c o l o n - %f\SVERIGES l®if LANTBRUKSUNIVERSITET
Genetic Analysis of Boran, Friesian and Crossbred Cattle in Ethiopia
Mekonnen Haile-Mariam
Dissertation Uppsala 1994 o o i o ' i g 3\ SVERIGES LANTBRUKSUNIVERSITET
Genetic Analysis of Boran, Friesian and Crossbred Cattle in Ethiopia
Mekonnen Haile-Mariam
Institutionen for husdjursfdradling Rapport 113 och sjukdomsgenetik Publication No. 113
Swedish University of Agricultural Sciences Uppsala 1994 Department of Animal Breeding ISSN 0347-9706 and Genetics isrn SLU-HFS-R--113-SE Swedish University of Agricultural Sciences De sartment of Animal Breeding and Genetics Box 7023 S-7a50 07 UPPSALA
ISI N 91-576-4893-X
Printed at SLU Info/Repro, Uppsala 1994 ABSTRACT
This thesis, comprising six papers based on Ethiopian data, deals with the evaluation of Boran cattle and their crosses regarding reproductive, survival, and growth traits, and of dairy cattle of Friesian origin and their crosses with Boran regarding lactation milk yield, lactation length, calving interval (Cl) and abortion rate. In addition, estimates of genetic parameters and genetic and environmental trends for growth, milk yield and reproductive traits are included.
The results from the analysis of data collected between 1977 and 1985 from the Abemossa Cattle breeding and improvement ranch of the Ministry of Agriculture were: the weight of Boran, F1 and 3/4 Friesian calves were similar at birth but at weaning the F7 and 3/4 Friesian calves were heavier than the Boran by 19 and 22 kg, respectively. The significance of the interaction effect between breed and season of birth showed that Boran calves maintained uniform growth rate regardless of their birth season, while the highest fluctuation in weaning weight was observed in the 3/4 crossbred calves. The weights of Boran cattle at 1, 2, and 3 years of age were 179, 269 and 338 kg, respectively. The pre-weaning calf mortality rates of Boran and F-, calves were 4.0 and 2.9%, respectively. The comparison of Boran, FT Boran-Friesian and 3/4 Friesian heifers showed that the F^ required relatively fewer services per conception and also calved at a younger age than the other two breed groups. The mean CIs were 465, 552, 525 and 487 days for Boran cows naturally mated with Boran and Friesian bulls, and Boran and F1 cows artificially inseminated with Friesian semen, respectively.
Heritability estimates based on 24 years of data of naturally bred Boran cattle from the same ranch, ranged from 0.062 to 0.075 and from 0.037 to 0.043 for age at first calving (AFC) and Cl, respectively. The c^-value (the ratio of permanent environmental variance to total) for Cl ranged between 0.028 and 0.031. Estimates of genetic and environmental correlation between AFC and first Cl were -0.054 and -0.176, respectively. The h2-value of Cl among young cows was low (0.002 to 0.015) compared with that in mature cows (0.093). For birth weight (BW) a direct heritability (h/ ) of 0.24 and a maternal heritability (hm2) of 0.08 were estimated. In the case of weaning weight (WW) and yearling weight (YW) the estimates were 0.29 and 0.34 for hd2 and 0.06 and 0.05 for hm2, respectively. The c2-values were 0.14 and 0.05 for WW and YW, respectively. Estimates from bi- and tri-variate analyses were similar to those from univariate analysis but the estimates for hd2 of YW were relatively higher. The across trait correlation estimates were low to medium with the exception of high direct genetic and permanent environmental correlations between WW and YW. The genetic antagonism between direct and maternal effects was also strong (-0.33 to -0.68).
The difference between the minimum and maximum direct breeding values was more than 50 kg for both WW and YW, while the variation as well as the annual changes in estimated direct and maternal breeding value for BW were near-zero. The annual changes in direct and maternal breeding value were 0.32 and 0.02 kg I for WW and 0.22 and -0.005 kg for YW, respectively. Whereas the maternal environmental trends were weak, the variation in the direct environmental effects was the highest. Moreover, the mean generation interval was calculated to be 6.7 years, which means that on the average 3.3 generations of selection occurred in a 24-year selection period. During this time span, inbreeding increased to an average of 1.7% in animals bom in 1985.
The analysis of data on lactation milk yield, lactation length, Cl and abortion (including stillbirth) rate, collected from dairy cattle of Friesian origin and their Fj, 3/4 and 7/8 crosses with Boran at a University farm, showed that the overall unadjusted mean lactation milk yield, lactation length, Cl and abortion rate were 4,058 kg, 324 days, 437 days and 7.3%, respectively. As a deviation from Friesians, the Fj crosses had -2365 kg, -75 days, and -29 days in lactation yield, lactation length and Cl, respectively. The abortion rate was relatively low (5.5%) in the FI crosses compared with the overall least-squares mean of 8.3%. Heritability and c2- values were 0.08 and 0.24 for lactation yield, 0.0 and 0.20 for lactation length and 0.0 and 0.02 for Cl, respectively.
The effects of environmental factors in particular year effects were highly sign ficant for all traits. This was related mainly to shortage of feed during years of drought. In the Boran cattle breeding scheme, genetic gain in aggregate breeding value was 0.2% of the mean WW (169 kg) per year. This low response desjiite the good opportunity for selection was mainly due to selection on the basis of phenotypic performance, colour and conformation and also due to the high! genetic antagonism between maternal and direct genetic effects. Finally, it is recommended that improvement through breeding of indigenous as well as of crossbred or grade cattle in Ethiopia should be based on nucleus breeding systems in selected areas. Due to the extreme variations in cattle production systems and access to services, breed improvement programmes have to be flexible and carefully planned. CONTENTS Page
PUBLICATIONS INCLUDED IN THE THESIS 1
INTRODUCTION 2
BACKGROUND INFORMATION ON CATTLE PRODUCTION 4 I. Cattle Production Systems 4 II. Cattle Breeding Activities 5 III. Major Constraints 7
SUMMARY OF THE PAPERS INCLUDED 8 I. Productivity of Boran cattle and their Friesian crosses at Abemossa 8 Ranch, Rift Valley of Ethiopia. I. Reproductive performance and pre-weaning mortality. II. Productivity of Boran cattle and their Friesian crosses at Abemossa 10 Ranch, Rift Valley of Ethiopia. II. Growth performance. III. Estimates of direct and maternal (co)variance components of growth 11 traits in Boran cattle. IV. Estimates of genetic and environmental trends of growth traits in 12 Boran cattle. V. Genetic and environmental effects on age at first calving and calving 13 interval of naturally bred Boran (zebu) cows in Ethiopia. VI. Genetic and environmental effects on performance of dairy cattle 14 of Friesian origin and their Boran (zebu) crosses at Alemaya, Ethiopia.
GENERAL DISCUSSION 16 I. Improving Beef Production 16 1. Introduction 16 2. Planning a selection scheme 17 3. Crossbreeding for beef or for dairy and beef as a by-product 23 II. Improving Milk Production 24 1. Introduction 24 2. Production of crossbred heifers 25 3. Improving crossbred or grade cattle populations 26
CONCLUDING REMARKS 29
ACKNOWLEDGEMENTS 31
REFERENCES 32
SUPPLEMENT: Publications I-VI
I PUBLICATIONS INCLUDED IN THE THESIS
The present thesis is based on the following publications, which will be referred to by roman numerals I to VI.
I. Haile-Mariam, M., Banjaw, K., Gebre-Meskel, T. and Ketema, H. 1993. Productivity of Boran cattle and their Friesian crosses at Abemossa Ranch, Rift Valley of Ethiopia. I. Reproductive performance and pre-weaning mortality. Trop. Anim. Hlth. Prod. 25:239-248.
II. Banjaw, K. and Haile-Mariam, M. 1994. Productivity of Boran cattle and their Friesian crosses at Abemossa Ranch, Rift Valley of Ethiopia. II. Growth performance. Trop. Anim. Hlth. Prod. 26:49-57.
III. Haile-Mariam, M. and Kassa-Mersha, H. 1994. Estimates of direct and maternal (co)variance components of growth traits in Boran cattle. Accepted for publ. J. Anim. Breed, and Genet.
IV. Haile-Mariam, M. and Philipsson, J. 1994. Estimates of genetic and environmental trends of growth traits in Boran cattle. To be submitted.
V. Haile-Mariam, M. and Kassa-Mersha, H. 1994. Genetic and environmental effects on age at first calving and calving interval of naturally bred Boran (zebu) cows in Ethiopia. Anim. Prod. 58:329-334.
VI. Haile-Mariam, M. 1994. Genetic and environmental effects on performance of dairy cattle of Friesian origin and their Boran (zebu) crosses at Alemaya, Ethiopia. To be submitted.
1 INTRODUCTION
Ethiopia, with its about 30 million head ranks first in Africa and eighth in the world with regard to the size of its cattle population. For the 3.7 million cows milked, the average yield is estimated to be about 206 kg per year. The average cattle off-take is 7.4% per annum with an average dressed carcass weight of 110 kg (FAO, 1992). Though their productivity in terms of milk and beef are low, cattle play a significant role in the economy of the country by providing hides, draught power and manure for fuel and fertilizer. In Ethiopia unfavourable climatic conditions and the large cattle population contribute to scarcity of feed, overgrazing, frequent drought and high incidence of animal diseases, etc. In addition, socio-economic factors including lack of infrastructure and organisational problems limit productivity and the provision of services and marketing as well as transfer of technology. Simultaneously there is lack of appropriate technology to improve animal nutrition, health, breeding and management. Research and development work in these areas, particularly to assist small traditional farmers, is inadequate.
Generally, breed improvement efforts in Ethiopia have so far been limited to crossing Zebu with temperate cattle mainly for dairy production (Kebede, 1992; Kiwliwa et al, 1983). Studies on indigenous cattle have mainly taken place in comjection with crossbreeding programmes (IAR, 1972; Schaar et al., 1981). The main reason for neglecting the improvement of indigenous cattle is the opinion thatithey are poor milk producers and that their improvement through selection will itake a long time and hence they should be used for crossbreeding purposes only.
However, there are several reasons that justify indigenous cattle improvement programmes: First, studies in Ethiopia (Kebede, 1992) and elsewhere (Madalena et ai, 1990) have shown that the productivity of Ft crosses between Zebu and tern aerate breeds is better in a stressful environment than that of crosses with a higl er level of temperate inheritance. This suggests that if either continuous Ft production or rotational crossbreeding between zebu and temperate cattle is to be used as a breeding strategy, selection within the indigenous cattle would increase productivity in the crosses still further. Secondly, the present indifference to the indigenous cattle does not consider the fact that 'improved' zebu bulls could be used on grade crossbred cows in order to maintain the exotic inheritance at the desired level. Thirdly, draught and beef production will to a large extent depend on indigenous cattle, due to their ability to survive low nutritional regimes and drought and their resistance to tropical diseases and other environmental stresses. Also the fact that the economics of beef production is influenced to a great extent by the animal's ability to survive and reproduce, rather than rapid growth, favc urs indigenous animals. Genetic improvement of indigenous cattle could also be regarded as a means of ensuring their future use and survival. There are many circumstances where genetic improvement can be compatible with conservation (Rege and Baker, 1993). Another rationale for genetic improvement could be to avoid indiscriminate breeding and inbreeding, particularly in areas where large numbers of male animals are castrated for draught purposes (Smith, 1988).
In Ethiopia there are several indigenous cattle types distributed over different parts of the country (Alberro and Haile-Mariam, 1982). Hence, aspiring to improve all of them may not be economically justifiable. Thus, attempts to characterize and identify the most promising breeds for genetic improvement should be one of the priorities. One such important breed which is also the concern of this report is the Boran cattle.
Boran cattle are hardy and relatively productive animals which may have a significant role to play in arid and semi-arid regions of the world (FAO/UNEP, 1980). They were also considered by FAO to be one of five breeds which should be given priority for further development and conservation (Philipsson, 1992b). From their origin in south and southeastern Ethiopia, Kenya and Somalia they are now found in several African countries including Tanzania, Uganda, Zambia, Zaire (Alberro, 1986). Although they are originally owned by nomadic pastoralists who depend on milk for their livelihood, their role as beef and draught animals as well as for milk production in a crossbreeding system is increasing. Results of studies in East Africa have shown that the potential of Boran cattle for beef production under ranch conditions is also considerable (Trail et al, 1984).
The number of Boran cattle in Ethiopia is close to one million (Alberro, 1986). Recognizing their economic importance, the Ministry of Agriculture established in 1959 a Boran cattle breeding programme with the objective of improving beef production through selection. In 1972, a crossbreeding programme to produce F} Friesian-Boran heifers for dairy was also started.
To take advantage of the potential of both indigenous and exotic animals, attempts to improve indigenous cattle while using them for crossbreeding should be a way forward. Hence this thesis presents information on the comparative performance of indigenous and exotic cattle and their crosses as well as some genetic parameter estimates which are required in the design and application of practical breeding programmes. The thesis comprises six articles dealing with:
1) the evaluation of Boran cattle and their crosses under different reproductive management regimes (artificial insemination and natural service) for calving interval and age at first calving; 2) the evaluation of Boran cattle and their crosses for growth and calf mortality up to weaning and for number of services per conception; 3) estimation of genetic parameters and genetic and environmental trends for growth traits (up to one year of age) and reproductive traits (age at first
3 calving and calving interval); 4) the evaluation of dairy cattle of Friesian origin and their crosses with Boran (F,, 3/4 and 7/8 Friesian) for lactation milk yield, lactation length, calving interval and abortion rate.
The main objective of this thesis is to present a brief summary of the papers included. This is followed by a discussion of the papers and some other relevant literature with the aim of assessing their implication for future cattle breeding work. Before that, however, some general information about cattle production in Ethiopia is presented.
BACKGROUND INFORMATION ON CATTLE PRODUCTION
I. Cattle Production Systems
i The climatic conditions in Ethiopia vary from humid tropical in the western lowlands through mild subtropical in the highlands, to the arid tropical conditions of the eastern lowlands. Due to this extreme variation in climatic conditions as welljas variations in feed and water supplies and population density, the livestock production systems and their objectives vary considerably. Generally speaking, two major systems with some intermediate types can be identified.
Highlands The highlands, at an altitude exceeding 1500 metres, occupy about 40% of the total area of the country (Gryseels and Anderson, 1983), yet they support about 92% of the luman and 78% of the cattle population (MOA, 1984). From an economic view point (Getahun, 1978) three different zones can be identified in the highlands, all o^ which are characterized by subsistence smallholder mixed farming system.
High potential cereal/livestock zone: This zone comprises the central highlands, the northeastern highland valley and the Lake Tana basin. In this zone, cattle are used for draught purposes, and for milk and meat production. The relative importance of cattle production as compared to cereals varies within the zone. In the (entral highlands, increased productivity based on crossbred animals could be achieved (Gryseels and Anderson, 1983). In areas close to big cities such as Addis Aba ?a there is already a tendency to concentrate on dairy production. In many area s included in this zone, improper drainage limits crop production and for the sam i reason livestock production might be limited by internal parasites.
Low potential cereal/livestock zone; The low-potential cereal/livestock zone consists of the degraded highlands which are climatically part of the high- potential cereal/livestock zone and the plateaux and escarpments lying between the lowlands and the highlands. This is an area where crop yields are low and there is acute shortage of arable land and productive pasture (Gryseels and Anderson, 1983). Due to the lack of pasturage, fanners in this zone keep only draught animals. In this zone, little forage development has taken place and the main limiting factor for cattle production is lack of forage.
High potential horticulture/livestock zone: The high potential horticulture/livestock zone has a high annual rainfall, a long wet season and a moderately, warm climate (Getahun, 1978). This is an area primarily producing coffee, ensete (false banana), chat (Catha edulis), root crops etc. Cattle in this zone are kept primarily for milk and meat production as the hoe is the main agricultural tool (Gryseels and Anderson, 1983). However, the use of oxen for draught purposes is also increasing in this zone. Due to the availability of abundant forage the potential for increased meat and milk production is high. Nevertheless, the major limiting factor in this area might be parasitic diseases.
Lowlands In the lowlands the primary aim of cattle production is milk and milk products. Cattle have social and cultural functions and they also serve as a means of insurance against adversities. Male and extra female animals are occasionally sold to buy grain and other goods. Due to the increasing human population, and the ability to irrigate part of the lowlands, the size of the area held by nomads might decrease in the future. For example, areas in the lowlands of Arsi which were used for grazing are now being cultivated (Tacher and Jahnke, 1992). The potential of the lowlands could be exploited to a large extent by integrating it into the highland crop/livestock production system.
H. Cattle Breeding Activities
In Ethiopia there are a number of organisations involved in livestock research and development. Here some of their activities related to cattle breeding will be highlighted.
National Artificial Insemination Centre (NA1C) The National Artificial Insemination Centre was established in 1981 after reorganizing small local AI stations. NAIC is responsible for production and distribution of semen as well as importing of bulls and semen. It also recruits bull calves of Friesian and Jersey breeds locally if their dams' production exceeds 4,000 and 2,500 kg per 305-day lactation, respectively. In addition the Centre has crossbred (3/4 temperate-zebu and F: crosses) and indigenous bulls (Fogera, Boran) which are selected on the basis of their own birth and weaning weight and condition. Between 1984 and 1993, the Centre produced about 316,800 doses of semen of which about 190,000 doses were distributed. In 1993 the Centre produced 35,096 doses of semen mainly from Friesian bulls and distributed 27,108 doses
5 (Zewde et al., 1994).
Ranches The Ministry of Agriculture has ranches in different parts of the country. These are involved in indigenous cattle improvement programmes as well as in production of crossbred (F: temperate-zebu) heifers to be distributed to farmers. The Ministry also has a Jersey herd of 165 animals which was established with the objective of providing breeding bulls for crossbreeding purposes for farmers.
Smallholder dairy projects. There are a number of dairy development activities assisted by different organisations, established with the objective of assisting small farmers. Although they have now ceased activities, the Chilalo Agricultural Development Unit and the Wolaita Agricultural Development Unit, were the two pioneer organizations which were engaged in smallholder dairy development in Ethiopia. The Dairy Rehabilitation and Development Project and the Fourth Livestock Project are currently providing extension services to small farmers in many parts of the country. The Selale Peasant Dairy Development Project, financed by the Finnish International Development Agency (FINNIDA), was established with the objective of assisting the country to develop a livestock base for milk production. These protects and other activities related to livestock production are coordinated or executed by the Ministry of Agriculture.
Livestock related research institutes The Institute of Agricultural Research (IAR) which sponsors and coordinates all agricultural research in the country has undertaken a comprehensive crossbreeding research programme with the main objective of comparing different com binations of indigenous zebu and European breeds in regimes with contrasting env ronmental and 'farming conditions' in order to determine the best adapted crossbred groups for each region (Kebede, 1992). Teaching institutes like the Ale:naya University of Agriculture were also involved in crossbreeding and eva uation of indigenous animals at Alemaya, Debre-Zeit and Gonder stations. The International Livestock Centre for Africa (ILCA) is also involved in research relaled to assessment of milk productivity of cattle under small-farmers conditions (Gryseels and Anderson, 1983). It is also assisting the national organisations in data analysis.
State dairy farms There are about 12 dairy farms operated by the Ministry of State Farms Development which own the largest exotic and/or crossbred cattle population in the country. In the future these farms are expected to have the potential to serve as a source of improved dairy animals. In addition the Ministry is responsible for the collection, processing and distribution of milk and milk products. III. Major Constraints
There are several factors which hinder cattle production in Ethiopia. Some of the most important ones are highlighted here. However, it is worth recognizing that the combined effect of several factors working together is more important than individual factors as such.
Feed resources The major problem limiting cattle production in Ethiopia is lack of adequate feed. This not only limits productivity but it might threaten the very survival of the animals, particularly during years of drought and in the dry season in some areas. During the last 10 to 20 years the increase in livestock population and decrease in the communal grazing area due to increases in human population has resulted in overstocking and overgrazing. This situation is aggravated by the fact that grazing land is communally owned and livestock are individually owned.
Animal diseases Animal diseases, including external and internal parasites, limit particularly the productivity of imported or crossbred animals. In the south and southwest of the country, blood parasite infections such as trypanosomiasis have become more rampant (Slingenbergh, 1992). In the traditional production system in the central highlands, however, mortality rates of indigenous cattle are relatively low (Mukasa-Mugerwa et al, 1989) except when death waves occur as the result of drought or epizootic diseases (Tacher and Jahnke, 1992). Therefore, the effect of parasites or other diseases on the level of productivity (Tilahun, 1990) is more critical than on mortality as such.
Management Lack of management skills and technical know-how is another reason for the low productivity of cattle in Ethiopia. The adoption of newly introduced technologies could prove to be difficult. Increased efficiency in all aspects of animal production can be achieved if the animals and the resources available are managed properly. For example, making breeding or calving seasons match the seasonal availability of pasture, is one of the management strategies to follow. Efficiency in reproductive management is another area of importance, particularly on farms established to produce crossbred heifers and on all farms where Al is used. On- farm research to ascertain the reasons for differences in productivity between farms could help to identify promising locally adapted management systems (Peters, 1992).
Genetic potential Although the indigenous cattle of Ethiopia are well adapted and to some extent disease tolerant, their genetic potential for milk and meat production is low. The fact that in the highlands most of the relatively fast growing males are castrated
7 1 and used for draught purposes might suggest that only a small number of poorly performing bulls are left for breeding, perhaps, resulting in inbreeding and lower production. Studies on the extent of inbreeding are not available, however.
