Journal of Biogeography (J. Biogeogr.) (2009) 36, 78–87

ORIGINAL ARTICLE

Climate change drives speciation in the southern rock agama (Agama atra) in the Cape Floristic Region, South Africa Belinda L. Swart1, Krystal A. Tolley2 and Conrad A. Matthee1*

1

Evolutionary Genomics Group, Department of Botany and Zoology, University of Stellenbosch, Stellenbosch, South Africa and 2Molecular Ecology and Evolution Program, South African National Biodiversity Institute, Kirstenbosch Research Centre, Cape Town, South Africa

ABSTRACT

Aim Vicariance has played a major role in the evolution of the southern rock agama, Agama atra (Reptilia: Agamidae), and it is hypothesized that habitat shifts will affect small-scale patterns of gene flow. The Cape Floristic Region (CFR) is known for high levels of diversity and endemism; thus we set out to investigate whether genetic structuring of CFR populations of A. atra corresponds to regional environmental shifts. Location Cape Fold Mountains and the Cape Floristic Region of South Africa. Methods The phylogeographical structure of 116 individuals of A. atra was determined by making use of 988 characters derived from two mitochondrial DNA fragments (control region and the NADH dehydrogenase subunit 2 coding region, ND2). Most animals originated from the CFR, but to gain a better understanding of the processes and patterns of dispersal within the species, 17 additional specimens from outside the CFR were also included and analysed in a phylogenetic context. Results Parsimony and Bayesian analyses revealed four distinct CFR clades (Cape clades) associated with geography. Phylogenetic analyses suggest that populations of A. atra in the CFR region are not entirely isolated from other populations, because some individuals from outside the CFR were nested within the four main Cape clades. The combined mitochondrial DNA data set revealed 59 distinct haplotypes in the CFR. Analysis of molecular variance (amova) confirmed the high degree of genetic structure among the Cape clades, with more than 75% of the genetic variation found among the geographical areas. A spatial amova suggested that a ‘central clade’ originally defined as one of the four Cape clades may contain several additional populations. The main cladogenesis of A. atra within the CFR is estimated to have taken place c. 0.64– 2.36 Ma.

*Correspondence: Conrad A. Matthee, Evolutionary Genomics Group, Department of Botany and Zoology, Stellenbosch University, Private Bag X1, Matieland, Stellenbosch 7602, South Africa. E-mail: [email protected]

78

Main conclusions Agama atra shows at least four distinct genetic provinces within the CFR region, which highlights the conservation importance of this biologically diverse area. The dates of separation among the clades coincide well with the documented Pleistocene climate fluctuations, which might have contributed towards the isolation among lineages; the congruent genetic structure of A. atra with other CFR taxa further supports vicariance as a main isolating factor. Keywords Agamidae, Cape Fold Mountains, coalescence, control region, landscape genetics, lizard, mtDNA, ND2, phylogeography, southern Africa.

www.blackwellpublishing.com/jbi doi:10.1111/j.1365-2699.2008.01988.x

ª 2008 The Authors Journal compilation ª 2008 Blackwell Publishing Ltd

Speciation in the southern rock agama INTRODUCTION The Cape Floristic Region (CFR) represents one of six floral kingdoms and is regarded as a global hotspot for biodiversity (Cowling & Holmes, 1992; Myers et al., 2000). It has been suggested that floral speciation in this region is driven by the unique ‘island-style’ habitat formations that differ in, among other aspects, climate, soil and topography (Linder, 2003). The establishment of the fynbos biome (representing the dominant CFR vegetation type) is complex (Linder, 2005), but some of the major plant radiations have been linked to climatic fluctuations during the Pliocene and Pleistocene (Linder et al., 1992; Midgley et al., 2001; Linder, 2003). In particular, seasonally arid conditions intensified c. 3 Ma in the western CFR (deMenocal, 2004), which is currently within the winter rainfall zone. It has been suggested that the dry, fire-prone habitat potentially allowed some floral groups to radiate rapidly, while mesic pockets remained species-poor (Linder, 2005). As an extension to that hypothesis, it is possible that aridification coupled with fire cycles contributed towards creating more suitable habitat for some faunal species. Thus, climatic fluctuations during the Pleistocene could have fragmented and shifted species distributions repeatedly, thereby

contributing towards allopatric faunal speciation in the region (Tolley et al., 2006; Price et al., 2007). The CFR is inhabited by a unique collection of herptofauna, comprising no fewer than 186 species (28% of which are endemic; Baard et al., 1999). Fragmented distributions have been reported for frogs (e.g. Arthroleptella; Channing et al., 1994) and chameleons (e.g. Bradypodion; Tolley et al., 2004, 2006), but the spatial patterns of genetic and phenotypic variation within and among the remainder of the herptofauna in the CFR are not well understood. The rich biodiversity of CFR reptiles, coupled with naturally fragmented habitat patches, provides a potentially valuable system for studying evolutionary processes driving genetic differentiation among southern African reptiles. In the present study, we investigate patterns of genetic variation in the CFR of a lizard considered to be a rupicolous generalist, the southern rock agama, Agama atra, Daudin 1802 (Reptilia: Agamidae). The species is endemic to southern Africa and has a widespread distribution (Fig. 1; Branch, 1988). It prefers to perch on rocky outcrops or large boulders and tends to avoid areas with dense vegetation (Burrage, 1974). Previous investigations into the phylogeographical structure of A. atra suggested the existence of three reciprocally

