Results

Reproductive performance
Body weight
Lactation milk yield
Butterfat
Dairy productivity
Initial indications of production on smallholder farms

Reproductive performance

Introduction. Reproductive performance is a trait of outstanding importance in dairy cattle enterprises. The size of the calf crop is all-important for herd replacement, and the production of milk depends heavily on reproductive activity. Possible genetic improvement in virtually all traits of economic importance is closely tied to reproductive rate.

Differences in breeding efficiency are largely due to environment, although between breeds heredity also plays a part in the variation of reproductive performance. The best cows are clearly those that have their first calf at an early age and have minimum calving intervals thereafter.

Age at first calving. The mean age at first calving for 524 heifers born on the station was 32.9 ±0.3 months, with a coefficient of variation of 22%.

The analysis of variance shown in Table 1 indicates that breed group, year of birth and breed group x season of birth interactions significantly affected age at first calving.

Table 1. Analysis of variance of age at first calving.

* = P<0.05
** = P<0.01

The least squares mean estimates of age at first calving are shown in Table 2¹. The ¾ Exotic ¼ Arsi grades calved significantly earlier (31.3 months) than all other breed groups, whose individual ages at first calving ranged from 33.6 to 35.7 months. It appears that Arsi cattle, when managed well, can express their genetic potential for early maturity. Mahadevan (1966) observed that irrespective of whether cattle were of Indian, African, European or crossbred origin, their mean age at first calving under a given tropical environment was essentially the same and ranged from 3 to 4 years. The present study seems to indicate that in the subtropical highlands environment, given reasonably good management, age at first calving can be reduced to between 30 and 36 months.

1 Sample means can differ substantially from computed least squares means, as the latter are computed by adjusting for unequal subclass numbers.

Table 2. Estimated least squares means for age at first calving (months).

Within variable groups, row means followed by the same letter do not differ significantly (P<0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

The significant year effects (Table 2) indicate that age at first calving ranged from 27.5 months in 1970 to 43.3 months in 1976. Above-average ages at first calving occurred more during later years, a trend which could be attributed to factors such as reduced availability of supplementary feed and problems associated with obtaining semen.

Calving interval. The calving interval is the period between two consecutive parturitions, and ideally should be in the region of 12 to 13 months. The calving interval is thus closely matched to a yearly production cycle and influences the amount of milk a cow is likely to produce in a given period. The mean calving interval for 1099 records was 459 ±4 days with a coefficient of variation of 28%.

The analysis of variance shown in Table 3 indicates that breed group, year of calving, lactation number and breed group x season interactions significantly affected calving interval.

Table 3. Analysis of variance of calving interval.

*   = P<0.05
** = P<0.01

Estimated least squares means of calving interval are shown in Table 4. The longest calving interval (525 days) occurred among the 7/8 Friesian 1/8 Local breed group, these being the highest grade. The ½ Jersey ½ Arsi (403 days) and the ½ Exotic ½ Arsi (393 days) had significantly shorter calving intervals than the pure Arsi (439 days), both these groups sharing common Jersey germplasm. All crosses with the Zebu were Friesian, and their calving intervals did not significantly differ from those of the pure Zebu. The superiority of Jersey crosses over Friesian crosses with respect to reproductive performance and fitness has often been noted in the literature (Khishin and El-Issawi 1954; Marples and Trail 1966; Kumar 1969). A comprehensive review of the subject (Kiwuwa 1974) concluded that large sized Bos taurus breeds (Friesian, Brown Swiss, Red Dane) and their crosses with Bos indicus manifest longer calving intervals than smaller sized breeds (Jersey, Criollo, Guernsey) or their crosses with the indigenous cattle.

The effect of year of calving (Table 4) suggested significant increases in calving intervals between 1970 and 1978. To determine any linear trend in calving interval over the complete period from 1968 to 1980, the regression of the least squares constants on year of calving (represented as 1 to 13) was calculated. The regression coefficient was 13 days, significant at the 1% level, indicating that calving interval has increased by 13 days per year over the 13-year period.

