A Molecular Study of Phylogenetic Relationships and Evolution of Antipredator Strategies in Australian Diplodactylus Geckos, Subgenus Strophurus

Home , Crenadactylus ocellatus, Diplodactylinae, Diplodactylus galeatus, Diplodactylus pulcher, Nephrurus vertebralis, Orraya

Blackwell Science, LtdOxford, UKBIJBiological Journal of the Linnean Society0024-4066The Linnean Society of London, 2004? 2004 821 123138 Original Article

ANTIPREDATOR STRATEGIES IN STROPHURUS J. MELVILLE ET AL. Biological Journal of the Linnean Society, 2004, 82, 123–138. With 5 figures

A molecular study of phylogenetic relationships and evolution of antipredator strategies in Australian Diplodactylus geckos, subgenus Strophurus

JANE MELVILLE1,2*, JAMES A. SCHULTE, II2 and ALLAN LARSON2

1Department of Sciences, GPO Box 666E, Museum Victoria, Melbourne, VIC 3000, Australia 2Department of Biology, Washington University, St Louis, MO 63130, USA

Received 26 February 2003; accepted for publication 17 December 2003

We present phylogenetic analyses of the lizard genus Diplodactylus subgenus Strophurus using 1646 aligned positions of mitochondrial DNA sequences containing 893 parsimony-informative characters for samples of 12 species of Strophurus and 19 additional Australian gecko species. Sequences from three protein-coding genes (ND1, ND2 and COI) and eight intervening transfer RNA genes were examined using parsimony, maximum-likelihood and Bayesian analyses. Species of Strophurus appeared to form a monophyletic group with the possible exception of S. taenicauda. Strophurus has evolved two distinct defence/display characteristics: caudal glands, which expel an unpalatable substance, and striking mouth colours. Caudal glands appeared to have arisen once in a common ancestor of Strophurus, with dermal augmentation of caudal glands characterizing a subclade within the subgenus. Evolution of yellow and dark-blue mouth colours in Strophurus occurred in the context of diurnal activity and may be interpreted as an augmentation of defensive behavioural displays. Molecular divergence suggests that arboreality evolved in a common ancestor of Oedura and Strophurus approximately 29 Mya and that the caudal glands of Strophurus arose approximately 25 Mya. © 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 82, 123– 138.

ADDITIONAL KEYWORDS: Australia – defensive display – Gekkonidae – Gekkota – molecular systematics – Pygopodidae – mitochondrial DNA – Reptilia – Sauria.

INTRODUCTION est genus within Diplodactylini, but prior phylogenetic hypotheses suggest that it is not a monophyletic group We examined phylogenetic relationships among diplo- (Greer, 1989). dactyline geckos in Australia with emphasis on Diplo- The 16 species placed in subgenus Strophurus pos- dactylus subgenus Strophurus, a group distinguished sess caudal glands that eject a sticky noxious sub- from its closest relatives by arboreality and diurnal stance through skin ruptures on the mid-dorsal line of activity. Diplodactylus and related genera (Crenadac- the tail to repel predators (Rosenburg & Russell, tylus, Oedura, Rhynchoedura) belong to the taxon 1980). Strophurus is diagnosed also by ﬂeshy cloacal Diplodactylini, one of two major subgroups of the spurs, inscriptional ribs approaching the ventral line, gekkonid subfamily Diplodactylinae (Fig. 1) recog- and absence of the lateral pair of postcloacal bones nized by Kluge (1967a, b, 1987) and Bauer (1990). Mol- (Böhme & Sering, 1997). Russell & Rosenburg (1981) ecular phylogenetic work by Donnellan, Hutchinson & initially placed D. ciliaris, D. elderi, D. michaelseni, Saint (1999) supports monophyly of Diplodactylini but D. spinigerus, D. strophurus, D. taenicauda and provides no strong support for monophyly of the other D. williamsi in the subgenus Strophurus. Rosenburg, subgroup, Carphodactylini. Diplodactylus is the larg- Russell & Kapoor (1984) later discovered caudal glands in D. rankini and added this species to Stro- phurus, while also suggesting that D. wilsoni should *Corresponding author. E-mail: [email protected] be included in this subgenus. Since then, D. assimilis,

124 J. MELVILLE ET AL.

Carphodactylini Diplodactylini

Australia New Caledonia New Zealand Hoplodactylus Naultinus Bavayia Rhacodactylus Eurydactylodes Pseudothecadactylus Phyllurus Nephrurus Underwoodisaurus Carphodactylus Diplodactylus Lucasium Rhynchoedura Crenadactylus Oedura Strophurus

Diplodactylinae

Figure 1. Schematic diagram of hypothetical relationships within Diplodactylinae as proposed by Greer (1989), based primarily on Kluge (1967b). Lucasium has been synonymized with Diplodactylus, and Phyllurus was split into three genera (Orraya, Phyllurus and Saltuarius) in subsequent studies (Kluge, 2001; Hoskin, Couper & Schneider, 2003).

D. intermedius, D. jeanae, D. mcmillani, D. robinsoni, Another striking defence/display feature of most D. taeniatus, D. wellingtonae and D. wilsoni have been Strophurus is a conspicuous mouth colour, either added to Strophurus (Greer, 1989; Cogger, 2000; dark blue or yellow (Greer, 1989). Geckos have a Kluge, 2001), which is given generic status by some characteristic defensive posture that usually involves authors. Although we regard these species tentatively lifting and arching the body, extending the throat, as being part of Diplodactylus, we refer to them in the and slowly waving the raised tail from side to side. remainder of this manuscript using Strophurus in Within Diplodactylinae, this behaviour has been place of the genus name to distinguish them from documented in Nephrurus spp. (Galliford, 1978), other Diplodactylus. Oedura tryoni, Phyllurus platurus, and S. taeniatus Geckos of the subgenus Strophurus can fire from the (Schmida, 1973). Many species open their mouths dorsal surface of their tail thin streams of viscous tail- during these displays, exposing the mouth colour gland secretions aimed by curving the tail (Rosenburg (Bustard, 1964). A blue mouth occurs in S. assimilis, & Russell, 1980; Greer, 1989). Preliminary biochemi- S. ciliaris, S. intermedius, S. rankini, S. spinigerus, cal characterization shows that the substance con- S. strophurus, S. wellingtonae and S. williamsi; a tains protein and glycoprotein and is similar among yellow/orange mouth colour occurs in S. ciliaris and species (Rosenburg et al., 1984). Previous studies sug- S. taeniatus, whereas S. elderi has the pink mouth gest that the caudal glands are evolutionary modifica- characteristic of most Diplodactylini (Greer, 1989). tions of fat bodies that occupy a similar position in We present a molecular phylogenetic study of 12 of many other lizards, including other Diplodactylini the 16 recognized species of Strophurus (excepting (Rosenburg et al., 1984; Greer, 1989). Most Strophurus S. jeanae, S. michaelseni, S. robinsoni and S. wilsoni) lack caudal fat bodies, their place being filled by cau- to examine evolutionary patterns of caudal glands and dal glands (Rosenburg & Russell, 1980). Caudal mouth colours. Sequences reported include a 1853-bp glands have evolved separately in the diplodactyline segment of the mitochondrial genome extending from genus Eurydactylodes from New Caledonia (Böhme & the protein-coding gene ND1 (subunit one of NADH Sering, 1997). dehydrogenase), through the genes encoding tRNAIle,

