Portal de Periódicos da CAPES

Taxonomic Revision of the South American Subspecies of the Turtle Kinosternon scorpioides

1997; The Society for the Study of Amphibians and Reptiles; Volume: 31; Issue: 4 Linguagem: Inglês

10.2307/1565602

1937-2418

Mario R. Cabrera, Sonia Edith Colantonio,

Primate Behavior and Ecology

-The taxonomic status of Kinosternon scorpioides seriei Freiberg, and K. s. carajasensis Cunha is reviewed. The characters from their original descriptions were reassessed. Morphometric distinction was evaluated with principal components, UPGMA cluster, and discriminant function analyses. The presumed distinctive characters showed extensive overlap among both subspecies and the nominal race, and no new character could be found to discriminate among the three subspecies. The taxonomic significance of their supposed allopatry is discussed. It is suggested that both K. s. seriei and K. s. carajasensis be placed in the synonymy of K. s. scorpioides. The South American subspecies of Kinosternon scorpioides include K. s. scorpioides (Linne, 1766), K. s. seriei Freiberg, 1936, and K. s. carajasensis Cunha, 1970. Kinosternon s. pachyurum Muller and Hellmich, 1936, is regarded as a synonym of K. s. seriei (Freiberg, 1938; Berry, 1978; Cei, 1993). These forms are supposedly well-delimited geographically: Kinosternon s. seriei inhabits the Gran Chaco, K. s. carajasensis occurs only in the Serra dos Carajas, State of Para, Brazil, and the nominal form is widespread in northern South America (Pritchard and Trebbau, 1984). However, subspecific identification of specimens without locality data is often difficult, due to overlapping of characters. Kinosternon s. seriei and K. s. carajasensis were described on small, perhaps inadequate samples. As a result, their validity, or that of their characters, has been doubted repeatedly (e.g., Pritchard, 1979; Vanzolini et al., 1980; Vanzolini, 1981; Pritchard and Trebbau, 1984; Cabrera, 1995). In his description of K. s. seriei, Freiberg (1936) gave not a diagnosis, but a brief description based on three specimens representing both sexes, with measurements and illustrations. Under the subheading Observaciones Freiberg summarized the features that he felt would distinguish this subspecies (see below). Cunha (1970) described K. s. carajasensis on the basis of six individuals, juvenile and adult, representing both sexes, four of them forming the type series. He provided a diagnosis, with measurements and some drawings, comparisons to related forms, and a map of the Serra dos Carajas. Later, Cunha et al. (1985) treated this form as a synonym of K. s. scorpioides with a brief justification: ' .. presently admitted as a synonym of (K.) scorpioides scorpioides, notwithstanding showing marked variations (our translation). In order to assess the validity of, or relationships among these three South American races of K. scorpioides, we conducted this review. MATERIALS AND METHODS This study was approached from two levels: (1) morphological features relevant to the recognition of these subspecies, according to their original descriptions, were analyzed and compared; (2) a numerical analysis was performed, including both diagnostic and other mensural characters from the shell. The samples comprised turtles with carapace length >110 mm from throughout the range of the nominal race in Brazil, plus all available specimens of K. s. seriei in the South American collections surveyed, and all the type series of 507 This content downloaded from 157.55.39.135 on Sun, 03 Jul 2016 06:25:29 UTC All use subject to http://about.jstor.org/terms M. R. CABRERA AND S. E. COLANTONIO K. s. carajasensis, except one juvenile (see Appendix 1). Shell and scute measurements (to the nearest 0.5 mm) were recorded, including proximal dorsal width of the nuchal (=cervical) scute (PN); distal dorsal width of the nuchal scute (DN); midline length of the nuchal scute (NL); maximum width of the first vertebral (=central) scute (WVI); maximum carapace length (CL); maximum carapace width (CW); midline length of gular scute (GL); midline length of the plastral forelobe (PFL); maximum length of the plastral hindlobe (PHL); maximum plastron length (PL); height of the anterior shell opening, measured from the ventral plane to the medial anterior border of the carapace (AH). Because the interfemoral (IF) and interanal (IA) seams are frequently asymmetrical in this species, the length of the medial border of the right and left of each pair of scutes for each specimen was measured, and the respective means were calculated. Maximum measurements are not always coincident with midline measurements. To facilitate comparisons with earlier papers, our scute terminology follows conventional use, in spite of the probable homology of plastral scutes as proposed by Hutchison and Bramble (1981). The specimens of the samples vary considerably in CL (mm): K. s. scorpioides (x = 141.1 + 12.9, range = 116-166, N = 40); K. s. seriei (x = 156.6 + 17.4, range = 135-181, N = 9); K. s. carajasensis (x = 125.3 ? 3.5, range = 122-129, N = 3). Therefore, all characters were standardized for body size by linear regression of each variable against CL, using the residuals in the statistical calculations. The use of residuals has been controversial (see review in Jungers et al., 1995). However, we submitted our raw data to DM RAW (ratios formed by dividing each variable by the geometric mean, the latter computed as the nth root of the product of all n variables; Jungers et al., 1995), as an alternative size-adjustment method, and the results obtained were virtually identical to those using residuals. The use of residuals as input variables was preferred because it has proved useful in taxonomic studies of turtles (e.g., McCord and Iverson, 1991; Iverson and McCord, 1994). ANOVA and MANOVA tests were performed between K. s. scorpioides and K. s. seriei; K. s. carajasensis was excluded due to small sample size. A principal components analysis (PCA) of the character correlation matrix, a UPGMA cluster analysis using Squared Euclidean Distance, and a canonical discriminant function analysis (DFA) were performed using the residuals of each character, with the exception of CL, PN, DN, and NL. The last three characters were excluded because they showed no linear relationship to CL. Only one male specimen (the type) and two females of K. s. carajasensis were available for analysis. Therefore, these specimens were included as ungrouped (i.e., classified a posteriori) in the DFA of K. s. scorpioides versus K. s. seriei. In all cases males and females were analyzed separately to reduce the effects of sexual dimorphism. All morphometric analyses were conducted using SPSS (SPSS Inc., 1990) software.