Zootaxa 1363: 23–38 (2006) www.mapress.com/zootaxa/
ISSN 1175-5326 (print edition)
Copyright © 2006 Magnolia Press
ISSN 1175-5334 (online edition)
ZOOTAXA
A review of the systematics of the genus Bradypodion (Sauria: Chamaeleonidae), with the description of two new genera COLIN R. TILBURY 1,4*, KRYSTAL A. TOLLEY 2,4 & WILLIAM R. BRANCH 3 1
No. 2 The Bend, P.O. Box 347, Nottingham Road, 3280, KZN, South Africa. E-mail:
[email protected];
[email protected] 2 Molecular Systematics Laboratory, South African National Biodiversity Institute, Kirstenbosch Research Centre, P/Bag X7, Claremont 7735 South Africa. E-mail:
[email protected] 3 Port Elizabeth Museum, P.O. Box 13147, Humewood 6013, South Africa. E-mail:
[email protected] 4 Evolutionary Genomics Group, Department of Botany and Zoology, University of Stellenbosch, South Africa *Corresponding author
Abstract The taxonomic history and composition of the genus Bradypodion as construed by Klaver & Böhme (1986) and new morphological and molecular data relevant to the taxonomy of the group is reviewed. The combined evidence strongly supports a formal rearrangement of the group into three distinct genera. Bradypodion, type species Chamaeleo pumilus Daudin 1802, is retained for the southern African species. Two new genera are erected to accommodate additional well-diagnosed clades within central and east African species previously referred to Bradypodion. Species of the “fischeri complex” are assigned to Kinyongia gen. nova, whilst the endemic Mulanje chameleon is placed in the monotypic genus Nadzikambia gen. nova. Key words: Chamaeleonidae, Bradypodion, Phylogeny, new genera, mitochondrial and nuclear DNA
Introduction The first species of Bradypodion to be formally described was Chamaeleo pumilus Daudin 1802. Fitzinger (1843) subsequently designated pumilum as the type species for a new genus, Bradypodion. Over the course of the next 133 years, the taxonomy of Bradypodion pumilum and its cogeners underwent a convoluted series of taxonomic shifts including being assigned to two other genera viz: Microsaura and Lophosaura (Grey 1864, Boulenger 1887, Werner 1902, Methuen & Hewitt 1914, Fitzsimons 1943, Loveridge
Accepted by S. Carranza: 5 Oct. 2006; published: 23 Nov. 2006
23
1363
ZOOTAXA
1363
1957, Hillenius 1959, Mertens 1966 and Klaver 1973). Raw 1976 and Klaver & Böhme 1997 give good summaries of the taxonomics of the genus. Up until 1976, B. pumilum and all its cogeners were included within the genus Chamaeleo. Finally Raw (1976), citing a number of features purported to be diagnostic for the genus including its unique cranial structure and viviparity, resurrected the genus Bradypodion and assigned all the South African dwarf chameleons to this genus as full species. Klaver & Böhme (1986) after assembling all the currently available information, proposed a systematic revision of the family Chamaeleonidae based largely on hemipenal characteristics, including the available evidence accrued on lung morphology, chromosomal and osteological data. They recognized five genera and two sub-genera, four of these from Africa and three from Madagascar and its adjacent Indian Ocean islands. Although the genus Bradypodion was retained by Klaver & Böhme (1986), they did not consider the diagnostic features of Bradypodion sensu Raw (1976) to be restricted to only South African species. Rather, they expanded the genus to include a group of east and central African chameleons, loosely termed the “fischeri complex” (Hillenius 1959, Böhme & Klaver 1990). This group currently includes Bradypodion fischeri (Reichenow), B. tenue (Matschie), B. uthmoelleri (Müller), B. xenorhinum (Boulenger), B. carpenteri (Parker), B. adolfifriderici (Sternfeld), B. excubitor (Barbour), B. mlanjense (Broadley), B. oxyrhinum Klaver & Böhme, and B. tavetanum (Steindachner). Under this rearrangement, Bradypodion now comprised a morphologically heterogeneous, widespread group of species (Fig. 1) that was acknowledged by Klaver & Böhme (1986) to be undiagnosable by any morphological synapomorphies. Although many authors have adopted the phylogeny of Klaver & Böhme (1986), and most new species descriptions since 1986 by other authors adopted their taxonomic classification (Tilbury 1991, 1998; Lutzmann & Nečas 2002; Nečas et al. 2003; Nečas 2004), the diagnosis and content of Bradypodion remains contested (e.g. Branch 1998). As time has passed and new data has accrued, there is a growing conviction that the inclusion of the central and east African species within an expanded genus Bradypodion is not supported. Whereas in the past the Southern African forms around pumilum were once considered to be at most subspecies of pumilum, there is now good evidence that the genus comprises a monophyletic group of distinct evolutionary taxonomic units (Tolley et al. 2004, 2006; Tolley & Burger 2004).