ANTIPREDATOR STRATEGIES IN STROPHURUS 125 tRNAGln, tRNAMet, ND2 (NADH dehydrogenase sub- loops (dihydrouridine (D), TYC (T) and variable Trp Ala Asn unit two), tRNA , tRNA , tRNA , OL (origin of loops) of some tRNA genes, and some intergenic light-strand synthesis), tRNACys and tRNATyr, to the sequences, were excluded from the phylogenetic protein-coding gene, COI (subunit I of cytochrome c analyses. oxidase). In addition, we sampled eight other species within Diplodactylini, including representatives of three previously recognized species groups of Diplo- PHYLOGENETIC ANALYSIS dactylus, and eight species in Carphodactylini. We also Phylogenetic trees were estimated using PAUP* beta included in our analysis a pygopodid lizard, Lialis version 4.0b10a (Swofford, 2002) with 100 heuristic jicari, because prior studies suggest a phylogenetic searches using random addition of sequences under affinity between pygopodids and diplodactyline geckos the parsimony criterion. Bootstrap resampling was (Donnellan et al., 1999). Two Australian gekkonine applied to assess support for individual nodes using taxa, Heteronotia binoei and Gehyra variegata, and a 500 bootstrap replicates with ten heuristic searches previously published sequence of the south-east Asian featuring random addition of sequences. Decay indices species, Gekko gecko (Macey et al., 1999), were (= ‘branch support’ of Bremer, 1994) were calculated included as outgroups. for all internal branches of the tree using TREEROT version 2a (Sorenson, 1999). We refer to bootstrap percentages and decay indices as heuristic measures of MATERIAL AND METHODS branch support because they provide useful information regarding the amount of character support for SPECIMEN INFORMATION individual branches but they do not constitute statis- Details of the voucher specimens from which DNA was tical tests of hypotheses. extracted are given in Table 1. Maximum likelihood was used to estimate a phylogenetic tree that maximized the probability of the observed data. ModelTest version 3.06 (Posada & LABORATORY PROTOCOLS AND ALIGNMENT OF Crandall, 1998) was used to compare the goodness- DNA SEQUENCES of-fit of different models of sequence evolution to the Genomic DNA was extracted from liver, muscle or data, and to generate optimal likelihood settings. blood using Qiagen QIAamp tissue kits. Different Simultaneous optimization of maximum-likelihood primer combinations were used to amplify from parameters and phylogenetic hypotheses for this genomic DNA (Table 2). Amplifications of genomic data set was computationally impractical. The best- DNA were conducted using a denaturation step at fit model parameters were fixed and the overall 94∞C for 35 s, annealing at 53∞C for 35 s, and exten- most parsimonious tree(s) were used as starting sion at 70∞C for 150 s, with 4 s added to the extension trees for branch swapping in 25 heuristic searches per cycle for 30 cycles. Negative controls were run for with random addition of taxa to find the overall best all amplifications. Amplified products were purified likelihood topology. Bootstrap resampling was on 2.5% Nusieve GTG agarose gels and reamplified applied to assess support for individual nodes using under similar conditions. Reamplified double- 100 bootstrap replicates with ten heuristic searches stranded products were purified on 2.5% acrylamide featuring random addition of sequences. Model gels (Maniatis, Fritsch & Sambrook, 1982). Template parameters were fixed for bootstrap analyses, as DNA was eluted from the acrylamide passively over above. 3 days with Maniatis elution buffer (Maniatis et al., Bayesian analysis was used to estimate a phyloge- 1982). Cycle-sequencing reactions were run using a netic tree using many of the default values in Promega fmol DNA sequencing system with a dena- MrBayes 2.1 (Huelsenbeck & Ronquist, 2001). All turation step at 95∞C for 35 s, annealing at 45–60∞C analyses were initiated from random starting trees for 35 s, and extension at 70∞C for 1 min, for 30 cycles. and run for 2 000 000 generations using four incre- For manual sequencing, reactions were run on Long mentally heated Markov chains. Values of the likeli- Ranger sequencing gels for 5–12 h at 38–40∞C. For hood model selected from the best-fit model of automatic sequencing, reactions were run on an nucleotide substitution using ModelTest were esti- ABI373 system. mated from the data and initiated using flat priors. Alignment of tRNA genes was based on secondary Trees were sampled every 100 generations, resulting structural models (Kumazawa & Nishida, 1993; in 20 000 saved trees. To ensure that Bayesian analy- Macey & Verma, 1997). Secondary structures of ses reach stationarity, the first 5000 saved trees were tRNAs were inferred from primary structures of the discarded as ‘burn-in’ samples following Leaché & corresponding tRNA genes using these models. Reeder (2002). Three analyses were run indepen- Unalignable regions in the three length-variable dently beginning with different starting trees to

Table 1. Museum numbers of voucher specimens from which DNA was extracted, GenBank accession numbers and localities

Voucher specimen Museum # GenBank# Locality

Carphodactylus laevis QMJ8944 AY369017 Lake Barrine, Queensland Crenadactylus ocellatus SAMAR22245 AY369016 10 km south of Barrow Creek, Northern Territory D. conspicillatus ANWC6160 AY369012 80 km west of Alice Springs, Northern Territory D. galeatus ANWCR6161 AY369009 5 km west of Ross River Homestead, Northern Territory D. pulcher WAMR146811 AY369011 Goongarrie Station, Western Australia D. stenodactylus ANWC6151 AY369010 Old Andado Homestead, Northern Territory D. vittatus SAMAR28400 AY369013 25.3 km north-west of Iron Knob, South Australia Gehyra variegata ANWCR6138 AY369026 Old Andado Homestead, Northern Territory Heteronotia binoei ANWCR6147 AY369027 22 km west of Glen Helen Roadhouse, Northern Territory Lialis jicari TNHC 59426 AY369025 New Guinea Is. Irian Jaya, Merauke, Indonesia N. levis SAMAR19968 AY369018 Indooroopilly Outstation, South Australia N. laevissimus WAMR146821 AY369020 Yeo Lake Road, Western Australia N. milii SAMAR19395 AY369022 Taylors Island, South Australia N. vertebralis WAMR146822 AY369019 Banjawarn Station, Western Australia N. wheeleri wheeleri WAMR146823 AY369021 Wydgee Station, Western Australia Oedura marmorata SAMAR34209 AY369015 Lawn Hill, Queensland Saltuarius cornutus SAMAR29204 AY369023 Wiangaree, New South Wales Pseudothecadactylus lindneri AMSR90195 AY369024 Liverpool River, Northern Territory Rhynchoedura ornata ANWCR6141 AY369014 Native Gap, Stuart Highway, Northern Territory S. assimilis SAMAR20750 AY368999 Port Kenny, South Australia S. ciliaris SAMAR29884 AY368996 Curtin Springs, Northern Territory S. ciliaris aberrans WAMR146817 AY368997 Yeo Lakes Road, Western Australia S. elderi SAMAR29924 AY369000 Curtin Springs, Northern Territory S. intermedius SAMAR22768 AY369001 Uro Bluff, South Australia S. mcmillani AMSR126185 AY369008 Mitchell Falls, Western Australia S. rankini SAMAR22889 AY369002 Warroora Homestead, Western Australia S. spinigerus SAMAR22882 AY369003 37 km east of Kalbarri, Western Australia S. strophurus WAMR97160 AY369004 Kalli Homestead, Western Australia S. taeniatus NTMR014930 AY369005 Sangster’s Bore, Tanami Desert, Northern Territory S. taenicauda ABTC27723 AY369006 5 km east of Miles, Queensland S. wellingtonae WAMR146819 AY368998 Banjawarn Station, Western Australia S. williamsi SAMAR25518 AY369007 Danggali CP, South Australia

Sequences were aligned for phylogenetic analysis with a sequence of Gekko gecko (Macey et al., 1999), AF114249, from Phuket Province, Thailand. SAM = South Australian Museum, Adelaide, QM = Queensland Museum, Brisbane, AM = Australian Museum, Sydney, WAM = Western Australian Museum, Perth, NTM = Northern Territory Art Gallery and Museum, Darwin, ANWC = Australian National Wildlife Collection, CSIRO, Canberra, Australia, and TNHC = Texas Natural History Collection, Austin, Texas.

check that searches did not become trapped on local than an alternative or whether the differences in tree optima. For all three runs, ln-likelihood scores con- length could be attributed to chance alone (Larson, verged on similar values. Sampled trees from all 1998). The Wilcoxon signed-ranks test was conducted three runs were combined to yield 45 000 saved trees. as a two-tailed test using PAUP* (Felsenstein, 1985). These trees were used to generate a 50% majority- The SH test of Shimodaira & Hasegawa (1999), a rule consensus tree in PAUP* and the percentage of likelihood-based test of topologies, was used to evalu- trees having a particular clade represented that ate the statistical signiﬁcance of the favoured tree rel- clade’s posterior probability (Huelsenbeck & ative to alternative hypotheses. The statistical power Ronquist, 2001). of the SH test appears comparable to that of the two- A Wilcoxon signed-ranks test (Templeton, 1983; tailed Wilcoxon signed-ranks test (Townsend & Lar- Felsenstein, 1985) was applied to examine whether son, 2002). Phylogenetic topologies were constructed the most parsimonious tree was signiﬁcantly shorter in MacClade (Maddison & Maddison, 1992) and used

Table 2. Primers used in this study

Human position Gene Sequence Reference

L3914-amp ND1 5¢-GCCCCATTTGACCTCACAGAAGG-3¢ Macey et al. (1998) L4178 ND1 5¢-CAACTAATACACCTACTATGAAA-3¢ Macey et al. (1997a) L4437-amp tRNAMet 5¢-AAGCAGTTGGGCCCATRCC-3¢ Macey et al. (1997b) H4419b tRNAMet 5¢- AAGCAGTTGGGCCCATRCC-3¢ Macey et al. (2000) L4419a tRNAMet 5¢-AAGCAATTGGGCTCATACC-3¢ Macey et al. (1997a) L4882b ND2 5¢-TGACAAAAAATTGCNCC-3¢ Macey et al. (2000) H4980-amp ND2 5¢-ATTTTTCGTAGTTGGGTTTGRTT-3¢ Macey et al. (1997a) L4831 ND2 5¢-TGACTTCCAGAAGTAATACAAGG-3¢ Macey et al. (1997a) L5556b tRNATrp 5¢-GCCTTCAAAGCCCTAAA-3¢ Macey et al. (1997a) H5617b tRNAAla 5¢-CTGAATGCAACTCAGACACTTT-3¢ Macey et al. (1997a) H5934a-amp COI 5¢-AGRGTGCCAATGTCTTTGTGRTT-3¢ Macey et al. (1997a) H6159-amp COI 5¢-ATAATTGGAGCCCCAGACATAGC-3¢ Weisrock et al. (2001)