(a) A. knobeli Northern-central Southern African clade Southern-eastern Southern African clade

(b)

Figure 1 (a) The distribution of Agama knobeli and the two Agama atra clades in southern Africa (reproduced with permission from Matthee & Flemming, 2002), showing the location of reference samples from outside the Cape Floristic Region (CFR, letters; see Appendix S1 for further details); (b) CFR sampling localities of A. atra used in this study (numbers; see Appendix S1). Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

79

B. L. Swart et al. monophyletic mitochondrial DNA clades (Fig. 1; Matthee & Flemming, 2002), partly supported by allozymes, morphology and life-history data (Mouton & Herselman, 1994; Flemming, 1996). It was suggested that one of the clades (Namibian clade) be recognized as a distinct species, A. knobeli, and it will be treated as such in the present study. The distribution of the three Agama clades corresponds broadly to patterns observed in some other vertebrates, and vicariance has been hypothesized to be the main driving force behind the genetic patterns (Matthee & Robinson, 1996; Lamb & Bauer, 2000; Matthee & Flemming, 2002; Miller-Butterworth et al., 2003; Smit et al., 2007). While broad-scale patterns have been established, a finescale investigation into each clade of A. atra can enhance our understanding of the effects of habitat shifts on patterns of gene flow. This is especially important in the CFR, where high diversity and endemism are thought to have been a result of climatic fluctuations (Linder et al., 1992; Midgley et al., 2001; Linder, 2003; Forest et al., 2007). Given that the CFR is a known hotspot for biodiversity, sampling effort in the present study concentrated on the ‘south-eastern clade’ of Matthee & Flemming (2002), which is geographically centred within the CFR. We hypothesized that the origins of the clade correspond to the establishment of the CFR biome during the Pliocene, and that genetic patterns within this clade correspond to more recent regional environmental shifts. MATERIALS AND METHODS Specimens examined Tissue samples (tail clips) of 114 individuals of A. atra and two of A. knobeli from 49 localities throughout South Africa were used in this study (Fig. 1; see Appendix S1 in Supporting Information). Ninety-nine samples from 37 localities were collected in the CFR and preserved in 96% ethanol. The additional 17 samples were from 12 localities (lettered localities in Fig. 1a) mostly outside the CFR; most originated from a previous investigation by Matthee & Flemming (2002). The latter allowed for a phylogenetic interpretation and also increased the total geographical coverage. Molecular techniques Total genomic DNA was extracted using proteinase K digestion and a phenol/chloroform technique (Sambrook et al., 1989). The mitochondrial DNA (mtDNA) NADH dehydrogenase subunit 2 coding region (ND2) was amplified using primers L4437 (Macey et al., 1997a) and H5934 (Macey et al., 1997b), and an annealing temperature of 53C following standard polymerase chain reaction (PCR) cycling procedures (Palumbi, 1996). Species-specific control region (CR) primers were designed using primer walking: the initial PCR and sequencing was done using the L15162 cytochrome b primer (Palumbi & Kessing, 1991) and a newly designed speciesspecific 12S ribosomal RNA primer (H1204 – 5¢-ACA AGC 80

CTA TAC ATG CAA GC-3¢). This c. 2600-bp region spans the entire CR and allowed for Agama CR primers to be designed (L15850 – 5¢-TAC TGC CTC TAA CCT CAA CC-3¢ and H698 – 5¢-GCT TGC ATG TAT AGG CTT GT-3¢). primer ver. 3 software (Rozen & Skaletsky, 2000) was used for primer selection, and primer numbers correspond to positions on the human mitochondrial genome (Anderson et al., 1981). Internal primers were also designed to facilitate amplification of problematic DNA templates (L15895 – 5¢-AGC TTA ATA CAA AGC GCA GT-3¢ and H592 – 5¢-CAC ATG ATC TTT CCA AGA CC-3¢). Polymerase chain reaction products were visualized on 0.8% agarose gels containing ethidium bromide. Gel purification was carried out using the Wizard gel extraction kit (Promega, Madison, WI, USA). The purified products were cyclesequenced using the BigDye terminator kit version 3.0 (Applied Biosystems, Foster City, CA, USA) and analysed on a 3100 ABI automated sequencer (Applied Biosystems). Sequences were edited with sequence navigator ver. 1.01 (Perkin Elmer, Boston, MA, USA) and alignment was performed in clustalX (Thompson et al., 1997) using default parameters. Phylogenetic analyses Two of the 116 specimens analysed in a phylogenetic context were representatives of the closely related sister species A. knobeli (Matthee & Flemming, 2002), and were used as the outgroup taxon. Two methods of phylogenetic reconstruction were implemented on the combined data set: parsimony and Bayesian inference. Separate analyses of the two markers revealed broadly congruent results and are not presented here (for details see Swart, 2006). The parsimony analysis was implemented in paup* ver. 4.0b10 (Swofford, 2002) using equal weighting of characters, the heuristic search option with tree–bisection–reconnection (TBR) branch swapping, and 100 random additions of taxa. Nodal support was assessed by 1000 bootstrap replicates. A Bayesian analysis was performed using MrBayes ver. 3.1.1 (Huelsenbeck & Ronquist, 2001). ModelTest ver. 3.6 (Posada & Crandall, 1998) was run to investigate the evolutionary model that best fit the data set. Both the Akaike information criteria (AIC) and likelihood ratio test (LRT) specified a twoparameter HKY model + I + G, and MrBayes was run specifying two rate categories with uniform priors for all parameters, and included the gamma distribution and the proportion of invariable sites. Four data partitions were created, and each was allowed to run with separate values for the model parameters: a single data partition for CR, and three partitions for ND2 (first, second and third codons separately). Four independent Markov chain Monte Carlo searches of 10 million generations each were performed for the combined data set. Trees were saved every 1000 generations and burn-in was determined by examination of the split frequencies and the log-likelihood scores for stationarity. The first 4 million generations (4000 trees) were excluded as burn-in. The

