Morphological Species Delimitation in The Western Pond Turtle (Actinemys): Can Machine Learning Methods Aid in Cryptic Species Identification?

Abstract

As the discovery of cryptic species has increased in frequency, there has been an interest in whether geometric morphometric data can detect fine-scale patterns of variation that can be used to morphologically diagnose such species. We used a combination of geometric morphometric data and an ensemble of five supervised machine learning methods (MLMs) to investigate whether plastron shape can differentiate two putative cryptic turtle species, Actinemys marmorata and Actinemys pallida. Actinemys has been the focus of considerable research due to its biogeographic distribution and conservation status. Despite this work, reliable morphological diagnoses for its two species are still lacking. We validated our approach on two datasets, one consisting of eight morphologically disparate emydid species, the other consisting of two subspecies of Trachemys (T. scripta scripta, T. scripta elegans). The validation tests returned near-perfect classification rates, demonstrating that plastron shape is an effective means for distinguishing taxonomic groups of emydids via MLMs. In contrast, the same methods did not return high classification rates for a set of alternative phylogeographic and morphological binning schemes in Actinemys. All classification hypotheses performed poorly relative to the validation datasets and no single hypothesis was unequivocally supported for Actinemys. Two hypotheses had machine learning performance that was marginally better than our remaining hypotheses. In both cases, those hypotheses favored a two-species split between A. marmorata and A. pallida specimens, lending tentative morphological support to the hypothesis of two Actinemys species. However, the machine learning results also underscore that Actinemys as a whole has lower levels of plastral variation than other turtles within Emydidae, but the reason for this morphological conservatism is unclear.

Fig. 1

Geographic distribution of specimens used for comparing the hypothesized subdivisions of the A. marmorata complex. Each hypothesized scheme has two or more possible classes (see Table 1 for explanation of schemes). Sample size differs between schemes because of variance in our ability to confidently assign museum specimens to the schemes. The number of localities shown on each map is less than the number of specimens sampled because some localities produced multiple specimens. A. marmorata, marm; A. pallida, pall; central coast ranges, CCR; San Joaquin valley, SJ; Baja peninsula, Baja; Sierra foothills, Foothill.

Fig. 2

Depiction of general plastron shape of A. marmorata and position of the 19 landmarks used in this study. Anterior is toward the top of the figure.

Fig. 3

Principal components analysis scatterplots summarizing plastron shape variation in two of the datasets used in this study. (A) Scatterplot of the first two PCA axes from the eight emydid species datatset. (B) Scatterplot of the first two PCA axes from the Trachemys subspecies dataset. There are clear distinctions between the different species or subspecies in both datasets. Point colors correspond to the categories within each dataset; point size is proportional to individual centroid size. Parenthetical values in axis labels are the percentage of total variance accounted for by the axis.

Fig. 4

Principal components analysis scatterplots summarizing plastron shape variation in the A. marmorata dataset. Each panel corresponds to one of the six different classification schemes analyzed in this study (Table 1). Point color corresponds to categories within each scheme and class names correspond to geographic regions. “N/A” specimens could not be assigned confidently to a class in given scheme. Point size is proportional to individual centroid size, except for N/A specimens. Parenthetical values in axis labels are the percentage of total variance accounted for by the axis.

Fig. 5

Scatterplots comparing morphological variation present in a single well-sampled locality (89 specimens; Trinity County, CA, USA) to the remainder of the A. marmorata dataset. (A) Scatterplot of the first two principal component axes for the full A. marmorata dataset. (B) Scatterplot of shape scores (sensu Drake and Klingenberg 2008) from a multivariate regression of shape versus log₁₀ altitude. (C) Scatterplot of shape scores from a multivariate regression of shape versus log₁₀ slope. (D) Scatterplot of shape scores from a multivariate regression of shape versus flow rate. Specimens from the Trinity County locality are shown in blue; specimens from other localities in gray.