Evidence Suggesting Nonmonophyly of Bradypodion Some of the morphological attributes of South African Bradypodion are incongruent with northern putative cogeners. All South African Bradypodion have a midline gular crest, usually formed of prominent scaly lobes, whilst none of the described “fischeri complex” possess such ornamentation. Moreover, South African Bradypodion lack rostronasal projections, whilst all except three species of the “fischeri complex” have such features. 24
© 2006 Magnolia Press
TILBURY ET AL.
The hemipenes of Southern African Bradypodion and the “fischeri complex” essentially display the same plesiomorphic configuration of a calyculate truncus with apical ornamentation based around two pairs of denticulate rotulae. Hemipenial morphology is thus phylogenetically uninformative with respect to relationships between these groups. However, the Mulanje Chameleon, B. mlanjense, the most southerly of the “fischeri complex”, has a unique hemipenal structure (see below).
FIGURE 1. Combined locality map of Bradypodion sensu lato demonstrating geographic distribution patterns of two broad groups within the genus sensu Klaver & Bhme 1986. Red dots indicate records for Southern African Bradypodion and black dots indicate records for the “fischeri complex”. Data locality points are based on specimens from seventeen museum collections.
A REVIEW OF BRADYPODION
© 2006 Magnolia Press
25
ZOOTAXA
1363
ZOOTAXA
1363
Bauer (1997) noted that all species of Southern African Bradypodion have a pigmented parietal peritoneum. He therefore questioned the inclusion of central and east African species, which lack parietal peritoneal pigmentation, within Bradypodion. Furthermore, all South African Bradypodion are viviparous, whilst east and central African species assigned to Bradypodion are oviparous. Although viviparity also exists in several East African chameleons allied to the species around the Chamaeleo (Trioceros) bitaeniatus Fischer and Chamaeleo (Trioceros) werneri Tornier groups (sensu Hillenius 1959), Townsend & Larson (2002) specifically rejected the hypothesis that all viviparous chameleons formed a monophyletic group. Based on mtDNA analysis, they offered strong support for at least two separate origins of viviparity (Townsend & Larson 2002). In all species groups where viviparity occurs (Bradypodion, Chamaeleo (Trioceros) bitaeniatus group, and Chamaeleo (Trioceros) werneri group), or is present in an intermediate phase [e.g. Rhampholeon (Bicuspis) marshalli Boulenger (Humphreys 1990)], this reproductive strategy is always associated with pigmentation of the parietal peritoneum. The internal structure of the lungs of the South African Bradypodion (Type B – Klaver 1981, Klaver & Böhme 1986) is largely similar to that of the east African “fischeri complex”, as well as that of the Malagasy genera Calumma and Furcifer. Essentially the lungs of all these species are small and consist of a relatively simple sac with a varying number of small septae present on the anterior, dorsal, antero-ventral and cephalic walls. The lungs of the South African species, however, lack external appendages or trailing diverticulae, although these are present in most other typical chameleons examined to date. Whilst the presence of a gular pouch has been verified in all species examined of the South African Bradypodion [i.e. damaranum (Boulenger), occidentale (Hewitt), pumilum and ventrale], this accessory respiratory structure is absent in all the species of the “fischeri complex” examined to date (adolfifriderici, fischeri, tenue and xenorhinum; Klaver 1973, 1979, 1981). The marked difference in cranial morphology between the South African Bradypodion and all other typical chameleons has been cited as evidence of a high level of divergence by some authors (Metheun & Hewitt 1914, Raw 1976), but not others (Klaver 1973, Klaver & Böhme 1986). In the South African Bradypodion the lateral crests of the head that extend posteriorly from the supraorbital ridge, form over the lateral edge of the parietal bone, and the temporal crests outline the position of the postfrontal and squamosal bones. In the species of the “fischeri complex”, and as far as can be ascertained in most other species of typical chameleons, the lateral crest of the head is formed along the line of the squamosal bone. As such the lateral crests of the South African Bradypodion are not homologous with the lateral crests of B. fischeri (Reichenow) and other species of typical arboreal chameleons (Raw 1976). The broad parietal bone in South African Bradypodion uniquely recurves posteriorly and on each side has a supra-temporal process which projects inferiorly to make contact with the ascending process of the squamosal bone. In B. fischeri and all other genera of typical chameleons, the parietal bone is narrowed
26
© 2006 Magnolia Press
TILBURY ET AL.