Marketing and infrastructure Lack of roads and other services limit the ability of farmers to sell their products during some parts of the year - at least in remote areas. The fact that the lack of a market and inappropriate pricing policies have limited production was evident from the increase in the supply of milk to Dairy Development Enterprises (DDE) when the price of milk was recently increased slightly (DDE, 1993). However, it was mainly farmers around Addis Ababa who may have benefited from this price rise. The selling of milk could be difficult in some parts of the country during the long fasting season. The attempt by the Ministry of Agriculture to organise peasant farmers into user groups to process their milk into dairy products with longer shelf-life might have its greatest influence in such areas. The same factors which affect marketing of animals or animal products also hinder livestock development activities including the Al and veterinary services.
SUMMARY OF THE PAPERS INCLUDED
I. Productivity of Boran Cattle and their Friesian Crosses at Abernossa Ranch, Rift Valley of Ethiopia. I. Reproductive performance and pre- weaning mortality
In this study the reproductive performance (calving interval, age at first calving and number of services per conception for first pregnancy) and pre-weaning moi lality rate of Boran cattle and their crosses with Friesian, based on data collected from the Abemossa Boran cattle breeding and improvement ranch, were ysed. The data were collected between 1977 and 1985. As cows were artificially inseminated as well as naturally served by Boran or Friesian bulls, the data on calving interval and age at first calving (AFC) of each breed and mating group were analysed separately:
1. Boran cows naturally mated with Boran bulls (B-NS-B); 2. Boran cows naturally mated with Friesian bulls (B-NS-F); 3. Koran cows artificially inseminated using Friesian semen (B-AI-F); 4. Fjj cows artificially inseminated using Friesian semen (Fr AI-F); 5. ’ijhree-quarter Friesian heifers artificially inseminated using Friesian semen for AFC only (3/4 F-AI-F).
The model used included the effects of year and season for all traits. In addition the effects of parity in the case of calving interval, and breed group in the case of r.iumber of services per conception for first pregnancy (NSC) were considered. The model for pre-weaning calf mortality rate (all recorded deaths and stillbirths) included year and season of birth, parity, sex, breed and the interaction of breed with year.
The results showed that NSC was 1.81,1.61 and 1.69 for Boran, F1 Boran-Friesian and 3/4th Friesian heifers, respectively. Heifers conceiving in the main wet season (July to October) required significantly (P<0.05) fewer NSC than those conceiving during the other seasons. The observed increase in NSC for those which conceived during the dry and short wet seasons may be related to the occurrence of high temperatures and feed scarcity.
The difference in AFC of the four breeding groups was marked. The crossbreeds (F1 and 3/4) which were artificially inseminated calved 13 to 15 months earlier than the Boran heifers. On the other hand, artificially inseminated Boran heifers (B-AI-F) calved 1.9 months later than those naturally mated (B-NS-B) and this difference was also statistically significant (P<0.05).
The mean calving intervals were 465,552,525 and 487 days for cows in the mating groups of B-NS-B, B-NS-F, B-AI-F and Fr AI-F, respectively. Calving interval was significantly influenced by year of calving (P<0.01) in all mating groups but the effect of calving season was significant only in Boran cows which were naturally mated with Boran bulls and those artificially inseminated. In both groups the shortest calving intervals were observed for cows calving during the main wet season, showing the advantage of planning the calving seasons. This was due to the availability of adequate pasture during this season which enables the cows to be in good condition during and after calving for reconception in the ensuing breeding season.
The pre-weaning calf mortality rate was 4.0 in Boran and 2.9% in F-, crosses and was significantly influenced by parity, sex and interaction of breed with year of birth. The higher survival rate of F-j calves, which was marginally significant (P<0.20), was probably due to the effect of individual heterosis.
The results from this study with regard to AFC of crossbreeds, NSC of all heifers and the pre-weaning calf mortality rate are comparable to those in other reports from the tropics. However, the AFC of Boran heifers was high and the calving interval of all cows was relatively long. Both year and season had an important influence on calving interval, while AFC was influenced by year only. Thus, improvement in both traits could be achieved by adopting better management as well as by crossbreeding.
Although it is difficult to make direct comparisons between breeds as they were usually confounded with mating type, the F: crossbreeds required fewer NSC, showed relatively lower mortality, lower AFC and shorter calving interval than the pure Boran.
9 H. Productivity of Boran Cattle and their Friesian crosses at Abemossa Ranch, Rift Valley of Ethiopia, n. Growth Performance
This paper evaluates the effects of genetic and environmental factors on the growth performance of Boran cattle and their crosses at the Abernossa ranch. The data! used included 4,197 births and 2,441 weaning weight records of Boran, Fj Boran-Friesian and 3/4 Friesian calves. In addition 390 adjusted one-year, 177 adjusted 2-year and 364 adjusted 3-year weights of Boran cattle were analysed to estimate the influence of genetic and environmental factors. The least-squares procedures of Harvey (1977) were applied to analyse the data using both fixed and mixed models. The model used included the fixed effects of breed of calf, year and season of birth, sex and parity as well as the interaction of season and year with breed and the random effects of sires and cows within sires. In the case of weaning and post-weaning weights both the linear and quadratic partial regressions on age were also fitted.
The imain results were: - Boran, F: and 3/4 calves weighed 25.2, 25.4 and 25.7 kg at birth and 158, 177 and 180 kg at weaning, respectively. The total weight gain from birth to wjeaning was 132,151 and 154 kg while the daily weight gain was 535,638 and 642 g for the Boran, Fj and 3/4th Friesian calves, respectively; - Weights of Boran cattle at 1, 2, and 3 years of age were 179,269 and 338 kg, respectively; - Weaning weight of calves was significantly (P<0.01) affected by all factors studied; - At weaning, animals bom during the short wet season (March to June) were significantly heavier than those bom in the main wet season (July to Oct.) and tl ose bom during the dry season took an intermediate course; - T lie interaction between breed and season of birth was also highly significant <0.02). The highest fluctuation in weaning weight with regard to season of b: rth was observed for the 3/4 and ?! crossbred calves. Moreover, when bom d uring the main wet season, 3/4 crosses weighed 5.6 kg less than the F1 calves, though they showed the highest growth rate when bom during the other two seasons. This change in ranking between the Ft and 3/4 Friesian crosses depending on their season of birth indicates that crossbreeds acquiring more than half of their genes from Friesian cattle require feed supplementation in periods of feed scarcity in order to maintain their superiority and that this requirement increases with the level of Friesian inheritance. The fact that Boran calves maintained a uniform growth rate regardless of their birth season is an indication of their adaptability. - Animals bom during the short wet season were 5.5% and 4.8% heavier at w eaning and 2-year of age, respectively, than the overall mean; The reason for the difference in weaning weight among the breed groups could be c ue partly to differences in additive genetic effects. In addition the difference
10 between crossbreeds and Borans reflects the effect of individual heterosis, since both breed groups suckled Boran cows. However, part of the reason for the difference between the 3/4 and F: crosses is attributed to mainly maternal heterosis.
From this study the following conclusions were made: - The growth performance of the animals is influenced by both genetic and environmental factors; - Systematic environmental factors such as year and season of birth, parity, sex and age at which animals are weighed had an influence on body weight; - Year of birth was a major source of variation in the weights at different ages; - Considering the weaning weight of the three genetic groups, the F1 calves grew 12,2% faster than the Boran and had a growth rate comparable to the 3/4 crosses; - Crossbreeding with temperate breeds could increase the efficiency of beef production under environments like Abemossa Ranch.
III. Estimates of direct and maternal (co)variance components of growth traits in Boran cattle
The objective of this study was to present estimates of variance and covariance components and the resulting genetic and phenotypic parameters among growth traits of Boran cattle, using REML (restricted maximum likelihood) from univariate and multivariate analysis.
The data included pedigree and weight records of animals born between 1959 and 1985 at the Abemossa Cattle breeding and improvement Ranch in Ethiopia. Birth weight (BW), weaning weight (WW) and yearling weight (YW) for 5,256,4,082 and 2,417 calves, respectively, were available. The fixed effects fitted were year-season of birth or weighing, parity and sex for all traits. Age of the animal at weighing and its inbreeding coefficient were also fitted as covariables for WW and YW. The random effects in the complete model were the animals' additive genetic effect, the maternal additive genetic effect and their covariance and the permanent environmental effects of the dam.
In addition to the complete model described above, five incomplete models, which ignored one or two of the maternal effects and/or which assumed no covariance between the direct and maternal genetic effects were applied to each trait. Estimates of (co)variance components were also obtained from bi- and tri-variate analyses using selected models. The analysis was carried out using the Derivative- Free Restricted Maximum Likelihood (DFREML) computer package of Meyer (1993a).
11 The results showed that for BW, a model, which included direct and maternal genetic effects and their correlation, was selected as 'appropriate'. In the case of WW and YW the 'appropriate' model was one which also included a random effect of permanent environment of the dam. For BW a direct heritability (h/) of 0.24; and a maternal heritability (hm2) of 0.08 were estimated. In the case of WW and1 YW the estimates were 0.29 and 0.34 for h f and 0.06 and 0.05 for fcm2, respectively. The ratio of permanent environmental variances to the total (c2) were 0.14 and 0.05 for WW and YW, respectively. Estimates from bi- and tri-variate analyses were similar to those from univariate analysis but the estimates for h / were relatively higher. Parameter estimates based on WW and YW together, as well as the trivariate analysis appeared to have accounted for selection bias on WW and thus resulted in a higher h / estimate for YW. The across trait correlation estimates were low to medium with the exception of high direct genetic and permanent environmental correlations between WW and YW.
Generally, parameter estimates of Boran cattle were within the range of those for other breeds. However, the genetic antagonism between direct and maternal effects seems to be stronger (-0.33 to -0.68). In addition the amount of variation, particularly the phenotypic variation, observed for all traits was low compared with other tropical breeds with similar means.
IV. Estimates of genetic and environmental trends of growth traits in Boran cattle
The{objective of this study was to assess the genetic change made at the Abemossa Catffle breeding and improvement Ranch in Ethiopia. The data on birth (BW), weaning (WW) and yearling (YW) weights of animals bom between 1959 and 1985 werie analysed using univariate REML (restricted maximum likelihood). The number of animals in the pedigree was 5,882 animals and those with a weight reccjrd were 5,255 for BW, 4,082 for WW and 2,417 for YW. The model included the fixed effects of year-season of birth or weighing, parity and sex for all traits. Age of the animal at weighing and its inbreeding coefficient were also fitted as covariables for WW and YW. The random effects were the animals' additive genetic effect, the maternal additive genetic effect and their covariance and the permanent environmental effect of the dam for WW and YW. However, the model for BW did not include the permanent environmental effects of the dam,
Thejmean generation interval was calculated to be 6.75 years which means that on the average 3.29 generations of selection occurred in a 24-year selection period, during which inbreeding increased to an average of 1.7% in animals bom in 1985. The regression of estimated direct and maternal breeding values on year of birth wasj-0.002 and 0.003 kg for BW, 0.32 and 0.02 kg for WW and 0.22 and -0.005 kg for YW, respectively. The aggregate breeding value, which is the sum of the maternal and direct breeding values, showed an increase of 0.34 and 0.21 kg per year for WW and YW, respectively. Whereas the maternal environmental trends were weak, the variation in the direct environmental effects was the highest. Generally, a lower level of annual genetic trend was estimated when breeding values were regressed on generation coefficients than on year of birth.
The genetic trend estimates indicated the possibility to increase growth traits in Boran cattle. Despite the good opportunity for selection as exemplified by the more than 50 kg difference between the maximum and minimum breeding values for WW and YW, the maximum genetic gain in aggregate breeding value was 0.2% of the mean WW (168.9 kg) per year. This is to be expected, due to the relatively high level of genetic antagonism between direct and maternal effects in Boran cattle and also problems in the execution of the selection scheme with relatively long generation intervals. Furthermore, selection was based on phenotypic performance, colour and conformation rather than on breeding value estimates. The marked variation between years in direct environmental trend suggests that improvement in environmental conditions should be part of the genetic improvement programme if the advantage from the selection activity is to be realized. In addition, measures to reduce both the rate of inbreeding and generation intervals need to be considered.
V. Genetic and environmental effects on age at first calving and calving interval of naturally bred Boran (zebu) cows in Ethiopia
The main objective of this study, as opposed to Paper II which concentrated on the effect of fixed factors on fertility of Boran cattle and their crosses under two reproductive management regimes (artificial insemination and natural service), was to determine genetic parameters for age at first calving (AFC) and calving interval (Cl) of naturally mated Boran cows. In addition, estimates of the environment and genetic trends using 24 years of data were reported.
The data were analysed using the Derivative Free Restricted Maximum Likelihood (DFREML) package of Meyer (1989) by fitting an animal model to obtain estimates of genetic parameters. The genetic parameters were subsequently used to obtain Best Linear Unbiased Predictions (BLUP) of breeding values of animals. All available pedigree information was included.
The mean AFC and Cl were 41.8 months and 442 days, respectively. The results showed that the ft2-value for AFC was 0.062, and 0.075 when estimated on the original and selected data, from which cows that did not calve after 5 years of age were deleted, respectively. The h2-va\ue for Cl was 0.037, and 0.043 when estimated on the original and selected data, from which cows with AFC higher than 5 years and Cl longer than 2 years were deleted, respectively. The
13 corresponding c2 (ratios of permanent environmental variance to total) were 0.031 and 0.028 for the respective datasets.
Estimates of h2 from bivariate analysis (AFC and the first Cl) were 0.065 and 0.012 for AFC and first Cl, respectively, while their genetic and environmental correlations were -0.054 and -0.176, respectively. The h2 values estimated in a bivariate analysis of the first three CIs were 0.015, 0.002 and 0.093 for the first, second and third Cl, respectively. The low /z2 for young cows might have been due to their vulnerability to environmental stress, compared with relatively mature cows. The annual genetic change for both traits was not significant, whereas the solutions for year effects were highly variable. The regression of the solutions of year effects for AFC on year of birth showed that it increased by about 10 days while the variation in Cl was due to random year to year fluctuation.
The [conclusions drawn from this study were as follows: - Genetic parameter estimates in this herd are within the range of those reported for tropical cattle; - Due to the fact that the h2 of AFC is close to zero and the possibility of negative genetic correlation with Cl and that the h2 of different Cl varied considerably, any improvement of these traits by selection would be rather difficult; - Because variation in AFC could arise from differences in growth rate, age at the onset of puberty and the ability to conceive (Galina and Arthur, 1989) - which means that the h2 observed reflects variations in a number of interrelated traits - it was suggested that alternative traits such as scrotal circumference which is fc vourably correlated with semen characteristic and a good indicator of age at puberty in both sexes (Meyer et al, 1990) might be considered; \e other limitation of both AFC and Cl as measures of fertility is that they e: cclude animals which are not calving, which might justify a shift to recording o :her traits such as calving rate or days to calving where in the latter case cows n)t calving are assigned a predicted value, derived from threshold theory (Nptter, 1988). In the short term, however, improvement in fertility in this herd has to come from improved management including better data recording, timely culling of cows with long Cl and high AFC and using bulls with good semen characteristics.
VI. Genetic and environmental effects on performance of dairy cattle of Friesian origin and their Boran (zebu) crosses at Alemaya, Ethiopia
This study evaluated the performance of dairy cattle of Friesian origin imported fror r Kenya and their crosses with Boran using 23 years of data. The animals were kept at the Alemaya University of Agriculture Dairy Farm (500 km east of Addis Abe ba) which has an altitude of 1,980 m above sea level and an annual rainfall of abo it 800 mm. The objectives of this study were to provide parameter estimates and to compare breed groups on the basis of milk production and reproduction traits and to look into some environmental factors affecting them.
The data used in the study were compiled from the farm records which included information on milk yield, breeding, calving and pedigree as well as health records. All lactations except those shorter than 50 days were used to study lactation milk yield and lactation length. The numbers of records were 742, 587 and 876 for lactation milk yield, calving interval and abortion rate (including stillbirth), respectively. The data was collected between 1965 and 1987 and the breed groups were those of Friesian origin and their F-[ and 3/4 crosses and Grade Friesian (those with 7/8 and above Friesian inheritance).
The overall mean unadjusted lactation milk yield, lactation length, calving interval and abortion rate were 4,058 kg, 324 days, 437 days and 7.3%, respectively. The deviation of the FT crosses from that of Friesian origin was -2,365 kg for lactation yield, -75 days for lactation length and -29 days for calving interval. The lactation yield of the grade Friesians was similar to that of the Friesian but that of the 3/4 crosses was 897 kg below that of the Friesians. The incidence of abortion was relatively low (5.5%) in the F} crosses, compared with the overall least-squares mean of 8.3%. Abortion rate was influenced by season, with 41.5% of the 63 instances occurring in the dry (November to January) period during which the animals were fed mainly hay which might have been mouldy.
Heritability and c2 (the ratio of permanent environmental variances to total) were 0.08 and 0.24 for lactation yield, 0.0 and 0.20 for lactation length and 0.0 and 0.02 for calving interval, respectively. The inbreeding level of all animals (except the F^ in the herd increased to a mean of 3.84%, assuming that the level at the beginning was zero. The influence of inbreeding coefficient on performance was not significant, however.
The conclusion of this analysis was that although the productivity of cattle of Friesian origin was significantly higher than that of the F7 and 3/4 crosses, the fact that their yield decreased by 66 kg per annum while there was no negative genetic trend indicates that they require improved management and a higher level of feeding in order to maintain their superiority. The main reason for the decline in the milk yield seems to be associated with scarcity of feed due to drought. Moreover, due to a high abortion rate, long calving intervals, high age at first calving and a small number of lactations per cow (on average 3) particularly in Friesians and higher grades, it was rather difficult to increase or even maintain the population size at the farm.
15 GENERAL DISCUSSION
One feature of the papers included in this thesis is that the environmental effects, particularly the year effects, are considerable. Generally the phenotypic performance of the animals deteriorated during the study period. The main reason foil the between year variation is the occurrence of drought. The first cycle of drought occurred between 1973 and 74 and again it was repeated between 1983 and 85. The rainfall in 1980 was also below average at both locations. Another reason seems to be the instability which occurred between 1976 and 1977 (Paper VI) in the Alemaya area and between 1974 and 1975 (Papers IV and V) in the Abemossa area. Deterioration in animal performance over study periods was observed in almost all animal performance evaluation studies in Ethiopia (Haile- Mariam and Mekonnen, 1987; Kebede, 1992; Negussie, 1992; Negash, 1994). This suggests that at least part of the reason for the deterioration observed is the fact that new improvement projects usually start with high input which cannot be sustained for a long time. The lesson to be learned from these is perhaps that new projects before they start should give enough attention to local environmental constraints, level of resource, training of local staff, etc.
Another feature of the data used in these papers is that they come from government farms. Almost all data used to analyse breeding work in Ethiopia so far | are from government farms or research stations (e.g., Haile-Mariam and , M ?konnen, 1987; Kebede, 1992; Negussie, 1992; Negash, 1994; Hassen, 1994). Cc rtsequently they are not representative of the farming conditions in the country. T1‘ is highlights that there is a need to collect data from the majority of the animals in the hands of small farmers.
As. previously mentioned, a subsistence smallholder in Ethiopia expects to use his ca tie for draught purposes and also to produce milk and manure and to slaughter thorn for beef when the need arises. However, efforts to improve cattle productivity through breeding in Ethiopia can still be viewed from two perspectives: one which emphasises improvement of indigenous cattle for beef production and the other which emphasises the use and improvement of in iigenous and crossbred (temperate x indigenous) cattle mainly for dairy pr oduction.
Improving beef production Introduction
A] ?proximately 98% of the cattle population in Ethiopia is indigenous. In the fo reseeable future they will be the major producers of meat and draught power. Indigenous animals are owned by peasant farmers or pastoralists and improving their productivity will benefit their owners who represent more than 85% of the total human population. r
In attempting any indigenous breed improvement, all technical and other issues should be considered at the planning stage to justify the investment and to overcome some of the drawbacks already seen in some selection programmes. After reviewing selection works in tropics, Taneja (1990) reported lack of appropriate analysis of breeding objectives and lack of periodic evaluation of breeding schemes as primary reasons for their limited success. In Paper IV, for example, it was shown that the genetic progress observed from the selection scheme of Boran cattle in Ethiopia was at the most 0.2% of the mean (168.9 kg) per annum for weaning weight. This limited genetic response despite the good prospects for selection (Paper IV) was due to lack of appropriate selection criteria and lack of appropriate statistical analysis to separate genetic effects from environmental effects. Consequently, selection at Abernossa was based on phenotypic performance, colour and conformation rather than on estimated breeding values. The other reason could be the relatively high level of genetic antagonism between direct and maternal effects in Boran cattle and the relatively long generation intervals. In this thesis, some of the important issues that should be considered when planning breed improvement schemes are discussed with regard to the Boran cattle breeding scheme.
2. Planning a selection scheme
The two major steps that should be considered when planning a selection programme involve definition of the breeding objective, including recording of traits and estimation of genetic parameters and formulating and optimising of the scheme; the latter including prediction of breeding values as well. # Definition of objectives Breeding objectives are extremely important when designing a selection programme, though they are difficult to define for the tropics. This is due to the fact that animals are multipurpose and there is limited knowledge of the production systems to determine economic values. Moreover, setting economic values for adaptive traits which are crucial in the tropics is rather difficult. Estimates of genetic parameters for some important traits are presented in Paper III and IV for Boran cattle. However, for reasons given above deriving economic weights is not attempted. Once the required information have become available approaches like those of Kasonta and Nitter (1990) or of Dempfle (1992) may be used to obtain such estimates.
Assuming that the specific objective of the Boran cattle breeding programme at Abemossa was to increase beef production, the following discussion will focus on some of the important traits that should be considered. Traits that affect beef production are final weight (growth), carcass quality, survival and reproduction. Although carcass quality should be an important trait, it is expensive to measure
17 (Renand et al.r 1992) and under the present conditions in Ethiopia there is no quality payment. Emphasis on quantity only might therefore be justified.