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

Speciation in the southern rock agama remaining 6000 trees were used to construct a 50% majority rule consensus tree in paup* ver. 4.0b10 indicating the posterior probabilities for nodes. Uncorrected p-distances between main groupings identified by the tree were estimated using mega ver. 2.1 (Kumar et al., 2001). Population analyses To examine genetic relationships among populations within the CFR, median-joining haplotype networks were constructed using network ver. 4.1 (Bandelt et al., 1999). Networks were initially drawn separately for the two markers (Swart, 2006), and included all CFR individuals (n = 99) for ND2 (549 bp) and CR (439 bp). Haplotype (h) and nucleotide diversity (p) were calculated in arlequin ver. 3.0 (Excoffier et al., 2005). The distribution of mitochondrial variation was further investigated by an analysis of molecular variance (amova, Excoffier et al., 1992) for both FST and FST. Groupings tested by amova were based on an a priori hypothesis of population structure that included clades recovered in the median-joining network and supported by the phylogenetic analyses. To identify genetically distinct geographical groupings that might represent populations, samova ver. 1.0 (spatial analysis of molecular variance; Dupanloup et al., 2002) was used. One hundred simulated annealing processes were performed for a possible number of K populations, ranging from 2 to 10. To determine whether genetic distance (using FST as a proxy) correlated with geographical distance, the Mantel test (Mantel, 1967) was performed in mantel ver. 1.18 (http:// life.bio.sunysb.edu/morph). The data were grouped a posteriori using the most well differentiated sampling localities suggested by the samova. The geographical distances (straight-line distances) were estimated in ArcView gis ver. 3.2 (ESRI, 1999) using the central point in each geographical region. Estimation of divergence times Divergence times between CFR clades were estimated using the program mdiv (Nielsen, 2002). This program is based on the coalescent theory and uses a Bayesian approach to estimate the likelihood parameters of h (effective population size), M (migration rate) and T (time since divergence) between pairs of populations, assuming all populations diverged from a common ancestral population. The pairwise comparisons were run under a finite-site model for 10 million generations each with a burn-in period of 1 million generations (Tmax and Mmax = 10). All pairwise runs were repeated three times to examine whether separate runs converged on the same posterior distributions for the parameters. To confirm the best estimate of posterior distribution, the values of h, M and T were plotted, and credibility intervals (the interval that contains 95% of the posterior probability distribution) were estimated for each parameter where possible. Pairwise estimates of T were converted to years before present using t = Th/2l, where l is the mutation rate. The

mutation rate was first assumed to be 0.65% per lineage per Myr, as suggested for agamid lizards (Macey et al., 1998), and a generation time of 3 years. However, for a more comprehensive analysis, a range of lower rates were also used (0.25% and 0.50% per lineage per Myr) and these rates were estimated directly from the p-distances and a dated phylogeny derived for the sister family to the Agamidae, the Chamaeleonidae (Tolley et al., 2008). Given the criticisms associated with using a constant molecular clock (Bromham & Penny, 2003; Kumar, 2005), we examined whether a constant mutation rate could reasonably be applied to the data set. The unconstrained phylogeny was compared with the same phylogeny with the molecular clock enforced in paup* 4.0b10 (Swofford, 2002). The log-likelihood values of both trees were compared by the ln L ratio test. RESULTS The phylogenetic analyses showed strong support for a monophyletic southern–eastern southern African clade, consistent with the findings of Matthee & Flemming (2002; Fig. 2). This southern–eastern clade contained four strongly supported monophyletic Cape clades (northern CFR, central CFR, Limietberg and Cape Peninsula clades) and one from the Transkei region (east of the CFR; localities labelled h in Fig. 2). Remaining single samples of the southern–eastern clade from outside the CFR were scattered throughout the clade, and localities f, g, j and l were nested within the central CFR clade (Fig. 2). Phylogenetic associations among these clades were not well resolved, and further subdivision within the central CFR clade lacked phylogenetic support. Population differentiation and geographical structure Because networks drawn for individual gene fragments were largely congruent, the data were combined into a single median-joining network (Fig. 3). A total of 59 mtDNA haplotypes (988 bp total: 439 bp of CR and 549 bp of ND2) were identified in 99 CFR samples of A. atra (h = 0.981 ± 0.005; p = 0.0214 ± 0.011). Consistent with the phylogenetic analysis, the haplotype network indicated four main CFR Cape clades differing by at least 25 site changes among haplotypes. The central CFR clade appears to be partitioned into three smaller subclades, although only clade B is strongly supported by Bayesian posterior probability (Fig. 2). When sequence divergences were compared among the four Cape clades (Table 1), the ND2 region consistently showed larger sequence divergences (maximum pairwise value 5.34% ± 0.88) than the CR (4.47% ± 0.94). Within each clade, average sequence divergence was less than 1% for both markers. To investigate the possibility of additional structure within the central CFR, amova was conducted to determine if the observed clades could be statistically supported at a population level. The amova showed that all three subclades within the central CFR clade (A, B and C in Fig. 4) are significantly