Primers are designated by their 3¢ ends, which correspond to the position in the human mitochondrial genome (Anderson et al., 1981) by convention. H and L designate primers whose extension produces the heavy and light strands, respectively. Primers used to amplify are designated by ‘amp’ following the primer name. Positions with mixed bases are labelled with the standard one-letter code: R = G or A, Y = C or T, N = G, A, C, or T.

as constraints in PAUP*. A search, using the maxi- Among tRNA genes, several loop regions were mum-likelihood settings determined previously in unalignable, as were non-coding regions between MODELTEST and starting from a neighbour-joining genes. Parts of the D- and T-loops for the genes encod- tree, was used to estimate a phylogenetic tree that ing tRNATrp (positions 1417–1426 and 1460–1467), maximized the probability based on the alternative tRNAAla (positions 1502–1503 and 1539–1542), hypothesis. An SH test, using 10 000 replications, was tRNACys (positions 1686–1692, 1698–1699 and 1725– then run to determine whether the favoured likelihood 1737) and tRNATyr (positions 1762–1769 and 1801– tree was signiﬁcantly better than an alternative or 1807) were excluded from analyses. Parts of the D-loop whether their differences could be attributed to from the genes encoding tRNAIle (positions 133–141) chance alone. and tRNAGln (positions 263–265) were excluded from analyses. Parts of the T-loop from the genes encoding Met Asn RESULTS tRNA (positions 334–341) and tRNA (positions 1586–1587) were excluded from analyses. Non-coding SEQUENCE ALIGNMENT sequences between ND1 and tRNAIle genes (positions Twenty-six new sequences of Australian geckos, rep- 95–119), tRNAIle and tRNAGln genes (positions 193– resenting ~1853 bases of the mitochondrial genome, 205), ND2 and tRNAMet genes (positions 355–356), are reported. These sequences were aligned for phylo- ND2 and tRNATrp genes (positions 1392–1403), genetic analysis with a sequence of Gek. gecko (Macey tRNATrp and tRNAAla genes (positions 1481–1486), et al., 1999). Of the 1646 unambiguous sites in 27 tRNAAla and tRNAAsn genes (positions 1558–1560), Asn Cys aligned sequences, 1136 were variable and 893 were tRNA gene and OL (positions 1635–1637), tRNA Cys parsimony-informative. gene and OL (positions 1672–1673), tRNA and All base positions in protein-coding genes were tRNATyr genes (positions 1745–1749), and tRNATyr and alignable. Part of the ND1 gene was excluded from COI genes (positions 1822–1823) were not used. analyses (positions 1–19) because a shorter region of Excluded regions comprised 205 of the 1853 aligned the mitochondrial genome was sequenced in L. jicari positions, less than 12% of the aligned sequence posi- (1834 bp). Gaps were placed in ND2 gene sequences tions. Percent sequence divergences between all pairs at codon 308 (positions 1350–1352) for all species of aligned sequences are presented in Table 3. except Carphodactylus laevis, Nephrurus levis, N. milii, Pseudothecadactylus lindneri, S. strophurus, Gek. gecko, L. jicari, H. binoei and Geh. variegata. NUMBER OF INFORMATIVE CHARACTERS AND TESTS FOR SUBSTITUTIONAL SATURATION Several unalignable regions in OL were excluded from the analyses (positions 1642–1643 and 1648– Because our molecular data contained a large number 1665). of parsimony-informative characters both within

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 82, 123–138 128 J. MELVILLE ET AL. ; = = = 1.49 1.30 1.52 1.34 1.44 1.59 1.54 1.70 1.55 1.28 1.41 1.15 1.37 1.08 1.15 1.44 1.50 1.52 1.35 1.31 1.32 1.40 1.38 1.33 1.35 1.29 1.37 1.48 1.49 1.41 1.45 1.42 ; NM CO SW ; .32 – ; lindneri

levis

taenicauda intermedius S. Nephrurus S.

= = = NL SI SD ; ; ; Pseudothecadactylus

= elderi PL vittatus S. ;

D. =

taeniatus = S. SE

= ; model) are shown above the diagonal. CL ; DV

ST + ; I + assimilis S.

Saltuarius cornutus =

= stenodactylus strophurus D. SA

. PC S. ; = ;

= DS ; SS ; wellingtonae galeatus S.

= Heteronotia binoei =

spinigerus = Oedura marmorata S. SO

= ; ; = HB ; OM SP ; ; pulcher D.

= wheeleri rankini DP S. N. ciliaris aberrans ;

= = S. Gehyra variegata

= = SR NW ; ; SB GV ; ; mcmillani laevissimus S. Lialis jicari

= = = LJ SM NS ; Diplodactylus conspicillatus ; Strophurus ciliaris

;

= = DC SC ; ; Gekko gecko vertebralis

= N.

= GG ; airwise comparisons of haplotypes among Australian Diplodactylinae species and outgroup taxa NV P ; Rhynchoedura ornata Rhynchoedura

20 0.20 0.21 0.19 0.19 0.20 0.19 0.18 0.20 0.19 0.21 0.19 0.20 0.15 0.20 0.19 0.19 – 0.43 0.41 0.55 0.67 0.75 0.79 0.76 0.72 0.58 0.61 0.61 0.69 1.45 1.16 = SC SB SO SA SE SI SR SP SS ST SD SW SM DG DS DP DC DV RO OM CO CL NL NV NS NW NM PC PL LJ GV HB GG milii N. williamsi able 3.