Final weight: Final weight is genetically correlated with both weaning weight (0.83) and post-weaning weight (0.86) in temperate cattle (Renand et al, 1992). The genetic correlation between direct weaning and yearling weight observed (Paper III) in Boran cattle was also high. However, considering yearling weight as a post- weaning weight in the herd at Abernossa ranch or in any tropical cattle where animals are not ready for slaughter before they are 2 years of age might be questionable. First, yearling weight is still influenced by maternal genetic and environmental factors (Paper III). Secondly, its correlation, for example, with live weight at slaughter might not be high enough, particularly in view of the correlation values among traits observed in Paper in where it was shown that the across traits genetic correlations for Boran cattle were relatively small compared with literature estimates from temperate cattle (Meyer, 1993b). Hence, further study to identify a suitable post-weaning weight as selection criterion is suggested.
Measuring growth performance and improving it through selection could be successful, though certain limitations must be considered carefully. First, the genetic correlation between maternal and direct effect appears to be more antagonistic (Paper m) in Boran cattle. Secondly increasing growth which usually results in increased mature weight is likely to have some drawbacks in tropical environments. The consequence of increased growth rate on the overall productivity of the animals in a variable environments should be carefully considered. Increased growth will result in increased mature size which might mal e the animals less adapted and more susceptible to drought, heat load or other env: ronmental hazards (Frisch and Vercoe, 1984) and/or in increased weight of breeding cows thus increasing their maintenance requirement. The implication of such improvements may be assessed using approaches such as systems analysis (Bladcburn and Cartwright, 1987).
The fact that genetic correlation between maternal and direct effects is antagonistic and that selection for increased growth could result in increased mature size may sug; *est that establishing two lines (maternal and paternal lines) might have some advantages. If, for example, the correlation of maternal genetic effects with mature size- (which is essentially direct) is low compared with that of direct effects, the maternal line should be selected on the basis of estimated maternal breeding values. However, there are certain limitations to this approach: the genetic variance of the maternal effects in Boran cattle (Paper HI) is rather small, due to which limited success is anticipated. Establishing two lines would also mean a small number of animals per line, given the same scale of operation. In addition, it nr ight complicate the programme as well. When possible selection for components of growth less related to mature size should be emphasised under tropical conditions (Menissier and Frisch, 1992). That means data both on growth and mature size should be collected. To avoid the consequent increase in generation interval, components of growth traits which can be measured early in life and at the same time are less or not correlated with mature weight should be identified and used as selection criteria (Frisch and Vercoe, 1984). As procuring weighing scales could be both difficult and expensive, measurements such as heart-girth and height at withers (Karlsson, 1979) might be recorded instead of weight, once appropriate prediction equations have been established.
However, the search to identify indicator traits or measurements which can be used as selection criteria for growth and which are less correlated with mature size is essential. Lee and Haley (1990) found that sheep selected for large testis size (adjusted for body weight) grew faster and reached maturity earlier but had smaller mature size than those selected for small testis size. Such animals thus would have a relatively lower maintenance requirement and would also be fatter at any given time, making them relatively drought resistant as well (Robertson, 1982). Similarly, Kachman et al. (1988) reported that mice selected for age at puberty (early rapid growth) grew faster without a sizeable concomitant increase in mature weight.
Reproduction. Reproductive efficiency may be defined broadly as the number of calves weaned per cow per year of production and per lifetime of a cow. In the studies included in this thesis, genetic analysis of calving interval and age at first calving were given for Boran cattle (Paper V). The phenotypic mean and environmental factors affecting pre-weaning mortality rate were also presented in Paper I. The level of pre-weaning mortality which was defined to include stillbirth rate was comparatively low in the Boran herd under study. Although monitoring pre-weaning mortality is important, omitting it from the selection criteria at this stage could be justified for two reasons: first, the fact that the level was low may indicate that the genetic variation is also low. Secondly, it is difficult to obtain estimates of genetic parameters measured on similar cattle breeds or environments. For the time being, by excluding pre-weaning mortality it might be possible to keep the programme simple.
The data available from the Boran cattle breeding ranch in this study were not suitable to analyse important reproductive traits such as calving rate and 'stayability7 (i.e. the number of calves produced in a lifetime of a cow). Recording the number of cows at the beginning and end of each breeding season and the disposal date and reason, of breeding cows could have improved the quality of the data. Age at first calving is a trait affected by age at puberty, growth and age at exposure for breeding (Paper V) which makes selection based on it less reliable. Its heritability was also low. Furthermore, reproductive traits are largely
19 influenced by random environmental factors (Menissier and Frisch, 1992). It was also shown in Paper V that the fr2-value of first and second calving interval was lower than the third, which suggested that selecting at younger age for calving interval is rather difficult. In Paper V it was suggested that data on scrotal circumference, which is a useful indicator trait to detect age at puberty in males and] related females (Notter, 1988), should be collected for future use. Indicators like scrotal circumference might also be interesting with regard to their association with mature size (Lee and Haley, 1990). They are also cheap and easy to record, compared with other possible physiological indicators.
Another point worth considering with regard to reproductive performance is its relationship with growth traits. When selection is for growth, a poorer reproductive performance would be expected if the association between the two group of traits were antagonistic. MacNeil et al (1984) reported a genetic correlation of 0.16 between age at puberty and post-weaning daily gain (which is little influenced by maternal factors). Similarly a study which involved a long-term selection based on 365-day weight suggested that selection for weight alone will result in delayed sexual maturity and shorter herd life (Luesakul-Reodecha et al, 1986).
In Paper V it was shown that there was no negative genetic trend in reproductive performance as measured by age at first calving and calving interval after 24 years of selection for growth. However, the level of genetic trend in growth traits was also low in the herd (Paper IV) because selection was not strictly based on estimated breeding values. Although such relationship should be estimated from covariance analysis, here the simple correlations between the estimated breeding value of the two groups of traits might give some indication. Table 1 shows that selection on direct breeding value of WW and YW could decrease age at first cahing, but it might slightly increase calving interval.
Table 1. Correlation between estimated breeding values for growth traits (maternal and direct weaning and yearling weight) and reproductive traits (calving interval and age at first calving)
Trait Maternal WW+ Direct WW Maternal YW* Direct YW
Ago at first calving 0.175 -0.253 0.188 -0.225 Calving interval 0.197 0.154 0.070 0.167 fweaning weight; ^yearling weight.
Alt lough the results in Table 1 do not completely agree with MacNeil et al (1984) anc Luesakul-Reodecha et al (1986), it is advisable that both reproductive and sur /ival traits should be considered, at least in a long-term selection programme
20 for increased beef production. This requires collection of data and estimation of genetic parameters for growth traits, including mature weight and reproductive and survival traits. In the meantime, estimates available from this study (Papers III and V) and from the literature (Koots et al., 1994) could be used to predict breeding values.
Formulating arid optimising a selection programme Estimates of breeding values are presented in Paper IV for growth traits and in Paper V for reproductive traits. By estimating the breeding values of all animals it was possible to estimate the genetic trend which was important for monitoring the effectiveness of the selection programme. However, formulating and optimising breeding schemes is an important subject which was not considered in detail in the Papers included in this thesis.
Nevertheless, some biological parameters required for planning and optimising the scheme were estimated. Generation interval and inbreeding coefficients with some comments to optimise them were presented in Paper TV. Estimates of reproductive performance (age at first calving and calving interval) (Papers I and V) and pre- weaning mortality (Paper I) which affect the rate of genetic progress as well as its dissemination are dealt with.
In Paper IV, lack of proper planning was assumed to be one of the reasons for the limited genetic progress observed in the Boran herd. It was shown that no animals were selected from among those bom in some years, while selecting the best animals from each calf-crop may have been the optimum. At the same time, although the proportion selected was small and the selected animals were used for a long time leading to an increased generation interval (Paper IV), those selected were not necessarily superior when compared with their contemporaries. This can be seen from Table 2 where the estimated breeding values of non foundation sires or dams which were widely used are presented.
Possible selection intensities among both males and females and accuracy of selection may be estimated roughly if decisions on mating methods (artificial insemination versus natural service) and selection methods (performance versus progeny testing) are made. Partly to make the programme simple, natural service and performance testing might be assumed. At the initial stage, where the main objective is selection for growth, the BLUP-animal model could give reasonable accuracy. The accuracy of selection for maternal breeding value of growth traits might be low. However, accuracy could still be reasonable due to inclusion of all relationships among animals, by multi-trait evaluation and by the fact that station records are fairly accurate.
21 Table 2. Breeding value estimates of non-foundation widely used sires and dams.
Sires* Dams* i 1 Mean Max Min Mean Max Min i BW(ke) Maternal 0.207 0.545 -0.220 0.038 0.353 -0.378 Direct “0.348 0.357 -1.253 0.286 1.121 -0.223 Aggregate -0.140 0.747 -0.708 0.324 0.891 0.203 WWCke) Maternal 1.678 4.978 -2.950 -0.150 5.430 -4.500 Direct 3.164 15.460 -11.293 4.834 14.816 -4.153 Aggregate 4.842 14.033 -6.315 4.792 11.805 -5.733 YWtae) Maternal 0.466 2.360 -3.395 -0.717 3.540 -4.096 Direct 1.779 14.685 -6.990 4.765 12.169 -2.040 Aggregate 2.245 11.290 -5.070 4.050 8.995 1.400
t=7|sires with 1,567 progeny; $=18 dams with 187 progeny.
The optimum age for selection could not be determined from any of the studies incl ided in the present thesis. However, there should be two stages of selection. The first stage could be at weaning where animals are selected on the basis of mat ?rnal and direct breeding value for pre-weaning growth. The second stage cou d be between about 18 months to a slaughter age, depending on results from future analysis of such a trait. Another alternative could be a multi-trait selection whf reby animals at weaning are selected on their own and all their relatives' recc rds for weaning weight and their relatives' records for post-weaning growth until their own post-weaning growth record is available. However, much of the discussion on optimising and formulating the breeding scheme given here is based on assumptions. Final decision should depend on the result of the comparison of alte rnative breeding strategies, based on computer simulation e.g., by the gene flow method (Kasonta and Nitter, 1990).
Although the above discussion is focused on Boran cattle improvement for beef production, the issues raised are also valid when considering improvement of other indigenous breeds. An important point which should be considered with regard to indigenous cattle improvement is the need to establish sizeable breeding herds with accurate recording. This should be done for some of the important breeds or groups of cattle. In the absence of any organised national recording scheme, genetic improvement might be possible by establishing nucleus breeding her^s and selecting within them (Smith, 1988). Selection of animals to be used as foundation stock should be based on population screening (Timon, 1993) and/or when possible on indicator traits. Precautionary measures to avoid the disadvantages of nucleus breeding schemes such as possible genotype-environment interaction between selection and commercial conditions and risk of disease (Smith, 1988) should be considered. Before establishing such centres, the production system which might be peculiar to each breed and locality should be studied. Based on such studies, the right breeding objectives should be set. Thereafter the required data should be collected. This should be accompanied with data processing facilities and procedures to disseminate the improved genetic material.
3. Crossbreeding for beef or for dairy and beef as a by-product
The potential of crossbreeding for beef production is well documented in Africa (Thorpe et al, 1981; Gregory et al, 1984; Trail et al, 1984) but there is no large- scale crossbreeding activity in Ethiopia. In Papers I and II which describe a programme designed to produce crossbred heifers for dairy production, it was shown that F-j (Friesian x Boran) crossbred calves were 12% heavier at weaning than Boran. Moreover, the pre-weaning mortality of F: crossbred calves was relatively lower than that of Boran calves (Paper I). Experimental results from Alemaya University of Agriculture (Wagner et al, 1969) and IAR (O'Donovan et al, 1978) have also shown that crossbred calves outperformed pure indigenous for beef production.
However, when productivity was based on weight of calf at weaning, pre-weaning calf survival and calving rate, the advantage of the crossbreeds was not maintained. The reason is that the cows which produced crossbred calves had long calving interval because artificial insemination was usually used (Haile-Mariam, 1993) due to poor adaptability of Bos taurus bulls (Renand et al, 1992). Furthermore, in the highlands, the growth of crossbred calves (3/4 temperate and F2) with the exception of the F-j was not markedly better than the indigenous calves, mainly due to their susceptibility to internal parasites (Kebede, 1992). The same study showed that the advantage of the crossbreeds decreased with age. Therefore, for crossbreeding to be used widely, the efficiency of artificial insemination and management of the calves must be improved. It is also worth noting that Ethiopia has a number of indigenous cattle breeds and that they can be used for crossbreeding to exploit both heterosis and breed complementarity for beef, dairy and draught.
23 13. Improving milk production 1. Introduction
In the preceding section the discussion focused on improvement of indigenous cattle mainly for beef production, but improving the milk production potential of some of the local breeds is also important. In areas where the production systems and level of husbandry are such that the introduction of Bos taurus inheritance is not advisable or not possible for some reason, the indigenous cattle will remain thernajor producers of milk. In addition, their improvement for milk production is important because it is known that crossbreeds from 'improved' zebu are better than non-improved zebu (Syrstad, 1988). Moreover, poorly performing temperate or high-grade cattle could be 'down-graded' by mating them with 'improved' zebu when the need arises. The improvement of indigenous cattle could also be based on j\ucleus breeding units. Initially limited work with one or two breeds could be attempted. When such a plan is adopted, appropriate selection criteria - perhaps also including beef production traits - should be set up. Another breeding approach for milk production in areas where breeds with temperate inheritance are not recommended could be to cross local less productive breeds with 'more productive' zebu breeds. For example, it was reported by Mahadevan et al (1962) thajt crossbreeding East African Zebu with Sahiwal increased milk yield by up to 64% in Kenya.
For semi-intensive or intensive milk production in the highlands, however, indigenous cattle breeds in Ethiopia are found to be inferior to their crosses with temperate dairy cattle (Schaar et al, 1981; Kiwuwa et al, 1983; Kebede, 1992), Under improved management in the highlands, it was shown that pure Friesians coi ld produce more milk than any of their crosses, though the result was in favour of Fj crossbreeds when survival rate (eg., abortion rate) and reproduction rate were also considered (Paper VI). The difficulty of maintaining purebred temperate dairy cattle in the Ethiopian highlands is documented in Paper VI, In addition, Hassen (1994) observed a loss due to abortion, stillbirth and disease of up to 62% of the Jersey animals bom in one state farm near Addis Ababa. The sane study showed that the mortality rate among Jersey animals imported as in- cal: heifers had reached 29.4% when about 3 years had passed since their arrival in ^Ethiopia. Similar results were reported by Negash (1994) for dairy cattle of Holstein-Friesian origin in the Central Highlands of Ethiopia. The conclusion is that dairy production based on pure temperate cattle breeds in Ethiopia is difficult if their survival and reproductive rate cannot be improved considerably.
In the preceding paragraph it was noted that milk yield of indigenous cattle is low while the adaptability of temperate cattle is poor. Crossbreeding where the adaptability of indigenous cattle is combined with high production capacity and good temperament of temperate cattle is therefore an attractive option. However, along with breed improvement, milk marketing and processing facilities and a
24 sound milk pricing policy as well as advice on better feeding and management of cattle and improved animal health services should be made available.
Once the above issues are considered, the choice of which temperate breed to cross with which indigenous breeds should be given attention. Differences among Bos indicus breeds in Ethiopia for crossbreeding seem to be small (Kiwuwa et al., 1983; Kebede, 1992). However, the breeds considered in both studies were small and further studies on many of the indigenous breeds are still necessary. Among temperate breeds it seems that results both from Ethiopia (Kebede, 1992) and elsewhere (Syrstad, 1990) suggest that Friesian crosses are superior to other breeds for milk yield in both poor and good environments. However, Kiwuwa et al (1983) showed that Jersey crosses were comparable to Friesian crosses when productivity was measured on the basis of annual fat-corrected milk yield per unit metabolic weight of cows, albeit few observations. In remote areas where the sale of liquid milk is difficult and where feed is in short supply, the Jersey cross might be an alternative.
As dairy development based on crossbred and grade cattle is recent in Ethiopia (Kiwuwa et al, 1983), it is necessary to increase their numbers and at the same time plan their improvement for sustainable use.
2. Production of crossbred heifers
In the near future the basis of dairy production in Ethiopia would likely be F: (temperate x local) crossbreeds in order to exploit both breed differences and heterosis among breeds. If so, cheap and efficient ways of producing and distributing crossbred cattle should be given attention.
Using ranches The main source of crossbred heifers in Ethiopia for small farmers has been the ranches run by the Ministry of Agriculture. However, these ranches show rather low reproduction rates (Paper I) due to which the number of heifers produced was not adequate to meet the demand. As a result, for example, the Selale Peasant Dairy Development Project financed by FTNNIDA is importing heifers from Kenya at a high cost. Producing crossbred animals at ranches and distributing them only to farmers who are capable of managing them has the advantage of avoiding indiscriminate crossbreeding, which might prove to be disappointing. In the case where the ranches are to be used as a source of heifers, their efficiency has to be improved and several options should be explored. However, when the involved farmers get the crossbred animals on loan, as is the case in Ethiopia, this could prove to be too expensive for a smallholder, particularly if the animal he got were to die before calving.
25 Encouraging private farmers to produce crossbred heifers - say, under ranching conditions in the lowlands, for sale in the highlands - could be an option. To minimize transport costs and utilize the cattle populations in several areas, small ranches may be established in as many localities as possible. Nevertheless, the role of Boran cattle as a dam line for crossbreeding will still be significant.
Using farmers' cattle Another perhaps cheaper way of increasing crossbred heifers is by encouraging individual farmers to cross their zebu cows with temperate breeds. This was one of the ways which helped to increase the population of grade cattle in Kenya (Stotz, 1979). The Ministry of Agriculture in Ethiopia has encouraged farmers to participate in such efforts. An example is a cattle collection centre (at Chacha some 100 km to the north of Addis Ababa) where farmers keep their zebu animals at a station until they are inseminated and pregnancy tested.
Another point that should be explored is the use of crossbred and purebred bulls from some of the State Dairy Farms for crossing with indigenous cattle (Brannang andjFersson, 1990). However, this requires some sort of preliminary evaluation of crossbred animals produced in this way, versus those produced by Al, before extensive natural service can be used to increase the population of crossbred heifers. Moreover, the implications of any extensive crossbreeding on the indigenous cattle genetic resource of the country should be carefully assessed.
3. mproving crossbred or grade cattle populations
While increasing the crossbred or grade herds, a breeding strategy for improving or a : least maintaining their productivity should also be instituted. Discussion on breeding for dairy production in Ethiopia has hitherto been concerned with mak ing decisions regarding the level of temperate inheritance (Kiwuwa et al. 1983; Keb ide, 1992). The experience to be gained from other tropical countries with regard to selecting within crossbred populations is rather limited (Kropf and Cha :ko, 1992) with few exceptions.
Breeding strategies It was noted earlier that Fx crosses between Bos taurus and Bos indicus outperform other breed groups, particularly under poor to average management conditions. The j management and feeding in Ethiopia in the foreseeable future is expected to remain at a rather poor level. The best option, at least technically, for small farmers might be to use F: crosses if their continuous supply at a reasonable price could be insured. In some countries it is possible to organise production of F7 crossbred animals on marginal lands (Madalena et al, 1990). In Ethiopia, however, continuous F-, production would be rather difficult because of the numbers reqi ired and the subsistence nature of agriculture. If this system is considered as an alternative then the improvement of the indigenous cattle should be emphasised because 'improved7 zebu are better than non-improved for crossbreeding (Syrstad, 1988) as well.
Another breeding strategy that could be considered is criss-crossing of Bos indicus and Bos taurus cattle with the objective of maintaining a relatively high level of heterosis. Under relatively low management conditions criss-crossing was found to give reasonable profitability in Brazil (Madalena et al., 1990). In Ethiopia such a system would have another advantage, i.e. animals with a higher level of Bos indicus inheritance could be used largely for draught or beef production while those with a higher Bos taurus level could be used mainly for milk production. However, this requires reliable identification and recording and the availability of pure temperate bulls or semen and improved zebu bulls at a reasonable price. The limited experience of Al personnel with the recommendation of the Ministry of Agriculture to use semen of bulls with different levels of Friesian inheritance on crossbred cows of different levels in Ethiopia shows that implementing such a system would rather be difficult. The other problem with two-breed criss-crossing is that it is difficult to synchronize the management and production level when the proportion of Bos taurus varies between 1/3 and 2/3 from one generation to another (Trail and Gregory, 1981).
Another option is the formation of a three-breed cross i.e., one breed of zebu crossed with two temperate breeds. McDowell (1985) reported that three-breed crosses are less productive than two-breed crosses in the tropics. In Ethiopia the 1/4 Jersey, 1/4 Friesian and 1/2 Arsi (Zebu) were among the low producing crosses while the 1/2 Friesian, 1/4 Jersey and 1/4 Arsi were among the high producers, though the number of observations of the former breed group was small (Kiwuwa et al, 1983). Based on a larger dataset collected from the same project (Arsi Rural Development Unit) as that of Kiwuwa et al. (1983) it was shown that the reproductive performance of three-breed crosses was better than those of two-breed crosses with 75% temperate inheritance (Negussie, 1992).
