Study of bivalve larval growth using image processing

The understanding of larval development mainly lies on growth difference studies, for which, a large amount of size data is necessary. Due to the conventional technique, used for their acquisition, showing numerous limits, the possibilities offered by the image analysis technique have been investigated. This methodological substitution supposed the equivalence of size measured with both techniques, meaning the existence of a homology between anatomical descriptors, length (L(a)) and width (W(a)), and geometrical descriptors L(c) and W(c). Pecten maximus, Patinopecten yessoensis, Ruditapes decussatus and Ruditapes philippinarum showed such a homology. In contrast, Crassostrea gigas did not exhibit a constant orientation of its highest dimension and, consequently, the analysis based on L(c) or W(c) would generate incoherent results with respects to those obtained from anatomical descriptors. Nevertheless, this problem may be circumvented by using the larval area and its equivalent diameter. The complete automatization of the system is discussed. (C) 1998 Elsevier Science B.V. All rights reserved.