GENETIC-IMPROVEMENT OF FARMED TILAPIAS - GROWTH-PERFORMANCE IN A COMPLETE DIALLEL CROSS EXPERIMENT WITH 8 STRAINS OF OREOCHROMIS-NILOTICUS

A complete diallel cross experiment was carried out with eight strains of Nile tilapia (Oreochromis niloticus). The strains represented four wild populations collected from various locations in Africa and four populations that had been reproduced over a large number of generation s for tilapia farming in Asia. Growth performance was recorded in a to tal of 23 739 individually tagged progeny of the 64 different strain c ombinations after a grow-out period of about 90 days in seven differen t test environments representing applied farming systems in the Philip pines. Least square means of body weight at harvest were computed for each strain combination within and across test environments, and addit ive genetic effects, maternal (reciprocal) effects and non-additive ge netic effects (heterosis) were estimated. The least square mean hetero sis of both reciprocals of all strain crosses across all test environm ents was 4.3%, and the cross that expressed the largest non-additive g enetic effect showed a least square mean heterosis of 14%. However, on ly seven out of the 22 crosses that expressed a significant heterosis were better performing than the best pure strain, and the largest gain was then about 11%. The least square mean heterosis within test envir onments ranged from 0 to 9.6%, and the largest heterosis was observed in some of the test environments with poor growth performance. Signifi cant reciprocal effects were observed, showing that some trains perfor med better as sire ?,trains and others as dam strains. Significant gen otype (strain combination) by test environment interactions were also detected. The interactions were mainly associated to an expected varia tion in the magnitude of the differences between genotypes that was pr oportional to the mean performance in the test environments. However, some examples of rr-ranking of the genotypes were observed, mainly ass ociated to the non-additive component of performance. It was concluded that specific crossing schemes. possibly involving specialised sire a nd dam lines, may improve growth performance in Nile tilapia. but that this improvement will be quite marginal and should be accompanied by selection within the parent strains. Furthermore, the growth performan ce of such hybrids may be more sensitive to genotype by environment in teractions affecting the non-additive component of performance. A regu lar selection program based on additive genetic performance will norma lly result in a comparable genetic improvement after a short period of repeated selection. and may then continue beyond the results obtained in the present cross experiment without the complications caused by t he laborious dissemination procedures required by a crossbreeding prog ram and the problems caused by a possibly increased environmental sens itivity in the hybrids. (C) 1998 Elsevier Science B.V.