The remaining alternative is formation of a composite breed based on crossbred cattle. The need to determine the optimum proportion of Bos indicus to Bos taurus inheritance in order to establish and select within the synthetic breed seems to be less important (Franklin, 1986; Madalena, 1989). The difference in the level of production between Fv 5/8 temperate-3/8 Zebu and 3/4 temperate-1 /4 Zebu is quite small (Madalena et al, 1990; Kiwuwa et al, 1983; Kebede, 1992; Syrstad, 1990). Syrstad (1990) also showed that there is no environment by breed group interaction between F-j and the back crosses to temperate breed for milk production, though Madalena et al (1990) demonstrated that the profitability of F1 crosses was better than other crosses, particularly when the management level was low. Another important result worth noting is that the performance of progeny from inter se mating of F: (Kebede, 1992; Syrstad, 1988) and 5/8
27 t crosses (Madalena et al., 1990) in the absence of selection has been rather poor.
However, there seems to be some advantage of using F, crosses in the formation of a: synthetic breed. First the selection could start one generation earlier. Secondly F, crosses are more stress-resistant than back crosses or 5/8. The formation of Australian Milking Zebu was based on Fj Sahiwal (or Sindhi)-Jersey crosses (Hayman, 1974). A similar programme in Brazil (Madalena, 1989) based on crossbred cattle (cows with 1/2 to 3/4 and bulls with 1/2 to 7/8 temperate inheritance) was quite successful in developing a 'breed' called Brazilian Milking crossbred cattle (MLB). However, among 5/8,3/4 and 7/8 temperate bulls the 5/8 were found to be better in tick resistance (Teodoro et al, 1984 cited by Madalena, 1989). In India there is a programme to develop a breed called Frieswal based on 3/8 to 5/8 Bos taums inheritance with the rest coming from Sahiwal at the Military Farms (Mudgal and Arora, 1994).
Finally for the few more progressive farmers who can afford to provide a high level of management, direct importation of semen might be economically feasible (Mpofu et al, 1993b). Where adapted animals are required for farmers with poor to average management level and where there is difficulty in organising any sort of milk recording scheme (a point which will be discussed in a later section), the useiof imported semen from tropically developed breeds like Australian Milking Zebu (Hayman, 1974), Australian Friesian Sahiwal (Alexander and Tierney, 1990) etc. should be considered. Initially, however, comparative evaluation of such breeds with the locally produced crossbred or grade cattle in the country should be carried out.
Mil): recording and data analysis It is clear that a well designed selection programme within a crossbred population regc rdless of the level of temperate inheritance but close to 50% has much in its favc ur. Normally, such a scheme require an organised milk recording programme. Exp *rience from other developing countries indicates that a full-fledged recording sche me and progeny testing might be difficult to organise under Ethiopian conditions. The effectiveness of such recording schemes was questioned regarding both technical (Mpofu et al, 1993a; Philipsson, 1992a; Chauhan, 1992) and economic (Mpofu et al 1993b) efficiency. After a detailed review of the exercise in India, Chauhan (1992) argued for a limited progeny testing scheme of crossbred bulls in institutional herds (university, research, Military Farms, etc.) or for the esta )lishment of special progeny testing stations. In some cases, selection of young bull 5 based on dam's production and using them for a limited period could be an easier alternative (Philipsson, 1992a).
In I thiopia the existing state dairy farms and some progressive private farms could be used as nucleus herds. Arrangements for screening (Timon, 1993) some of tl e outstanding animals from other herds based on production and/or indicator
28 traits (Kropf and Chacko, 1992) could be accommodated into such a programme. Initially milk recording schemes within nucleus breeding units or otherwise should start in localities which have a reasonable number of crossbred dairy cattle, access to milk market and where a reasonable level of Al, veterinary and other essential services could be provided.
If a domestic genetic improvement programme is to be considered, a functioning milk recording scheme is essential at least within the nucleus herds and if possible in other cooperating herds in the vicinity of the nucleus. When considering the establishment of nucleus breeding units, the accuracy of evaluation, possible genotype by environment interaction, and model for data analysis are important (Chauhan, 1992). In addition, ways of accounting for possible sources of error such as allowing the calf to suckle its dam, the possibility of day to day variation in milk yield due to inconsistency in feed supply and other environmental factors, the small herd size, etc, should be considered if village herds are to be included in the recording scheme. Frequency of milk recording that could give reasonable accuracy should be decided on the basis of data generated from the area itself. Usually the correlation of lactation yield estimated from once per month yield records with the actual lactation yield record is found to be low in tropical conditions (Lindstrom, 1976; Vaccaro et al, 1994). Consequently Vaccaro et al. (1994) suggested that it might be prudent to record lactation length, which can be recorded easily and is more reasonably correlated with lactation yield than is lactation yield itself. However, the h2-value of lactation length is usually lower than that of lactation yield (e.g. Paper IV). More frequent recording as well as combining small herds in a village into groups and using an appropriate model could be important (Chauhan, 1992).
CONCLUDING REMARKS
The studies included in this thesis and the discussion presented suggest that there is a need for careful planning of all aspects of breed improvement schemes, from selecting the breeds to be improved to the dissemination of the improved genetic material. At the same time the environmental components - particularly the effect of year - was a major source of variation. The analysis of the Boran cattle improvement programme suggests that the deterioration in the environmental conditions, the absence of clearly set selection criteria and the lack of proper planning were the major reasons for the limited the success of the programme. However, the scheme at Abernossa demonstrated that it is possible to make genetic progress in growth traits even under difficult conditions and that this progress can be increased considerably provided that the available data is used for breeding value estimation and selection.
The fact that almost all the data analysed in Ethiopia with regard to breed
29 w
improvement is coming from state farms or research stations suggests that there is a need to collect data from the majority of the animals which are in the hands of small traditional farmers. In addition, the quality of the data that are collected even from state or research farms needs to be improved. Particularly data on reproductive traits and survival rates as well as on post-weaning and cow weights need to be collected. For the time being weaning and yearling weight may be used as a selection criteria for growth traits.
There is a need for research to assess the effect of genetic improvement in growth traits on adaptability, mature size and survival of indigenous cattle in variable environments. In addition, studies to identify indicator traits of growth or milk yield performance which are easy and cheap to measure should be considered. The possibility of increasing growth without concomitant increase in mature size is worth investigating.
The importance of indigenous animals as dam lines for crossbreeding for milk production and as major source of beef and draught power should also be emphasised. This also involves the identification of breeds which should be given priority for improvement, Furthermore, the advantages of crossbreeding between the indigenous breeds should be carefully evaluated.
Milk production, particularly in the vicinity of large population centres, will be largely based on crossbred or grade cattle. Considering the availability of different services and also the level of infrastructure, improvement of both indigenous and crossbred populations would probably be based on nucleus breeding systems. The establishment of such nucleus herds should start in some selected areas on a limited scale.
Ina easing the efficiency of the Al programme should be given attention. Selective insemination of fanners' animals could have a much greater impact on increasing the crossbred dairy cattle population than using ranches. However, in areas with limited feed supply and/or serious disease situations these problems need to be con sidered in parallel with the introduction of crossbred or temperate cattle.
Finally, the success of any breed improvement work in Ethiopia, where there are extreme variations in livestock production system, climate and access to services, will depend on its flexibility. The programme should be assessed carefully before implementation and periodically thereafter. Moreover, it is important to recognize that! for any breeding programme to be accepted by the small traditional farmers in the highlands it should also satisfy their need which is primarily draught power. Therefore the challenge for those who recommend crossbreeding for dairy production in the Ethiopian Highlands could be to balance these two important functions of cattle.
30 ACKNOWLEDGEMENTS
Financial support for this study was received from the Swedish Agency for Research Cooperation with Developing Countries (SAREC). The assistance of the following organisations towards the completion of this thesis is acknowledged: - the Ethiopian Animal and Fishery Resources Development Main Department; - the Department of Animal Breeding and Genetics , SUAS; - the Alemaya University of Agriculture, Department of Animal Science; - the International Livestock Centre for Africa.
I wish to express sincere gratitude to: - Dr Jan Philipsson, my supervisor, for reviewing the papers included in the thesis and the thesis itself and for his encouragement and assistance; - Prof. Birgitta Danell, Head of the Department of Animal Breeding and Genetics, for her interest and assistance in many ways; - Prof. Jan Rendel for his support and for reviewing the thesis; - Dr Eskil Brannang for his support and useful comments on the earlier version of the thesis; - Dr Birgitta Malmfors for her support and interest; - Ms Siw Karlsson for her support in many ways and for excellent secretarial assistance while preparing the thesis. - Mr Max Brandt for linguistic revision.
The assistance and cooperation of colleagues at the Department of Animal Breeding and Genetics is appreciated. Particular thanks go to Mr Anders Dahlin, Dr Thore Henningsson, Dr Thorvaldur Arnason, Mr Agust Sigurdsson, Mr Lieuwe Appel, Dr Erling Strandberg, Dr Georgios Banos, Dr Feng Gu, Mr Rolf Lund, Mr Dan Englund, Mr Peter Bodin and Mr. Hailu Kassa-Mersha.
In Ethiopia I wish to express sincere gratitude to the former and present staff members of the Department of Animal Science of AUA. In particular many thanks to Dr Kano Banjaw for his interest and cooperation. During my recent field tour in Ethiopia I received valuable assistance from numerous people, including: Dr Goshu Mekonnen, Mr Adnan Beker, Mr Sisay Asers and Mr Sisay Gezehagn at Alemaya; Dr Tekaligne Mamo, Dr Daniel Kifitasa, Mr Teshome Shenkoru, MS Alem-Eshet and Ms Ayalenesh at Debre-Zeit; Mr Hizikias Ketema and other staff members at the Ministry of Agriculture in Addis Ababa and at Abemossa, Gobe, Wolaita, Fichie, Asela and Debre-Berhane; Mr Biruke Yemane, Mr Gebre-Micheal Meles, Mr Tesfaye and Dr Emiru Zewde at the National Artificial Insemination Centre; Mr Abebe Lemma at DDE, Addis Ababa; Mr Abdinasir Ibrahim at Awassa.
Finally, the support and interest of my family, relatives and friends generally in my education was a source of inspiration to me to complete this thesis.
31 REFERENCES
Alberro, M. 1986. Ethiopia: The Boran cattle and their tribal owners. World Anim. Rev. 57:30-39. Alberro, M. and Haile-Mariam, S. 1982. The indigenous cattle of Ethiopia. World Akim. Rev. 41;42: 2-10; 27-34. Alexander, G.L and Tierney, M.L. 1990. Selection methods in the development of the AFS breed of tropical dairy cattle. Proc. 4th World Congr. Genetics Applied Livestock P rod., July 23-27 1990, Edinburgh, Scotland, 14:263-266. Blackburn, H.D. and Cartwright, T.C. 1987. Simulated production and biological efficiency of sheep flocks in a shifting environment. }. Anim. Sci 65:399-408. Brannang, E. and Persson, S. 1990. Ethiopian Animal Husbandry: A Handbook. SLU, Uppsala, Sweden. Chauhan, V.P.S. 1992. A review of the procedures used for genetic evaluation of cattle and buffalo bulls in India. World Rev. Anim. Prod. 27:(2)82-90. Dairy Development Enterprise (DDE), 1993. Annual Report for 1984-85 (Ethiopian calendar), DDE, Addis Ababa. Dempfle, L. 1992. Report on genetic improvement of trypanotolerant livestock in west and central Africa. Proc. of the FAO expert consultation on the genetic aspects of trypanotolerance. 3-4 Sept., 1991. FAO animal production and health paper 94. Rome, pp 77-97. FAO (Food and Agriculture Organization). 1992. Production Year-book, 1991, Vol. 45, Rome. FAO/UNEP. 1980. Technical Consultation on Animal Genetics Conservation and W Management. FAO, Rome. Fraiildin, I.R. 1986. Breeding ruminants for the tropics. Proc. 3rd World Congr. Genetics Applied to Livestock Prod., July 16-221986, Lincoln, Nebraska, 11:451- 461. Frisoh, J.E. and Vercoe, J.E. 1984. An analysis of growth of different cattle g motypes reared in different environments. J. Agric. Sci. Camb. 103:137-153. Gali na, C.S. and Arthur, G.H. 1989. Review of cattle reproduction in the tropics. Parts 1 and 2. Anim. Breed. Abstr. 57:583-590; 679-686. Getahun, A. 1978. Zonation of the highlands of tropical Africa: The Ethiopian highlands. Working paper, ILCA, Addis Ababa. Gregory, K.E., Trail, J.C.M., Sandford, J. & Durkin, J. 1984. Crossbreeding cattle in beef production programmes in Kenya I. Comparison of purebred Boran and Boran crossed with the Charolais, Ayrshire and Santa Gertrudis breeds. Trop. Anim. Hlth and Prod., 16:181-186. Gryseels, G. and Anderson, F.M. 1983. Research on farm and livestock productivity in the central Ethiopian highlands: Initial results, 1977-1980. ILCA, Addis Ababa. Hai e-Mariam, M., 1993. Productivity of Boran cows calving Boran and Ft Friesian c lives at Abemossa Ranch, Ethiopia. Proc. VII World Conference Animal Production, June 28 - July 2, 1993, Edmonton, Alberta, Canada, vol. 2:208-209. Haile-Mariam, M. and Mekonnen, G. 1987. Reproductive performance of Fogera cattle and their Friesian crosses. Eth. J. Agric. Sri. 9:95-114. Harvey, W.R. 1977. User's guide for least-squares and maximum likelihood computer program. Ohio State University, Columbus, USA. Hassen, Y. 1994. Evaluation of the performance of Danish Jersey cattle at Ada Berga State Farm, Ethiopia. M.Sc. Thesis. Alemaya University of Agriculture, Dire-Dawa, Ethiopia. Hayman, R.H. 1974. The development of the Australian Milking Zebu. World Anim. Rev. 11:31-36. IAR (Institute of Agricultural Research). 1972. Progress report for the period of April 1971 to March 1972. IAR, Addis Ababa, p. 153. Kachman, S.D., Baker, R.L. and Gianola, D. 1988. Phenotypic and genetic variability of estimated growth curve parameters in mice. Theor. Appl. Genet. 76:148-156. Karlsson, U. 1979. Correlated responses of selection for growth rate in Swedish dual-purpose cattle breeds. I. Consequences in mature cow size. Acta Agric. Scand. 29:295-303. Kasonta, J.P. and Nitter, G. 1990. Efficiency of nucleus breeding scheme in dual purpose cattle in Tanzania. Anim. Prod. 50:245-251. Kebede, B. 1992. Estimation of additive and nonadditive genetic effects for growth, milk yield and reproduction traits of crossbred (Bos taurus x Bos indicus) cattle in the wet and dry environments in Ethiopia. PhD. Thesis, Cornell University. Ithaca, NY. pp. 235. Kiwuwa, G.H., Trail, J.C.M., Kurtu, M.Y., Worku, G., Anderson, F.M. and Durkin, J. 1983. Crossbred dairy cattle productivity in Arsi Region, Ethiopia. ILCA Research Report No. 11. ARDU and ILCA, Addis Ababa, Ethiopia. Koots, K.R., Gibson, J.P., Smith, C. and Wilton, J.W. 1994. Analyses of published genetic parameter estimates for beef production traits. 1. Heritability. Anim. Breed. Abstr. 62:309-338. Kropf, W. and Chacko, C.T. 1992. Merits of using exotic cattle breeds for performances improvement in the tropics. 43rd Annual meeting of the EAAP, Madrid, 14-17 Sept., 1992. Lee, G.J. and Haley, C.S. 1990. Body weight adjusted testis size as a selection criterion to improve production efficiency in sheep. Proc. 4th World Congr. Genetics Applied Livestock Prod., July 23-271990, Edinburgh, Scotland, 16:370- 373. Lindstrom, U.B. 1976. Milk recording in developing countries. World Anim. Rev. 19: 34-42. Luesakul-Reodecha, C., Martin, T.G. and Nelson, L.A. 1986. Effects of long-term selection for 365-day weight on maternal performance of beef cows. Proc. 3rd World Congr. Genetics Applied Livestock Prod., July 16-22 1986, Lincoln, Nebraska, 12:205-209. MacNeil ,M.D., Cundiff, L.V., Dinkel, C.A. and Koch, R.M. 1984. Genetic correlations among sex-limited traits in beef cattle. /. Anim. Sci. 58:1171-1180. IMI ------"" I Institute of Agricultural 33 L i t> • » y Madalena, F.E. 1989. Cattle breed resource utilization for dairy production in Brazil. Rev. Brasil Genet. 12, (3-Suppl.): 183-220. Madalena, F.E.; Teodoro, R.L.; Lemos, A.M.; Monteiro, J.B.N. and Barbosa, R.T. 1990. Evaluation of strategies for crossbreeding of dairy cattle in Brazil. J. Dairy S(k. 73:1887-1901. Mahadevan, P., Galukande, E.B. and Black, J.G. 1962. A genetic study of the Sahiwal grading-up scheme in Kenya. Anim. Prod. 4:337-342. McDowell, R.E. 1985. Crossbreeding in tropical areas with emphasis on milk, health and fitness. /. Dairy Sci., 68:2418-2435. Menissier, F. and Frisch, J.E. 1992. Genetic improvement of beef cows. In: Jarrige, R. and BGranger,C. (eds). Beef cattle production. World Anim. Sci. C5. Elsevier, Amsterdam, pp. 55-82. Meyer, K. 1989. Restricted maximum likelihood to estimate variance components for animal models with several random effects using a derivative-free algorithm. Genetic Selection and Evolution 21:317-340. Meyier, K., 1993a. DFREML version 2.1 Programs to estimate variance components by restricted maximum likelihood using a derivative-free algorithm. User notes. Animal Genetics and Breeding Unit, University of New England, Armidale, NSW. Mimeo. Meyer, K., 1993b. Estimates of direct and maternal correlations among growth triits in Australian beef cattle. Livest. Prod. Sci. 38:91-105. Meyer, K., Hammond, K., Parnell, P.F., Mackinnon, M.J. and Sivarajasingam, S. 1^90. Estimates of heritability and repeatability for reproductive traits in Australian beef cattle. Livest. Prod. Sci. 25:15-30. MOA (Ministry of Agriculture) 1984. Livestock sub-sector review. Main report. JV inistry of Agriculture, Addis Ababa. Mpofu, N., Smith, C. and Burnside, E. B. 1993a. Breeding strategies for genetic in iprovement of daily cattle in Zimbabwe. 1. Genetic evaluation. /. Dairy Sci. 76: 1163-1172. Mpofu, N., Smith, C., Van Vuuren and Burnside, E. B. 1993b. Breeding strategies for genetic improvement of dairy cattle in Zimbabwe. 2. Economic evaluation. }. Dairy Sci. 76:1173-1181. Mudgal, V.D. and Arora, C.L. 1994. Frieswal project: present status and expectations for the future. World Anim. Rev. 79:(2)46-50. Mukasa-Mugerwa, E., Bekele, E. and Tessema, T. 1989. Type and productivity of indigenous cattle in central Ethiopia. Trop. Anim. Hlth. Prod. 21:120. Negash, M. 1994. Reproductive performance of a Holstein-Friesian dairy cattle at Holetta, Shoa, Ethiopia. M.Sc. Thesis. Alemaya University of Agriculture, Dire- Dawa, Ethiopia. Negussie, E. 1992. Reproductive performance of local and crossbred dairy cattle at the Asella Livestock farm, Arsi, Ethiopia. M.Sc. Thesis. Alemaya University ojf Agriculture, Dire-Dawa, Ethiopia. Notter, D.R. 1988. Evaluating and reporting reproductive traits. Proc. Beef L improvement Federation, Ann. Conference, Albuquerque, Beef Improvement Federation, pp. 21-42. O'Donovan, P.B., Gebrewolde, A., Kebede, B. and Gallal, E.S.E. 1978. Fattening studies with crossbred (European x Zebu) bulls. 1. Performance on diets of native hay and concentrate. J. Agric. Sci. Camb. 90: 415-429. Philipsson, J. 1992a. Sustainable dairy cattle breeding systems for less developed countries - examples of problems and prospects. 43rd Annual Meeting of the EAAP, 14-17 Sept., 1992, Madrid. Philipsson, J. 1992b. Practical issues for the conservation and improvement of priority breeds: A global review of the genetic resources of cattle. Expert consultation on the management of global animal genetic resources. 3-4 April., 1992. FAO animal production and health paper 104, Rome, pp. 129-155. Peters, K.J. 1992. Experience in the development of livestock farming systems in Sub-Saharan Africa. 43rd Annual Meeting of the EAAP, 14-17 Sept., 1992, Madrid. Rege, J.E.O. and Baker, R.L. 1993. Conservation of Indigenous African breeds. Issues and Options. International Conf. on the Convention on Biological Diversity: National interests and global imperatives. Jan. 26-29, 1993, Nairobi, pp. 1-14. Renand, G., Plasse, D. and Andersen, B.B. 1992. Genetic improvement of cattle for growth and carcass traits. In: Jarrige, R. and Beranger,C. (eds). Beef cattle production. World Anim. Sci.. C5. Elsevier, Amsterdam, pp.87-104. Robertson, A. 1982 . Genetics aspects of growth. Proc. World Congr. Sheep and Beef Cattle Breeding, vol. 1, 19-23 June, 1982, New Zealand, pp. 427-437. Schaar, J., Brannang, E. & Meskel, L.B. 1981. Breeding activities of the Ethio- Swedish integrated rural development project. 2. Milk production of zebu and crossbred cattle. World Anim. Rev., 37:31-36. Slingenbergh, J. 1992. Tsetse control and agricultural development in Ethiopia. World Anim. Rev. 70-71:30-36. Smith, C. 1988. Genetic improvement of livestock, using nucleus breeding units. World Anim. Rev. 65:2-10. Stotz, D. 1979. Smallholder dairy development in past, present and future in Kenya. Doctors Dissertation, Hohenheim University, Hohenheim. Syrstad, O. 1988. Crossbreeding for increased milk production in the tropics. Norwegian Agricultural Sciences 2:179-185. Syrstad, 0 . 1990. Dairy cattle breeding in the tropics: the importance of genotype x environment interactions. Livest. Prod. Sci. 24:109-118. Tacher, G. and Jahnke, H.E. 1992. Beef production in tropical Africa. In: Jarrige, R. and Beranger,C. (eds). Beef cattle production. World Anim. Sci. C5. Elsevier, Amsterdam, pp. 419^436. Taneja, V.K. 1990. Advances in dairy cattle genetics and breeding -system analysis of animal breeding work in tropics. Proc., 4th World Congr. Genetics Applied to Livestock Prod., July 23-27 1990, Edinburgh, Scotland, 14:365- 368. Thorpe, W., Cruickshank, D.K.R. and Thompson, R. 1981. Genetic and
35 environmental influences on beef cattle production in Zambia. 4. Weaner Production from purebred and reciprocally crossbred dams. Anim. Prod., 33:165- 177. Tilahun, G. 1990. Pathogenesis of Dictyocaulus filaria in Ethiopian sheep: Effect of repeated infections. Eth. J. Agric. Sci. 1253-62. Timon, V.M. 1993. Strategies for sustainable development of Animal agriculture - An FAO prospective. Proc. of the FAO Expert Consultation held in Rome, Italy, 10-14 Dec. 1990. FAO Anim. Prod, and Health paper 107. Trail, J.C.M. & Gregory, K.E. 1981. Sahiwal Cattle: An Evaluation of their potential contribution to milk and beef production in Africa. ILCA Monograph 3, ILCA, Addis Ababa, Ethiopia. Trail, J.C.M., Gregory, K.E., Durkin, J. and Sandford, J. 1984. Crossbreeding cattle in beef production programmes in Kenya. II. Comparison of purebred Boran and Boran crossed with the Red Poll and Santa Gertrudis breeds. Trap. Anim. Hlth and Prod., 16:191-200. Vaecaro, L.P.de. 1990. Survival of European daily cattle breeds and their crosses with Zebu in the tropics. Anim. Breed. Abstr. 58:475-494. Vaccaro, Lv Vaccaro, R., Verde, O., Mejias, H, P6rez, A., Rios, L., and Romero, E. 1994. An improvement program for tropical dual purpose cattle. Proc., 5th World Congr. Genetics Applied Livestock Prod., Aug. 7-12 1994, Guelph, Ontario, Canada. 20:313-318. Wagner, D.G., Holland, G.L., Mogess, T. 1969. Crossbreeding studies with Ethiopian beef cattle using improved semen. East African Agricultural and Fwestry J. 34: 426-432. Zewde, E., Sinegiorgis, Mv Yemane, B. and Mune, A. 1994. Production of bull semen at Kaliti. Draft report National Artificial Insemination Centre, Kaliti, Ethiopia. I r
r Trop. Anim. Hllh Prod. (1993) 25, 239-248 PRODUCTIVITY OF BORAN CATTLE AND THEIR FRIESIAN CROSSES AT ABERNOSSA RANCH, RIFT VALLEY OF ETHIOPIA. I. REPRODUCTIVE PERFORMANCE AND PRE-WEANING MORTALITY
M. Haile-M ariam1, K. Banjaw, T. Gebre-Meskel2 and H. Ketema 2
Alemaya University of Agriculture, PO Box 138, Dire Dawa, Ethiopia
SUMMARY Data collected on number o f services per conception (NSC), age at first calving (AFC), calving interval and calf pre-weaning mortality rate of Boran cattle and their Friesian crosses were evaluated. The NSC was 1-81, 161 and 169 for Boran, Fi Boran-Friesian and three-quarter Friesian heifers respectively. The AFC varied from 31-5 months in the Fj crosses to 46-8 months in the Boran heifers artificially inseminated with Friesian semen. The mean calving intervals were 465, 552, 525 and 487 days for Boran cows naturally mated to Boran and Friesian bulls, Boran and Fj c o w s artificially inseminated with Friesian semen respectively. The pre- weaning calf mortality rate was 3-4% and was significantly influenced by parity, sex and interaction of breed with year of birth. Year as well as season had in general significant effects on the traits considered. However, the variation between mating or breeding groups was also large and it was in favour of the Ft crosses.
