Artículos de revistas
A new application of morphometrics in a study of the variation in uncinal shape present within the Terebellidae (Polychaeta): a reevaluation from digital images
Registro en:
Cahiers De Biologie Marine. Cahiers De Biologie Marine, v. 48, n. 3, n. 229, n. 240, 2007.
0007-9723
WOS:000250028400002
Autor
Garraffoni, ARS
Camargo, MD
Institución
Resumen
In this study, the morphometric approach was used to establish distinct morphological groups in regard to uncini shape within the four subfamilies of Terebellidae (Polychaeta). To achieve this objective, 24 distances were measured, based on photographs of the three uncini dissected from segment 7 and another three on segment 16, from 31 species, 2 belonging to Trichobranchinae, 2 to Polycirrinae, 6 to Thelepodinae and 21 to Terebellinae. Those distances were based on 7 real landmarks and 6 'extrapolated' landmarks, and measurements of the area, perimeter, greatest length, shortest length, diagonal perimeter, horizontal perimeter and vertical perimeter. The multidimensional approach to assess the similarity and dissimilarity of the distinct uncini patterns was done using n-MDS, ANOSIM and PCA. A prediction model was developed to identify the Terebellidae subfamily based on neural network. The n-MDS performed on the 31 Terebellidae species recognized three distinct groups: one group composed of species belonging to the subfamilies Thelepodinae-Terebellinae and two other distinct groups formed by the species of Trichobranchinae and Polycirrinae. The permutation tested by ANOSIM confirmed the trends observed in the n-MDS. The ordination of the two PCA axes explained 64.20% of the variance on the PCA realized on uncini of segment 7 and 65.05% of the variance on the PCA realized on uncini of segment 16. The PCA showed that most of the measurements made on the uncini were important to obtain a better classification of the data. Thus, these three different patterns can be assumed to be three different character states that define the overall uncini shape and can be used in future cladistics analyses. Finally, the model developed based on the neural network showed good success in classifying the four subfamilies. 48 3 229 240