Parturition number effects indicated decline in calving interval from first to fourth parturition, the mean intervals being 510, 454, 423 and 396 days respectively. Shorter calving intervals at later parturition stages are a function of selective culling against repeat breeder cows and were as expected for a well managed herd.

Breeding efficiency. Calculated as a percentage, breeding efficiency (BE) is a measure based on the regularity of calving and the age at which cows first calve. If an animal calves late for the first time its maintenance cost as a fraction of total cost tends to increase and its life-time production decreases. BE provides for objective comparisons between breeds with respect to their suitability/adaptability for growth and reproduction in a given environment. Its derivation requires a choice of the desirable age at first calving and the desirable length of subsequent calving intervals.

The average age at first calving at Asela was 33 months, ranging from 31 to 36 months among breed groups (Table 2). Previous literature on the subject (Kiwuwa 1974) indicated that age at first calving of Bos taurus or their crosses with Bos indicus throughout the tropics was around 32 months, and the results in Table 2 concur with these observations.

Assuming 13 months (396 days) as the upper limit of an ideal calving interval, and 32 months (960 days) to be the optimal age at first calving, the following formula was used to derive the BE index:

BE = [(N - 1) 396 + 960]/(age in days at each successive calving)

Where: BE is the breeding efficiency coefficient; (N - 1) is the number of calving intervals in N gestations; 396 is the upper limit of the desirable calving interval (days); and 960 is the upper limit of the desirable age at first calving (days). The estimated coefficients were expressed as percentages by multiplying by 100.

The mean BE index for 1269 available calving records was 95 ±0.6 % with a coefficient of variation of 23%. BE in the Asela herd was thus below 100%, indicating that either the breed groups or the standards of herd management were not adequate for optimal growth and regular reproduction.

Table 4. Estimated least squares means for calving interval (days).

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

The analysis of variance in Table 5 demonstrates significant breed, year, season and parity influences on BE.

Table 5. Analysis of variance of breeding efficiency.

*   = P<0.05
** = P<0.01

Estimated least squares means of BE are given in Table 6. Mean BE estimates were lowest (64 - 68) among the indigenous (Zebu, Arsi) breed groups and highest (98 - 107) among the higher ¾ and 7/8 grades. Although there were no significant differences in BE between half-breds and the ¾ to 7/8 higher grades, the half-breds tended to have rather lower BEs than the higher grades.

The significant year effects in Table 6 indicate that between 1971 and 1974 BE at Asela was above 100%, but dropped substantially from 1976 to 1981. To determine the linear trend in BE over the complete period from 1971 to 1981, the regression of the least squares constants on year of calving (represented as 1 to 11) was calculated. The regression was - 6.6%, significant at the 1% level, indicating that BE has decreased by 6.6% per year over the 11-year period.

The significant (P <0.05) seasonal effects indicate that the BE index was higher (94 to 95%) among records starting from January to May than those starting from June to December (Table 6). Parity effects on BE indicated significantly higher (93 - 94%) indices in third and fourth parturitions than in first and second ones (88 - 89%). BE gradually improved from the first to the fourth parturition and above. Improvement in the fertility of older cows following systematic culling at the younger ages could be the key. These results follow closely the decrease in calving intervals from the first to the fourth parturition and above.

Table 6. Estimated least squares means for breeding efficiency coefficient.

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05).

Body weight

Cow postpartum weight. A total of 1310 cow postpartum weight records, taken within 30 days of calving, were available. The mean weight was 276 ±1.4 kg, with a coefficient of variation of 18%.

The analysis of variance shown in Table 7 indicates that breed group, year of calving, season of calving, lactation number and breed x season interactions significantly affected body weight.

Estimates of least squares means of postpartum body weight are given in Table 8. Large and significant breed differences were shown to exist. The other indigenous Zebus were significantly heavier (309 kg) than the Arsi (236 kg). In general higher grades were heavier (306 - 336 kg) than F₁ crossbreds, the exception being the ½ Friesian ½ Zebu cross, which weighed 328 kg, about the same as the ¾ and 7/8 Friesian crosses. The lightest crosses, 269 kg and 278 kg, were the ½ Jersey ½ Arsi and ½ Exotic ½ Arsi breed groups, indicating the influence of both the Arsi and the Jersey.

