2.5. Morphological data analysis
A mixed data matrix was created with all morphological characters
analysed. For multivariate analysis, a logarithmic transformation was
made for quantitative variables (number of oral tentacles, number of
longitudinal vessels in folds of the branchial sac, number of
longitudinal vessels between the folds of the branchial sac, number of
gonads and number of stomach folds). The height and width were not used
for this analysis because these two characters are highly variable due
to the elastic nature of the animal and the amount of water it contains.
Nevertheless, correlation was tested among these two characters and the
rest of the studied ones, and no significant relation was found (data
not shown). Multivariate analyses were used to determine affinities
between specimens of Cnemidocarpa verrucosa based on a
morphological character matrix. Nonmetric multidimensional scaling
(NMDS) was performed using three dimensions and Gower distance.
Nonparametric multivariate analysis of variance (PERMANOVA) was used to
assess differences between the groupings obtained in NMDS, each term of
the analysis was tested using 9999 permutations. PAST 3.16 free
statistical software was used for all the morphological data analyses
(Hammer, Harper, & Ryan, 2001).