RO S. = Crenadactylus ocellatus T Proportional sequence divergence (uncorrected) is shown below the diagonal, and likelihood distances (GTR Carphodactylus laevis RO 0.20OM 0.21 0.20CO 0.22 0.20 0.23CL 0.20 0.21 0.23 0.20 0.20NL 0.25 0.24 0.20 0.18 0.25 0.26NV 0.23 0.20 0.20 0.26 0.26 0.22 0.20 0.27NS 0.20 0.25 0.27 0.22 0.21 0.27 0.19 0.24 0.27NW 0.26 0.23 0.21 0.28 0.20 0.24 0.27 0.26 0.26 0.22 0.21NM 0.28 0.19 0.25 0.27 0.26 0.24 0.23 0.20 0.26 0.24 0.18 0.24PC 0.27 0.26 0.26 0.23 0.21 0.26 0.25 0.18 0.24 0.26 0.25 0.25 0.22PL 0.24 0.21 0.26 0.25 0.20 0.24 0.25 0.25 0.26 0.22 0.26 0.20 0.26 0.24 0.20LJ 0.24 0.24 0.26 0.24 0.26 0.23 0.26 0.22 0.26 0.23 0.20 0.25 0.24 0.25 0.25GV 0.26 0.23 0.24 0.26 0.22 0.26 0.23 0.22 0.25 0.24 0.27 0.24 0.25 0.23 0.24 0.26 0.21 0.32 0.26 0.24HB 0.22 0.23 0.26 0.26 0.25 0.27 0.25 0.23 0.27 – 0.33 0.26 0.23 0.20 0.23 0.24 0.31GG 0.25 0.26 0.26 0.24 0.24 0.26 0.33 0.29 0.23 0.21 0.22 0.25 0.31 0.26 0.25 0.26 0.36 0.24 0.23 0.25 0.33 0.27 0.24 – 0.22 0.25 0.32 0.43 0.28 0.24 0.29 0.36 0.24 0.24 0.24 0.32 0.26 0.23 0.22 0.26 0.31 0.53 0.27 0.26 0.28 0.36 0.24 0.24 0.25 0.32 0.29 0.23 0.22 0.25 0.30 0.61 0.26 0.56 0.26 0.27 0.34 – 0.24 0.24 0.33 0.29 0.25 0.23 0.25 0.30 0.78 0.29 0.66 0.25 0.28 0.34 0.24 0.26 0.31 0.27 0.24 0.23 0.25 0.31 0.81 0.28 0.73 0.28 0.27 0.34 0.25 0.26 0.33 0.28 0.24 0.63 0.23 – 0.30 0.75 0.27 0.77 0.27 0.26 0.35 0.25 0.24 0.33 0.27 0.26 0.68 0.24 0.30 0.73 0.27 0.72 0.26 0.24 0.34 0.23 0.25 0.31 0.27 0.26 0.72 0.24 0.30 0.64 0.26 0.65 0.26 – 0.58 0.34 0.27 0.26 0.32 0.24 0.24 0.65 0.23 0.29 0.57 0.26 0.55 0.25 0.58 0.35 0.27 0.27 0.33 0.17 0.25 0.73 0.25 0.30 0.56 0.24 0.63 0.26 0.57 0.33 0.25 0.26 0.33 – 0.23 0.26 0.64 0.25 0.30 0.75 0.17 0.60 0.22 0.52 0.35 0.24 0.27 0.32 0.26 0.28 0.58 0.25 0.31 1.42 0.15 0.68 0.20 0.46 0.36 0.25 0.27 0.33 0.22 0.39 0.57 0.24 0.30 1.08 0.23 – 1.41 0.19 0.45 0.35 0.24 0.26 0.33 0.20 0.37 0.71 0.24 0.31 0.37 1.10 0.20 0.64 0.34 0.22 0.27 0.33 0.20 0.53 1.32 0.24 0.31 0.37 – 0.59 0.35 0.24 0.25 0.38 0.33 0.21 0.71 1.13 0.25 0.31 0.56 1.20 0.34 0.24 0.26 0.40 0.33 0.18 0.74 0.27 0.30 0.74 1.10 0.35 0.24 0.24 0.57 0.33 0.34 – 1.40 0.27 0.30 0.80 0.34 0.23 0.27 0.70 0.32 0.52 1.16 0.26 0.31 1.43 0.35 0.21 0.28 0.79 0.34 0.72 0.26 0.31 1.21 0.42 0.34 – 0.28 1.37 0.33 0.77 0.25 0.32 0.62 0.35 0.27 1.14 0.33 1.44 0.23 0.31 0.59 0.36 0.24 0.33 1.13 0.52 – 0.31 1.27 0.35 0.23 0.32 0.54 0.30 1.03 0.37 0.25 0.33 1.42 0.30 0.64 0.36 – 0.31 1.15 0.31 1.19 0.33 0.33 0.30 1.07 0.35 – 1.39 0.30 0.32 1.04 0.26 0.34 – 0.81 0.31 0 SBSO 0.08 –SA 0.11 0.11SE 0.13 – 0.15 0.14SI 0.15 0.20 0.15 0.15 0.22SR 0.24 – 0.12 0.16 0.18 0.25 0.13SP 0.12 0.13 0.17 0.22 0.14 0.13 – 0.19 0.17SS 0.12 0.21 0.12 0.14 0.17 0.24 0.12 0.22ST 0.13 0.15 0.13 0.18 0.24 0.14 0.19 0.28 – 0.15 0.12 0.17SD 0.15 0.34 0.12 0.17 0.28 0.16 0.07 0.10 0.15 0.15 0.13 0.19SW 0.18 0.36 0.08 – 0.26 0.17 0.09 0.29 0.07 0.20 0.18 0.19 0.11 0.13SM 0.34 0.16 0.08 0.41 0.06 0.21 0.19 0.35 0.11 0.14 0.17 0.17 0.14 0.21 0.39 0.07DG – 0.28 0.18 0.43 0.13 0.14 0.30 0.17 0.15 0.23 0.46 0.21 0.16 0.17 0.41 0.20 0.12DS 0.14 0.34 0.19 0.15 0.30 0.43 0.23 0.25 0.17 0.47 0.21 0.12 0.20 – 0.36DP 0.17 0.19 0.15 0.10 0.40 0.34 0.31 0.18 0.45 0.21 0.08 0.23 0.42 0.16 0.20 0.17 0.29 0.42 0.09 0.34 0.17 0.21 0.45DC 0.19 0.08 0.30 0.39 0.17 0.20 – 0.35 0.40 0.27 0.28 0.36 0.19 0.22 0.48 0.18 0.08 0.10 0.20DV 0.35 0.17 0.19 0.34 0.53 0.36 0.36 0.36 0.17 0.22 0.45 0.19 0.13 0.28 0.21 0.40 0.17 0.19 0.41 0.61 0.18 0.36 0. 0.35 – 0.21 0.57 0.19 0.30 0.14 0.33 0.21 0.38 0.19 0.18 0.39 0.70 0.34 0.43 0.38 0.19 0.67 0.18 0.22 0.17 0.35 0.20 0.50 0.16 0.19 0.37 0.76 0.36 0.40 0.32 0.21 0.77 0.20 0.26 – 0.40 0.19 0.30 0.59 0.19 0.19 0.39 0.77 0.35 0.36 0.47 0.21 0.82 0.19 0.34 0.38 0.20 0.38 0.68 0.17 0.19 0.38 0.67 0.44 0.41 0.57 0.20 0.78 0.20 0.40 0.35 0.20 0.42 0.78 0.27 – 0.20 0.47 0.60 0.40 0.40 0.68 0.21 0.74 0.19 0.40 0.40 0.20 0.43 0.75 0.34 0.21 0.56 0.63 0.39 0.51 0.71 0.20 0.63 0.20 0.40 0.36 0.20 0.47 0.66 0.33 0.18 0.62 0.60 0.42 0.62 0.72 0.22 0.67 0.39 – 0.36 0.48 0.20 0.50 0.57 0.41 0.20 0.68 0.71 0.40 0.71 0.65 0.21 0.64 0.41 0.44 0.59 0.23 0.44 0.57 0.37 0.20 0.65 1.43 0.52 0.70 0.51 0.21 0.80 0.43 0.37 0.65 0.20 0.45 0.38 0.54 0.34 – 0.61 1.14 0.56 0.70 0.55 0.18 1.45 0.44 0.53 0.70 0.21 0.33 0.30 0.74 0.39 0.51 0.65 0.64 0.52 0.21 1.23 0.42 0.57 0.67 0.18 0.48 0.30 1.37 0.33 0.52 0.70 0.58 0.66 – 0.42 0.44 0.69 0.60 0.22 0.57 0.21 1.12 0.47 0.49 0.73 0.57 1.29 0.38 0.45 0.72 0.55 0.15 0.68 0.43 0.59 0.61 0.62 0.52 1.08 0.55 0.40 0.70 0.54 – 0.70 0.23 0.39 0.65 1.28 0.53 0.69 0.59 0.37 0.73 0.52 0.65 0.33 0.53 0.69 1.07 0.59 1.39 0.76 0.41 0.59 0.63 0.69 0.49 0.65 0.69 0.50 1.15 0.34 0.84 0.55 0.59 1.31 0.56 0.45 0.70 0.62 0.72 0.48 0.79 0.58 0.55 1.08 0.60 0.62 0.73 0.53 1.38 0.46 0.71 0.66 0.70 0.54 0.65 0.72 0.58 1.13 0.55 0.62 0.71 1.30 0.63 0.90 0.72 0.53 0.66 0.60 0.69 1.07 1.14 0.89 0.57 0.66 0.80 0.58 0.65 1.01 0.90 0.62 1.32 0.85 0.73 0.56 0.86 0.56 1.08 0.82 1.37 0.55 0.70 0.74 0.78 1.04 0.53 0.71 1.44 0.64 0.70 0.64 1.18 0.70 1.35 0.77 0.63 1.07 1.41 0.72 1.18 1.41 1.17 SC – 0.10 0.15 0.19 0.21 0.16 0.17 0.16 0.23 0.23 0.33 0.14 0.28 0.37 0.36 0.41 0.38 0.38 0.39 0.37 0.49 0.61 0.70 0.74 0.72 0.66 0.57 0.59 0.57 0.72 1.31 1.12

