PHENOTYPIC AND GENETIC VARIATION IN LAMBS' GROWTH USING LINEAR MODELS

ABSTRACT
Data used in the present study was collected during 1992 to 2005 for two different sheep breeds, Farafra (2559 records) and Saidi (1539 records). The aim was to determine the efficiency of three Linear models [Linear with two parameters (L1), Linear with three parameters (L2) and second order polynomial (Quadratic) models] to the growth curves of Farafra and Saidi lambs by using monthly records of live weight from birth to 540 days of age. Farafra breed was raised in Mallawi research station and Saidi breed was raised in Seds research station belonging to Animal Production Research Institute (APRI). 
The models were evaluated according to coefficient of determination (R2), Akaike's information criterion (AIC) and Schwarz's Bayesian information criterion (BIC). In both breeds, all models fitted the data, with high R2 ranged from 98.7 to 99.5 for males and 98.8 to 99.7 for females. The L1 model gave the best R2 value which was 99.7 in Saidi females, while the Quadratic model gave the lowest R2 value of 98.7 in Farafra males.
Absolute Growth Rate (AGR) obtained from L1, L2 and Quadratic models for Farafra were 0.14, 0.13 and 0.10, respectively, while for Saidi were 0.13, 0.12 and 0.10, respectively in the first month and decreased gradually to reach 0.03, 0.04 and 0.04 for Farafra and 0.03, 0.04 and 0.03 for Saidi at 540 days of age, respectively. All growth curve parameters were significantly influenced by sex, type of birth, age of dam, year and season of birth (P<0.01), except the effect of age of dam on parameters B and C in Quadratic model for Saidi sheep that was not significant (P>0.05). 
Parameters A, B and C were moderate to highly heritable in Farafra lambs (h2=0.24 to 0.74). Genetic correlations between parameters were all positive and ranged from 0.35 to 1.00. In Saidi lambs, parameters had low to moderate heritabilites (0.01 to 0.48). Genetic correlations between parameters ranged from −0.87 to 1.00.
The results of this study suggest that L1 (Y=A+Bt0.5) for growth monitoring can be useful with both breeds under condition of both stations.