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

81

B. L. Swart et al. 11 17 17 16 16 17, 15 11 10 10 37

1.00 94

0.97

1.00 100 1.00 100

1.00 98 1.00 100 1.00 100

1.00 96 1.00 97 a a

1 4 6 2, 1, 6 1 1 1 1 3 3 6

e

b b

g 36, 30, 35, 29 29 27 29 31 28 30 33 34 35 35 29 32 5 8 19 18 l 23 23 27 25 25, 26 27 27 25 25 24, 27 22 21 7 20, 7 j f

i h 1.00 h h 100 12 13 13 14 14 9 9 9 k

Central CFR

Southerneastern Southern African clade

Cape Peninsula Limietberg

Northern CFR

c d

Northern-central Southern African clade A. knobeli

5 changes

different at a population level, with more than 85% of the overall genetic variation among the assemblages (FST = 0.865, P < 0.001; FST = 0.874, P < 0.001). All pairwise comparisons were significant (Table 2). To investigate further the structure within the central CFR clade, a spatial analysis of molecular variance (samova) was run. The proportion of population variance was maximized when the central CFR clade was partitioned into seven populations (FCT = 0.71; Fig. 4). This suggests that while the network and the phylogenetic analysis indicated only three central clades, population structure may be at a finer level within this geographical region. The remaining Cape clades were not included in the samova because these clades represent strongly supported monophyletic assemblages in the phylogenetic tree. The Mantel test revealed no isolation by 82

Figure 2 A parsimony phylogram for Agama atra including the additional samples from other regions in southern Africa (letters) and Cape Floristic Region (CFR) haplotypes (numbers; see Appendix S1 for further details). Bayesian posterior probabilities are indicated above the branches; parsimony bootstrap support values below.

distance (r = 0.17, P = 0.76) among the seven populations identified by the samova. Estimation of divergence times Divergence times were estimated for the four main clades within the CFR that were consistently found and strongly supported by all analyses. Additional sampling at a finer geographical scale will be required to resolve confidently the divergence times of additional subclades within the central CFR clade. The results of the ln L ratio test suggested that a constant clock could not be rejected for these data (k = 20.005; d.f. = 74; v2 = 95.08; P > 0.05). The coalescent-based analysis (Table 3) suggests that the northern CFR and Limietberg clades have the oldest coalescent, dating c. 0.95–2.60 Ma. Most

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

Speciation in the southern rock agama

3 Central CFR Clade A

7

3

Figure 3 Median-joining network for Agama atra showing six main clades: Cape Peninsula, northern Cape Floristic Region (CFR), Limietberg and three central CFR clades. Haplotypes are indicated by circles, with the size of the circle proportional to the number of individuals sharing that haplotype. Connecting branches are proportional to the number of base changes, except where indicated by the actual number on the branches. Median vectors are indicated by black dots and represent intermediate missing haplotypes.

4

3 2

7 2

3

4

Central CFR Clade C

2 2 14

2 13

2

8

8

Cape Peninsula clade

3 3

2 3

22 2

Limietberg clade

One base change

Table 1 Average uncorrected sequence divergences among the four mtDNA clades (control region below diagonal; ND2 above diagonal); standard errors given in parentheses.

Northern CFR Limietberg Cape Peninsula Central CFR

5

3

Northern CFR clade

Central CFR Clade B

2

Northern CFR (%)

Limietberg (%)

Cape Peninsula (%)

Central CFR (%)

– 3.54 (0.81) 2.17 (0.58) 3.13 (0.70)

4.53 (0.84) – 4.07 (0.91) 4.47 (0.94)

4.55 (0.86) 4.70 (0.83) – 1.12 (0.39)

5.34 (0.88) 4.18 (0.08) 4.44 (0.79) –

neous divergence among the clades apart from the relatively recent Cape Peninsula–central CFR divergence. DISCUSSION

of the other clades appear to have diverged within a similar time period. The coalescent of the Cape Peninsula and the central CFR clade is much more recent, between 0.64 and 1.67 Ma. Estimates of migration rates between clades are small, suggesting historically low gene flow between these clades (Table 3). The analysis suggests a more or less contempora-

In the present study, strong support was found for the existence of a distinct southern–eastern clade of A. atra in southern Africa, consistent with the study of Matthee & Flemming (2002). Comprehensive sampling within this clade shows substantial structure among geographical regions and four distinct A. atra mtDNA clades (Cape Peninsula, northern CFR, Limietberg, central CFR). There were no shared haplotypes between any of the four CFR areas, although the coalescent analysis suggests a potentially low level of connectivity among these clades. The population-level analyses uncovered a finer population structure and suggest the existence of at least another three subclades within the central CFR clade. Only one of these subclades (B) is

Figure 4 Geographical distribution of the six Cape clades of Agama atra identified by a median-joining network and supported by amova: white circles = northern Cape Floristic Region (CFR) clade; white square = Limietberg clade; white triangles = Cape Peninsula clade; black circles = central CFR clade, divided into clades A, B and C. Dotted lines indicate geographical distribution of the seven A. atra populations within the central CFR clade identified by samova. The position of Beaufort West (outside the CFR) is indicated by a solid white circle. Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

83

B. L. Swart et al. Table 2 Pairwise FST estimates (below diagonal) and FST values (above diagonal) among the four Agama atra clades; significance values (P) in parentheses.