Table 7. Analysis of variance of cow postpartum weight.

* = P<0.05
** = P<0.01

Year effects on mature body weight (Table 8) indicated significant differences between the early (1969 - 1970) and the later years (1971 - 1980). From 1971 onwards, however, no definite trends could be detected. The season of calving, from March to May, had significantly lower (293 kg) body weights than the others (304 - 305 kg). This season is preceded by dry weather conditions from November to February, and the lower weight during this March - May season is probably due to lower pasture availability during the later stages of pregnancy. Body weights increased by significant amounts from the first (268 kg) to the fourth (331 kg) parturition and above.

Calf birth weight. A total of 1111 birth weights of female calves, recorded within 24 hours of calving, were available. The mean weight was 24.7 ± 0.13 kg, with a coefficient of variation of 18%.

The analysis of variance shown in Table 9 indicates that breed group, year of birth, lactation number of dam, and breed x season interactions significantly affected birth weight.

Estimated least squares means of birth weight are presented in Table 10. Breed differences in birth weight followed similar trends to those in postpartum weight. Heaviest mean birth weights were among the ½ Friesian ½ Zebu, ¾ Friesian ¼ Zebu and 7/8 Friesian 1/8 Local crosses (27.1 - 28.4 kg), while the lightest were among the Arsi (21.5 kg) and the ½ Jersey ½ Arsi crosses (21.9 kg).

Table 8. Estimated least squares means for cow postpartum weight.

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05).

Table 9. Analysis of variance of calf birth weight.

* = P<0.05
** = P<0.01

No clear trends were indicated by year of birth effects on birth weight. Parity effects on birth weights were significant after the second parturition. While there was no significant difference in birth weight between the first and second parturitions, both were significantly lower than those from later parturitions.

Table 10. Estimated least squares means for birth weight of female calves (kg).

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

Lactation milk yield

Introduction. The common procedure in data validation for dairy records is to exclude all those curtailed by sale, slaughter or sickness of the animal. However, temperament effects associated with milk letdown interference among indigenous cattle cause problems concerning the best way to handle such records. Some workers have excluded short lactations due to milk letdown interference; others have argued that since the phenomenon is breed-characteristic, such records should be included in the analysis. Clear definition of milk letdown refusal and short lactation per se needs further study. Kiwuwa et al (1983) have examined the data and the conditions under which certain animals have displayed apparent milk letdown interference at Asela. They concluded that lactations in which peak yields were reached (generally between 21 and 60 days) and cows later voluntarily dried up should be regarded as normal records, if no extraneous factors led to the cessation of milk yield. Thus, in the current comparison between breed groups, all records that exceeded 75 days in lactation length were used in the analyses provided that such records were not later affected by slaughter, sale or sickness.

Analyses of milk production characteristics. Milk production parameters were total lactation yield, lactation length, yield per day of lactation, and dry period. The values covered 1371 milk yield/lactation length records and 1032 dry period records which satisfied the "above 75 days lactation period" criterion. The discrepancy between the number of lactations and the number of dry periods was due to one more record for lactations being available than for dry periods, while animals culled after their first lactation did not have a dry period recorded.

The analyses of variance for total lactation milk yield, lactation length, milk yield per day of lactation, and dry period are given in Table 11.

Estimated least squares means of total lactation milk yield, lactation length, milk yield per day of lactation, and dry period are indicated in Table 12.

Total lactation milk yield. The mean total lactation milk yield was 1775 ±26 kg with a coefficient of variation of 55%. The analysis of variance in Table 11 indicates that breed group, year of calving and parturition number significantly affected lactation milk yield.

