Data preparation and analyses

Data preparation
Data analyses

Data preparation

A number of factors were identified in the preliminary screening of the data at Asela that had a bearing on the analysis techniques to be used.

First was the largely disproportionate representation of the indigenous breed groups in the trial. The majority were Arsi, with smaller numbers of Fogera, Barca and Boran. The latter three breeds belong to the group characterized by the large East African Zebu. Additional crossbred Friesian x Boran heifers had also been purchased from Kenya to increase the number of Boran crosses. Thus it was decided that the Fogera, Barca and Boran animals could best be treated as a single Zebu group larger in body size than the Arsi.

Second, there had been from 1969 to 1973 irregular changes in breeding policy, milking and recording practices at Asela. Hand milking was practiced from 1968 to 1972, then discontinued for the crossbreds but continued for the Arsi breed. Breeding policy had changed from crossbreeding with Jersey to crossbreeding with Friesian. The production of F₂ animals was dropped and straight upgrading was adopted. It was therefore decided to combine various genetic groups into the most appropriate grades of Bos taurus germplasm. This procedure would make it possible to determine environmental and breed group differences.

Breed groups. Data were available on ten breed groups:

1. The Arsi, which are the local dominant type found in the Arsi Region of Ethiopia; they are small in size (200 - 250 kg), and were the major foundation stock used in the crossbreeding scheme.

2. The Zebu, which consisted of three types Barca, Fogera and Boran - native to other provinces of Ethiopia; they are larger framed, with heavier body sizes (300 - 350 kg) than the Arsi.

3. ½ Jersey ½ Arsi crossbreds, which were the products of first crosses between the Jersey (Bos taurus) and the Arsi (Bos indicus) breeds. Crossing between these two breeds was discontinued after the early phases of the crossbreeding scheme.

4. ½ Friesian ½ Arsi crossbreds, which were the products of first crosses between Friesian (Bos taurus) and Arsi (Bos indicus) breeds. Crossing between the two breeds continued as the first step in the upgrading that is still being practiced today in Ethiopia.

5. ½ Friesian ½ Zebu crossbreds, which were the products of first crosses consisting of Friesian x Barca, Friesian x Fogera or Friesian x Boran. The Barca, Fogera and Boran are larger sized and originated in regions other than Arsi.

6. ½ Exotic ½ Arsi grades, which developed as a result of intercrossing between ½ Jersey ½ Arsi females and ½ Friesian ½ Arsi bulls. This group thus carried 25% of Jersey breed germplasm.

7. ¾ Friesian ¼ Arsi grades, which developed as a backcross of ½ Friesian ½ Arsi females to purebred Friesian bulls.

8. ¾ Friesian ¼ Zebu grades, which developed as a backcross of ½ Friesian ½ Barca, ½ Friesian ½ Boran, and ½ Friesian ½ Fogera females to purebred Friesian bulls.

9. ¾ Exotic ¼ Zebu grades, which resulted from upgrading ½ Exotic ½ Arsi cows to purebred Friesian bulls. This breed group thus retained 1/8 of the Jersey breed germplasm and 5/8 of the Friesian.

10. 7/8 Friesian 1/8 local grades, which consisted of a small number of high grade cattle of 7/8 Friesian 1/8 Arsi end 7/8 Friesian 1/8 Zebu that were grouped together.

Seasonal classification. Based on the rainfall records from 1968 to 1977, the months of the year were grouped into five subseasons covering the first and second parts of the dry season, the first and second parts of the long wet season, and the short wet season:

Cow performance traits. Individual records were built up for each cow and each parturition. These gave breed group, number and date of birth (when known) of the cow, and the current parturition date, previous parturition date, lactation milk yield, mean butterfat percentage, drying off date, sex of calf, calf birth weight, cow weight at parturition and next parturition date. From these data, the age at first calving, calving interval, breeding efficiency, lactation length, milk yield per day of lactation, length of dry period, total fat yield, fat-corrected milk yield, annual milk yield per cow, annual fat-corrected milk yield per cow and annual fat-corrected milk yield per unit metabolic weight of cow were additionally computed, as indicated in each appropriate section under the heading Results.

Data analyses

All characters were analysed by least squares procedures (Harvey 1977) using fixed models. Unequal and disproportionate subclass numbers gave unbalanced factorial designs for which conventional analyses of variance techniques were not applicable. Typical models used included the fixed effects of breed group, year of birth or parturition, parity number, and the interaction between breed group and season. The specific factors included in the model used will be evident when the results are presented for each character analysed. The residual mean square was used as the error term to test the significance of all differences evaluated. Linear contrasts of least squares means were computed to determine the significance of differences between groups. More comparisons were made using the least squares means than there are independent degrees of freedom. Therefore, all of the comparisons are not independent, and the error rate over the entire set of comparisons may be different from that indicated by the level of probability. Tests of significance associated with the linear contrasts, although not independent, can be taken as guides as to whether the observed values could have occurred by chance.