INTRODUCTION The Abernossa Boran cattle improvement programme was launched in Ethiopia by the Ministry of Agriculture in 1959 with the purchase of about 350 female and 11 male Borans from the Sidamo Administrative region. In 1972, a crossbreeding pro gramme between Boran and Friesian breeds was added with the objective of produc ing F| crossbred dairy heifers. A considerable amount of information was collected by the Ranch on the per formance of both the Boran and its crosses over the years. However, the evaluation of the comparative performance of purebred and crossbred animals had not been undertaken. Kassa-Mersha and Arnason (1986) and Arnason and Kassa-Mersha (1987) reported non-genetic factors and genetic parameters of growth of the pure bred Boran. The present study reports factors affecting the reproductive performance o f Boran cattle and their crosses when artificial and natural service are used. In addition the pre-weaning calf mortality rate of Boran and their Friesian crosses at the Ranch was analysed.
MATERIALS AND METHODS Study area The Abernossa Ranch is located about 180 km south of Addis Ababa in the Rift Valley, at a latitude of 7 015'N and a longitude of 38°45'E at an altitude of 1,700 metres.
’Addressee for reprints: Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, S 750 07 Uppsala, Sweden. 2Ministry of Agriculture, Addis Ababa, Ethiopia.
239 240 HAILE-MARIAM ET AL.
The average annual rainfall of the area is 824 mm (1977 to 1985) but it was below 600 mm in 1980,1984 and 1985. The main wet season during which 54% of the annual precipitation occurs is between July and October. Thirty seven percent of the rain falls between March and June (short wet season), while the months of November to February are the dry season. The mean maximum and minimum temperatures are 27*1° and 12 0°C respectively, with the highest temperatures occuring between March and June.
Herd management Gows in the cross breeding unit (CBU) and Boran breeding unit (BBU) were managed similarly with the exception of the difference in the mating system. In the BBU, Boran bulls were run with Boran cows all the year round at the ratio of one bulljper 50 cows. However, from 1981 a seasonal mating scheme was introduced (September to December). Boran heifers which served as replacements in both units were available for mat ing when they were 24 months, whereas F; Boran X Friesian heifers were bred by artificial insemination (Al) at 18 months of age. In both cases the heifers needed to attain a minimum body weight of 250 kg. However, this plan was not adhered to. The ;F[ crosses were back-crossed to the Friesian through Al. They were then preg- nancty tested and sold between 3 and 6 months of pregnancy. All calves were allowed to suckle their dams up to approximately 8 months of age. After weaning animals werei kept in groups based on sex and age and were allowed to graze on natural pas ture (without supplementation. However, some concentrates (oilseed cake and wheat bran) were available during years of critical feed shortage.
Data analysis Data collected between 1977 and 1985 on the reproductive performance (age at first calving, calving interval and number of services per conception for first preg- nandy) and pre-weaning calf mortality rate, of both Boran and their Friesian crosses, were analysed by least-squares procedures (Harvey, 1977). As cows were arti- ficia ly inseminated as well as naturally served by Boran or Friesian bulls, the data on calvi qg interval and age at first calving (AFC) of each breed and mating group were analysed separately: 1. B)ran cows naturally mated to Boran bulls (B-NS-B). 2. B nan cows naturally mated to Friesian bulls (B-NS-F). 3. Boran cows artificially inseminated by Friesian semen (B-AI-F). 4. F 1 cows artificially inseminated using Friesian semen (Fj-A I-F). 5. Tiree-quarter Friesian heifers artificially inseminated using Friesian semen (three- q iarter F-A I-F ) for AFC only. The model used included the effects of year and season for all traits. In addition the effet ts of parity in case of calving interval, and breed in case of number of services per conception for first pregnancy (NSC) were considered. The model for pre-weaning calf mortality (all recorded deaths and stillbirths) included year and season of birth, parity, sex, breed and its interaction with year. The residual mean square was used as the error term to test the significance of differences evaluated among groups. Linear contrasts of least-squares means were computed to determine the significance of differences within groups for all characters. CATTLE PRODUCTIVITY IN ETHIOPIA 241
RESULTS Number of services per conception for first pregnancy (NSC) Although breed did not have a significant effect on NSC, Boran heifers required 0-2 more services than the F! crosses. This difference approached significance (P < 010). Heifers which conceived in the main wet season (July to October) required sig nificantly (P < 0-05) fewer NSC than those conceiving during the other seasons. Year of conception had a significant effect(P < 0-01) on NSC (Table I).
AFC The difference in AFC of the 4 breeding groups (Table II) was marked. The cross breeds (F] and three-quarter) which were artificially inseminated calved 13 to 15 months earlier than the Boran heifers. In a separate analysis, which included all Boran heifers, B -A I-F heifers calved at 46-9 months of age, whereas B-NS-B were 45-0 months old at first calving, showing statistical difference (P < 0 05) due to mating system. The F, crosses born during the short wet season were significantly (P < 0-01) younger at first calving by 1-2 and 1-4 months than those born during the main wet and the dry seasons, respectively.
Calving interval Calving interval was significantly influenced by year of calving (P < 0 01) in all
T able I
Estimated least-square means and standard errors (s.e.) for number of services per conception for first pregnancy
'Variable No M ean ±s.e.
Overall mean 835 1-70 ± 0 07
Breed group * Boran 140 1-81 ± 0 1 0 F, Crosses 652 1-61 ± 0 0 5 | Friesians 43 1 -69 ± 0 1 6
Season of conception ** M ain wet 358 1 -58 ± 0 08a Short wet 252 1 -74 ± 0-10b D ry 225 1 -79 ± 0-09b
Y ear o f conception *** 1978 19 1 -22 ± 0-25ab 1979 56 1-97 ± 0-15C 1980 111 1-45 ± 0-12ab 1981 212 1-45 ± 0-09ab 1982 186 191 ± 0-09c 1983 62 2-49 ± 0-14d 1984 106 l-72±0-llbc 1985 83 1-38 ± 0-133
Within variable groups, means with the same letter do not differ significantly (P > 0-05). * = P < 0-10, ** = P < 0 05, *** = P < 0 01, 242 HAILE-MARIAM ET AL.
Table II
Least-square means and standard errors (s.e.) for age al first calving (months) i i B-NS-BB-AI-F F, -AI-F |F-AI-F
Variable N o Mean ± s.e. No Mean ± s.e. N o Mean ± s.e. No Mean ± s.e.
Overall mean 242 45-2 ± 0-4 188 46 8 ± 0-8 602 31-5 ± 0-2 42 32-7 ± 0-9
Season o f birth n.s. n.s. ** n.s. Main wet 110 45-3 ± 0-5 82 45 8 ± 0 -9 196 31-7 ±0 3b 11 34-4 ± 1-7 Short wet 44 45-7 ± 0-8 71 46-7 ± 10 183 30-7 ± 0-3a 19 33-4 ± 1 -3 Dry 88 44-7 dr 0-6 35 4 8 0 ± 1-2 223 32-1 ±0>3b 12 30 6 ± 1-6
Year of birth ** n.s. *** 1977 41 42-4 ± 0-8a 6 5 2 0 ± 2 -6 64 3 4 0 ± 0-6b 4 38-4 ± 2-7b 1978 37 46-9 ±0 8c 37 46-9 ± 10 91 306 ± 0-5a 6 32-5±2-3b 19|79 33 47-2 ± 0-9c 62 44-5 ± 0-8 150 30 6 ± 0-4a 11 32-4 ± l-6b 1980 47 4 5 -5 ± 0 -8 bc 57 45-5 ± 0 -9 143 30-2 ± 0-4“ 11 33-4 ± l<6b 1981 54 43-7 ± 0-7ab 21 45 0 ± 1-4 52 30-3 ± 0-6a 4 25-4 ± 2-7a 1982 30 45-6 ± 1-0** 5 47-0 ± 2-9 102 33-2 ± 0-4b 6 34 0 ± 2-3b
With in variable groups, means with the same letter do not differ significantly (P > 0 05). n.s. ■= not significant. * —P < 0 05. ** = P < 0 01. mating groups but the effect of calving season was significant only in Boran cows whibh were naturally mated to Boran bulls and those artificially inseminated. Calving interval generally decreased for all cows except the F l} as parity increased from the first to the ninth and later parturitions. However, there was a slight rise in calving intervals of B-NS-B and B-N S-F cows calving for the fifth and sixth timi; (Table III).
Preweaning calf mortality rate Iihe overall calf mortality rate was 3-4% and was significantly (P < 0*01) affected by parity, sex and the interaction of breed with year of birth (Table IV). Iihe mortality rate of calves born to first calvers was significantly (P < 0 01) higher by 2-7 to 3*2% than the other parities. Also the mortality rate of male calves was significantly (P <0*01) greater by 1*7% units than that of female calves.
DISCUSSION NSC As in the present study Tegegne et al. (1981) and Alberro (1983) observed fewer NSC for crosses as compared with zebu breeds of Ethiopia. The observed increase in NSC for those which conceived during the dry and short wet seasons may be reU.ted to the occurrence of high temperatures and feed scarcity. A similar increase in WSC during the dry months of the year was reported by Tegegne et al. (1981) anc Swensson et al. (1981) working on zebu and zebu-temperate crosses in Ethio pia, Thatcher (1974) also showed that high temperature decreases fertility and embryonic viability thereby increasing the NSC. The marked differences between the years could be due to several factors such as variation in the amount of rainfall through its effects on feed supply, availability of feed supplementation, temperature CATTLE PRODUCTIVITY IN ETHIOPIA 243
T able III
Least-squares means and standard errors (s.e.) for calving interval (days)
B-NS-B B-NS-F B--AI-F F, - A I - F
V ariable No Mean ± s.e. No Mean ± s.e. No Mean ± s.e. No Mean ± s.e.
Overall mean 1289 465 ± 4 505 552 + 1 7 1181 525 ± 5 91 487 ± 19
Parity of cow ** ** ** n.s. 1 227 511 ± 8 a 12 601 ± 3 2 b 153 567 ± 10c 21 487 ± 3 1 2 199 468 ± 9b 65 567 ± 20b 179 538 ± 9b 18 551 ± 3 4 3 -4 375 437 ± 5a 206 522 + 18a 371 514 ± 7 b 25 496 ± 29 5 -6 261 451 ± 8ab 155 528 ± 17a 332 500 ± 7ab 20 531 ± 30 7 -8 155 445 ± 10ab 57 512 ± 20a 138 489+ IIa 7 425 ± 52 9 + 72 427 ± 15a 10 509 ± 35a 8 448 ± 42a - -
Season of calving ** n.s. ** n.s. Main wet 636 444 ±5a 149 558 ± 18 457 511 ± 6 a 15 472 ± 38 Short wet 257 479 ± 8b 176 559 ± 18 338 523 ± 7ab 31 518 ± 2 8 Dry 396 473 ± 6b 180 541 + 19 386 540 ± 7b 45 471 ± 2 7
Y ear o f calvine ** ** ** ** 1977 156 488 ± 10b 220 494 ± l l b 267 488 ± 8a 30 414 ± 30a 1978 189 446 ± 9a 121 4 5 6 + 12a 198 491 ± 10ab 11 450 ± 45a 1979 256 445 ± 8 a 81 465 ± 14a 142 491 ± l l ab 11 431 ± 4 3 a 1980 211 488 ± 6b 63 620 ± 15° 100 613 ± I2d 10 583 ± 42a 1981 195 474 ± 9b 4 691 ± 6 9 ° 96 506 ± 13b 7 455 ± 55a 1982 119 446 ± 12a 8 482 ± 35ab 207 547 ± 8C 11 528 ± 4 1 ab 1983 163 471 ± l l b 8 558 ± 3 5 b 171 537 ± 9C 11 548 ± 40b
Within variable groups, means with the same letter do not differ significantly (P > 0 05). ** = P < 0 01, n.s. = not significant. and the efficiency of the inseminators. The effect of feed supplementation and temperature on NSC was reported by Swensson et al. (1981) and Thatcher (1974) respectively. However, a drawback of measuring fertility as NSC is that animals which do not conceive are excluded. If the proportion of such animals vary among breed groups the differences in NSC may not exactly reflect the fertility rate. The above results should be interpreted with some caution because of this.
AFC The F t and three-quarter calved about a year earlier than the Boran, indicating there is a positive effect of crossbreeding and appropriate mating system on AFC. Three-quarter Friesian heifers had a longer AFC than the F[ crosses (Table II) which is in agreement with reports by Osman and Russell (1974) in Sudan and Katpatal et al. (1978) in India. The AFC in this study for the crossbreeds (Fj and three-quarter) compares well with results in Ethiopia by Swensson et al. (1981) and Alberro (1983). However, the AFC of Boran heifers both naturally served and artificially inseminated was higher than values reported by Gregory et al. (1984) for Boran cattle in Kenya. Kassa-Mersha and Amason (1986) working on Boran cattle born at Abemossa estimated that heifers had an average AFC of 41-5 months. This may sug gest lowering the AFC of Boran heifers by reducing the age of first service and by improving the general management. This is substantiated by Boran heifers attaining 244 HAILE-MARIAM ET AL.
Table IV
Least square means and standard errors (s.e.) for pre-weaning calf mortality rate (% )
Variable N o. Mean ± s.e.
Overall mean 2909 3-4 ± 0 4
Breed of calf * Boran 1377 4 0 ± 0-5 Fj crosses 1532 2-9 ± 0 -5
Parity of calf ** 1 488 5-6 ± 0-8b 2 395 2 - 9 ± 0 9 a 3-4 703 2-7 ± 0‘5a 5 -6 642 2-5 ± 0-6“ 7 + 681 2-4 ± 0-7a Season of birth * Main wet 1384 2-3 ± 0-5 Short wet 676 3-9 ±0 6 Dry 849 4-0 ±0 7
Year of birth n.s. 1979 543 4 -5 ± 0 '8 1980 597 3-6 ± 0 -7 1981 328 1-7 ± 1-0 1982 374 4-3 ± 0-9 1983 372 3-2 ± 0 -9 1984 365 3*1 ± 0 -9 1985 330 3-5 ± 0-9
Sex o f calf ** Female 1497 2-6 ± 0-5a Male 1412 4-3 ± 0-5b
Within variable groups, means with the same letter do not differ significantly (P > 0-05). * = P < 010, ** = P < O'Ol, n,s. = not significant. a body weight of 244 kg at 2 years of age (Haile-Mariam, 1987). Boran heifers may thu$ be mated at 2 years in order to calve at between 3 and 3*5 years. The fact that heifers born during the short wet season calved earlier than the aver- agelAFC might be due to the 13-3 and 21 -8 kg heavier weaning weight of calves born during this season than those born during the dry and the main wet seasons respectively (Haile-Mariam, 1987). This argument is supported by McDowell and Leining (1978) who observed that AFC is influenced by the onset of puberty which itself is affected by both the physiological and anatomical development (weight) of the janimal. Although the variation between the years in AFC was highly significant (Table II), the effect of birth year on AFC did not show any consistent trend across the jdifferent mating groups. Factors such as inconsistency of age at first breeding, which varied between 18 and 36 months, scarcity of feed and other managerial differ ences might have contributed to the variations in AFC over the years.
Calving interval The calving interval values for B-NS-B and Fj cross cows (Table III) are in gen CATTLE PRODUCTIVITY IN ETHIOPIA 245 eral agreement with the findings of Osman and Russell (1974) and Trail et al. (1985). The values for B -N S -F and B -A l-F were close to those reported by Ward et al. (1988) for Kenana cattle in the Sudan and lie within the ranges of those given by Rein hardt (1982) for Boran cattle in Kenya. Trail and Gregory (1981), Gregory et al. (1984) and Trail et al. (1984) found mean calving intervals of 413, 412, and 421 days respectively for Boran cattle in Kenya. The longer calving intervals of Fi cows compared to that of B-NS-B cows (Table III) may be attributed more to the inadequacy of the insemination programme (poor oestrus detection, improper time of insemination etc.) than to the breed since shorter intervals for the F! Friesian-zebu crosses than for zebu have been reported by Tegegne et al. (1981) and Alberro (1983) in Ethiopia. The long intervals between calvings of artificially inseminated Boran cows (Table III) may be related to diffi culties in heat detection and occurrence of silent as well as night heats coupled with short heat periods commonly manifested by zebu cows (Trail et al., 1971). The calving interval of 552 ± 17 days for B -N S-F cows is longer than for those artificially inseminated. This may partly be due to the long calving interval of B -N S-F cows in the first parity and those calving in 1981 which were few in number (Table III). In addition, if artificial insemination failed to get a cow in calf in the CBU she was mated to a Friesian bull naturally and was considered as naturally mated. The decrease in calving interval with parity may be attributed to selective culling of repeat breeders and retaining those with short intervals which is reflected in the small number of cows attaining a high parturition number. The shortest calving interval of B-NS-B and B -A I-F cows calving during the main wet season is due to the availability of adequate pasture during this season, which enables the cows to be in good condition during and after calving for reconcep tion in the following breeding season. Dunn et al. (1969) also reported that low level of precalving and/or postcalving feeding had a significant negative effect on subse quent reproduction. The longer calving interval of cqws of all mating groups calving in 1980 (Table III) could be due to the lower rainfalljhat year. However, the longer calving interval of B-N S-F cows which calved in 1981 is perhaps due to the few observations.