Estimated least squares means of lactation milk yield are shown in Table 12. Lactation milk yields of the indigenous breed groups (Arsi and Zebu) were not significantly different. Neither were those of the ½ Jersey ½ Arsi, ½ Friesian ½ Arsi and ½ Exotic ½ Arsi groups. Within the higher grades, the ¾ Friesian ¼ Arsi, ¾ Friesian ¼ Zebu, ¾ Exotic ¼ Arsi and 7/8 Friesian 1/8 Local were not significantly different and produced at virtually the same level as the ½ Friesian ½ Zebu breed group. The F₁ crosses produced significantly more milk (123%) than the indigenous Arsi and Zebu, and the higher grades outyielded the equivalent F₁ crosses (+ 19%) with the exception of the ½ Friesian ½ Zebu. Crossbreeding the indigenous cattle with the Bos taurus breeds had more than doubled milk yields of the F₁ generation, and further upgrading to ¾ Exotic had almost tripled the yields.

The significant year effects (Table 12) indicated that lactations in the period 1976- 1979 outyielded those of other years. Yields in second and subsequent lactations were at least 9% higher than the first, with the mean increase being 11 %.

Lactation length. The mean lactation length was 364 ±4 days, with a coefficient of variation of 39%. The analysis of variance in Table 11 indicates that breed group, year of calving and parturition number significantly affected lactation length.

Table 11. Analyses of variance of total lactation milk yield, lactation length, milk yield per day of lactation and dry period.

* = P<0.05
** = P<0.01

Table 12. Estimated least squares means for total lactation milk yield, lactation length, milk yield per day of lactation and dry period.

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

Estimated least squares means of lactation length are shown in Table 12. The overall mean of 350 days was longer than the 305 days generally accepted as the ideal period. The shortest lactation lengths were among the Arsi (272 days), Zebu (303 days), ½ Exotic ½ Arsi (282 days) and ½ Jersey  ½ Arsi (334 days) breed groups, which did not differ significantly from each other. Friesian grades had lactation lengths of 378 days in both ½ Friesian ½ Zebu and ¾ Friesian ¼ Zebu, and 411 days in the 7/8 Friesian 1/8 Local grade. Thus all breed groups with Friesian genes tended to have longer lactation lengths.

There were no clear trends in year effects on lactation length. Table 12 indicates, however, that lactations in the period 1974 - 1980 were longer than those between 1968 and 1973.

The significant parity effects on lactation length show a definite reduction from the first lactation (406 days) to the fourth (305 days) lactation and above. This trend could be the result of a gradual improvement in the fertility of individual cows together with the culling of repeat breeder animals.

Milk yield per day of lactation. The mean milk yield per day of lactation was 5.0 ±0.06 kg, with a coefficient of variation of 44%. The analysis of variance in Table 11 indicates that breed group, year of calving, parturition number and breed x season of calving significantly affected milk yield per day of lactation.

Estimated least squares means of milk yield per day of lactation are shown in Table 12. The breed differences followed the same pattern as for total lactation milk yield, except that whereas the F₁ crosses produced significantly more milk (107%) than the indigenous Arsi and Zebu, the higher grades did not significantly outyield the equivalent F₁ crosses.

From parturition numbers one to three, yield increased significantly at each stage.

Dry period. The mean dry period was 97 ±2 days, with a coefficient of variation of 78%. The analysis of variance in Table 11 indicates that breed group, year of calving and season of calving significantly affected dry period.

Estimated least squares means of dry period are shown in Table 12. Dry periods were significantly shorter in all crosses (70 - 108 days) than in the indigenous Arsi and Zebu (154- 165 days) breed groups. Year effects indicated shorter dry periods from 1969 to 1975 than from 1975 to 1979. An exception seems to have been 1968, with a mean dry period of 164 days, possibly due to management problems during the initial stages of the crossbreeding scheme.

Significant season effects on dry period indicate that cows calving from March to May (short rains) experienced a longer dry period (115 days) than those calving during other seasons (87 to 108 days). Cows calving from March to May face increasingly dry and hot weather during the last third of their lactation (November to February), and as a result climatic and nutritional stress may bring about an early curtailment of their lactation. Although not significant, a longer dry period (108 days) was found after the first lactation than after later lactations (95 - 98 days).

Under the conditions that prevailed at Asela station, initial crossbreeding between the Arsi or Zebu (Bos indicus) and the Friesian or Jersey (Bos taurus) cattle had at least doubled, and a further generation of upgrading tripled, milk yields over those of the indigenous cattle. Although the best F₁ cross was the Friesian x Zebu, the F₁ Friesian x Arsi did not significantly out-yield the F₁ Jersey x Arsi. Upgrading to ¾ and 7/8 Friesian inheritance had produced significant increases in milk yield, but lactation periods were longer. The Friesian x Zebu cross did not show any difference in milk yield between the F₁ and the ¾ Friesian.