Northern CFR Limietberg Cape Peninsula Central CFR A Central CFR B Central CFR C

Northern CFR

Limietberg

Cape Peninsula

Central CFR A

Central CFR B

Central CFR C

– 0.942 0.921 0.938 0.921 0.905

0.945 – 0.951 0.942 0.918 0.900

0.927 0.955 – 0.903 0.879 0.858

0.941 (< 0.001) 0.944 (< 0.001) 0.904 (< 0.001)

0.927 0.923 0.884 0.605

0.918 0.913 0.873 0.574 0.656

(< (< (< (< (<

0.001) 0.001) 0.001) 0.001) 0.001)

(< 0.001) (< (< (< (<

0.001) 0.001) 0.001) 0.001)

Table 3 Pairwise estimates of migration rates (M) and time since divergence (T) among the four main Cape clades of Agama atra based on the mtDNA sequence data using the program mdiv [a range of mutation rates was applied (l = 0.25–0.65% per lineages per Myr)]. Northern CFR Limietberg l 0.65% 0.50% 0.25%

Limietberg

Cape Peninsula

M = 0.02 (0.00–1.54) T = 0.95 Ma T = 1.23 Ma T = 2.46 Ma

Cape Peninsula l 0.65% 0.50% 0.25%

M = 0.02 (0.00–0.96)

M = 0.02 (0–6.76)

T = 0.95 Ma T = 1.23 Ma T = 2.47 Ma

T = 1.00 Ma T = 1.30 Ma T = 2.60 Ma

Central CFR l 0.65% 0.50% 0.25%

M = 0.04 (0.02–0.6)

M = 0.08 (0.02–0.84)

M = 0.28 (0.06–1.8)

T = 0.91 Ma T = 1.18 Ma T = 2.36 Ma

T = 0.90 Ma T = 1.17 Ma T = 2.33 Ma

T = 0.64 Ma T = 0.84 Ma T = 1.67 Ma

geographically distinct, whereas the other two subclades (A and C) contained individuals from opposite ends of the study area (Fig. 4). This may be an artefact of geographical sampling within a subset of the range of A. atra, or may instead reflect retention of ancestral polymorphisms or perhaps long-distance dispersal. The samova for the central CFR clade suggested seven populations that are consistent with geography (Fig. 4). Sequence divergence is high among the four reciprocally monophyletic Cape clades (c. 5%) and equals that found between some lizard species (Macey et al., 1998; Matthee et al., 2004; Tolley et al., 2004). Despite this, there is little evidence of substantial morphological differences among these clades (Flemming, 1996; Flemming & Mouton, 2000). The detection of several unexpected genetic assemblages could indicate that A. atra comprises a number of hitherto undescribed evolutionary lineages. 84

(< 0.001) (< 0.001) (< 0.001) (< 0.001) (< 0.001)

0.649 (< 0.001) 0.562 (< 0.001)

(< (< (< (<

0.001) 0.001) 0.001) 0.001)

(< (< (< (< (<

0.001) 0.001) 0.001) 0.001) 0.001)

0.658 (< 0.001)

All analyses showed substantial structuring in the western CFR, an area known for its high diversity in other taxonomic groups (Linder, 2003). Three A. atra clades (northern CFR, Cape Peninsula and Limietberg) have contact points in this area, and all are within the present-day winter rainfall zone. The establishment of a robust winter rainfall zone in the Pliocene with its core in this area (Chase & Meadows, 2007), as well as Pleistocene interglacial and glacial fluctuations, have been proposed as major driving forces in floral speciation in the CFR (Linder et al., 1992; Midgley et al., 2001; Linder, 2003). These climatic shifts may similarly have affected the diversity and distribution of A. atra in the region. Bauer (1999) proposed that limited dispersal within fragmented mountain habitats has strongly influenced phylogeographical patterns in rupicolous lizards. The same processes may be reflected in the distribution of other southern African mountain-dwelling taxa such as the elephant shrew, Elephantulus edwardii (Smit et al., 2007) and the red rock rabbit, Pronolagus rupestris (Matthee & Robinson, 1996). There are no readily apparent extant geographical barriers between the four Cape clades of A. atra, but past barriers to gene flow may have influenced present-day patterns. Some of the differentiation is strong, and consistent with historical fragmentation. This is particularly applicable to explain the large genetic divergences among the four main Cape clades. It is plausible to link the differentiation among the clades to fragmentation due to historical climate changes, as suggested for other CFR fauna (e.g. chameleons, Tolley et al., 2006; cicadas, Price et al., 2007). Within the CFR, numerous taxa also show a northern clade usually centred around the Cederberg (localities 1–3 in Fig. 1), a mountain chain known for its high degree of endemism (Linder & Mann, 1998; Linder, 2003). Coalescent analyses suggest that the diversification events among the four Cape clades range between 0.64 and 2.47 Ma, which corresponds to climatic upheaval in the Pleistocene. Agama atra prefers mesic to arid climatic conditions, with sparse vegetation (Burrage, 1974). Because the winter rainfall zone was fragmented into several isolated dry areas during the Pleistocene (Barrable et al., 2002; Carr et al., 2006; Chase & Meadows, 2007), this could have contributed towards the establishment of the high genetic diversity in the west. The Cape Peninsula clade, which is separated from the nearest extension of the Cape Fold mountains by a plain of flat sands c. 50 km wide (the Cape Flats), represents a special case.