0.5 ocellatus, (4) Ps. lindneri and (5) a clade containing all sampled Diplodactylini (Rhynchoedura ornata, 0.4 O. marmorata, Diplodactylus spp. and Strophurus spp.) except C. ocellatus (bootstrap 94%, decay index 0.3 All uncorrected pairwise differences 11; Fig. 3). Strophurus was monophyletic with the exclusion of S. taenicauda, and monophyly of the sub- 0.2 Uncorrected differences without silent genus as a whole cannot be rejected using our statis- 3rd position changes 0.1 tical tests (Table 4). 3rd position silent transitions The earliest divergence within the clade containing Pairwise proportional difference 0 Strophurus excluding S. taenicauda produced two 0 0.5 1 1.5 2 major subgroups: (1) S. taeniatus and S. mcmillani Maximum likelihood distance (bootstrap 89, decay index 11), and (2) a clade containing all remaining Strophurus (bootstrap 76, decay Figure 2. Heuristic tests for substitutional saturation. index 7). Within subgroup (2) the basal split separated Total percentage divergence between pairs of aligned sequences and 3rd position silent transitions are plotted S. elderi from a clade containing the remaining species against maximum-likelihood distances, which correct for (bootstrap 85%, decay index 8), which formed three multiple substitutions occurring among the sequences. subclades: (1) S. assimilis and S. strophurus (bootstrap 100%, decay index 33), (2) S. intermedius, S. rankini, S. spinigerus and S. williamsi (bootstrap Diplodactylinae and between the ingroup and out- 86%, decay index 6) and (3) S. ciliaris, S. ciliaris aber- group taxa, a simple lack of characters cannot explain rans and S. wellingtonae (bootstrap 100%, decay index the weak support observed for many branches in the 19). A grouping of S. intermedius, S. rankini and parsimony phylogeny. The number of phylogenetically S. spinigerus also received strong heuristic support informative characters (893) was 28 times the number (bootstrap 97%, decay index 9). of taxa sampled (33). We plotted maximum-likelihood Within the nominate subgenus of Diplodactylus, distances vs. observed differences between paired sampled members of the D. vittatus species group sequences to test for substitutional saturation. The (D. galeatus, D. pulcher and D. vittatus) were strongly strong non-linear curve, in both uncorrected distances grouped with D. conspicillatus, formerly assigned to and silent transitions, suggests substitutional satura- the D. conspicillatus species group but grouped here tion. The relationship between maximum-likelihood with D. pulcher. Our results suggest merging the distances and uncorrected divergences between D. conspicillatus and D. vittatus groups into one spe- paired sequences was a linear increase until approxi- cies group. Our phylogenetic topology confirms Greer’s mately 20% sequence divergence, after which the (1989) prediction that D. stenodactylus (representing relationship became strongly non-linear (Fig. 2). A the D. stenodactylus group) is grouped with genus large gap in pairwise sequence divergence occurred Rhynchoedura, although this result was not statisti- between approximately 27% and 30%, which can be cally robust. attributed to the large sequence divergence between The five species of Nephrurus sampled were the ingroup (Diplodactylinae) and outgroup (Gekkon- grouped with strong support (bootstrap 100%, decay inae) taxa. index 25), consistent with monophyly of this genus. Our molecular phylogenetic results support Bauer’s (1986, 1990) and Greer’s (1989) hypothesis that PHYLOGENETIC RELATIONSHIPS USING PARSIMONY the species with greatest phalangeal reduction A single most parsimonious tree (6038 steps) resulted (represented here by N. levis, N. laevissimus and from analysis of 1646 aligned base pairs of 33 gecko N. vertebralis) form a clade (bootstrap 100%, decay species, containing 893 phylogenetically informative index 25). base positions (Fig. 3). Monophyly of a group containing all sampled Australian Diplodactylinae and L. jicari received strong heuristic support (bootstrap PHYLOGENETIC INFERENCES USING 100%, decay index 49). Within Australian diplodac- MAXIMUM-LIKELIHOOD AND BAYESIAN ANALYSES tyline geckos, relationships among basal lineages Among 56 likelihood models examined using the leading to the following taxa were not adequately shortest tree from the parsimony analysis, the most resolved to be distinguished from a polytomy: (1) complex model (GTR + I + G) had the highest likeli- L. jicari, (2) a moderately supported group (bootstrap hood and was significantly favoured over alternatives 82%, decay index 9) containing all Carphodactylini by hierarchical likelihood-ratio tests. Model parame- species sampled (Nephrurus spp., C. laevis and Saltu- ters were: alpha = 0.530; proportion of invariant arius cornutus) except Ps. lindneri, (3) Crenadactylus sites = 0.147; substitution rates R(a) = 0.670,

84 Strophurus wellingtonae 100 7 Strophurus ciliaris aberrans 19 Strophurus ciliaris 67 Strophurus intermedius 3 97 87 9 Strophurus rankini 86 85 7 6 Strophurus spinigerus 8 Strophurus williamsi 76 100 Strophurus assimilis 7 33 Strophurus strophurus 94 8 Strophurus elderi 89 Strophurus taeniatus 11 Strophurus mcmillani 1 92 Diplodactylus galeatus 100 9 Diplodactylus vittatus 56 5 20 100 Diplodactylus pulcher 33 Diplodactylus conspicillatus 94 72 Strophurus taenicauda 11 7 Oedura marmorata 63 Diplodactylus stenodactylus 3 4 Rhynchoedura ornata 1 Crenadactylus ocellatus Pseudothecadactylus lindneri Carphodactylus laevis Nephrurus levis 79 100 25 9 84 Nephrurus vertebralis 90 7 10 Nephrurus laevissimus 100 82 100 Nephrurus wheeleri 49 9 25 Nephrurus milii Saltuarius cornutus Lialis jicari 57 Gehyra variegata 1 Gekko gecko Heteronotia binoei

100 changes

Figure 3. Phylogram of the most parsimonious tree, indicating phylogenetic relationships within Australian Diplodactyl- inae. Bootstrap values are presented above the branches and decay indices are below the branches.

R(b) = 3.533, R(c) = 1.117, R(d) = 0.491, and of a group containing all Australian Diplodactylinae R(e) = 4.910; empirical base frequencies A = 0.367, and L. jicari (bootstrap 100%). Within Diplodactyli- C = 0.352, G = 0.085, and T = 0.196. Twenty-ﬁve nae were three major subdivisions: (1) Ps. lindneri, random-addition heuristic searches using the parsi- (2) a group containing all other Carphodactylini mony tree as the starting topology yielded a single sequenced and L. jicari (bootstrap 74%) and (3) a optimal tree with a log-likelihood score of -25 724.13 group containing all Diplodactylini species sequenced (Fig. 4). (bootstrap 91%). Within the Diplodactylini, Maximum-likelihood bootstrap resampling using C. ocellatus was the sister taxon to a well-supported 100 bootstrap replicates (Fig. 4) supported monophyly clade containing all remaining species (bootstrap

Table 4. Hypotheses tested by using the Wilcoxon signed-ranks test and SH test with molecular data

Wilcoxon signed-ranks test SH-test

Hypothesis N Z P Diff - ln L P

A. Non-monophyly of 1. Nephrurus, Carphodactylus, Saltuarius 284 0.47 NS 6.18 NS 2. Diplodactylini 228 0.18 NS 7.08 NS B. Monophyly of 1. Strophurus 63 1.13 NS 8.19 NS 2. Carphodactylini – Nephrurus, Carphodactylus, Saltuarius 33 2.61 0.015 13.39 0.047 Pseudothecadactylus 3. Strophurus – except S. elderi 117 2.61 0.005 38.15 0.033 4. Blue mouth – all Strophurus except S. elderi, S. ciliaris 231 5.68 <0.001 174.25 <0.001 aberrans, S. taeniatus 5. Blue mouth – all Strophurus except S. elderi, S. ciliaris 311 8.36 <0.001 346.84 <0.001 aberrans, S. taeniatus plus Saltuarius 6. Yellow mouth – group S. ciliaris aberrans and S. taeniatus 138 5.44 <0.001 64.59 <0.001

For the Wilcoxon signed-ranks test, the null hypothesis was that the shortest tree(s) corresponding to the conditions listed are not significantly longer than is the overall most-parsimonious tree (N = number of characters that differ in minimum number of changes on the two trees, Z = normal approximation derived from the Wilcoxon test statistic, P = significance level for a two-tailed test, NS = result not significant). For the SH test, the null hypothesis was that the maximum- likelihood tree corresponding to the conditions listed is not significantly different from the overall optimal maximum- likelihood tree. (Diff - ln L is the difference in ln-likelihood between the alternative hypothesis as stated and the overall maximum-likelihood tree).

98%). All Diplodactylus and Strophurus except HYPOTHESIS TESTING USING ALTERNATIVE D. stenodactylus formed a clade (bootstrap 100%). All TOPOLOGIES Strophurus except S. taenicauda formed a clade with strong support (bootstrap 95%). Groupings within Although Diplodactylini (Crenadactylus, Diplodacty- Strophurus were identical to those from parsimony lus, Oedura, Rhynchoedura and Strophurus) was analysis with the same subgroups receiving strong recovered as a monophyletic group in both parsimony heuristic support. and likelihood analyses and supported 100% by Baye- Topological differences between results of parsi- sian analysis, this grouping was not statistically mony and likelihood analyses were limited to parts of robust according to tests based on parsimony and the tree where neither analysis resolved branches maximum likelihood (Table 4). with strong support, making their results essentially Parsimony and likelihood tests rejected monophyly equivalent. of Carphodactylini (Carphodactylus, Nephrurus, Bayesian analysis performed using the GTR + I + G Saltuarius and Pseudothecadactylus) because of nucleotide substitution model and parameters esti- strong support for placing a pygopodid, L. jicari, mated from the sequence data included 60 000 saved within this group (Table 4). Within Carphodactylini, trees, 15 000 of which were considered ‘burn-in’, leav- genera Carphodactylus, Nephrurus and Saltuarius ing 45 000 trees. A 50% majority-rule consensus tree formed a monophyletic group in all analyses with of the 45 000 trees had the same topology as that moderate to strong heuristic support, although the of the maximum-likelihood tree with a mean log- results in Table 4 indicate this grouping not to be sta- likelihood of -25 780.60 and variance of 40.58 (Fig. 4). tistically robust. All groupings that received strong heuristic support All phylogenetic analyses found Strophurus not to from parsimony and likelihood analyses occurred in form a monophyletic group, but the tests in Table 4 99–100% of the 45 000 trees from the Bayesian anal- indicate that the data could not statistically reject ysis. Comparison of branch support as assessed using monophyly of Strophurus. This result is consistent parsimony, likelihood and Bayesian criteria was con- with the observation that the branches that must be sistent with the suggestion that Bayesian posterior broken to place S. taenicauda with the clade contain- probabilities may overestimate support for branches ing all other Strophurus received very weak support in the tree (Suzuki, Glazko & Nei, 2002). by parsimony and likelihood criteria, although two got

