Mitochondrial DNA Corroborates Taxonomy of the South American Chelid Turtles of the Genera Platemys and Acanthochelys
Abstract
A phylogenetic analysis of partial sequences of 2 mitochondrial genes, NADH dehydrogenase subunit 4 (ND4) and cytochrome b (cyt b), supports the generic distinction and sister taxon relationship of Platemys and a monophyletic Acanthochelys. Acanthochelys radiolata is the most distinct member of the genus Acanthochelys and is sister to a clade that includes A. spixii, A. pallidipectoris, and A. macrocephala. Parsimony and maximum likelihood analyses suggested A. spixii (A. pallidipectoris, A. macrocephala), but this was not strongly supported in bootstrap analysis parsimony and maximum likelihood analyses and low Bayesian posterior probabilities. Evidence of molecular diversity within morphologically defined species suggests that there may be additional cryptic species in this group.
Previous to Iverson (1992), the South American chelid turtles of the Platemys/Acanthochelys complex comprised a single genus (Platemys, Wagler 1830) containing 5 species. The Turtle Taxonomy Working Group (2007) revised this taxonomy so that the genus Platemys currently includes only P. platycephala, and the 4 remaining species are now considered Acanthochelys (Gray 1873). Platemys occurs throughout the Amazonian river drainage in Brazil, Venezuela, Colombia, Ecuador, Peru, Bolivia, Guyana, Suriname, and French Guyana (Iverson 1992). Two subspecies are currently recognized: P. p. platycephala, which is broadly distributed, and P. p. melanonota (Ernst 1983b), which is known only from a few river systems in Peru and Ecuador (Ernst and Barbour 1989). The 4 species of Acanthochelys (Turtle Taxonomy Working Group 2007) are distributed in southern South America. Acanthochelys radiolata has a disjunct range in Brazil including a narrow eastern coastal strip and a geographically isolated western population (Rhodin et al. 1984a). Acanthochelys spixii occurs in eastern Brazil and adjacent Argentina and Uruguay (Ernst 1983c; Rhodin et al. 1984a). Acanthochelys macrocephala occurs in central Bolivia and adjacent Brazil (Rhodin et al. 1984b). Acanthochelys pallidipectoris is geographically isolated, occurring only in the Chaco region of northern Argentina and perhaps in southern Bolivia and Paraguay (Rhodin 1982; Ernst 1983a; Rhodin et al. 1984a). The present distribution of all 5 species in the complex is shown in Figure 1.
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0800.1
Standard karyotypes for all 5 species were reported by McBee et al. (1985); the group is characterized by a remarkable level of chromosomal variation. The diploid number (2n) ranges from 48 to 64, and the number of autosomal arms (fundamental number [FN]) ranges from 60 to 64. Acanthochelys spixii and A. pallidipectoris have indistinguishable karyotypes (2n = 50, FN = 62), whereas A. radiolata (2n = 50, FN = 63) and A. macrocephala (2n = 48, FN = 60) each have unique karyotypes. Platemys platycephala has 2n = 64 and FN = 64 and also possesses a unique form of ploidy mosaicism. It is 1 of only 2 known sexually reproducing organism to possess diploid–triploid mosaicism (Bickham et al. 1985; Bickham and Hanks 2009).
A previous study of allozymes examined the phenetic and phylogenetic relationships among the species of this complex. Derr et al. (1987) showed P. platycephala to be the most divergent taxon in the group based on their phenetic analysis. It had more than twice the number of autapomorphic character states of any of the other species of the complex. Moreover, McBee et al. (1985) suggested the possibility of partitioning the complex into 2 genera based on the high degree of karyotypic differentiation between P. platycephala and the other 4 species. Phylogenetic analysis of the allozyme data showed a sister relationship between A. pallidipectoris and A. spixii that is consistent with these 2 taxa sharing identical karyotypes (McBee et al. 1985). However, allozymes failed to resolve phylogenetic relationships among the remaining taxa. Previous phylogenetic analysis of 12S and 16S rRNA genes of chelids demonstrated a sister relationship between Platemys and Acanthochelys; however, only 2 of the 5 species, Platemys platycephala and Acanthochelys pallidopectoris, were included in those analyses (Seddon et al. 1997; Georges et al. 1998).