Pre-weaning calf mortality rate The pre-weaning mortality rates are comparable to those reported by Trail and Gregory (1981) and Trail et al. (1984) for Boran cattle in Kenya. The 1-1% higher survival rate of Fj calves was marginally significant (P < 0-10) and it was probably due to the effect of individual heterosis for the trait (Preston and Willis, 1970). Trail and Gregory (1981) showed that the viability of Sahiwal calves up to weaning was 56% and was significantly (P < 0 05) lower than the 80 to 90% viability of Ayrshire and Ayrshire-Sahiwal crosses. The reason for the higher mortality of calves of first parity could be due to a higher frequency of dystocias and stillbirths among calves from young cows and also as a consequence of their sometimes inadequate milk pro duction compared to mature cows. The higher mortality rate (Table IV) of calves born during the generally dry months (dry and short wet) is probably caused by the tendency of zebu dams to reduce or terminate milk supply in times of stress (Willis, 1974). The interaction of breed with year was evident from the increased mortality rate of Boran calves over the years, except during 1985, while that of the crossbreds showed no clear trend. Boran calves born in 1982, 1983 and 1984 had mortality rates of 6-0, 5-0 and 5-5% respectively as compared to the mean of 4 per cent. The high mortality 246 HAILE-MARIAM ET AL. rate of 1982 has no specific explanation but that of calves bom in 1983 and 1984 may partly be because'the low rainfall of 1984 and 1985 resulted in nutritional stress and because of increased calving difficulty and stillbirth rates as a result of increase in birth weight of Boran calves from 24-2 kg in 1978 to 26-0 kg in 1984. During this period birth weight of Fj calves varied between 25.-2 and 25-6 kg (Haile-Mariam, 1987). Laster et al (1973) also reported that each kg increase in birth weight tended to increase dystocia by 2-3% and calf losses within 24 hours post partum were 3-7 times higher in calves experiencing difficulty at birth than those bora without difficulty. The higher mortality rate of male calves might be because their greater size at birth caused dystocia and a higher proportion of calves dying at birth (Preston and Willis, 1970).
| CONCLUSION AFC of crossbreds, NSC of all heifers and the pre-weaning calf mortality rate is comparable with other reports in the tropics. However, the AFC of Boran heifers and the living interval of all cows is relatively long. Both year and season had an import ant influence on calving interval while AFC was influenced by year only. Thus, improvement in both traits could be achieved by adopting better management as well as by crossbreeding. Allowing cows in the BBU to calve during the main wet season has decreased the calving interval by close to 7% compared to those cows calving during the other seasons. Similarly, applying this to cows in the crossbreeding herd and concentrating the effort of inseminators and the management during the 3 to 4 months of the breed ing season could help to decrease the calving interval further. Although it is difficult to make direct comparisons between breeds as they were usually confounded with mating type, the F( crossbreds required fewer NSC, showed relatively lower mortality, lower AFC and shorter calving interval than the pure Boran.
ACKNOWLEDGEMENTS The authors are grateful to Mr A. R. Sayers for his co-operation in the analysis of the data at the International Livestock Centre for Africa, Addis Ababa. The Eth opian Animal and Fishery Resources Development Main Department is acknowledged for allowing us to use the data. The assistance of Dr J. Philipsson and other members of the Department of Animal Breeding and Genetics in the preparation of this manuscript is also acknowledged.
Accepted for publication September 1992
REFERENCES
A lb^ r r o , M. (1983). Comparative performance of F| Friesian x zebu heifers in Ethiopia. Animal Produc tion, 37, 247-252. A r n a s o n , T. & Kassa-M ersha, H. (1987). Genetic parameters of growth of Ethiopian Boran cattle. Animal Production, 44, 201-208. D u n n . T . G ., I n g a l l s , J. E., Z im m e r m a n , D . R. & W il t b a n k , J. N. (1969). Reproductive performance of 2- year-old Hereford and Angus heifers as influenced by pre- and post-calving energy intake. Journal o f Animal Sciences, 29, 719-726. CATTLE PRODUCTIVITY IN ETHIOPIA 247
G r e g o r y . K. E„ T rail. J. C. M .. S a n d f o r d . J. & D u r k in . J. (1984). Crossbreeding cattle in beef production programmes in Kenya. I. Comparison of purebred Boran and Boran crossed with the Charolais, Ayrshire and Santa Gertrudis breeds. Tropical Animal Health and Production, 16, 181-186. H a ile-M a r ia m . M. (1987). Evaluation of reproduction and growth performance of Boran cattle and their crosses with Friesians at Abemossa, Ethiopia. MSc. Thesis, Alemaya University of Agriculture. DireDawa, Ethiopia. H a rv e y . W. R. (1977). User’s guide for least-squares and maximum likelihood computer program. Ohio State University, Columbus, USA. K assa-M e rsh a , H, & A r n a so n . T. (1986). Non-genetic factors affecting growth of Ethiopian Boran cattle. World Review Animal Production, 22, 45-55. K a tpa ta l, B. G., K a u sh ik , S. N. & S in g h a l . R. A. (1978). Factors affecting age at first calving in Holstein- Sahiwal crossbreds at varying levels of Holstein breeding. Indian Journal of Animal Sciences. 48. 245-248. L a ster , D. B„ G l im p. H. A., C u n d if f . L. V. & G r e g o r y . K. E. (1973). Factors affecting dystocia and the effects of dystocia on subsequent reproduction in beef cattle. Journal o f Animal Science, 36. 6S5-705. M cD o w el l. R. E. & L e in in g . K. B. (1978). Handling of European cattle in the tropics for reproductive ef ficiency. Paper presented at II Congresso International Da Raca Chianina, Sao Paulo, Brazil. O sm a n . A. H. & R ussell. W. S. (1974). Comparative performance of different grades of European-zebu crossbred cattle at Ghurashi Dairy Farm, Sudan. Tropical Agriculture, (Trinidad), 51. 549-558. P resto n . T. R. & W illis, M. B. (1970). Intensive beef production. Pergamon Press, Oxford. UK. Reinhardt. V. (1982). Reproductive performance in a semi-wild cattle herd (Bos indicus). Journal o f Agri cultural Sciences. (Cambridge). 98. 567-569. Sw en sso n . C , S c h a a r . J., B r a n n a n g . E. &. M eskel. L. B. (1981). Breeding activities of the Ethio-Swedish integrated rural development project. 3. Reproductive performance of zebu and crossbred cattle. World Animal Review, 38, 31-36. T e g e g n e . A., G a lla l. E. S. E. & K ebed e. B. (1981). A study on the reproduction of local and F, crossbred (European x zebu) cows. I. Number of services per conception, gestation length and days open till conception. Ethiopian Journal o f Agricultural Science, 3. 1-14. T h a t c h e r . W. W. (1974). Effects of season, climate and temperature on reproduction and lactation. Journal o f Dairy Science. 57. 360-368. T rail. J. C. M. & G r e g o r y . K. E. (1981). Characterization of the Boran and Sahiwal breeds of cattle for economic characters. Journal of Animal Science, 52. 1286-1293. T rail. J. C. M., Sa c k e r . G. D. & F ish er. 1. L. (1971). Crossbreeding beef cattle in western Uganda. 1. Performance of Ankole, Boran and zebu cows. Animal Production, 13, 127-141. T r a il. J. C. M., G r e g o r y . K. E.. D u r k in . J. & S a n d f o r d . J. (1984). Crossbreeding cattle in beef production programmes in Kenya. II. Comparison of purebred Boran and Boran crossed with the Red Poll and Santa Gertrudis breeds. Tropical Animal Health and Production, 16, 191-200. T r a il. J. C. M .. M u r r a y . M ., S o n es. K., J ibbo. J. M . C , D u r k in . J. & L ig h t . D . (1985). Boran cattle main tained by chemoprophylaxis under trypanosomiasis risk. Journal of Agricultural Science, (Cambridge) 105, 147-166. W a r d . P. N ., Sa ee d . A. M.. L ig h t . D. & W ilson. R. T. (1988). Reproductive performance of Kenana cows in Sudan. Tropical Agriculture. (Trinidad), 65, 73-76. W illis. M. B. (1974). Beef cattle production in developing countries. (Proceedings). (Ed. A. J. Smith). Centre for Tropical Veterinary Medicine, University of Edinburgh, Scotland.
PRODUCTIV1TE DU BETAIL BORAN ET DES SES CROISEMENTS AVEC LA RACE FR1SONNE AU RANCH D'ABERNOSSA, DANS LA VALLEE DU RIFT (ETHIOPIE). I. PERFORMANCE DE REPRODUCTION ET MORTALITE AVANT SEVRAGE Resume— Les auteurs one estime les donnees obtenues sur le nombre de montes avant la conception, l’age au premier velage. I’intervalle entre deux mises bas et le taux de mortalite avant sevrage sur le betail Boran et les croisements frisons au ranch d’Abernossa (Ethiopie). Le nombre de montes etait de 1,81, 1,61 et 1,69 pour les genisses Boran, F, Boran x frisonne et 3/4 frisonne respectivement. L’age au premier velage a varie de 31,5 mois pour le croisement F| a 46,8 mois pour les genisses Boran artificiellement insemi- nees avec du sperme frison. L'intervalle moyen au velage etait de 465, 552, 525 et 487 jours pour les femelles Boran fecondees en monte naturelle respectivement par des taureaux Boran et Frison, et des vaches Boran et F| inseminees artificiellement avec du sperme frison. Le taux de mortalite avant sevrage etait de 3,4p.l00. II etait in fluence de fa?on significative par le nombre de parturitions anterieures, le sexe et l'interaction raciale 248 HAILE-MARIAM ET AL. avec Tannee de naissance. L’annee aussi bien que la saison, avaient en regie generate des effets signifi- catifs sur les caracteres consideres. Neanmoins la variation entre les groupes de monte et de naissance etait jegalement large mais jouait en faveur du croisement F(. I PROtJUCTIVIDAD DE GANADO BORAN Y SUS CRUCES CON FRIESIAN EN EL RANCHO DE ABERNOSSA EN EL VALLE DE RIFT DE ETIOPIA. I. DESEMPENO REPRODUCCTIVO Y MORTALIDAD PREDESTETE Resumen— Se evaluaron los datos sobre el numero de servicios por concepcion (NSC), edad a la primera paricion (EP), intervalo entre partos y tasa de mortalidad predestete, del ganado Boran y sus cruces con Friesian. El numero de servicios por concepcion fue 1-81, 1*61 y 1-69 para Boran, Fj Boran/Friesian y tres ouartos novillas Friesian, respectivamente. La EP vario de 31-5 meses en los cruces F |, hasta 46-8 me- ses eh las novillas Boran inseminadas artificialmente con semen Friesian. El intervalo entre partos fue de 465, 552, 525 y 487 dias para vacas Boran servidas por monta natural con toros Boran y Friesian, vacas Boran y vacas F| inseminadas artificialemente con semen de Friesian, respectivamente. La tasa de morta lidad'predestete fue de 3-4% y fue significativamente influenciada por el numero de partos, sexo e interac- cion :de raza con el ano de nacimiento. El ano como tambien le estaciones tambien tuvieron efecto significativo sobre las variables consideradas. Sin embargo, la variacion entre los grupos de empadre fue mayor, siendo en favor de los cruces F |. II Trop. Anim. Hlih Prod. (1994) 26, 49-57 PRODUCTIVITY OF BORAN CATTLE AND THEIR FRIESIAN CROSSES AT ABERNOSSA RANCH, RIFT VALLEY OF ETHIOPIA, n. GROWTH PERFORMANCE
K. Banjaw an d M . H aile-M a r ia m '
Alemaya University of Agriculture, Box 138, Dire Daw a, Ethiopia
SUMMARY Birth (4,197) and weaning (2,441) weight data on Boran, FI Boran Friesian and three quarter Friesian calves as well as adjusted one-year (390), 2-year (177) and 3- year (364) weights on Boran cattle were analysed to estimate the influence of genetic and environmental factors. Boran, F) and three quarter calves weighed 25-2, 25-4 and 25-7kg at birth and 157-5, 176-7 and 179-9 kg at weaning, respectively. All factors included in the analy sis and their interaction had significant effects on both traits with the exception of the effect o f season of birth and its interaction with breed group on birth weight. Weight of Boran cattle at one-, 2- and 3-years of age were 179, 269 and 338 kg, respectively. Heritability values calculated on the basis of paternal half-sibs were 0-32, 0-24, 0-48, 0-29 and 0-24 for birth, weaning, one-year, 2-year and 3-year weights, respectively. The study indicated that F1 crosses were 12-2% heavier at weaning than Boran calves. O f the environmental factors considered, year of birth was found to be the major source o f variation mainly due to variation in the amount of rainfall between years.
INTRODUCTION The background information on the Boran cattle breeding programme at the Abernossa Ranch in Ethiopia has been described by Haile-Mariam et al. (1993). This paper evaluates the effects of genetic and environmental factors on the growth performance of Boran cattle and their crosses at the same ranch.
MATERIALS AND METHODS Calves were weaned in a batch system, usually between 210 and 270 days old. Selected Boran females were retained as replacements in both the pure and cross breeding units. Boran bulls for breeding purpose were selected on their pedigree and parental information as well as on their own birth and weaning weight and con formation. However, the selection was not based on adjusted weights for systematic factors and there were no defined selection criteria. For the present study, information on 1,715 Boran, 2,307 Ft and 175 three- quarter crossbred calves born between 1977 and 1985 as well as weaning weights of 1,204 Boran, 1,171 F[ and 66 three-quarter crossbred calves were available. Adjusted one, 2, and 3-year weights were also estimated for 390, 177 and 364 purebred Boran animals, respectively. Weights at birth and weaning were recorded for most animals but postweaning weights were not routinely recorded. Also animals not required for breeding were usually sold after weaning.
1 Addressee for reprints: Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, S 750 07 Uppsala, Sweden.
49 50 BANJAW AND HAILE-MARIAM
The data were analysed by the least-squares procedure of Harvey (1977) using both fixed and mixed models. The model used included the fixed effects of breed of calf, year and season of birth, sex and parity as well as the interaction of season and|year with breed. In the case of weaning and postweaning weights both the Unear and quadratic partial regressions on age were also fitted. However, the final analysis on one, 2 and 3-year weight was run on weights adjusted to the respective ages using both the linear and quadratic regression coefficients. This procedure was assumed to adjust for the difference in weighting ages. The residual mean square was used as an error term to test the significance for each character analysed. Linear contrasts of least square means were computed to determine the significance of differences between means. In addition, heritability values for the different categories of body weights studied were estimated by computing among and within sire variance compo nents for Boran calves only.
RESULTS Birt i weight a U1 factors studied, except season of birth and its interaction with breed, had a sign ificant (P < 0 01) effect on birth weight. ] ieifers calving for the first time produced lighter calves than those bom to cows in the hird to eighth parity. Birth weights of calves showed an increasing trend over the yeais, except during 1980. The regression ofleast squares means of birth weight of all calves on year of birth (numbered from 1 to 9) indicated that birth weight increased by Q- 14kg per annum (P < 0-05). However, the highly significant interaction between breed and year indicated that birth weight of Boran calves increased over the years, whije there was no clear trend in the case of crossbreeds. This was substantiated by the fact that the regression of birth weight of Boran calves on year was 0-24 kg and was!statistically significant (P < 0*01). Weaning weight Weaning weight of calves was significantly (/>< 0 ,01) affected by all factors stuc ied (Table I). The total weight gain from birth to weaning was 132-1, 151-3 and 154 3 kg while the daily weight gain was 534*6, 637*9 and 641-9 g for the Boran, Fi and three-quarter Friesian calves respectively. Young cows calving for the first time and old cows calving for the ninth and later times weaned the lightest calves. The highest weight was recorded for calves produced by (ows in their third to fourth parturition. \nimals bom during the short wet season were significantly heavier than those bor i in the main wet season and those bom during the dry season took an inter mediate course (Table I). The interaction between breed and season of birth was alsc highly significant (P < 0-01). The highest fluctuation of weaning weight due to the effect of season was observed for the three-quarter and Fj crossbred calves. More over, when bom during the main wet season three-quarter crosses weighed 5-6 kg less than the Fj calves while they showed the highest growth in the other 2 seasons (Fig. 1). Both year and its interaction with breed had a significant effect (P <0*01) on weaning weight. The highest weights were observed during years when rainfall was above average. The weaning weights of calves bom in 1980 followed by those of 197^9 was the lowest, mainly due to the low amount of precipitation during 1980. Both the linear and quadratic regression of weaning weight on age were significant BORAN PRODUCTIVITY IN ETHIOPIA 51
T a ble I
Estimated least-square means and standard errors {s.e.) for birth and weaning weight (kg)
Birth weight (kg) Weaning weight (kg)
Variables No. Mean ± s.e. No. Mean ± s.e.
Overall mean 4197 25-43 ± 0-05 2441 171-4 ± 1-2 Breed ** ** Boran 1715 25-17 ±0-05a 1204 157-5 ± l-0a F] crosses 2307 25-39 ± 0 04b 1171 176-7 ± 0 -9 b 3/4 Friesian 175 25-73 ±0-14c 66 179-9 ± 3-0b ** Parity ** 1 658 25-14 ± 0 -0 7 a 396 167-2 ± 1 -5a 2 622 25-41 ± 0-08b 364 174-2 ± l-6bc 3 -4 1136 25-59 ±0-07c 663 178-4 ± 1 4d 5 -6 932 25-53 ± 0-07cb 458 176-5 ± l-5cd 7 -8 620 25-61 ± 0 -0 8 c 408 172-3 ± 1 -6b 9 + 229 25-36 ±0-12ba 152 166-7 ± 2-la Season of birth n.s. ** M ain wet 1817 25-43 ± 0 09 1192 159-0 ± 1 -9a Short wet 1073 25-38 ± 0 07 565 180-8 ± l -9c Dry 1307 25-48 ± 0 08 684 174-3 ± 2-lb Year of birth ** ** 1977 657 25-18 ± 0 -0 9 a 1978 476 25-04 ± 0 -1 7a 1979 662 25-35 ±0-1 lab 4 ll 160-6 ±~3-0ab 1980 599 25-00 ± 0 -1 3a 492 154-6 ± 2 -5 a 1981 340 25-15 ±0-20a 327 186-7 ± 3-0d 1982 361 25-24 ± 0 -1 8 ab 236 181-7 ± 3 -6d 1983 352 25-82 ±0-16c 290 175-1 ± 2-6cd 1984 384 25-55 ± 0 -13bc 364 174-4 ± 2-3cd 1985 366 26-54 ± 0-16d 321 166-6 ± 3-lb Sex of calf ** ** Fem ale 2107 25-21 ± 0 -0 6 a 1284 167-8 ± 1 3a Male 2090 25-65 ± 0 06b 1157 175-0 ± 1 3b
Regression on age linear 0-469 ± 0-0240** Regression on age quadratic -0-0019 ±0-0007**
Within variable groups, means with the same letter do not differ significantly (P > 0 05). ** = />< 0 01. n.s. = not significant (P > 0 05).
(P < 0-01). The coefficient estimated shows that for each day increase above the mean weaning age (238 days), there was an average increase of 469 g in weaning weight (Table I). The quadratic component of the regression was negative and the overall relationship is shown in Fig. 2.
Postweaning weights The least-squares means for adjusted one, 2 and 3-year weights of purebred Boran cattle are presented in Table II. The total weight gain from birth to one, 2 and 3-years of age were 154-8 ± 4-8, 244 ± 6 0 and 313-0 rfc 8-0 kg respectively, while the corresponding daily growth rates were 424 ± 13, 336 ± 9 and 286 ± 9 g respectively. Season of birth had no effect on the one and 3-year weights, but significantly (P < 0 05) affected the 2-year weight and animals born during the short wet season were 13 kg heavier than the overall mean. 52 BANJAW AND HAILE-MARIAM
Breed o Boran —f— F1 crosses —o— crosses
F ig. 1. Effect of season of birth on weaning weight of calves. ! Parity affected yearling weight significantly. Calves bom to cows in their third to sixth parity maintained their superiority of weaning weight up to one year of age. However, 2 and 3-year weights were not influenced by parity. Year of birth had a sign ficant (P < 0-01) effect on all the 3 postweaning weights (Table II). Similarly
19 0
180
o> 170 a> 5 S
160
150
210 220 230 240 250 260 270 Age (days)
F ig. 2. The regression of weaning weight of calves on age. BORAN PRODUCTIVITY IN ETHIOPIA 53
T able II
Estimated least-square means and standard errors (i.e.) for adjusted one. 2 and 3-year weights o f pure Boran cattle (kg).
One-year weight Two-year weight Three-year weight
Variables No Mean ± s.e. No Mean ± s.e. No. Mean ± s.e.
Overall mean 390 179 ± 5 177 269 + 6 364 338 ± 8
Parity ** n.s. n.s. 1 61 180 ± 5b 37 257 ± 9 55 333 ± 9 2 66 180 ± 5b 26 270 ± 9 45 341 ± 9 3 -6 183 185 ± 5b 73 279 + 7 175 342 ± 8 T 80 172 ± 5a 41 270 ± 8 89 337 + 8
Season of birth n.s. * n.s. M ain wet 193 178 + 5 99 261 ± 8a 194 344 ± 8 Short wet 128 174 ± 6 37 282 + 9b 66 334 + 9 Dry 69 186 + 5 41 264 ± 8a 104 337 ± 8
Year of birth ** ** ** 1977 21 315 ± 12c 29 348 + 10b 1978 50 194~±8d 38 268 ± llb 55 266 ± 10a 1979 50 179 ± 6cb 15 2 1 5 + 13a 43 344 ± 9b 1980 20 187 ± 7dc 32 301 ± 10c 61 367 ± 9c 1981 37 192 ± 6 d 16 291 ± llb c 61 351 ± 9b 1982 113 185 ± 6cd 3 267 + 31abc 69 349 ± 9b 1983 59 171 ± 6 b 32 257 ± 16ab 46 343 ± l l b 1984 39 159 ± 7a 20 236 ± 16a 1985 22 168 ± 9 a b _ _ _ I Sex o f calf * * ** *♦ Female 174 167 ± 5a 82 244 + 8a 336 308 ± 7a Male 216 192 ± 5b 95 294 + 7b 28 369 ± 10b Regression on ** *» n.s. age linear 0-233 ± 0-04 0169 + 0042 0 054 ± 0 004
Within variable groups, means with the same letter do not differ significantly (P > 0 05). * = P < 0 05. ** = PcOOl. sex affected weight at all ages with male calves being significantly heavier than females by 25-4, 50 0 and 61-0 kg at one, 2, and 3-years of age respectively. The linear regression coefficients of 1 and 2-year weights on the respective ages were 0-23 and 0-17 kg per day (P < 0 01). However, the value estimated for the 3- year age was not significant. A one day increase above 3-years of age increased weight by merely •« 0-05 kg.