77 Gehyra variegata 94 Gekko gecko Strophurus ciliaris 100 100 61 Strophurus ciliaris abberans 80 Strophurus wellingtonae 89 Strophurus intermedius 100 100 100 81 Strophurus rankini 85 93 100 99 Strophurus spinigerus 100 Strophurus williamsi 77 100 Strophurusassimilis 98 100 Strophurus strophurus 95 100 Strophurus elderi 76 Strophurus taeniatus 55 98 Strophurus mcmillani 90 Diplodactylus galeatus 96 100 100 Diplodactylus vittatus 84 100 99 100 Diplodactylus pulcher 100 Diplodactylus conspicillatus

77 Diplodactylus stenodactylus 100 98 Rhynchoedura ornata 100 Oedura marmorata 91 100 Strophurus taenicauda Crenadactylus ocellatus Carphodactylus laevis

Nephrurus levis 100 53 61 Nephrurus vertebralis 91 100 84 97 100 100 Nephrurus laevissimus 70 100 Nephrurus wheeleri 100 100 100 100 74 Nephrurus milii 91 Saltuarius cornutus Lialis jicari Pseudothecadactylus lindneri Heteronotia binoei

Figure 4. Optimal maximum-likelihood tree under the GTR + I + G model, which indicates phylogenetic relationships within Australian Diplodactylinae. Bootstrap values are presented above the branches and percentages calculated from 45 000 Bayesian trees representing posterior probability values are below the branches.

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 82, 123–138 ANTIPREDATOR STRATEGIES IN STROPHURUS 133 high Bayesian support. The hypothesis of a single evo- are well above the level (10%) at which mitochondrial lutionary origin of the distinguishing morphological DNA sequences are expected to show substitutional characteristics of Strophurus therefore cannot be saturation (Moritz, Dowling & Brown, 1987), and rejected by the data presented here. Both parsimony are therefore probably underestimates. The average and likelihood-based tests rejected monophyletic divergences of haplotypes among the six lineages of groupings of Strophurus species sharing the two Diplodactylini (Strophurus, the D. vittatus group, derived mouth colours, blue and yellow, and a group- O. marmorata, R. ornata, D. stenodactylus and ing of the blue-mouthed Saltuarius with blue- C. ocellatus) were above 20% (range 20–23%; Table 3), mouthed Strophurus (Table 4). These results suggest indicating that these lineages have undergone sepa- homoplastic evolution of mouth colour in diplodac- rate evolutionary change for at least 30 Myr. The tyline geckos. divergences within the Carphodactylini (C. laevis, Nephrurus and Saltuarius) were similar to those within Diplodactylini (range 22–25%; Table 3). The TAXONOMIC ISSUES IN DIPLODACTYLINAE sequence divergences between Carphodactylini, Diplo- Our results are consistent with molecular phyloge- dactylini, Lialis and Pseudothecadactylus were all netic work by Donnellan et al. (1999) in grouping greater than 25%, indicating that divergence of their Pygopodidae (L. jicari) with the gekkonid subfamily ancestral lineages exceeds 45 Myr. The divergence Diplodactylinae, but equivocal with respect to whether between the Gekkonidae outgroups and the Diplodac- the pygopodid forms the sister taxon to Diplodactyli- tylinae exceeded 30% (Table 3), as found by Macey nae or occurs phylogenetically nested within this sub- et al. (1999) for comparisons between Gekkoninae and family. Our phylogenetic analyses are consistent with Eublepharinae. These dates are generally compatible monophyly of Diplodactylini (Kluge, 1967b) although with previous estimates based upon immunological this result was not statistically robust. Monophyly of comparisons of albumins (reviewed by King, 1990). Carphodactylini was neither supported nor statistically rejected. DISCUSSION Our sampling was insufficient to assess monophyly of the large genus Diplodactylus, but our results sug- Our phylogenetic analyses considered together with gest that Diplodactylus is likely to be paraphyletic the morphological characters discussed by Greer with respect to Oedura and Rhynchoedura. A conser- (1989) suggest a single origin of caudal glands in vative taxonomy would consider these latter names Diplodactylus and a monophyletic subgenus Strophu- synonyms of Diplodactylus pending a more extensive rus. Our molecular phylogenetic analyses consistently phylogenetic survey of Diplodactylus. grouped all Strophurus species except S. taenicauda Within Carphodactylini, our study strongly sup- as a clade. The molecular data were equivocal regard- ports monophyly of genus Nephrurus, including ing the exact phylogenetic placement of S. taenicauda, N. milii, as proposed by Bauer (1986, 1990). Some pre- and did not reject its grouping with the other Strophu- vious work placed N. milii (synonymous with Under- rus. We found the sharing of caudal glands to woodisaurus milii) in genus Phyllurus (Kluge, 1967b; be a strong morphological synapomorphy grouping Russell, 1980), but our study strongly favours inclu- S. taenicauda with the other Strophurus, and this sion of N. milii within Nephrurus. grouping could be achieved on the parsimony tree (Fig. 3) by collapsing a single branch having a decay index of 1. S. taenicauda probably represents an early AGES OF PHYLOGENETIC DIVERGENCES phylogenetic divergence from the group containing the The mitochondrial segment analysed here has been remaining Strophurus, this divergence being suffi- found to evolve at a rate of approximately 1.3% diver- ciently old to be beyond the optimal resolving power of gence between lineages per Myr of evolutionary sepa- the mitochondrial DNA sequences reported here. This ration; this calibration has been confirmed among phylogenetic position for S. taenicauda also explains diverse vertebrates (see review by Weisrock et al., Kluge’s (1967b) observation that this species appears 2001), including gekkonid lizards (Macey et al., 1999). intermediate between Strophurus and other Diplodac- This calibration was used to estimate divergence tylus. Further testing of monophyly of Strophurus times for major lineages of Diplodactylinae using the should emphasize more slowly evolving molecular maximum-likelihood corrected distances in Table 3. sequences and detailed comparisons of caudal-gland Haplotype divergences among Strophurus species morphologies of S. taenicauda and other Strophurus. were substantial, indicating that evolutionary diver- Independent evolution of caudal glands has occurred gences among extant species occurred approximately elsewhere in Diplodactylinae in the New Caledonian 8–28 Mya, with sequence divergences ranging genus Eurydactylodes, although morphological differ- between 6% and 22% (Table 3). Most of these values ences are sufficient to reject homology of these caudal