Taxonomic conclusions are strongest when based on multiple lines of evidence. In this study, we complement existing morphological, chromosomal, and allozyme data with mtDNA sequence data including partial sequences of 2 genes (ND4 and cytochrome b [cyt b]) from all 5 recognized species of the Platemys/Acanthochelys complex. We test the hypothesis of a monophyletic Acanthochelys and examine the relationships among the 4 species of Acanthochelys. The data are interpreted in light of previous studies of the systematics of the group.
METHODS
We used archived tissue samples stored at −80°C that were examined in the allozyme and karyotypic studies of Derr et al. (1987) and McBee et al. (1985). Voucher specimens are all deposited in the Texas Cooperative Wildlife Collection, Texas A&M University. DNA was extracted utilizing standard proteinase K digestion and organic phase phenol/chloroform extraction (Hillis et al. 1996). Two different regions of mtDNA were amplified via polymerase chain reaction (PCR) for analysis. A 719-base-pair (bp) portion of the ND4 gene was amplified utilizing primers ND4 and Leu (Arevalo et al. 1994). A 433-bp portion of the cyt b gene was amplified with primers CytbGlu (Caccone et al. 1999) and CytbH15149 (Engstrom et al. 2007). PCR amplification conditions were as stated in the publication of each respective primer pair. Amplicons were checked for expected size by electrophoresis through 1% agarose gels. The fragments were then purified using Qiaquick spin purification columns (Qiagen, Valencia, CA). Sequences were visualized with an automated sequencer (Applied Biosystems 377, Foster City, CA) following the manufacturer protocols. Internal sequencing primers designed specifically for this project were utilized to confirm the sequences for the ND4-Leu region. Amplicons were sequenced in both directions and sequences were aligned in Sequencher v3.1.1 (Gene Codes Corp., Ann Arbor, MI).
Confirmed sequences were then submitted to GenBank (accession nos. EF535281–EF535304). The African pelomedusid Pelomedusa subrufa (GenBank NC00194) and the South American chelid Phrynops gibbus were utilized as the out-group taxa. Aligned sequences were analyzed by maximum parsimony, maximum likelihood, and neighbor-joining analyses in PAUPv4.0b10* using the default settings, except as described below (Swofford 2002). Additionally, Bayesian analysis was carried out in MrBayes v3.0 (Huelsenbeck and Ronquist 2001). Modeltest3.7 was utilized to determine the appropriate genetic distance model. Results determined the GTR + Inv model as the appropriate model for this data set. Bootstrap analysis was conducted utilizing a full heuristic search with 10,000 replicates.
RESULTS
ND4 Region
Among the 719 bases assayed per individual for the portion of the ND4 region sequenced, 184 (25.9%) were polymorphic across taxa among the in-group, and 148 were parsimony informative. Sequence divergence estimates ranged from 0.021 to 0.132 within the genus Acanthochelys, from 0.136 to 0.150 between the genera Acanthochelys and Platemys, and from 0.156 to 0.280 in comparison with the out-group taxa. There were a total of 34 nonsynonymous changes and 103 synonymous changes. The transition/transversion ratio was 3.14.
Cyt b Region
Of the 433 nucleotides sequenced per individual for the cyt b, 117 (29.8%) were polymorphic across taxa, and 94 of these were parsimony informative. Sequence divergence estimates ranged from 0.012 to 0.135 within the genus Acanthochelys, from 0.185 to 0.21 between the genera Acanthochelys and Platemys, and from 0.133 to 0.297 in comparisons with the out-group taxa. A total of 28 non-synonymous substitutions were found within the cyt b region. In addition, a total of 48 synonymous substitutions were found within the sequenced region of cyt b. The transition/transversion ratio for cyt b was 3.10.