posteriorly into a thin sagittal spur that connects at its posterior apex with the squamosal bones. Authors disagree as to whether these cranial features are plesiomorphic (Klaver 1981) or apomorphic (Hillenius 1986, 1988). In iguanids (de Queiroz 1987), e.g. Sphenodon (Rieppel 1992) and Chamaeleo (Rieppel 1993), a broad parietal skull table precedes the ontogenetic development of the narrow parietal condition found in adults, indicating that the broad parietal condition is a plesiomorphic feature and the narrow condition derived. Intergeneric relationships derived from both cladistic analysis of the cranial structure and genetic studies suggest that Brookesia is the sister taxon of all other chameleons. Bradypodion sensu stricto and Brookesia share several cranial features including the broad parietal bearing supratemporal processes and fused nasals (Rieppel 1981, 1987; Rieppel & Crumley 1997). Other observations on the cranial structure of B. pumilum may have taxonomic implications, and include the retention of Jacobsen’s organ (Brock 1941, Malan 1946, Engelbrecht 1950). The dataset for this plesiomorphic character in the Chamaeleonidae is, however, largely incomplete and further discussion is thus deferred. Hofman et al. (1991) examined relationships between purportedly closely related species of chameleons using microcomplement fixation and starch gel electrophoresis of a range of allozymes. They found that the east African taxon B. tavetanum was closer to the Malagasy genus Furcifer than it was to the South African Bradypodion (represented by thamnobates, setaroi (Raw) and melanocephalum) and concluded that there was no support to classify them in the same genus. More recent molecular studies have shown that there are at least two Bradypodion clades that are reciprocally monophyletic and show levels of divergence that are comparable to divergences between other recognised chameleon genera (Townsend & Larson 2002, Tolley et al. 2004.) As the morphological evidence reviewed above does not support the monophyly of Bradypodion, sensu Klaver & Böhme (1986), and since a number of inferences had been made in the literature that the current taxonomy of the genus Bradypodion was flawed, a molecular analysis was conducted with the objective of examining the evolutionary relationships between the East African and South African Bradypodion and all other extant chameleon genera.
Molecular analysis: Materials & methods To determine if taxa currently included within Bradypodion share a close evolutionary history with each other, a phylogenetic analysis of 51 chameleons was run. The analysis included all described forms of east and central African Bradypodion, plus several representatives from each of the other genera of chameleons (Table 1). Sequences from 20 of these individuals have been published previously (Table 1). DNA extraction, PCR amplification, and cycle sequencing of two mitochondrial gene fragments were carried out following standard procedures formerly outlined in Tolley et al. (2004) using the A REVIEW OF BRADYPODION
© 2006 Magnolia Press
27
ZOOTAXA
1363
ZOOTAXA
1363
28
© 2006 Magnolia Press
TILBURY ET AL.