Butterfat

Introduction. Knowledge of the fat content of milk and total fat yield of dairy cattle is vital where the dairy industry is involved in the production of milk byproducts such as butter, ghee, cream and cheese. The standard fat-corrected milk (FCM) method permits comparisons on a common energy basis per unit of milk produced. In the current study, actual milk yields were adjusted to a 4% butterfat standard.

Analyses of butterfat characteristics. The parameters analysed were butterfat percentage, total fat, total fat-corrected milk yield and fat-corrected milk yield per day of lactation. A total of 768 records were available. The analyses of variance for the four measures are shown in Table 13.

Table 13. Analyses of variance of butterfat percentage, total fat, total fat-corrected milk yield and fat-corrected milk yield per day of lactation.

* = P<0.05
** = P<0.01

Estimated least squares means of butterfat percentage, total fat, total fat-corrected milk yield and fat-corrected milk yield per day of lactation are indicated in Table 14.

Butterfat percentage. The mean butterfat percentage was 4.8 ±0.03, with a coefficient of variation of 15%. The analysis of variance in Table 13 indicates that breed group and year of calving significantly affected butterfat percentage. Breed differences in butterfat percentage (Table 14) indicate that the local Arsi, Zebu F₁ Jersey x Arsi breed groups recorded significantly higher (5.1 to 5.5%) percentages than the F₁ crossbreds and higher grades of the Friesian (4.1 to 4.5%). Thus crossing the indigenous cattle with the Jersey breed maintains butterfat percentage at levels similar to those of the indigenous breeds, but doubles total milk-yields. In remote rural areas, where the marketing of fresh milk is often a problem, the use of Jersey crosses for production of butter or ghee could well be desirable. Year trends in butterfat percentage (Table 14) suggested a gradual decline from 1969 to 1977.

Total fat and fat-corrected milk yield. The mean total fat yield was 72 ±1.4 kg, with a coefficient of variation of 53%, and the mean fat-corrected milk yield was 1724 ±33 kg, with a coefficient of variation of 54%. The analysis of variance in Table 13 indicates that breed group, year of calving and to a lesser degree season of calving significantly affected both traits. The ½ Friesian ½ Zebu breed group had a significantly higher fat yield (98 kg) and fat-corrected milk yield (236.7 kg) than all others except the ¾ Friesian ¼ Zebu. The local Arsi and Zebu had significantly lower fat yields (35 and 41 kg) and fat-corrected milk yields (805 and 922 kg) than all others. Seasonal influences indicated that cows calving during March to May yielded significantly lower fat yields (65 kg) and fat-corrected milk yields (1559 kg) than those calving during June to December.

Table 14. Estimated least squares means for butterfat percentage, total fat, total fat-corrected milk yield and fat-corrected milk yield per day of lactation.

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

Fat-corrected milk yield per day of lactation. The mean fat-corrected milk yield per day of lactation was 4.8 ±0.08 kg, with a coefficient of variation of 44%. The analysis of variance in Table 13 indicates that breed group, year of calving and parturition number significantly affected fat-corrected milk per day of lactation. All crossbreds had significantly higher yields than the two indigenous breeds (Table 14), but there were no differences between ¾-breds and ½-breds. There were significant increases in yield from first to third parturitions.

Dairy productivity

Introduction. Varying milk output over different lactation lengths and calving intervals makes it difficult to compare animal performances directly using the individual traits of lactation milk yield, lactation length, dry period and calving interval. As a first step it is more valuable to express yields over a regular annual cycle, amalgamating the milk production and reproductive performance traits. Milk from different breeds of cattle contains varying proportions of fat. Comparisons between breeds, taking into account the energy value of the milk produced, are thus even more valid.