© 2004 The Linnean Society of London, Biological Journal of the Linnean Society, 2004, 82, 123–138 134 J. MELVILLE ET AL. glands with those of Strophurus (Böhme & Sering, secreted from the caudal glands is sticky and odorous 1997). Our molecular phylogenetic results considered (Bustard, 1970; Richardson & Hinchliffe, 1983). How- together with those of Donnellan et al. (1999) suggest ever, there are very limited accounts of tail-squirting that caudal glands are an evolutionarily derived fea- behaviour being observed in an antipredator role. ture in Diplodactylinae, having arisen independently S. elderi kept in captivity with L. burtonis are not in two different groups that are phylogenetically consumed by the pygopodid (Bustard, 1970), and nested within different parts of this subfamily. It L. burtonis has been observed to drop the gecko after seems less likely that caudal glands are ancestral for receiving a dose of caudal secretion. Rosenburg & Rus- Diplodactylini as hypothesized by Richardson & sell (1980) found that chicks offered mealworms Hinchliffe (1983). smeared with the caudal secretions from Strophurus Greer (1989) grouped Oedura with Strophurus demonstrated an obvious response by either rejecting based on their shared arboreality and associated the food item or taking a relatively long time to con- adhesive toe pads. These groups differ from other sume it. Thus, available information indicates that the Diplodactylini in being arboreal rather than terres- caudal secretions may be unpalatable to potential trial. Our data neither support nor reject this group- predators. ing but are compatible with viewing the toe pads as Evolution of tail structures in Strophurus occurs in a synapomorphy of Oedura and Strophurus. The best the context of conflicting functional demands. In hypothesis based on the levels of phylogenetic diver- addition to the antipredation function of caudal secre- gence reported here is that arboreality originated in tions (Rosenburg & Russell, 1980), tails also appar- a common ancestor of Oedura and Strophurus ently serve a prehensile function in climbing (Pianka approximately 29 Mya, and that caudal glands origi- & Pianka, 1976) and are used in behavioural dis- nated in a common ancestor of all Strophurus plays. All of these roles are compromised by tail auto- approximately 25 Mya, followed by further morpho- tomy (Pianka & Pianka, 1976; Bauer & Russell, logical modifications of the tail in some lineages of 1994), especially the caudal-gland ejection mecha- Strophurus. nism, which is not completely restored in regenerated The caudal glands of Strophurus constitute a longi- tails (Rosenburg & Russell, 1980). Pianka & Pianka tudinal series of paired chambers (one pair per verte- (1976) found a very low rate of tail loss in S. ciliaris, bral segment) filled with a sticky, viscous substance S. williamsi and S. strophurus, whereas they found a fired through a series of mid-dorsal slits, or rupture high rate of both predation and tail loss in S. elderi zones, on the tail. Enlarged scales and/or spines (Pianka & Pianka, 1976). Considered in the light of surround the caudal rupture zones in S. ciliaris, our phylogenetic results, these comparisons suggest S. intermedius, S. rankini, S. spinigerus, S. williamsi, that the clade characterized by elaboration of spines S. assimilis, S. wellingtonae and S. strophurus. These and tubercles surrounding the caudal glands species were grouped as a clade with moderate (S. ciliaris, S. intermedius, S. rankini, S. spinigerus, support in our molecular phylogenetic analysis, S. williamsi, S. assimilis and S. strophurus; Fig. 3) suggesting that enlarged caudal scales/spines are a has evolved a strategy in which tail loss is mini- synapomorphy for the group. The age of this clade can mized, whereas other Strophurus, such as S. elderi, be estimated at approximately 16 Myr using the max- are more likely to depend on tail autotomy to escape imum-likelihood corrected sequence divergences and predators. Bauer & Russell (1994) suggest, following the rate calibration of Weisrock et al. (2001). Minton (1983), that S. elderi may avoid predation by Greer (1989) grouped the remaining five species of oozing its tail secretion prior to tail detachment, Strophurus (S. elderi, S. mcmillani, S. michaelseni, thereby entangling the predator’s mouthparts in the S. taeniatus and S. wilsoni) based on the synapomor- sticky detached tail. Our phylogenetic results provide phy of lacking femoral pores, and he noted that these a historical framework for further comparative stud- species also share small body size, longitudinally ies of the functional morphology of Strophurus tails striped patterns, and use of hummock-grass habitats. to evaluate the complex interrelationship of coevolved Our analyses consistently grouped S. mcmillani and traits involving caudal glands, caudal autotomy, pre- S. taeniatus as predicted and neither reject nor sup- dation levels and caudal augmentation (spines and port a grouping of S. elderi with these species; sam- tubercles). pling was unavailable for the other two species of this The molecular phylogenetic results and mor- grouping. Our results are compatible with Greer’s phological synapomorphies discussed above suggest (1989) phylogenetic interpretations for these charac- homoplastic evolution of the striking mouth colours, ters and species but offer only partial confirmation of dark blue and yellow, that are observed in Strophurus his predictions. but absent from other Diplodactylini (Table 5). Mouth The caudal glands in Strophurus almost certainly colours other than pink are uncommon in geckos, per- serve an antipredation function. The substance haps because most geckos are nocturnal. Visibility of

Table 5. Internal mouth colours observed in diplodactyline geckos and outgroups

Published mouth Species Museum observations Field observations colour Reference

Strophurus ciliaris blue/black blue/black Greer (1989) S. ciliaris aberrans yellow/orange yellow/orange yellow/orange Storr, Smith & Johnstone (1990) S. assimilis blue/black S. elderi pink pink pink Greer (1989) S. intermedius blue/black blue/black Greer (1989) S. mcmillani blue/black S. rankini blue/black blue/black Greer (1989) S. spinigerus blue/black blue/black Greer (1989) S. strophurus blue/black blue/black blue/black Wilson & Knowles (1988) S. taeniatus yellow/orange yellow/orange Schmida (1973) S. taenicauda blue/black S. wellingtonae blue/black blue/black Storr et al. (1990) S. williamsi blue/black blue/black Bustard (1964) Diplodactylus conspicillatus pink pink D. galeatus dark tongue D. pulcher pink pink D. stenodactylus pink pink D. vittatus dark tongue Oedura marmorata pink pink Rhynchoedura ornata pink pink Crenadactylus ocellatus pink pink Carphodactylus laevis pink Nephrurus levis pink pink N. laevissimus pink pink N. milii pink N. vertebralis pink pink N. wheeleri pink pink Saltuarius cornutus blue/black Pseudothecadactylus lindneri pink Lialis jicari dark tongue Gehyra variegata pink pink Heteronotia binoei pink pink Gekko gecko pink

Original observations (unreferenced) are by J.M. Species marked ‘dark tongue’ have a blue-black tongue in an otherwise pink mouth.

mouth colour, especially the dark blue in Strophurus, which is phylogenetically nested within this clade, has would be very limited at night. Members of Strophu- the pink mouth colour considered ancestral for diplo- rus are the only Australian geckos that have been dactyline geckos (Table 5). The dark-blue colour, observed to be active during the day, exposing them- therefore, may have two separate origins within Stro- selves to daylight (Ehmann, 1980) and even basking in phurus, and it has arisen independently in the genus full sun (reviewed by Greer, 1989). Thus, it is possible Saltuarius (Fig. 5) and Ps. australis (A. M. Bauer, that mouth colour, when used in defence displays, has pers. comm.; species not sampled). The yellow colours coevolved with a shift in activity patterns, providing a observed in S. ciliaris and S. taeniatus almost more striking display against diurnal predators. The certainly represent separate derivations (Fig. 5). dark-blue colour spans the deepest phylogenetic diver- Homoplastic evolution of mouth colour has been found gences within Strophurus (Fig. 5) and might be con- also in Australian agamid lizards (Melville, Schulte & sidered ancestral for Strophurus except that S elderi, Larson, 2001). Further comparative studies done in

Pink mouth Strophurus wellingtonae Strophurus ciliaris aberrans Blue/black mouth Strophurus ciliaris Strophurus intermedius Yellow/orange mouth Strophurus rankini caudal Strophurus spinigerus Blue/black tongue glands Strophurus williamsi Strophurus assimilis No data available Strophurus strophurus Strophurus elderi Equivocal Strophurus taeniatus Strophurus mcmillani Diplodactylus galeatus Diplodactylus vittatus Diplodactylus pulcher Diplodactylus conspicillatus Diplodactylus stenodactylus Rhynchoedura ornata Oedura marmorata Strophurus taenicauda caudal glands Crenadactylus ocellatus Carphodactylus laevis Nephrurus levis Nephrurus vertebralis Nephrurus laevissimus Nephrurus wheeleri Nephrurus milii Saltuarius cornutus Lialis jicari Pseudothecadactylus lindneri Gehyra variegata Gekko gecko Heteronotia binoei

Figure 5. Phylogenetic reconstruction of caudal glands and mouth colour in Strophurus species and Diplodactylini outgroups. Sources of information on mouth pigmentation are summarized in Table 5.

the context of these phylogenetic results should clarify ACKNOWLEDGEMENTS the biological role and potential adaptive significance We thank A. M. Bauer and J. Losos for useful com- of mouth colour in Strophurus. ments on the manuscript. J. R. Macey provided advice In conclusion, our results are consistent with the on taxon sampling and laboratory methods. Advice on hypothesis that caudal glands arose early in the evo- ecological classifications of species was given by E. lutionary history of Strophurus, over 20 Mya, and that Pianka. We thank S. Donnellan (South Australian dermal augmentation occurred subsequently within a Museum, Adelaide) and J. Wombey (Australian clade currently containing seven species (S. ciliaris, National Wildlife Collection, CSIRO, Canberra) for S. intermedius, S. rankini, S. spinigerus, S. williamsi, providing specimens. Financial support was provided S. assimilis and S. strophurus). Elaboration of caudal by NSF grants DEB-9318642, NSF DEB-9726064, glands and low rates of tail autotomy for this clade NSF DEB-0071337 and NSF DEB-9982736. conform to evolutionary strategies predicted for an actively functional tail (Vitt, Congdon & Dickson, 1977). We suggest that evolutionary diversification of REFERENCES mouth colour in Strophurus is associated with diurnal Anderson S, Bankier AT, Barrell BG, de Bruijn MHL, defensive displays. Our phylogenetic analyses of char- Coulson AR, Drouin J, Eperon IC, Nierlich DP, Roe acter variation in Strophurus make this group a good BA, Sanger F, Schreier PH, Smith AJH, Staden R, system for further study of evolution of adaptations Young IG. 1981. Sequence and organization of the human for arboreality and diurnal activity in geckos. mitochondrial genome. Nature 290: 457–465.

