Statistical Approach to Finding Probable Feeding Attraetants for Juvenile Yellowtail

Abstract

Ceneral evaluation of the attractive effect through the behavioral exploratoty and feeding observations can be expressed quantitatively by the attraction index gr, which is the product of the coefficients g and r of a logistic curve y=g/{1+exp[-r(j-a)]} in the entering number time-courses. Tis index showed a rough positive linear relation to only crude lipid contents amongst nine general chemical constituent contents in the four kinds of representative food animals during storage at low temperature. This finding was verified statistically by the significance test of correlation between the index gr, coefficient g or r and the contents of each of the nine constituents or the combination of two or three of them. The significance test of multiple and partial regression coefficients in any of the index and two coefficients on the combination of any two factors including the constituents, foodanimal dummy and storage days revealed that sugar and lipid contents (=x_i) are the probable factors having a close relation to the index gr(=y). because multiple regression coefficients (r_v•xixj ) and partial ones (both of r_vxi•xj and r_vxj•xi) of combination to cuttlefish dummy (x_j) were significant. The factor probably having a relation to coeffcients g and r was sought with the significance test of 3-factor combinations chosen by that of 2-factor ones. Amino acids-volatile acids-one of other constituents and amino acids-volatile organic acids-one of other ones combinations showed significant multiple and partial correlation to g and r respectively. These results suggested that amino acids showed one of the most important factors having a close relation to feed attraction.