Finally, maximum returns from a dairy operation depend on the use of animals with high milk output relative to maintenance cost over the annual cycle. The milk-output to feed-input ratio can be directly measured through regularly recording the feed given to individual animals, but not many farmers can keep such records. Thus when breeds are of different body weights, it is more appropriate to express milk yields in terms of a measure of body weight.

In this section, three progressive measures of productivity are constructed and analysed: annual milk yield per cow, combining reproductive performance and milk production; annual fat-corrected milk yield per cow, combining reproductive performance, milk yield and milk quality; and annual fat-corrected milk yield per unit metabolic weight of cow, combining reproductive performance, milk yield, milk quality and an estimate of maintenance cost.

Annual milk yield per cow. Annual milk yield per cow was calculated as total lactation milk yield ÷ calving interval (days) x 365. The mean annual yield for 1024 records was 1474 ±21 kg, with a coefficient of variation of 46%.

The analysis of variance shown in Table 15 indicates that breed group, year of calving, parturition number and breed group x season interactions significantly affected annual milk yield per cow.

Table 15. Analysis of variance of annual milk yield per cow.

* = P<0.05
** = P<0.01

Estimated least squares means of annual milk yield per cow are shown in Table 16. The ¾ grades achieved higher annual yields than the ½-breads, which in turn had higher yields than the indigenous breds. There was no significant difference between the Arsi and Zebu, while the 7/8 Friesian was not superior to the ¾ grades.

Parturition number effects indicated that annual yields increased significantly at each parturition from first to fourth and over.

Annual fat-corrected milk yield per cow. Annual fat-corrected milk yield per cow was calculated as total lactation milk yield adjusted to a 4% butterfat standard ÷ calving interval (days) x 365. The mean annual yield for 558 records was 1474 ±29 kg, with a coefficient of variation of 47%.

The analysis of variance shown in Table 17 indicates that breed group, year of calving, season of calving and parturition number significantly affected annual fat-corrected milk yield per cow.

Estimated least squares means of annual fat-corrected milk yield per cow are shown in Table 18. All crosses had significantly higher yields than the indigenous groups but, in contrast to annual milk yield, the annual fat-corrected milk yield of ¾ grades was not superior to that of ½-breds.

Table 16. Estimated least squares means for annual milk yield per cow (kg).

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

Parturition number effects again indicated that annual fat-corrected milk yield increased significantly at each parturition from first to fourth and over.

Annual fat-corrected milk yield per unit metabolic weight of cow. Annual fat-corrected milk yield per unit metabolic weight of cow was calculated as total lactation milk yield adjusted to a 4% butterfat standard ÷ calving interval (days) x 365 metabolic weight of dam (kg^.). The mean annual yield for 524 records was 20.8 ±0.4 kg, with a coefficient of variation of 44%.

Table 17. Analysis of variance of annual fat-corrected milk yield per cow.

* = P<0.05
** = P<0.01

Table 18. Estimated least squares means for annual fat-corrected milk yield per cow (kg).

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05).

The analysis of variance shown in Table 19 indicates that breed type, year of calving and season of calving significantly affected annual fat-corrected milk yield per unit metabolic weight of cow.

Table 19. Analysis of variance of annual fat-corrected milk yield per unit metabolic weight of cow.

* = P<0.05
** = P<0.0

Estimated least squares means of annual fat-corrected milk yield per unit metabolic weight of cow are shown in Table 20. Again, all crosses had significantly higher yields than indigenous cattle, but there was no suggestion that ¾ grades were superior to ½-breds. The breed groups with the highest dairy productivity were the ½ Jersey ½ Arsi, followed by the ¾ Exotic ¼ Arsi. Though not significantly different from some other breed groups, the trend seems to testify to the superiority of the Jersey genotype in overall dairy productivity.

Season effects on dairy productivity (Table 20) were significant. The seasons with the highest estimates were June-July and August-September. Both these rainy periods gave significantly higher dairy productivity estimates than the dry season (January - February) and the short rainy season (March - May).

Contrary to the findings with the previous two indices, parturition number had no significant effect when metabolic weight was brought in as a component of the productivity index.