Bauer AM. 1986. Systematics, biogeography and evolutionary Kumazawa Y, Nishida M. 1993. Sequence evolution of mito- morphology of the Carphodactylini (Reptilia: Gekkonidae) chondrial tRNA genes and deep-branch animal phylogenet- (New Caledonia, Australia). PhD Dissertation. University of ics. Journal of Molecular Evolution 37: 380–398. California at Berkeley. Larson A. 1998. The comparison of morphological and molec- Bauer AM. 1990. Phylogenetic systematics and biogeography ular data in phylogenetic systematics. In: DeSalle R, Schier- of the Carphodactylini (Reptilia: Gekkonidae). Bonner Zool- water B, eds. Molecular approaches to ecology and evolution. ogische Monographien 30: 1–218. Basel, Switzerland: Birkhäuser-Verlag, 275–296. Bauer AM, Russell AP. 1994. Is autotomy frequency reduced Leaché AD, Reeder TW. 2002. Molecular systematics of the in geckos with ‘actively functional’ tails? Herpetological Nat- eastern fence lizard (Sceloporus undulatus): A comparison of ural History 2: 1–15. parsimony, likelihood, and Bayesian approaches. Systematic Böhme W, Sering M. 1997. Tail squirting in Eurydactylodes: Biology 51: 44–68. Independent evolution of caudal defensive glands in a Macey JR, Larson A, Ananjeva NB, Fang Z, Papenfuss diplodactyline gecko (Reptilia, Gekkonidae). Zoologischer TJ. 1997a. Two novel gene orders and the role of light- Anzeiger 235: 225–229. strand replication in rearrangement of the vertebrate mito- Bremer K. 1994. Branch support and tree stability. Cladistics chondrial genome. Molecular Biology and Evolution 14: 91– 10: 295–304. 104. Bustard HR. 1964. Defensive behaviour shown by Australian Macey JR, Larson A, Ananjeva NB, Papenfuss TJ. 1997b. geckos, genus Diplodactylus. Herpetologica 20: 198–200. Replication slippage may cause parallel evolution in the sec- Bustard HR. 1970. Australian lizards. Sydney, Australia: W. ondary structures of mitochondrial transfer RNAs. Molecu- Collins Ltd. lar Biology and Evolution 14: 30–39. Cogger HG. 2000. Reptiles and amphibians of Australia. Macey JR, Schulte II JA, Larson A, Ananjeva NB, Wang Sanibel Island, Florida: Ralph Curtis Books. Y, Pethiyagoda R, Rastegar-Pouyani N, Papenfuss TJ. Donnellan SC, Hutchinson MN, Saint KM. 1999. Molec- 2000. Evaluating trans-Tethys migration: An example using ular evidence for the phylogeny of Australian gekkonoid acrodont lizard phylogenetics. Systematic Biology 49: 233– lizards. Biological Journal of the Linnean Society 67: 97– 256. 118. Macey JR, Schulte II JA, Larson A, Fang Z, Wang Y, Ehmann H. 1980. Diurnal perching by the southern spiny- Tuniyev BS, Papenfuss TJ. 1998. Phylogenetic relation- tailed lizard Diplodactylus intermedius. Herpetofauna 12: ships of toads in the Bufo bufo species group from the east- 37. ern escarpment of the Tibetan Plateau: A case of vicariance Felsenstein J. 1985. Conﬁdence limits on phylogenies with a and dispersal. Molecular Phylogenetics and Evolution 9: 80– molecular clock. Systematic Zoology 34: 152–161. 87. Galliford M. 1978. A brief study of Nephrurus stellatus in cap- Macey JR, Verma A. 1997. Homology in phylogenetic analy- tivity. South Australian Herpetology Group Newsletter 1978: sis: Alignment of transfer RNA genes and the phylogenetic 1–2. position of snakes. Molecular Phylogenetics and Evolution 7: Greer AE. 1989. The biology and evolution of Australian 272–279. lizards. Chipping Norton, Australia: Surrey Beatty and Macey JR, Wang Y, Ananjeva NB, Larson A, Papenfuss Sons. TJ. 1999. Vicariant patterns of fragmentation among Hoskin CJ, Couper PJ, Schneider CJ. 2003. A new species gekkonid lizards of the genus Teratoscincus produced by the of Phyllurus (Lacertilia: Gekkonidae) and a revised phylog- Indian collision: A molecular phylogenetic perspective and eny and key for the Australian leaf-tailed geckos. Australian an area cladogram for Central Asia. Molecular Phylogenetics Journal of Zoology 51: 153–164. and Evolution 12: 320–332. Huelsenbeck JP, Ronquist F. 2001. MrBayes: Bayesian Maddison WP, Maddison DR. 1992. MacClade, analysis of inference of phylogenetic trees. Bioinformatics 17: 754–755. phylogeny and character evolution, Version 3.0. Sunderland, King M. 1990. Chromosomal and immunogenetic data: A new Massachusetts: Sinauer Assoc. perspective on the origin of Australia’s reptiles. In: Olmo E, Maniatis T, Fritsch EF, Sambrook J. 1982. Molecular clon- ed. Cytogenetics of amphibians and reptiles. Basel, Switzer- ing: A laboratory manual. New York, NY: Cold Spring Har- land: Birkhäuser-Verlag, 153–180. bor Laboratory Press. Kluge AG. 1967a. Higher taxonomic categories of gekkonid Melville J, Schulte II JA, Larson A. 2001. A molecular phy- lizards and their evolution. Bulletin of the American Museum logenetic study of ecological diversiﬁcation in the Australian of Natural History 135: 1–60. lizard genus Ctenophorus. Journal of Experimental Zoology Kluge AG. 1967b. Systematics, phylogeny and zoogeography (Molecular and Developmental Evolution) 291: 339–353. of the lizard genus Diplodactylus Gray (Gekkonidae). Aus- Minton SA. 1983. Highlights in herpetology: It’s not all tralian Journal of Zoology 15: 1007–1108. snakes. Proceedings of the Indiana Academy of Science 92: Kluge AG. 1987. Cladistic relationships in the Gekkonoidea 70–76. (Squamata, Sauria). Miscellaneous Publications of the Moritz C, Dowling TE, Brown WM. 1987. Evolution of ani- Museum of Zoology, University of Michigan 173: 1–54. mal mitochondrial DNA: Relevance for population biology Kluge AG. 2001. Gekkotan lizard taxonomy. Hamadryad 26: and systematics. Annual Review of Ecology and Systematics 1–209. 18: 269–292.

Pianka ER, Pianka HD. 1976. Comparative ecology of twelve Storr GM, Smith LA, Johnstone RE. 1990. Lizards of West- species of nocturnal lizards (Gekkonidae) in Western Austra- ern Australia III. Geckos and pygopods. Perth, Australia: lia. Copeia 1976: 125–142. Western Australian Museum. Posada D, Crandall. KA. 1998. Modeltest: Testing the model Suzuki Y, Glazko GV, Nei M. 2002. Overcredibility of molec- of DNA substitution. Bioinformatics 14: 817–818. ular phylogenies obtained by Bayesian phylogenetics. Pro- Richardson KC, Hinchliffe PM. 1983. Caudal glands and ceedings of the National Academy of Sciences, USA 99: their secretions in the western spiny tailed gecko, Diplodac- 16138–16143. tylus spinigerus. Copeia 1983: 161–169. Swofford DL. 2002. PAUP*: phylogenetic analysis using par- Rosenburg HI, Russell AP. 1980. Structural and functional simony (*and other methods), beta, Version 4.0b10a. Sunder- aspects of tail squirting: A unique defense mechanism of land, Massachusetts: Sinauer Assoc. Diplodactylus (Reptilia: Gekkonidae). Canadian Journal of Templeton AR. 1983. Phylogenetic inference from restriction Zoology 58: 865–881. endonuclease cleavage site maps with particular reference to Rosenburg HI, Russell AP, Kapoor M. 1984. Preliminary the evolution of humans and apes. Evolution 37: 221–244. characterization of the defensive secretion of Diplodactylus Townsend T, Larson A. 2002. Molecular phylogenetics and (Reptilia: Gekkonidae). Copeia 1984: 1025–1028. mitochondrial genomic evolution in the Chamaeleonidae Russell AP. 1980. Underwoodisaurus Wermuth 1965, a junior (Reptilia, Squamata). Molecular Phylogenetics and Evolu- synonym of Phyllurus Schinz 1822. Journal of Herpetology tion 23: 22–36. 14: 415–416. Vitt LJ, Congdon JD, Dickson NA. 1977. Adaptive strate- Russell AP, Rosenburg HI. 1981. Subgeneric classiﬁcation gies and energetics of tail autotomy in lizards. Ecology 58: in the gekkonid genus Diplodactylus. Herpetologica 37: 86– 326–337. 92. Weisrock DW, Macey JR, Ugurtas IH, Larson A, Papen- Schmida GE. 1973. Geckos aus Nordwest-Australien. Die fuss TJ. 2001. Molecular phylogenetics and historical bio- Aquarien-und Terrarien Zeitschrift 26: 316–317. geography among salamandrids of the ‘true’ salamander Shimodaira H, Hasegawa M. 1999. Multiple compari- clade: Rapid branching of numerous highly divergent lineages sons of log-likelihoods with applications to phylogenetic in Mertensiella luschani associated with the rise of Anatolia. inference. Molecular Biology and Evolution 16: 1114– Molecular Phylogenetics and Evolution 18: 434–448. 1116. Wilson SK, Knowles DG. 1988. Australia’s reptiles. A photo- Sorenson MD. 1999. Treerot, Version 2. Boston, Massachu- graphic reference to the terrestrial reptiles of Australia. Mel- setts: Boston University. bourne, Australia: Collins, Angus and Robertson.