (1999) The role of guesswork in conserving the threatened sand skink. Conservation Biology, 13, 190 194. McDonald DB, Hamrick JL (1996) Genetic variation in some plants of Florida scrub. American Journal of Botany, 83, 21 27. Moritz C (1994) Applications of mitochondrial DNA analysis in conservation: A critical review. Molecular Ecology, 3, 403 413. Moritz C, Faith DP (1998) Comparative phylogeography and the identification of genetically divergent areas for conservation. Molecular Ecology, 7, 419 430. Mount RH (1963) The natural history of the red-tailed skink, Eumeces egregius (Baird). American Midland Naturalist, 70, 365 385. Mount RH (1965) Variation and systematics of the scincoid lizard, Eumeces egregius (Baird). Bulletin of the Florida State Museum, 9, 183 213. Mount RH (1968) Eumeces egregius (Baird). Catalogue of American Amphibians and Reptiles, 73, 1 2. Nei M (1987) Molecular Evolutionary Genetics. Columbia University Press, New York. Pressey RL, Humphries CJ, Margules CR, Vane-Wright RI, Williams PH (1993) Beyond opportunism: Key principles for systematic reserve selection. Trends in Ecology and Evolution, 8, 124 128. Rocha LA, Bass AL, Robertson DR, Bowen BW (2002) Adult habitat preferences, larval dispersal, and the comparative phylogeography of three Atlantic surgeonfishes (Teleostei: Acanthuridae). Molecular Ecology, 11, 243 252. Saitou N, Nei M (1987) The neighbor joining-method: A new method for reconstructing phylogenetic trees. Molecular Biology and Evolution, 4, 406 425. Schneider S, Roessli D, Excoffier L (2000) Arlequin, Version 2000: A Software for Population Genetics Data Analyses. Genetics and Biometry Lab, University of Geneva, Switzerland. Smith LL, Franz R, Dodd Jr CK (1993) Additions to the herpetofauna of Egmont Key, Hillsborough County, Florida. Florida Scientist, 56, 231 234. Storfer A (1999) Gene flow and endangered species translocations: A topic revisited. Biological Conservation, 87, 173 180. Sutton PE, Mushinsky HR, McCoy ED (1999) Comparing the use of pitfall drift fences and cover boards for sampling the threatened sand skink (Neoseps reynoldsi). Herpetological Review, 30, 149 151. Swofford DL (2000) PAUP: Phylogenetic Analysis Using Parsimony, Version 3.1.1. Smithsonian Institution, Washington, D.C. Taylor EH (1935) A taxonomic study of the cosmopolitan scincoid lizards of the genus Eumeces, with an account of the distribution and relationship of the species. University of Kansas Science Bulletin, 23, 1 643.

212 Tiebout III HM, Anderson RA (1997) A comparison of corridors and intrinsic connectivity to promote dispersal in transient successional landscapes. Conservation Biology, 11, 620 627. United States Fish and Wildlife Service (1987) Endangered and threatened wildlife and plants; determination of the threatened status for two Florida lizards. Federal Register, 52, 42658 42662. United States Fish and Wildlife Service (1991) Preliminary Project Proposal: Proposed Establishment of Lake Wales Ridge National Wildlife Refuge. United States Fish and Wildlife Service, Atlanta, Georgia. Vrijenhoek RC (1989) Population genetics and conservation. In: Conservation for the Twenty-First Century (eds. Pearl M, Western D), pp. 89 98. Oxford University Press, New York. Walker D, Avise JC (1998) Principles of phylogeography as illustrated by freshwater and terrestrial turtles in the southeastern United States. Annual Review of Ecology and Systematics, 29, 23 58. Waples RS (1991) Pacific salmon, Oncorhynchus spp., and the definition of a species under the Endangered Species Act. Marine Fisheries Review, 53, 11 22. Waples RS (1995) Evolutionarily significant units and the conservation of biological diversity under the Endangered Species Act. In: Evolution and the Aquatic Ecosystem (eds. Nielsen JL, Powers DA), pp. 8 27, American Fisheries Society, Bethesda, Maryland. Watts WA, Hansen BCS (1988) Environments of Florida in the Late Wisconsin and Holocene. In: Wet Site Archaeology (ed. Purdy BA), pp. 307 323. The Telford Press, Caldwell, New Jersey. Webb SD (1990) Historical biogeography. In: Ecosystems of Florida (eds. Meyers RL, Ewel JJ), pp. 70 102. University of Central Florida Press, Orlando, Florida.