ZOOTAXA
1363
A REVIEW OF BRADYPODION
© 2006 Magnolia Press
29
ZOOTAXA
1363
following primers for ND2: L4437b (Macey et al. 1997a) and H5934 (Macey et al. 1997b), and 16S: L2510 and H3080 (Palumbi 1996). A 718 bp portion of the nuclear gene RAG1 was amplified and sequenced using primers F118 and R1067 (Matthee et al. 2004). Standard PCR and sequencing were followed for this gene fragment, with PCR annealing temperature at 57°C. All new sequences have been deposited in GenBank (Table 1). Voucher specimens are as listed in Table 1. A Bayesian analysis of 2036 characters from the two mitochondrial genes (ND2, 856 bp and 16S, 462 bp) and the one nuclear gene (RAG1, 718 bp) was run. The outgroup consisted of two other acrodont lizards (Table 1, Agamidae: Leiolepis belliana and Agamidae: Calotes versicolor). The South African Bradypodion have previously been shown to be monophyletic (Tolley et al. 2004) so to reduce computational time, only four representatives from that group were included in the present analysis. Bayesian inference was used to investigate optimal tree space using MrBayes 3.1.0 (Huelsenbeck & Ronquist 2001). Modeltest 3.6 (Posada & Crandall 1998) was run to investigate the evolutionary model that best fits the data set. Both the AIC and LRT test specified the most complex model (GTR+I+G), so MrBayes was run specifying six rate categories with uniform priors for all parameters. Several data partitions were created, and each was allowed to run with separate values for the model parameters. A single data partition was created for 16S, although 24 bases (260–268 and 284–298) were removed due to poor alignment. There were three partitions for ND2: 1st, 2nd, and 3rd codons separately, and three partitions for RAG1: 1st, 2nd, and 3rd codons separately. To ensure the results converged on the same topology, the MCMC was run four times for 5 million generations each, with trees sampled every 100 generations. The first 90k generations (900 trees) were removed as burn-in after examination the average standard deviation of split frequencies ( 0.95 posterior probability were considered supported. A parsimony analysis was also run following the procedures outlined in Tolley et al. 2004, and excluded the 24 bases in 16S (as in the Bayesian analysis). One thousand bootstrap replicates were run to evaluate confidence in the nodes. The parsimony analysis (not shown) produced a single tree (5055 steps, CI 0.38, RI 0.55) with the overall same topology and supported nodes (>75% bootstrap) as the Bayesian analysis. Competing phylogenetic hypotheses were tested by comparing the parsimony tree length of the phylogeny presented in this study with 1) the enforced monophyly of the Bradypodion sensu Klaver & Böhme (1986), 2) the enforced monophyly of Bradypodion including only southern and east African taxa, 3) the enforced monophyly of the southern African Bradypodion and the Mulanje taxon. Significance was evaluated with the K-H and S-H tests in PAUP*4.0b10 (Swofford 2002).
30
© 2006 Magnolia Press
TILBURY ET AL.
Molecular analysis: Results
ZOOTAXA
1363 The South African Bradypodion form a strongly supported monophyletic group to the exclusion of all other chameleons, including east and central African Bradypodion (sensu Klaver & Böhme 1986). In addition, the east and central African Bradypodion (sensu Klaver & Böhme 1986) form a monophyletic group with strong support (Fig. 2). This clade includes all species in the “fischeri complex”. A fuller picture of phylogenetic relationships among all known species within this clade utilizing mitochondrial and nuclear genes, will be presented elsewhere (Tolley et al. in prep). Sequence divergences (uncorrected p-distances) between the East and Southern African Bradypodion clades is high for all genes, averaging 20% for ND2, 8% for 16S, and 4% for RAG1 (Table 2). These values of divergence are as high as those found among any other chameleon clades that are currently recognised as genera. Bradypodion mlanjense (sensu Klaver & Böhme 1986) does not fall into either the South African Bradypodion clade, or the east African Bradypodion clade, but forms a separate monotypic clade (Fig. 2). Sequence divergence between it and other clades ranges from 20–27% for ND2, 9 to 15% for 16S, and 4 to 9% for RAG1, which are values as high as those between any other currently established genera. Each of the competing phylogenetic hypotheses enforcing monophyly showed a significantly worse fit (p