Initial indications of production on smallholder farms

Introduction. Heifers of the ½ Jersey ½ Arsi, ½ Friesian ½ Arsi, ½ Friesian ½ Zebu and ¾ Friesian ¼ Zebu crossbreds were sold to private farmers, the farmer generally being allowed to purchase only one animal. Farmers were required to demonstrate their willingness to allocate at least a hectare for pasture, and plant an equal area with fodder beet as a supplementary dry-season feed.

They were also required to cooperate in record keeping under the guidance of an extension agent who periodically measured milk yield and associated performance traits. Farmers had access to concentrates, AI services and veterinary care.

Table 20. Estimated least squares means for annual fat-corrected milk yield per unit metabolic weight of cow (kg).

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.

Data used in the present analysis were obtained from 124 individual farmers in Sagure, South Asela, Dera, Lemu, Gobe and Ticho locations in the Chilalo District of Arsi Region (see Figure 1). The information was recorded between 1969 and 1975, after which land reform changes promoted community rather than individual development. Management techniques under producer cooperatives or farmers' associations differ from those of individual farmers; thus it seemed logical for dairy production activities after 1975 to be studied separately at a future date.

After excluding data that were incomplete due to sale, slaughter, sickness or interrupted recording schedules, a total of 232 records of milk yield and 103 records of reproductive performance were available for analysis. The majority of the farms maintained only a single animal as a result of the heifer sales policy outlined above. This meant that individual cow differences were directly confounded with herd differences, and the assumption had to be made that the basic one cow one farmer policy resulted in random allocation of cows over herds and that management differences among farmers occurred at random on farms with one or two cows.

Means and variations of lactation milk yield, lactation length, milk yield per day of lactation, dry period, calving interval and annual milk yield are indicated in Table 21.

The analyses of variance for the six traits are shown in Table 22 and the estimated least squares means are given in Table 23.

Milk yield, lactation length and milk yield per day. Breed group had a significant influence on all three traits, year of calving influenced lactation length and daily milk yield, while parturition number influenced both lactation and daily milk yields.

There were significant differences in total lactation milk yield between all four of the breed groups sold to farmers. The ¾ Friesian ¼ Zebu grades outyielded all others. The ½ Friesian ½ Zebu was superior to the ½ Friesian ½ Arsi, which in turn was superior to the ½ Jersey ½ Arsi. Significant breed differences in lactation length indicated that the ½ Jersey ½ Arsi cross had the shortest lactation, while the ¾ Friesian ¼ Zebu grade had the longest. Lactation lengths of ½ Friesian ½ Arsi and ½ Friesian ½ Zebu were not significantly different from each other. In terms of daily milk yield estimates, the ½ Jersey ½ Arsi and ½ Friesian ½ Arsi were not significantly different. Similarly, the ½ Friesian ½ Zebu cross produced as much as the ¾ Friesian ¼ Zebu grade. The results in Table 23 indicated that, at the F₁ crossbreeding level, the Friesian crossed with the Zebu gave significantly better milk yield performance than when crossed with the Arsi.

Year effects on lactation length and daily milk yield indicated shorter lactation lengths and higher daily milk yields in 1974 than in all other years. Parturition number effects on lactation and daily milk yield were not significant after the first parturition.

Dry period, calving interval and annual milk yield. Breed group, year and season of calving and parturition number had no significant effect on dry period or calving interval, but breed group and parturition number affected annual milk yield. The ½ Friesian ½ Zebu and ¾ Friesian ¼ Zebu crosses had significantly higher annual milk yields than the ½ Friesian ½ Arsi and ½ Jersey ½ Arsi crosses. First parturition annual yields were significantly lower than second and third.

Table 21. Means and variations of performance traits of crossbred dairy cattle on smallholder farms.

Table 22. Analyses of variance of total lactation milk yield, lactation length, milk yield per day of lactation, dry period, calving interval and annual milk yield per cow.

** = P<0.01

Table 23. Estimated least squares means for total lactation milk yield, lactation length, milk yield per day of lactation, dry period, calving interval and annual milk yield per cow yield per cow.

Within variable groups, row means followed by the same letter do not differ significantly (P < 0.05). If no letter is used it indicates the variable group did not show a significant difference in the analysis of variance.