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

Speciation in the southern rock agama It is possible that historical aridification in the western side of the CFR reduced plant cover considerably and thereby provided more suitable habitat for A. atra, particularly so in the Cape Peninsula region. Following an initial colonization event c. 64,000 years ago, the clade may have become isolated by repeated marine transgressions throughout the Pleistocene interglacial (Hendey, 1983a; Mucina & Rutherford, 2006). This isolation may have been enhanced by the subsequent development of dense vegetation on the Cape Flats during mesic periods (Hendey, 1983a,b). Genetic isolation of the Cape Peninsula is also mirrored in other faunal species such as in the isopod genus Mesamphisopus (Gouws et al., 2003) and the freshwater crab Potamonautes brincki (Daniels et al., 2001). This region is also well known for its high levels of plant and invertebrate endemism (Picker & Samways, 1996; TrinderSmith et al., 1996; Linder & Mann, 1998; Linder, 2003). The four A. atra clades found in the CFR highlight the necessity of further fine-scale sampling in this biodiverse region. These clades were not detected in previous broad-scale investigations of this species (Flemming & Mouton, 2000; Matthee & Flemming, 2002), and this study is one of the few faunal investigations conducted to date for the CFR. The congruence between the geographical distributions of the evolutionary lineages found in A. atra and that of other diverse taxa (Daniels et al., 2001; Gouws et al., 2003; Tolley et al., 2006; Price et al., 2007) may indicate the co-fragmentation of biota and further fine-scale analyses are needed. Similar phylogeographical patterns obtained across these studies probably resulted from fragmentation of habitat, but subtle differences in phylogeographical patterns among the taxa would be expected due to differences in life-history characteristics. ACKNOWLEDGEMENTS We would like to thank all the people who helped with the specimen collection in the field or provided tissue samples for this study: M. Burger, M. Cunningham, S. Davies, D. du Toit, A. Flemming, C. Henderson, E. le Roux, C. Oosthuizen, J. Sakwa-Makokha, E. Swartz, A. Turner and K. Whitaker. We would like to thank B. Chase for the sampling map; Cape Nature, the Eastern Cape Department of Tourism and Economic Affairs, for allowing access to their reserves; and all the reserve managers and field rangers for their assistance. This work was funded by the South African National Research Foundation with grants to C.A.M., and some field work funded by WWF Table Mountain Fund with a grant to M. Cunningham. David Hafner, Scott Keogh and an anonymous reviewer are thanked for providing valuable editorial inputs during the review process. REFERENCES Anderson, S., Bankier, A.T., Barrell, B.G., de Bruijn, M.H.L., Coulson, A.R., Drouin, J., Eperon, I.C., Nierlich, D.P., Roe, B.A., Sanger, F., Schreier, P.H., Smith, A.J.H., Staden, R. &

Young, I.G. (1981) Sequence and organization of the human mitochondrial genome. Nature, 290, 457–465. Baard, E.H.W., Branch, W.R., Channing, A.C., de Villiers, A.L., le Roux, A. & Mouton, P. le F.N. (1999) A review of the amphibians and reptiles of the Cape Floristic Region as indicators of centres of biodiversity, sensitive habitats and sites of special interest. Western Cape Nature Conservation Board, Stellenbosch. Bandelt, H.J., Forster, P. & Ro¨hl, A. (1999) Median-joining networks for inferring intraspecific phylogenies. Molecular Biology and Evolution, 16, 37–48. Barrable, A., Meadows, M.E. & Hewitson, B.C. (2002) Environmental reconstruction and climate modelling of the Late Quaternary in the winter rainfall region of the Western Cape, South Africa. South African Journal of Science, 98, 611–616. Bauer, A.M. (1999) Evolutionary scenarios in the Pachydactylus group geckos of Southern Africa: new hypotheses. African Journal of Herpetology, 48, 53–62. Branch, W.R. (1988) Field guide to snakes and other reptiles of southern Africa. Struik Publishers, Cape Town. Bromham, L. & Penny, D. (2003) The modern molecular clock. Nature Reviews Genetics, 4, 216–224. Burrage, B.R. (1974) Population structure in Agama atra and Cordylus cordylus in the vicinity of De Kelders, Cape Province. Annals of the South African Museum, 66, 1–23. Carr, A.S., Thomas, D.S.A., Bateman, M.D., Meadows, M.E. & Chase, B. (2006) Late Quaternary palaeoenvironments of the winter-rainfall zone of southern Africa: palynological and sedimentological evidence from the Agulhas Plain. Palaeogeography, Palaeoclimatology, Palaeoecology, 239, 147–165. Channing, A., Hendricks, D. & Dawood, A. (1994) Description of a new moss frog from the south-western Cape (Anura: Ranidae: Arthroleptella). South African Journal of Zoology, 29, 240–243. Chase, B.M. & Meadows, M.E. (2007) Late Quaternary dynamics of southern Africa’s winter rainfall zone. Earth Science Reviews, 84, 103–138. Cowling, R.M. & Holmes, P.M. (1992) Flora and vegetation. The ecology of fynbos, nutrients, fire and diversity (ed. by R.M. Cowling), pp. 23–61. Oxford University Press, Oxford. Daniels, S.R., Stewart, B.A. & Burmeister, L. (2001) Geographic patterns of genetic and morphological divergence amongst populations of a river crab (Decapoda, Potamonautidae) with the description of a new species from mountains streams in the Western Cape, South Africa. Zoologica Scripta, 30, 181–197. deMenocal, P.B. (2004) African climate change and faunal evolution during the Pliocene–Pleistocene. Earth and Planetary Science Letters, 220, 3–24. Dupanloup, I., Schneider, S. & Excoffier, L. (2002) A simulated annealing approach to define the genetic structure of populations. Molecular Ecology, 11, 2571–2581. ESRI (1999) ArcView 3.2. Environmental Science Research Institute, Redlands, CA, USA.