ND4 and Cyt b Combined
Analysis of the combined data set by heuristic parsimony yielded 470 rearrangements, with the 1 tree retained that had a score of 701 (Consistency Index = 0.779, Retention Index = 0.781). Bayesian analysis was conducted using GTR + Inv model. Total number of Markov chain–Monte Carlo generations conducted in the analysis was 1 million generations. Generations were sampled every 100th generation, and a burn-in of 200,000 generations was used. Generations were sampled until the standard deviation of the split frequencies was below 0.01. In the analysis of the combined sequence data from ND4 and cyt b, all phylogenetic analyses supported the sister relationship of Platemys to Acanthochelys (Fig. 2) as first proposed by McBee et al. (1985) and supported by Derr et al. (1987). These analyses also showed A. radiolata to be sister to the remaining members of Acanthochelys. Strict parsimony and strict maximum-likelihood analyses also showed A. pallidipectoris and A. spixii to be sister taxa, and together they form a clade sister to A. macrocephala. However, in the bootstrap analyses of the maximum parsimony tree and the maximum likelihood tree and in the Bayesian analysis, the relationships among A. spixii and A. pallidipectoris were unresolved (Fig. 2).
Citation: Chelonian Conservation and Biology 12, 1; 10.2744/CCB-0800.1
DISCUSSION
A sister relationship of Platemys and a monophyletic Acanthochelys is fully supported in all of the molecular sequence analyses conducted. This is in concordance with previous findings based on chromosome number, morphology, and allozymes. Within Acanthochelys, A. radiolata is the sister taxon to the remaining 3 species in all the analyses conducted, which is also roughly equivalent to the generic level designations in other Testudines (Caccone et al. 1999; Lenk et al. 1999; Feldman and Parham 2002; Spinks et al. 2004). The relationships of the remaining members of Acanthochelys are not fully resolved. In parsimony bootstrap analysis, the clade formed by A. macrocephala, A. pallidipectoris, and A. spixii is strongly supported (100%). However, the relationships among A. macrocephala, A. pallidipectoris, and A. spixii were unresolved in the bootstrap analysis, as the branches comprised an unresolved trichotomy.
In a phylogenetic analysis of allozyme data, Derr et al. (1987) suggested a sister relationship of A. pallidipectoris to A. spixii, and these 2 species share identical karyotypes (McBee et al. 1985). This was suggested in our neighbor-joining analysis but was supported by only 59% of the replicates in the bootstrap analysis. No support for this relationship was evident in any of the phylogenetic or likelihood analyses.
One specimen of A. radiolata in our analysis (AK1453) demonstrated a higher level of sequence divergence relative to other specimens of A. radiolata. This specimen exhibited divergence estimates of 3.99% and 4.25% from AK1452 and AK6594, respectively, whereas the level of sequence divergence between AK1452 and AK6594 was only 0.78%. In comparison to other members of Acanthochelys, the level of sequence divergence exhibited by AK1453 in comparison with other members of A. radiolata is equivalent to the sequence divergences among A. macrocephala, A. pallidipectoris, and A. spixii. Previously published studies on Testudines have found similar levels of divergence between species (Lenk et al. 1999; Feldman and Parham 2002). All A. radiolata specimens were obtained via the pet trade, from which reliable locality data were unobtainable. The specimens were morphologically identified as A. radiolata by one of the authors (A.G.J. Rhodin). Presence of this level of mtDNA diversity within the morphologically defined species suggests that there may be additional cryptic species in this group. This potential of greater diversity warrants additional field and lab studies.
The molecular data supported the previously described sister relationship of the genus Platemys relative to members of the Acanthochelys genus with A. radiolata the sister taxon of the remaining 3 species of Acanthochelys. The relationships of the other 3 members of Acanthochelys were not resolved with the analysis of mtDNA data. However, the present data do not support (or refute) the sister taxon relationship of A. pallidipectoris and A. spixi that was proposed based on biochemical analyses (Derr et al. 1987). The inclusion of additional specimens of the underrepresented taxa in the molecular analysis of the genus Acanthochelys, as well as additional sequence data, could help further resolve the relationships among the species of this little known South American genus.
Present distribution of species from the genera Platemys and Acanthochelys.
Phylogenetic tree (maximum parsimony) showing the relationships among Platemys and Acanthochelys. Bootstrap support values (≥ 50%) are presented above the nodes (parsimony/maximum likelihood), and Bayesian posterior probabilities are presented below the respective nodes.
