Multidimensional analyses in studies on broiler chicken production results

The work presents a simultaneous multi-trait comparison of six commercial lines of broiler chickens. The analysis employed multidimensional statistical methods (comparative analysis, cluster analysis and profile analysis) which made it possible to rank and group objects into homogenous clusters and determine similarities between them. The study of individual commercial lines (Ross 508, Ross PM3, F-15, Flex, Hubbard and Hybro PN) included 12 consecutive hatching cycles at a poultry hatchery and then were continued when the birds were reared on a fattening farm. Comparative analysis demonstrated that no lines were allocated to group 1 (the best group). Group 2 consisted of five lines, of which Flex was the best because it had the highest percentage hatching of fertilised eggs and the chickens gained most weight during fattening. Cluster analysis yielded four groups of lines with common characteristics. The first cluster included Ross and Flex, which indicates that the lines were similar in terms of the internal structure and traits which describe them. Profile analysis demonstrated that Hubbard and Hybro PN were the most similar lines, whereas, Ross and Ross PM3 were the most dissimilar. Moreover, it was difficult to evaluate objects which were described using many traits. To sum up, comparative analysis was more useful for ranking the lines whereas cluster analysis and profile analysis for examining their similarity. Also, an application of multidimensional methods enables processing of statistical information (consisting of many traits) in order to help the poultry farmer make rational and appropriate decisions.