Application of the Self-Organizing Map to the study of skeletal anomalies in aquaculture: The case of dusky grouper (Epinephelus marginatus Lowe, 1834) juveniles reared under different rearing conditions

The setting up of effective rearing protocols for domestication of new candidate species for aquaculture and/or to enhance quality in widely reared species requires the availability of appropriate tools to detect patterns of covariation among rearing parameters and fish quality. In this framework, the pattern of occurrence of skeletal anomalies (SAs) in reared lots represents a proxy for quality, since the presence of SAs is associated with both a general lowering of performance and a negative image of aquaculture products in consumers. In this paper we explore the advantages of using Self-Organized Maps (SOMs) when dealing with the analysis of correlations between the pattern of SAs presence and rearing parameters in dusky grouper (Epinephelus marginatus, Lowe 1834) lots following two different experimental rearing approaches. SOMs were tested because the classic multivariate approach failed to produce meaningful results in the same dataset. A SOM was trained on a dataset containing the mean frequencies of 43 SAs occurring in 20 lots of dusky grouper sampled during three larval rearing cycles carried out in 2001, 2002 and 2004 in Italy, using two rearing approach: Green Water and Large Volume. A series of well-defined patterns were detected in SAs occurrence with respect to body regions. When SOM units were grouped into three clusters, a significant relationship was detected between lot origin (in terms of rearing approach) and SAs occurrence: The Large Volume methodology is to be considered more effective in enhancing the quality of dusky grouper larvae. This finding was independently validated by the superimposition of Hellinger distances obtained from the analysis of meristic counts. Finally, SOM visualized coherent and clear patterns of covariation between SAs occurrence and two crucial aspects or rearing: initial rearing density and final survival rate. We concluded that as a new ordination method SOMs afford effective representations of information gathered from patterns of SAs occurrence in aquaculture lots. Furthermore, SOM appears able to detect subtle but meaningful relationships between quality, as measured by independent descriptors such as SAs and meristic counts, and rearing parameters. It could be useful for quality assessment in both experimental and productive contexts, ultimately helping to reduce the incidence of SAs in aquaculture products and facilitating the identification of more effective approaches to the domestication of new species.