To compare the specimens identified as Leitoscoloplos panamensis, L. multipapillatus sp. nov. or Leitoscoloplos sp., 14 morphological features commonly used in orbiniid taxonomy were measured: body length to chaetiger 50 (Bln); thoracic width at chaetiger 10 (Tw), without parapodia; number of thoracic chaetigers (Tc); thoracic length (Tln); first chaetiger with branchiae (fBr); first neuropodium with bifurcate lobes (fBL); first chaetiger with interramal cirri (fIC); number of chaetigers with interramal cirri (cIC); first chaetiger with subpodal papillae (fSpP); number of chaetigers with subpodal papillae (cSpP); maximum number of subpodal papillae (mSpP); first chaetiger with stomach papillae (fStP); number of chaetigers with stomach papillae (cStP); maximum number of stomach papillae (mStP).
Orbiniids are very long, usually over 200 segments as adults (Eibye-Jacobsen, 2002) but, as all specimens in this study lacked their posterior regions, length measurements were standardized to 50 chaetigers. The measurements were made with an eyepiece micrometre in 38 specimens of L. panamensis collected in the Gulf of California, three syntypes of L. panamensis, 38 specimens of L. multipapillatus sp. nov., and 12 specimens of Leitoscoloplos sp. The 14 characters chosen were submitted to an exploratory analysis to identify general patterns such as their variability, which was performed with box diagrams and the Kruskal-Wallis test (KW) to identify differences between taxa; p values of less than 0.05 were considered statistically significant (Zar, 1996).
Based on the above mentioned morphometric measurements, a Principal Component Analysis (PCA) was carried out to evaluate the relationships between the three taxa under study, and to identify which variables best explained their variability (Hair et al., 1999). Later, a Discriminant Analysis using the forward stepwise method was applied to determine the best combination of variables that would discriminate between the taxa collected and the type material of L. panamensis. The standard statistic Wilks’ lambda (ranking from 1: no discriminatory power, to 0: perfect discriminatory power) was used to detect the statistical significance of the discriminatory power of the current model, where the groups must be differentiated at a significant level of p < 0.0001 to be considered true species. The characters’ selection for the model were made by the F-remove statistic, which selected the variables with F-values greater than 1. The partial Wilks’ lambda index was employed to evaluate the contribution of each character to the discrimination between groups, where 0 means a perfect discriminatory power. Next, a canonical analysis was carried out to compute the discriminant functions and to show how the morphological variables discriminate the specimens/species (Hair et al., 1999). All morphometric analyses were carried out using the software Statistica 7.0 for Windows.