ABSTRACT
The current research involved pure and reciprocal crosses of two African catfishes: Clarias gariepinus (Cl) and Heterobranchus longifilis (Ht) to give four groups namely Cl×Cl, Cl×Ht, Ht×Cl and Ht×Ht. Triplicate groups of progenies were reared for 56 days and length (total length) were taken weekly. Mortalities were recorded and growth shooters sorted. Data was corrected for missing values (mortality) using imputation algorithm. Data passed the Kaiser-Meyer-Olkin factor adequacy test (Measure of Sampling Adequacy (MSA)>0.8) and the Bartlett's test of sphericity (p<0.05) prior to principal component analysis, linear discriminant analysis and hierarchical clustering. For all samples, length between weeks 2 and 3 was 42.2% of variation –(dimension 1) which we call archikotic growth phase and from weeks 7 to 8 accounted for 11.1% (dimension 2) which we call telikotic growth phase are grouping factors. For non-growth shooters, length from weeks 1 to 5 (83.5% - dimension 1) named endiametic growth phase and from week 7 - 8 (6.3% - dimension 2) named as telikotic growth phase are the grouping factors. Growth shooters can be grouped based on length between week 2 and 6 (75% variation in dimension 1 - endiametic growth phase) and week 1 and 8 (10.2% variation in dimension 2 - architelic growth phase). Accuracy in the identification of progeny according to their family declined with uniformity in size. Grouping of the crosses also showed that there was a maternal influence on the clustering such that Cl×Cl and Cl×Ht clustered together as did Ht×Ht and Ht×Cl.