Genetic parameters Heritability (h2) analyses for birth weights of 1,464 and for weaning weights of 1,221 Boran calves born between 1977 and 1985 showed significant sire and dam within sire effect (P < 0-01). The h2 values for the 2 weights on the basis of paternal half-sibs were 0-32 ±0-18 and 0-24 ±0-16 respectively. Similarly the estimated h2 values for one, 2 and 3-year weights were 0-48 ± 0-20, 0-29 ± 0-24 and 0-24 ± 0-14 respectively. 54 BANJA W AND HAILE-MARIAM
DISCUSSION Birth weight Birth weight values in the present study are higher than those reported by Kebede and Gallal (1982) for Boran and their Friesian crosses in Ethiopia. The lower birth weight of calves from first calvers is consistent with the reports by Swensson et al. (1981) and Trail et al. (1985). The difference in birth weight between sexes was only 0-44 kg (P < 0*05) which is small when compared with reports from Boran cattle in Kenya (Ronningen et al., 1972). However, it was in agreement with 0*84 and 0*6 kg difference reported by Kassa-Mersha and Arnason (1986) and Kebede and Gallal (1982) respectively in Ethiopia.
Weaning weight The superiority of the three-quarter crosses over the Boran in weaning weight may be attributed to individual and maternal heterosis whereas that of the Fj crossbreeds oveij Boran calves reflects the effect of individual heterosis since both breeds suckled Boran dams. This is, of course, in addition to the difference in additive genetic vari ance for growth between them. The low weaning weight of calves born to the first and second calvers in particular may! be attributed to the higher nutritive requirement of the dams for their growth and body maintenance as well as for lactation. On the other hand, older cows in their nintn and higher parturition produced lighter weaners due to ageing. The trend in the effect of parity is in agreement with the findings of Trail et al (1971). Calves bom during the short wet season were heavier than those of the main wet season, indicating that they were reared and weaned during the latter season when feeding conditions for their dams was favourable. On the other hand, the inter mediate weight of those born in the dry season is because they were weaned into the short wet period after suckling their dams throughout the dry season. The present finding is in general agreement with the reports of Trail and Gregory (1981) in Kenya and Kassa-Mersha and Arnason (1986) in Ethiopia. r’he change in ranking between the F] and three-quarter (Fig. 1) based on their seas )n of birth indicates that crossbreeds with more than half of their genes from Friesian require feed supplementation in periods of feed scarcity to maintain their superiority. The fact that Boran calves maintained uniform growth rate during all seas 3ns indicates their adaptability to tropical conditions. This is in general agree- me t with the finding of Barlow and O’Neill (1980) in tropical Australia.
Pos(weaning weights loran calves born in the short wet season were heavier than those bom during the othdr 2 seasons at 2-years of age, which contradicts the report by Kassa-Mersha and Arnason (1986) where animals bom during the main wet season had heavier 2-year weights than those bom in the other seasons. However, the way months were classi fied jinto season was slightly different in their study. The lower one-year weight of calves bom in 1979, whose weight was affected by the drought of 1980, shows that yearling weight was affected more by the amount of rainfall of the year during which the calves attained one year of age than by that of tjie year of birth. Two and 3-year weights seem to be affected by the amount of rainfall of the year prior to which the animals attained the final respective ages. Thus* animals bom in 1979, 1983 and 1984 had the lowest 2-year weights, because BORAN PRODUCTIVITY IN ETHIOPIA 55 the rainfall during 1980, 1984 and 1985 was below average (Haile-Mariam et al., 1993). Similarly, animals born in 1978 were 2 years old in 1980 and therefore had a low adjusted 3-year weight, being affected by the drought of the latter year. The presence of fewer males than females and perhaps because there was more intense selection among males than in females (Table II) may have exaggerated the difference between the sexes. Even then these findings are in agreement with the reports of ILCA (1978) and Madureira et al. (1980). That increase in age beyond 3-years did not increase weight is in agreement with reports by Kassa-Mersha and Arnason (1986) in which female Boran animals reached maturity at approximately 3-years of age and at a body weight of 300 to 310kg.
Genetic parameters The h2 value of birth weight was in agreement with that reported by ILCA (1978), while that of weaning weight was lower than their estimates for Maure and Peul cattle of Mali. Arnason and Kassa-Mersha (1987) working on Boran cattle at Abernossa estimated h2 values of 0-11 and 0-22 for birth and weaning weight respectively. The discrepancy between their and our estimate of h“ for birth weight is fairly high but standard errors were also relatively high (0-18 vs. 0 06). Their estimate was based on 4,000 calves born over 24 years which may suggest that it is closer perhaps to the actual value. However, considering the average phenotypic increase of 0 24 kg per year, it is difficult to assume that it was such a low value unless the environmental trend greatly favoured birth weight. Furthermore, they included all parities except the first one in one group perhaps due to which parity did not have a significant effect, while parity divided into 6 classes was highly significant in our study. In another study Ronningen et al. (1972) estimated hr of 0-26 and 0-06 for birth and weaning weights respectively of Boran cattle in Kenya. The high standard error of the h2 estimates and perhaps the presence of some sort of bias due to selection in the present study makes the genetic parameter estimates only indicative. Although the random effect of sire was significant in all cases, the high standard error of the h2 estimates renders them as zero. However, apart from their high standard error the estimates are in general agreement to reports from other tropical areas. Plasse (1982) noted that the mean h2 value for zebu cattle in the American tropics was 0-45 for postweaning weights and it showed a decline after the animals had reached around 18-months of age. The growth performance of the animals in the present study is influenced by both genetic and environmental factors. Systematic environmental factors such as year and season of birth, parity, sex and age at which animals are weighed had an influence on body weight. Animals born during the short wet season were 5-5% and 4-8% heavier at weaning and 2-years of age respectively than their overall mean. Year of birth was a major source of variation in the weights at different ages. Considering the weaning weight of the 3 genetic groups, the Ft calves grew 12-2% faster than the Boran and had a comparable growth rate with that of the three-quarter crosses. Thus, this sug gests that crossbreeding with temperate breeds could increase the efficiency of beef production under environments like Abernossa Ranch.
ACKNOWLEDGEMENTS We thank the Ethiopian Animal and Fishery Resources Development Main Department for allowing us to use the data. Special thanks go to Mr Teferra 56 BANJAW AND HAILE-MARIAM
Gebre-Meskel, Hizikias Ketema, Desta Biliso and Diro Kebede for their co operation and assistance in the study. The data were analysed by Mr A. R. Sayers at the International Livestock Centre for Africa, Addis Ababa. We thank Dr J. Philipsson and other members of the department of animal breed ing! and genetics for their helpful suggestions on the paper.
Accepted for publication October 1992
REFERENCES
A r n a s o n , t . & K assa -M e r s h a , H. (1987). Genetic parameters of growth of Ethiopian Boran cattle. Animal Production, 44, 201-208. B a r l o w . R. & O’N e il l , G . H . (1980). Performance of Hereford and crossbred Hereford cattle in the sub tropics of New South Wales: Genetic analyses of preweaning performance of first-cross calves. Australian Journal of Agricultural Research, 31, 417-427. H a il e -M a r ia m , M ., B a n /a w . K ., G eb r e -M e s k e l , T. & K e t e m a , H . (1992). Productivity of Boran cattle and their Friesian crosses at Abemossa Ranch, Rift-Valley of Ethiopia. I. Reproductive performance and preweaning mortality. Tropical Animal Health and Production, In Press. H a r v e y , W. R, (1977). User’s guide for least-squares and maximum likelihood computer program, Ohio State | University, Columbus, USA. ILCA, International Livestock Centre for Africa (1978). Evaluation of the productivity of Maure and Peul cattle breed at the Sahelian Station, Niono, Mali. ILCA Monograph I, Addis Ababa, Ethiopia. K a s sa -M e r s h a , H. & A r n a s o n , T. (1986), Non-genetic factors affecting growth of Ethiopian Boran cattle. i World Review Animal Production, 2 2 ,4 5 -5 5 . K ebe [>e , B . & G a l l a l , E. S. E. (1982), A study of body weight from birth to 1 year of age in European-zebu crossbred cattle in Ethiopia. Animal Production, 34, 85-93. M ad j r e ir a , J. S ., S il v a , H . M ., F o n t e s . L. R. & S a m p a io , I. B. M. (1980). Some factors affecting body weight on Chiana x Nellore crossbreds on pasture. 3. Body weight at 365 days of age. Animal Breed ing Abstract, 48, 183. P lasj Ei D. (1982). Performance recording of beef cattle in Latin America. World Animal Review, 41,11-19. R 6 nj> in g e n , K ., L a m p k in , K . & G r a v ir , K . (1972). Zebu cattle in East-Africa. 2. Estimates of heritability and phenotypic correlation for some traits in Boran cattle. Swedish Journal o f Agricultural Research, 2. 218-228. S w e n s s o n . C., S c h a a r . J., B r a n n a n g , E. & M e sk e l, L. B. (1981). Breeding activities of the Ethio-Swedish integrated rural development project. 3, Reproductive performance of zebu and crossbred cattle. I World Animal Review, 38, 31-36. T r a ii , j . C. M. & G r e g o r y , K . E. (1981). Sahiwal cattle: an evaluation of their potential contribution to milk and beef production in Africa. ILCA Monograph 3, ILCA, Addis Ababa, Ethiopia. T r a m , J. C. M ., S a c k e r , G . D. & F is h e r . I. L. (1971). Crossbreeding beef cattle in western Uganda. 1. Per formance of Ankole, Boran and zebu cows. Animal Production, 13, 127-141. T r a u , J. C. M ., M u r r a y , M ., S o n e s . K., J ib b o . J. M . C., D u r k in , J. & L ig h t . D . (1985). Boran cattle main tained by chemoprophylaxis under trypanosomiasis risk. Journal of Agricultural Sciences, (Cambridge), 105, 147-166.
PRObUCTIVITE DU BETAIL BORAN ET DE SES CROISEMENT AVEC LA RACE FRISONNE AU RANCH D’ABERNOSSA, DANS LA VALLEE DU RIFT (ETHIOPIE). II. PERFORM ANCE D E CROISSANCE Resuijie— Pour apprecier l’influence des facteurs genetiques et du milieu, les auteurs ont analyse les donnees suivautes: poids a la naissance (4197 et au sevrage 2441) chez les veaux Boran, Boran x Frison F t, et 3/4 Frison. De meme, les poids du betail Boran a 1 an (390), 2 ans (177) et 3 ans (364) ont ete analyses. Les veauxj Boran, F| et 3/4 Frison pesaient respectivement 25,2-25,4-25,7 kg a la naissance et 157,5-176,7 et 17S>,9 kg au sevrage. Tous les fracteurs pris en compte pour Tanalyse des donnees et de leur interaction avaient des effets significatifs sur les deux caracteres a [’exception de la saison de naissance et de son inter action avec le groupe racial sur le poids a la naissance. Le poids du betail Boran a 1 an, 2 ans et 3 ans etait respectivement de 179, 269 et 338 kg. Les valeurs d’heritabilite calculee sur la base des demi-freres et demi- soeurs patemels etaient de 0,32-0,24,0, -48-0,29 et 0,24 pour la naissance, le sevrage, et les poids a 1, 2 et 3 ans respectivement. L’etude a releve que les croisements Fj etaient plus lourds de 12,2 p. 100 au sevrage que les veaux Boran. BORAN PRODUCTIVITY IN ETHIOPIA 57
Enfin, parmi les facteurs du milieu consideres, 1’annee de naissance est apparue comme la source principale de variation, due principalement a la variation de la pluviometrie d’une annee a l’autre.
PRODUCTIVIDAD DEL GANADO BORAN Y SUS CRUCES FRIESIAN EN EL RANCHO ABERNOSSA, EN EL VALE DE RIFT DE ETIOPIA. II. DESEMPENO EN EL CRECIMIENTO Resumen— Se analizaron los pesos al nacer (4,197) y al destete ((2,441) de ganado Boran, F] Boran Frie sian y tres cuartos terneros Friesian y los pesos ajustados al ano (390), a los 2 anos (177) a los 3 anos (364) de ganado Boran. Para estimar la influencia de factores geneticos y ambientales. El ganado Boran, F, y tres cuartos pesaron 25-2, 25-4 y 25-7 kg al nacim iento y 157 5, 176-7 y 179 9 kg al destete, respectivamente. Todos los factores incluidos en el analisis y su interaction tuvieron efecto signifi- cativo en ambos factores con la exception del efecto de la estacion sobre nacimientos y sus interacciones con grupo racial, sobre peso al nacer. El peso del ganado Boran a 1, 2 y 3 anos de edad fueron 179, 269 y 338kg, respectivamente. Los valores de heredabilidad para pesos calculados sobre la base de medio-hermanos pa- temos, fueron 0-32,0-24, 0-48, 0-29 y 0-24 para nacimento, destete, I ano, 2 anos y 3 anos, respectivamente. El estudio indico que los cruces F t fueron 12-2% mas pesadas al destete que las crias Boran. De los factores ambientales considerades el ano de nacimiento fue hallado ser el de mayor variacion principal- mente debidoals variacion in la precipitacion pluvial entre anos.
BOOK REVIEW
Ruminant Nutrition. The Tropical Agriculturalist Series. J. Chesworth. Macmillan, London. 1992. 170 pp. Soft back. £5.99. ISBN 0-333-57073-1.
This is another book in the award winning series, The Tropical Agriculturalist, from the Centre for Agricultural and Rural Cooperation (CTA), The Netherlands. At relatively low cost, books in this series provide a comprehensive coverage of specific topics up to first degree level. In this particular book on ruminant nutrition, the author presents a traditional treatment of the subject with chapters on water, minerals, vitamins, digestion and so on. At the same time there is inclusion of newer concepts such as protein degrad- ability. There is a useful chapter on nutritional planning and appendices dealing with methods of monitoring nutritional strategies and analysis of foods. Technically one can question the inclusion of sodium hydroxide as a method of treating roughages. Surely this is a totally inappropriate and even dangerous techni que for tropical systems. There is obviously justification for the inclusion of urea treatment of roughages but it is a little disappointing that the necessary circum stances for the practical application of the technique are not indicated. Given the level of readership targeted by books in this series it is pleasing to find a full glossary and for such a modest sized work the index is excellent. There are suffi cient illustrations of various kinds to lighten the text and overall this is a “good buy” .
D. Fielding Ill Estimates of direct and maternal (co)variance components of growth traits in Boran cattle
Mekonnen Haile-Mariam and Hailu Kassa-Mersha
Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, S-750 07, Uppsala, Sweden
Summary
Birth (BW), weaning (WW) and yearling weight (YW) data of Boran cattle from Abemossa ranch of the Ministry of Agriculture of Ethiopia were analysed to estimate genetic parameters due to maternal and direct effects. In the univariate analysis, six different models ranging from the 'simple' animal model (model 1) to a model which included direct and maternal genetic effects and their correlation and permanent environmental effects (model 6) were applied. After comparing the Log likelihood values, model 4 - which included direct and maternal genetic effects and their correlation for BW and model 6 for WW and YW - were selected as being appropriate. For BW, a direct heritability (hd2) of 0.24 and a maternal heritability (hm2) of 0.08 were estimated. In the case of WW and YW, the estimates were 0.29 and 0.34 for hd and 0.06 and 0.05 for hm2, respectively. The ratios of permanent environmental variance to the total (c2) were 0.14 and 0.05 for WW and YW, respectively. Estimates from the bi- and tri-variate analyses were similar to those from univariate analysis, but the estimates for hd2 were relatively higher. Particularly parameter estimates based on WW and YW together, as well as the trivariate analyses appeared to have accounted for selection bias on WW and thus resulted in a higher hd2 estimate for YW. The across trait correlation estimates were low to medium with the exception of high direct genetic and permanent environmental correlations between WW and YW. Generally, parameter estimates of Boran cattle were within the range of those for other breeds. However, the genetic antagonism between direct and maternal effects seems to be stronger (-0.33 to -0.68).
Introduction
In beef cattle birth weight and pre-weaning growth as well as early post-weaning weights are influenced by direct and maternal additive genetic variances and their covariances. These estimates may vary due to breed and/or environmental differences. Thus, variance component estimates in zebu cattle may differ from those of temperate cattle, for both environmental and genetic reasons. Knowledge of the size of the variances as well as the size and sign of the covariances is
1 T necessary in order to formulate appropriate breeding strategies for growth traits.
A Boran cattle breeding herd was established by the Ethiopian Ministry of Agriculture in 1959 to improve beef productivity through selection. The data collected at the ranch on birth and weaning weight were analysed by Arnason and Kassa-Mersha (1987). A smaller dataset covering the period between 1977 to 1985 was also analysed, by Banjaw and Haile-Mariam (1994). In both studies, genetic parameter estimates were obtained using Henderson's method 3. Although the former authors applied a more detailed model than the latter, they used the literature average for genetic correlation between direct and maternal effects and their analysis was concerned with growth up to weaning.
This paper presents estimates of (co)variance components and the resulting genetic and phenotypic parameters among growth traits of Boran cattle, using REML (restricted maximum likelihood) from univariate and multivariate analyses.
Ma erials and Methods boct tion and herd management
A description of the management, the area and the foundation stock is published elsejwhere (Arnason and Kassa-Mersha, 1987; Banjaw and Haile-Mariam, 1994; Haile-Mariam and Kassa-Mersha, 1994). Briefly, the ranch where the animals were located is about 180 km south of the capital, Addis Ababa. The average annual raiijfall is about 700 mm, most of it falling between July and October; the dry season lasts from November to February, The short rains occur between March and June.
All animals grazed on natural pasture with supplementation restricted to years of critical feed shortage. Animals to be used as replacement were selected on the basi s of their birth and weaning weight and their conformation and colour, as well as pedigree, in order to avoid inbreeding. However, there was no information on how these were combined to form an index, nor was there any official adjustment for mportant fixed effects.
Data and traits i The data used for this study included pedigree and weight records of animals born between 1959 and 1985. Except for the foundation animals, which were 322 plus 94 progeny bom on the ranch from unknown sires, all the other animals had complete pedigree records. Birth (BW), weaning (WW) and yearling (YW) weight data on 5,256, 4,082 and 2,417 calves, respectively, were available (Table 1). The disparity in the number of records available for each trait was due primarily to
2 I denna serie publiceras forsknings- In this series you will find research och forsoksresultat vid Lnstitutionen reports from the Department of Animal for husdjursforadling och sjukdoms- Breeding and Genetics within the genetik inom lantbruksvetenskapliga agricultural faculty, the Swedish fakulteten, Sveriges lantbruksuni- University of Agricultural Sciences. versitet. Tidigare nummer redovisas Earlier issues are listed below. nedan och kan i man av tiUg&ng Issues still in stock can be acquired anskaffas fran institutionen from the Department
1 Dim N.I. 1973. Genetic parameters and sire proof in purebred and crossbred dairy cattle. 2 Sundgren, P-E. 1973. Studies on pig performance testing. T 3 Danell 0 . 1973. Statistik for studerande i husdjursforadling. 4 Liljedahl, L-E. & Weyde C. 1974. Studier av besattningskontrollen for vaxphons - en jamforelse mellan ordinarie och forlangd testperiod. 5 Hansson I. 1974. Effect of Sex and Slaughter Weight on Growth Feed Efficiency and Carcass Characteristics of Pigs. T 6 Lissanework, B.M. 1974. Crossbreeding experiment with the Swedish polled breed. 7 Juneja, K. & Gahne, B. 1975. Blood groups and biochemical polymorphism in fish. 8 Sellei, J. 1975. Some characteristics of cattle red cells which influenced their reactivity in the haemolytic and agglutination tests. T 9 Ronningen, K. 1975. Estimation of non-additive genetic variation and crossbreeding effects and use of crossbreeding effects in animal breeding. 10 Ebbersten, K. & Skirman S. 1976. Fargforandringar och fargnedarvning hos svenska palsfar. 11 Lundstrom, K. 1976. Stress susceptibility and meat quality in Swedish Landrace and Yorkshire pigs. T 12 Darelius, K. & SkSrman, S. 1976. Crossbreeding for Mutton with Swedish Landrace. 13 Philipsson, J. 1976. Studies on calving difficulty, stillbirth and associated factors in Swedish cattle breeds. T 14 Dim, N.I. 1976. Studies on Dual-Purpose Dairy Cattle. T 15 Ekman-Bjaresten, I. 1976. Crossbreeding for Beef with Swedish Polled Cattle. T 16 Malmgren, S., Eriksson, J-A. & Ronningen, K. 1976. Algen som kottproducent. 17 Backman, I., Larsson, B., Hildingstam, J. & Carlsson, H. 1977. Smaltbarhetsforsok med regnbligslax (Salmo Gairdneri). Metodik och foderforsok. 18 Karlsson, R. 1977. Ekonomiska vikter i svinaveln. 19 Eriksson, J-A. 1978. Utnyttjande av stationskapadteten med hansyn tagen till falttest vid avelsvardering av svin. 20 Aspers, M. Sylvan S. Eriksson, J-A. & Wilhelmson, M. 1978. Styrd avskjutning i algstammen - en simuleringsstudie. 21 AhMn, K. 1978. Palsegenskaper hos lamm: Samband mellan bedomning p§ levande djur och beredda skinn samt objektiva matningar p3 Mrprover. 22 Wilton, J.W. & Danell, 0. 1978. Discounted Expressions of Traits in Beef Crossbreeding Programs. 23 Eriksson, J-A., Wilton, J.W. & Henningsson, T. 1978. Estimating Breeding Values for Rate of Gain of Beef Bulls in Sweden. 24 Brannang, E. & Lindkvist, G. 1978. Uppfodningsintensitet, inkalvnings&lder och mjolkavkastning - en serie tvillingstudier. 25 Ral, G. 1978. Berakning av ekonomisk lonsamhet for artificiell insemination under olika forutsattningar for svin. 26 Stress factors influencing live and carcass weight in lamb. Darelius, K. 1978. Weight losses and repeated weighing in lambs. Andersson, O. 1978. Changes in weight and quality of lamb carcasses due to delayed slaughter.