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

85

B. L. Swart et al. Excoffier, L., Smouse, P.E. & Quattro, J.M. (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics, 131, 479–491. Excoffier, L., Laval, G. & Schneider, S. (2005) Arlequin ver. 3.0: an integrated software package for population genetics data analysis. Evolutionary Bioinformatics (Online), 1, 47–50. Flemming, A.F. (1996) An analysis of the Agama atra complex (Reptilia: Agamidae) in Namaqualand and Southern Namibia. PhD Thesis, University of Stellenbosch, Stellenbosch. Flemming, A.F. & Mouton, P. le F.N. (2000) Geographic variation in sexual size dimorphism in the rock agama, Agama atra (Sauria: Agamidae). African Zoology, 35, 233–249. Forest, F., Grenyer, R., Rouget, M., Davies, T.J., Cowling, R.M., Faith, D.P., Balmford, A., Manning, J.C., Proches¸ , S., van der Bank, M., Reeves, G., Hedderson, T.A.J. & Savolainen, V. (2007) Preserving the evolutionary potential of floras in biodiversity hotspots. Nature, 445, 757–760. Gouws, G., Stewart, B.A. & Daniels, S.R. (2003) Cryptic species within the freshwater isopod Mesamphisopus capensis (Phreatoicidea: Amphisopodidae) in the Western Cape, South Africa: allozyme and 12S rRNA sequence data and morphometric evidence. Biological Journal of the Linnean Society, 81, 235–253. Hendey, Q.B. (1983a) Cenozoic geology and palaeogeography of the fynbos region. Fynbos palaeoecology: a preliminary synthesis (ed. by H.J. Deacon, Q.B. Hendey and N.N.J. Lambrechts), pp. 35–60. Council for Scientific and Industrial Research, Pretoria. Hendey, Q.B. (1983b) Palaeontology and palaeoecology of the fynbos region: an introduction. Fynbos palaeoecology: a preliminary synthesis (ed. by H.J. Deacon, Q.B. Hendey and N.N.J. Lambrechts), pp. 87–99. Council for Scientific and Industrial Research, Pretoria. Huelsenbeck, J.P. & Ronquist, F. (2001) MRBAYES: a program for the Bayesian inference of phylogeny. Available at: http:// mrbayes.csit.fsu.edu/index.php; last accessed 8 August 2008. Kumar, S. (2005) Molecular clocks: four decades of evolution. Nature Reviews Genetics, 6, 654–662. Kumar, S., Tamura, K., Jakobsen, I.B. & Nei, M. (2001) MEGA ver. 2.1. Molecular Evolutionary Genetics Analysis software, Arizona State University, Tempe, AZ, USA. Lamb, T. & Bauer, A.M. (2000) Relationship of the Pachydactylus rugosus group of geckos (Reptilia: Squamata: Gekkodidae). African Zoology, 35, 55–67. Linder, H.P. (2003) The radiation of the Cape flora, Southern Africa. Biological Reviews, 78, 597–638. Linder, H.P. (2005) Evolution of diversity: the Cape flora. Trends in Plant Science, 11, 536–541. Linder, H.P. & Mann, D.M. (1998) The phylogeny and biogeography of Thamnochortus (Restionaceae). Botanical Journal of the Linnean Society, 128, 319–357. Linder, H.P., Meadows, M.E. & Cowling, R.M. (1992) History of the Cape flora. The ecology of fynbos, nutrients, fire and diversity (ed. by R.M. Cowling), pp. 113–134. Oxford University Press, Oxford. 86

