Publications

A deterministic approach was used to compare breeding strategies utilizing local and imported semen for improvement of dairy cattle in Kenya. The local selection programs considered were a closed progeny testing scheme (CPT) and a young bull system progeny of local bulls (PLB). The continuous semen importation (CSI) and young bull system progeny of imported bulls (PIB) were the strategies based on imports. The breeding strategies were compared on the basis of selection response. The effect of genetic correlation (rg) and initial differences in genetic merit between the two populations was also examined. The CSI and PIB strategies ranked above CPT by, correspondingly, 35.1{%} and 25.0{%} when the rg between Kenya and the USA was assumed to be one. At a rg of 0.58, imports from countries which are 1.50 standard deviation (SD) above the local population were superior to CPT but equal to PLB. This means that if CSI is to be used at a rg of 0.58, then only semen from countries which are >1.50SD should be imported. At a rg of 0.70 and with an initial genetic difference of 1.25SD, the economic responses in CSI and PLB were equal. This indicates that semen importation is only justified when genotype–environment interaction is >0.70. It is concluded that wholesale importation of semen as a means of achieving sustainable genetic progress in Kenya is a less optimal solution and there is the need for an effective local selection program.

A deterministic approach was used to compare breeding strategies utilizing local and imported semen for improvement of dairy cattle in Kenya. The local selection programs considered were a closed progeny testing scheme (CPT) and a young bull system progeny of local bulls (PLB). The continuous semen importation (CSI) and young bull system progeny of imported bulls (PIB) were the strategies based on imports. The breeding strategies were compared on the basis of selection response. The effect of genetic correlation (rg) and initial differences in genetic merit between the two populations was also examined. The CSI and PIB strategies ranked above CPT by, correspondingly, 35.1{%} and 25.0{%} when the rg between Kenya and the USA was assumed to be one. At a rg of 0.58, imports from countries which are 1.50 standard deviation (SD) above the local population were superior to CPT but equal to PLB. This means that if CSI is to be used at a rg of 0.58, then only semen from countries which are >1.50SD should be imported. At a rg of 0.70 and with an initial genetic difference of 1.25SD, the economic responses in CSI and PLB were equal. This indicates that semen importation is only justified when genotype–environment interaction is >0.70. It is concluded that wholesale importation of semen as a means of achieving sustainable genetic progress in Kenya is a less optimal solution and there is the need for an effective local selection program.

Data consisting of 8922 weight records from 2642 Dorper sheep from 20 to 380 days of age spanning 1981–1996 were used to model the growth trajectory using random regression (RR) and estimate genetic parameters. The independent variables were orthogonal (Legendre) polynomials of age at weighing and orders of fit up to 3 were considered. Analyses were carried out fitting sets of RR coefficients due to direct and maternal additive genetic and environmental effects, and heterogeneous error variances. Direct additive genetic variance increased steadily along the trajectory from 0.9 at 20 days of age to 2.44 at 380 days of age. Maternal additive genetic variance increased steadily along the trajectory from 0.13 to 2.92. Direct permanent environmental effect variances increased steadily from a value of 0.66 at 20 days to 2.13 at 380 days of age. Maternal permanent environmental variances declined from 0.87 at 20 days of age to 0.01 at 200 days of age followed by an increase to 0.87 at 380 days. Direct genetic heritability estimates declined sharply from 0.13 at 20 days to 0.07 at 80 days of age then fluctuated thereafter. Maternal genetic heritability estimates increased along the trajectory up to 320 days of age after which it declined slightly. The genetic and phenotypic correlations decreased with increasing lag in time between the weights. The Dorper sheep growth was adequately modeled using RR models despite the limited data. This study has demonstrated the possibility of application of RR models for routine genetic evaluation of Dorper sheep in Kenya.

Data consisting of 8922 weight records from 2642 Dorper sheep from 20 to 380 days of age spanning 1981–1996 were used to model the growth trajectory using random regression (RR) and estimate genetic parameters. The independent variables were orthogonal (Legendre) polynomials of age at weighing and orders of fit up to 3 were considered. Analyses were carried out fitting sets of RR coefficients due to direct and maternal additive genetic and environmental effects, and heterogeneous error variances. Direct additive genetic variance increased steadily along the trajectory from 0.9 at 20 days of age to 2.44 at 380 days of age. Maternal additive genetic variance increased steadily along the trajectory from 0.13 to 2.92. Direct permanent environmental effect variances increased steadily from a value of 0.66 at 20 days to 2.13 at 380 days of age. Maternal permanent environmental variances declined from 0.87 at 20 days of age to 0.01 at 200 days of age followed by an increase to 0.87 at 380 days. Direct genetic heritability estimates declined sharply from 0.13 at 20 days to 0.07 at 80 days of age then fluctuated thereafter. Maternal genetic heritability estimates increased along the trajectory up to 320 days of age after which it declined slightly. The genetic and phenotypic correlations decreased with increasing lag in time between the weights. The Dorper sheep growth was adequately modeled using RR models despite the limited data. This study has demonstrated the possibility of application of RR models for routine genetic evaluation of Dorper sheep in Kenya.

Data on animal identification systems and selection criteria for sheep and goats were collected from the Rendille and Gabra communities in northern Kenya. These were then analysed through computation of indices, which represented a weighted average of all rankings of a particular trait or identification system. The three most important records kept were castration (indexþinspace}=þinspace}0.224), dates of birth (0.188) and entries into the flock (0.185). Identification was done through ear notching (0.409), branding (0.248), and coat colour of the animals (0.150). Characteristics with index ≥0.200 were considered more important and included big body size (Rendille, 0.260; Gabra, 0.251) and milk yield (Rendille, 0.206) for the buck's dam. Big body size (Rendille, 0.264; Gabra, 0.245) and offspring quality (Rendille, 0.252; Gabra, 0.265) were considered important attributes for the buck's sire. Important qualities for the ram's dam were big body size (Rendille, 0.246; Gabra, 0.216), offspring quality (Rendille, 0.200; Gabra, 0.235), fat deposition (0.233) among the Rendille and drought tolerance (0.246) among the Gabra. For the rams' sire, big body size (Rendille, 0.235; Gabra, 0.233), offspring quality (Rendille, 0.200; Gabra, 0.235) and fat deposition (Rendille, 0.203; Gabra, 0.220) were considered important. The results from this study imply that pedigree and performance recording have been practiced through own intricate knowledge and that pastoralists have deliberate selection criteria. This information is the cornerstone in the establishment of appropriate breeding programmes in the slowly changing pastoral systems.

{Data on animal identification systems and selection criteria for sheep and goats were collected from the Rendille and Gabra communities in northern Kenya. These were then analysed through computation of indices, which represented a weighted average of all rankings of a particular trait or identification system. The three most important records kept were castration (indexþinspace}=þinspace