The optimization of landmark data: a three-dimensional approach

CATALANO, SA; GOLOBOFF, PA; GIANNINI, N

San Miguel de Tucumán

Congreso; XXVII Meeting de la Sociedad Willi Hennig y VIII Reunión Argentina de Cladística y Biogeografía; 2008

Since the implementation of Farris (1970) optimization to analyze continuous characters (Goloboff et al 2006), there has been a renewed interest in using such characters in phylogenetic analyses, and in trying to extend phylogenetic analysis to quantitative descriptions of shape. All previous approaches were based on the optimization of values from some sort of unidimensional shape descriptor from the terminals. We consider that a more natural method should be based on explicitly estimating optimal ancestral shapes. Particularly, in the case of landmark configurations, we propose a generalization of Farris optimization to more than one dimension.  In this approach, each landmark in the ancestor can be placed on any position in a space that presents the same dimensionality than that on which the landmark data were defined. Under a 3D-optimization, the optimal landmark positions of ancestors are those which minimize the distances between homologous landmarks along branches.  Although not proposed as a final solution to the analysis of shapes in a parsimony context, our approach synthesizes well-known elements of phylogenetic analysis into a new perspective for the problem of phylogenetic analysis of shape.