Macey, J.R., Larson, A., Ananjeva, N.B. & Papenfuss, T.J. (1997a) Evolutionary shifts in three major structural features of the mitochondrial genome among iguanian lizards. Journal of Molecular Evolution, 44, 660–674. Macey, J.R., Larson, A., Ananjeva, N.B., Fang, Z.L. & Papenfuss, T.J. (1997b) Two novel gene orders and the role of light-strand replication in rearrangement of the vertebrate mitochondrial genome. Molecular Biology and Evolution, 14, 91–104. Macey, J.R., Schulte, J.A., Ananjeva, N.B., Larson, A., RastegarPouyani, N., Shammakov, S.M. & Papenfuss, T.J. (1998) Phylogenetic relationships among agamid lizards of the Laudakia caucasia species group: testing hypotheses of biographic fragmentation and an area cladogram for the Iranian Plateau. Molecular Phylogenetics and Evolution, 10, 118–131. Mantel, N. (1967) The detection of disease clustering and a generalized regression approach. Cancer Research, 27, 209– 220. Matthee, C.A. & Flemming, A.F. (2002) Population fragmentation in the southern rock agama, Agama atra: more evidence for vicariance in Southern Africa. Molecular Ecology, 11, 465–471. Matthee, C.A. & Robinson, T.J. (1996) Mitochondrial DNA differentiation among geographical populations of Pronolagus rupestris, Smith’s red rock rabbit (Mammalia: Lagomorpha). Heredity, 76, 514–523. Matthee, C.A., Tilbury, C.R. & Townsend, T.A. (2004) Phylogenetic review of the African leaf chameleons: genus Rhampholeon (Chamaeleonidae): the role of vicariance and climate change in speciation. Proceedings of the Royal Society B: Biological Sciences, 271, 1967–1975. Midgley, G.F., Hannah, L., Roberts, R., MacDonald, D.J. & Allsopp, J. (2001) Have Pleistocene climatic cycles influenced species richness patterns in the greater Cape Mediterranean Region? Journal of Mediterranean Ecology, 2, 137–144. Miller-Butterworth, C.M., Jacobs, D.S. & Harley, E.H. (2003) Strong population substructure is correlated with morphology and ecology in a migratory bat. Nature, 424, 187– 191. Mouton, P. le F.N. & Herselman, Y.M. (1994) Paradoxical reproduction and body size in the rock lizard, Agama atra atra, Namaqualand, South Africa. South African Journal of Zoology, 29, 199–203. Mucina, L. & Rutherford, M.C. (2006) The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19, South African National Biodiversity Institute, Pretoria. Myers, N., Mittermeier, C.G., da Fonseca, G.A.B. & Kent, J. (2000) Biodiversity hotspots for conservation priorities. Nature, 403, 853–858. Nielsen, R. (2002) MDIV software. Available at: http:// cbsuapps.tc.cornell.edu/mdiv.aspx; last accessed 8 August 2008. Palumbi, S.R. (1996) Nucleic acids II: the polymerase chain reaction. Molecular systematics (ed. by D.M. Hillis, C. Moritz and B.K. Mable), pp. 205–221. Sinauer Associates, Sunderland, MA, USA.

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

Speciation in the southern rock agama Palumbi, S.R. & Kessing, B.D. (1991) Population biology of the trans-arctic exchange: mtDNA sequence similarity between Pacific and Atlantic sea urchins. Evolution, 45, 1790–1805. Picker, M.D. & Samways, M.J. (1996) Faunal diversity and endemicity of the Cape Peninsula, South Africa – a first assessment. Biodiversity and Conservation, 5, 591–606. Posada, D. & Crandall, K.A. (1998) Modeltest: testing the model of DNA substitution. Bioinformatics, 14, 817–818. Price, B.W., Barker, N.P. & Villet, M.H. (2007) Patterns and processes underlying evolutionary significant units in the Platypleura stridula L. species complex (Hemiptera: Cicadidae) in the Cape Floristic Region, South Africa. Molecular Ecology, 16, 2574–2588. Rozen, S. & Skaletsky, H.J. (2000) Primer3 on the WWW for general users and for biologist programmers. Bioinformatics methods and protocols: methods in molecular biology (ed. by S. Krawetz and S. Misener), pp. 365–386. Humana Press, Totowa, NJ, USA. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular cloning: a laboratory manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA. Smit, H.A., Robinson, T.J. & Van Vuuren, B.J. (2007) Coalescence methods reveal the impact of vicariance on the spatial genetic structure of Elephantulus edwardii (Afrotheria, Macroscelidea). Molecular Ecology, 16, 2680–2692. Swart, B.L. (2006) The phylogeography of the southern rock agama (Agama atra) in the Cape Fold Mountains. MSc Thesis, University of Stellenbosch, Stellenbosch. Swofford, D.L. (2002) PAUP*. Phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sinauer Associates, Sunderland, MA, USA. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. & Higgins, D.G. (1997) The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research, 24, 4876–4882. Tolley, K.A., Tilbury, C.R., Branch, W.R. & Matthee, C.A. (2004) Evolutionary history and phylogenetics of the southern African dwarf chameleons, Bradypodion. Molecular Phylogenetics and Evolution, 30, 354–365. Tolley, K.A., Burger, M., Turner, A.A. & Matthee, C.A. (2006) Biogeographic patterns and phylogeography of dwarf chameleons (Bradypodion) in an African biodiversity hotspot. Molecular Ecology, 15, 781–793.

Tolley, K.A., Chase, B.M. & Forest, F. (2008) Speciation and radiations track climate transitions since the Miocene Climatic Optimum: a case study of southern Africa chameleons. Journal of Biogeography, 35, 1402–1414. Trinder-Smith, T.H., Cowling, R.M. & Linder, H.P. (1996) Profiling a besieged flora: endemic and threatened plants of the Cape Peninsula, South Africa. Biodiversity and Conservation, 5, 649–669. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 Specimens of Agama examined in this study, with general names of localities, locality letters or numbers (as in Fig. 1), decimal coordinates and GenBank accession numbers for each mtDNA fragment (ND2, CR). Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

BIOSKETCHES This research resulted from a Master’s study by Belinda Swart. She has a keen interest in phylogeography and is currently employed in the Department of Genetics at Stellenbosch University. Krystal Tolley heads the Molecular Ecology research programme at the South African National Biodiversity Institute and has a keen interest in herpetology. Conrad Matthee is a systematics researcher at Stellenbosch University and has been involved in phylogeographical research on mammals, invertebrates and lizards.

Editor: David Hafner

Journal of Biogeography 36, 78–87 ª 2008 The Authors. Journal compilation ª 2008 Blackwell Publishing Ltd

87