T = Thesis (Doktorsavhandling) 27 Seeger, P., Lundstrom, K. & Danell, 0 . 1978. Statistisk introduktion till Harvey’s program. 28 Engstrand, U., Lundstrom, K. & LSfgren, B. 1978. Programmering med SAS-76 och nfigot om den statistiska bakgrunden. 29 jDanell, 0.1978. Users' guide for DIGFE, a computer program for calculating discounted I gene flow expressions. 30 iRal, G. 1978. Studies on the biological and economic benefit obtainable by using crossbreeding and artificial insemination in pig production. T 31 Eriksson, J-A., Sylv&i, S. & Wilhelmson, M. 1979. Beskrivning for anvandare av datorprogram for simulering av populationsdynamik i hjortdjurspopulationer. 32 Malmfors. B. 1979. Meat and fat quality of boars gilts and castrates. T 33 Sylvan, S., Aspers, M. Eriksson, J-A. & Wilhelmson, M. 1979. Regulated harvesting of the moose population - a simulation study. 34 BrSnniJng, E., Wiktorsson, H., Andersson, M. & Pettersson, G. 1979. Korsningsforsok med SKB-rasen under tvS olika uppfodnings- och laktations-intensiteter. 35 Elofson-Bemstedt, A. & RSnningen, K. 1979. Matemell effekt- en Meraturstudie av skattningsmetoder och storlek. 36 Ojala, 1. & Tengroth, G. 1979. Juverhalsa och tjuvdiande hos fctr. 37 Elofson, L., Danell, B. & Philipsson, J. 1979. Harstamningsindex och ungtjursindex i mjolkboskapsaveln. 38 Sandberg, K. 1979. Studies on blood groups and genetic protein polymorphisms of the horse. T 39 Wilhelmson, M. 1979. Breeding Experiments with Japanese Quail (Cotumix c. japonica). T 40 Larsson, B. & Cedrins, R. 1979. Probleminventering inom omr&dena fiskavel, fiskens utfodring och fiskodlingsteknik. 41 Br&nin, I-L., Danell, 0 . & Wilhelmson, M. 1979. Koncessionsrenskotseln i Norrbottens 13n. En beskrivning av renskQtsel&ret, den forda statistiken och dess anvSndbarhet for produktionsstudier. 42 Danell, O. 1980. Studies concerning Selection Objectives in Animal Breeding. T 43 Arv£n, K. 1980. Svinens kroppsbyggnad benstallningar och rSrelser. 44 Wilhelmson, M. & Sylvan, S. 1980. Honlig fruktsamhet hos alg. Ett forslag till handbok. 45 Janson, L. 1980. Studies on Fertility Traits in Swedish Dairy Cattle. T 46 Andersson, K. 1980. Studies on Crossbreeding and Carcass Evaluation in Pigs. T 47 Lundstrom, K., Gahne, B. & Edfors-Lilja, I. (ed.). 1981. Immunogenetics in animal breeding. Proceedings from a post-graduate course. 48 StrBm, H. & Philipsson, J. 1981. Genetisk analys av arabhastmaterial. 49 Stigson, M. 1981. Studie av grisningsforlopp hos 16sg&ende suggor. 50 Eriksson, J-A. 1981. Best linear unbiased prediction of breeding values with regard to related contemporaries and selection of records. T 51 Danell, B. 1981. Evaluation of Sires on First Lactation Yield of Swedish Dairy Cattle. T 52 iSchaar, J. 1981. Casein Stability and Cheesemaking Properties of Milk; Effects of Handling, Mastitis and Genetic Variation. 53 Wilhelmson M. & Sylvan, S. 1981. Tekniska metoder for bevarande av genresurser - frysforvaring av embryoner. 54 Kurowska, Z., Ojala, I. & Danell, 0 . 1981. En metodstudie rfirande bestamning av palsskinnens tyngd. 55 Kurowska, Z. & Danell, 0 . 1982. FSrgnedarvning och farggeners effekt pci produktionsegenskaper hos ibx. 56 Sjaunja, L .-0 .1982. Studies on Milk Analysis of Individual Cow Milk Samples. T 57 Brannang, E., Darelius, K,, Gendron, E. & Ral, G. 1982. Energiansattning hos ungdjur och samband mellan mjolkproduktions- och kSttproduktionsegenskaper. 58 Larsson, B. 1983. Produktion av stor regnbSgslax for konsumtion.
T = = [Thesis (Doktorsavhandling) 59 Arnason, T. 1983. Genetic studies on conformation and performance of Icelandic toelter horses. T 60 Ral, G., Henningsson, T., Andersson, O. & Karlsson, U. 1984. "Scanningtekniken som metod att skatta slaktkroppsegenskaper hos ""levande notkreatur. 62 Lillpers, K., Wilhelmson, M. och Alarik, M. 1984. Variationer i honans varpmonster och vissa produktionsegenskaper i normala och forkortade dygn. 63 Eriksson, J-A. 1984. Beskrivning for anvandare av datorprogram for simulering av populationsdynamik i hjortdjurspopulationer -tillagg till Rapport 31. 64 Henningsson, T. 1985. Performance Testing for Beef Production Traits in Swedish Dual Purpose and Beef Cattle. T 65 Edfors-Lilja, I. 1985. Marker traits of disease resistance in the pig. Genetic studies of immune responsiveness and the intestinal receptor for E.coli K88. T 66 Johansson, K. 1985. Estimation of genetic parameters for use in the Swedish pig breeding programme. T 67 Strandberg, E. 1985. Estimation procedures and parameters for various traits affecting lifetime milk production: A review. 68 Urioste, J. 1986. Effekt av fodelsevikt och andra faktorer p& lammdodlighet i en forsoksbesattning med finullsf^r. 69 Nasholm, A. 1986. Viktsutveckling hos tackor av finullsras samt metoder att skatta vuxenvikt. 70 Strandberg, E. 1986. Inverkan av miljoeffekter p i avkastning, tomperiod och kalvningsintervall i de tre forsta laktationema hos mjolkkor. 71 Schaar, J. 1986. Variation in milk protein composition. Studies on K-casein and p-lactoglobulin genetic polymorphism and on milk plasmin. T 72 Lundeheim, N. 1986. Pig progeny station testing of disorders and production traits. T 73 Emanuelson, U. 1987. Genetic studies on the epidemiology of mastitis in dairy cattle. T 74 Urioste, Jorge. 1987. Reproductive traits in sheep breeding with emphasis on litter size as a threshold character. L 75 Edfors-Lilja, I. 1987. Department of Animal Breeding and Genetics - organization and activities. 76 Ericson, K. 1987. Crossbreeding effects between two Swedish dairy breeds for production and reproductive traits. L 77 Danell, 0. 1988. Theoretical aspects in the estimation of breeding values for all-or-none traits. 78 Ral, G., Berglund, B., Philipsson, J., Emanuelson, U. & Tengroth, G. 1988. Juver- och mjolk-barhetsegenskaper samt mjolkavkastning och mastitforekomst - effekter av ras och &lder samt inbordes samband. 79 Berglund, B. 1988. Calving performance, production and reproduction in early lactation. Studies of variation and interrelationships in Swedish dairy breeds under experimental conditions. T 80 Andersson-Eklund, L. 1988. Orsaker till att varmblodstravare inte kommer tillstart och sambandet mellan dessa orsaker och hastamas serumesteras (Es) typ. 81 Gates, P. 1988. Breed differences in forage intake as related to production increases in Swedish Red, Swedish Friesian, and Swedish Jersey cattle at Kungsangen experimental station (Hfa). MS 82 Strandberg, E. & Oltenacu, P.A. 1989. Economic Consequences of Different Times of Conception: A Simulation Study. 83 Danell, B., Janson, L. & Stromberg, L. 1989. Samtidigt urval for mjolkproduktion och fruktsamhet hos notkretur. Selektionseffekter vid olika forutsattningar - En simuleringsstudie. 84 Nasholm, A. 1989. Prediction of breeding values for mature weight in ewes. L
T = Thesis (Doktorsavhandling); L = Licentiate thesis (Licentiatavhandling); MS = Master of science thesis 85 Mahdy, E.A. 1989. Chromosomal localization of the major "histocompatibility complex (MHC) in some domestic animals by in situ hybridization. MS 86 Beyene, T. 1989. Performance of Arsi and crossbred sheep in the highlands of Arsi region, Ethiopia. MS 87 Hellander, E., Sjaunja, L-O. & Schaar, J. 1989. Citrathaltens variation i komjolk och leverantormjolk. 88 Andersson, L., Ral, G., Philipsson, J. & Jonsson, G. 1989. Variation i oliJka m&tt p& klovens exteridr och klovhomets innehSll av mineraler och aminosyror hos individprovade SRB- och SLB-tjurar. 89 Hansson, I. 1989. Ntftslaktkroppar, sammansattning och egenskaper. En rapport baserad P& styckningar utfSrda vid Avd. f8r k5ttvetenskap. 90 Andersson-Eklund, L. 1990. Associations of blood groups and blood protein polymorphisms with performance and production traits. L 91 SttUhammar, E-M. 1990. Genetic studies on male fertility in A.I. bulls. L 92 Bjorklund, T. 1990. Genetic studies on reproductive fitness in relation to effects of age and mutagen in Drosophila melanogaster. L 93 Peter sson, H. 1990. Genotype x Nutrition interactions in the performance testing of pigs. T 94 Thafvelin, B. 1990. The genetic variation in conformation of standardbred trotters and the relationship between conformation and performance. L 95 Engstrom, G. 1991. Genetic studies of reproductive fitness in relation to ageing in , Drosophila melanogaster and laying hens. T 96 Strandberg, E. 1991. Breeding for lifetime performance in dairy cattle. T 97 Kurowska, Z. 1991. Adjusting lamb weights for systematic effects in the Swedish sheep recording scheme. L 98 Lund£n, A. 1991. Marker genes and production traits in domestic animals. An association study with special reference to major histocompatibility complex genes. T 99 Chaudhary, R. 1992. Physical Gene Mapping in Pigs. Localization of the genes for PGD, ALB, TF, CP, and CS using in situ hybridization. MS 100 Andersson-Eklund, L. 1993. Genetic Markers and Quantitative Traits in Dairy Cattle. T 101 Setiabudi, R. 1993. Application of the Polymerase Chain Reaction (PCR) Technique for j Determination of Sex on the Cellular LeveL L 102 Villag6mez Zavala, D.A.F. 1993. Synaptonemal Complex Analysis of Chromosome j Translocations in Pigs and Cattle. T 103 Lagerkvist, G. 1993. Selection for Litter Size, Body Weight and Pelt Quality in Mink | (Mustek vison). T 104 JLillpers, K. 1993. Oviposition Patterns and Egg Production in the Domestic Hen. T 105 Gates, P.J. 1993. Sources of Variation in Litter Size in Sheep. L 106 jRydhmer, L. 1993. PIG REPRODUCTIVE GENETICS and correlations between j reproduction and production traits, T 107 jPetersson, C.J. 1993. Reindeer herd production - a modelling approach. T 108 Stem, S. 1994. Lean growth in pigs: Response to selection on high and low protein diets. T 109 Gu, F. 1994. In situ hybridization mapping of genetic markers in the porcine genome. T 110 Nasholm, A. 1994. Genetic studies on mature weight, maternal capacity and growth in Swedish finewool sheep. T 111 Lindhg, B., Danielsson, D.-A., Banos, G., Jansson, L. & Philipsson, J. 1994. Applied Breeding Policy 1981-1992 and its Genetic Effects in Two Swedish Dairy Breeds. 112 StSlhammar, H. 1994. Selection objectives and methods for in vivo evaluation of carcass traits based on performance testing of young dairy bulls. L
T = Thesis (Doktorsavhandling); L = Licentiate thesis (Licentiatavhandling); MS = Master of science thesis
I death or lack of that particular record. However, some animals were sold after weaning.
Table 1. Structure of the data, means, and measures of variation for birth (BW), weaning (WW) and yearling weight (YW) of Boran cattle
Item BW WW YW
No. of records 5256 4082 2417 No. of animals3 5847 5847 5847 5882 5882 5882 No. of sires 59 59 59 with progeny1’ 57 57 57 with their own record0 45 42 30 No. of dams 1246 1246 1246 with progeny13 1221 1055 878 with 1 progeny 294 268 288 with their own recordc 899 642 387 Weight (kg) Mean 23.6 168.6 191.5 S.D. 2.5 22.5 37.0 C.V.(%) 10.6 17.4 19.3 Age (days) Mean 0 252.5 385.0 S.D. - 22.5 67.1 C.V.(%) - 8.9 17.4 Inbreeding coefficient(%) Mean 1.17 1.18 1.09 S.D. 3.27 3.24 3.27 C.V.(%) 278 276 298 Year-season classes 79 73 59 Parity classes 6 6 6 Sex classes 2 2 2
Tor models 1 and 2 in the first line and for all other models on the second line. bNo. of sires (dams) with progeny in the data. T'lo. of sires (dams) which have their own and their progeny's record.
Birth weight was recorded within 24 hrs of birth, while the time at which weaning and post-weaning weights were recorded varied considerably. For this study, all weaning weights recorded between the age of 107 and 300 days were included in the WW analysis, while all post-weaning weights recorded between 320 and 600 days were considered as YW. The data structure, number of records, the mean weight and age and measures of variation for each trait are shown in Table 1.
3 Analysis
The fixed effects fitted in the analysis were year-season of birth or weighing, and parity and sex for all traits. The age of the animal at weighing and its inbreeding coefficient, calculated as described by Quaas (1976), were also fitted as covariables for tWW and YW. The complete model included the random effects of the animals' additive genetic effect the maternal additive genetic effect and their covariance, and the permanent environmental effect of the dam.
In addition to the complete model (designated as model 6) described above, five incomplete models, which ignored one or two of the maternal effects and/or which assumed no covariance between the direct and maternal genetic effects, were applied to each trait. The models were numbered, according to Meyer (1992a), as follows: (1) a 'simple' animal model which fitted direct genetic effects only, (2) an animal model including a permanent environmental effect of the dam, (3) an animal model which assumed that all the maternal influences were due to genietic effects only but which ignored the covariance between the two genetic effects, (4) an animal model with maternal genetic effects and a covariance between direct and maternal genetic effects, (5) an animal model with maternal genetic and permanent environmental effects. The model which was deemed 'ap ?ropriate' among the above was selected by comparing the values of the log like lihoods (log L).
REML (co)variance component estimates were also obtained from bivariate animal mo iel analysis, considering two traits at a time. For these analyses, model 4 for BW land model 6 for WW and YW were used, as indicated by the results from uni variate analysis. Finally, a trivariate analysis which included all traits was run.
The analysis was carried out using the Derivative-Free Restricted Maximum Likelihood (DFREML) computer package of Meyer (1993a). The same procedures as Outlined by Meyer (1992a) for univariate, and by Meyer (1993c) for bi- and tri variate analyses were followed. A convergence criterion level of Iff8 was assumed. Totil heritability (hT2) defined as by Willham (1972) was also calculated.
Res ults Umvariate analysis
Tat le 2 shows estimates of (co)variance components and genetic parameters for eac i trait when analysed according to six different models. The genetic covariance between direct and maternal genetic effects was always negative. In all cases the direct h2 was higher than the maternal h2. When the analysis was carried out on the complete model (model 6) the maternal h2 was about one-third of the direct h2 in he case of BW, but about one-fifth in the case of WW and YW. The total phenotypic variance as well as the genetic variance was small, particularly in the cas^ofBW. Table 2. Estimates of (co)variances and parameters from univariate analysis of birth (BW), weaning (WW) and yearling weight (YW) of Boran cattle according to different models
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 BW v 0.941 0.801 0.652 1.038 0.651 1.036 o j - - 0.154 0.362 0.149 0.366 a ? 3.375 0.081 3.459 3.265 3.454 3.253 Gp2 4.316 4.282 4.266 4.327 4.261 4.316 K 2 0.218 0.187 0.153 0.240 0.153 0.240 K 2 -- 0.036 0.084 0.035 0.085 -- - ^dm - -0.078 -0.079 --- - rdm -0.553 -0.551 c2 - 0.019 -- 0.001 0.0 h j 0.218 0.187 0.189 0.164 0.188 0.165 Hog L -8.2 -7.0 -2.9 0.0 -2.9 0.0 WW 188.29 105.86 87.90 124.11 96.59 142.65 c m2 -- 85.64 115.38 19.14 31.59 0 / 150.57 122.23 114.34 157.55 112.90 170.19 G m2 - - 22.77 45.97 11.41 24.01 a £2 347.12 344.18 354.68 333.70 348.35 318.57 oP2 497.69 491.97 491.79 495.83 490.46 495.15 hd2 0.302 0.248 0.232 0.318 0.230 0.344 h m2 - - 0.046 0.093 0.023 0.048 Cdm -- - -0.083 - -0.087 rdm --- -0.486 - -0.677 c2 - 0.052 - - 0.036 0.052 h T2 0.302 0.248 0.279 0.239 0.253 0.237 log L -6.7 -3.4 -3.1 -1.7 -2.6 0.0 +a/ , am2, oe2 and op2: direct and maternal genetic, error and phenotypic variances, respectively; hd2, hm2, hT2 and rdm : direct, maternal and total heritability and genetic correlation, respectively; cdm and c2: ratios of the genetic covariance between direct and maternal and of maternal environmental variance to aP2, respectively; Hog L is log likelihood expressed as deviation from model with highest value. 1 institute of Agricultural 1 5 Research 1 L i b r x I Ignoring the permanent environmental effect (model 4) gave results similar to those of model 6, where it was induded for BW. For both WW and YW there was a tendency toward increased maternal h2 when the permanent environmental variance was ignored (Table 2). The maternal h2 was almost equal to the direct h2 for WW when analysed according to model 4. Estimates according to model 4, where ctc2 is important, inflated om2 and proportionately reduced o/, as compared with model 6. Models which included the genetic correlation between direct and maternal effects (models 4 and 6) increased both hd2 andh j considerably for all traits. At the same time the log L values (Table 2) were relatively high, indicating that models which included genetic correlations were better fitted to the data. In the case of YW, model 6 was marginally better than model 4 (%2 value of 3.4 against a tabulated valiie of 3.84 significant at a 5% level of probability). But according to model 4 the h j 'increased due to substitution from c2, even as compared with the h j value of WW estimated according to the selected model (model 6). Across models, the hj2 remained fairly constant, with the exception of model 1 where a higher hT2 was estimated. Estimates of sampling errors of the parameters were high in the case of YW and were lowest in the case of WW (Table 3). Within trait they were relatively large for maternal parameter estimates vis-^-vis direct effects. The sampling errors were also high for cdm (the proportion of the genetic covariance between direct and maternal effects to total variance). Although the parameter estimates for BW were based on the largest dataset, the sampling errors were still high. Taple 3. Genetic parameter estimates with their approximate sampling error from the 'appropriate' model Trait Model \ 2 hm2 cdm c2 Est. s.e. Est. s.e. Est. s.e. Est. s.e. BW 4 0.240 0.086 0.084 0.033 -0.078 0.046 - WW 6 0.290 0.078 0.064 0.030 -0.077 0.039 0.135 0.024 YW 6 0.344 0.104 0.048 0.035 -0.087 0.049 0.052 0.027 jbreviations as in Table 2. Bivariate analysis Genetic parameter estimates from bivariate analysis CTable 4) were similar to those obt lined from univariate analysis (Table 3). However, the /z2-values of BW and YW showed a slight increase. The increase in that of YW might be related to the elimination of bias which has been introduced due to selection based on WW. However, that of BW is difficult to explain. The direct genetic correlation (Table 5) estimates for BW and WW (0.37) were lower than those of BW and YW (0.45). The maternal genetic correlation between BW and YW was small and negative, whereas that of BW and WW was positive (Table 5). Correlation estimates among traits were low to medium, except that of the direct genetic correlation between WW and YW (0.75) and their permanent environmental correlation (0.99). The bivariate estimates of genetic correlations between direct and maternal effects (Table 4) for WW (-0.35 to -0.47) and YW (-0.49 to -0.55) were slightly less antagonistic than that from univariate analysis (Table 2). However, in the case of BW the estimates from bivariate analysis (Table 4) were as strong as that of the univariate estimate (Table 2). Table 4. Estimates of phenotypic and error variances and genetic parameters for each trait from bivariate analysis of BW, WW and YW Parameters Traits V a,2 h f K 2 hT2 rdm c2 BW + WW 4.352 3.234 0.266 0.087 0.178 -0.572 - BW + YW 4.411 3.030 0.335 0.095 0.207 -0.654 - WW + BW 496.64 400.0 0.260 0.086 0.197 -0.472 0.121 WW + YW 423.77 228.45 0.342 0.080 0.296 -0.346 0.096 YW + BW 496.35 322.65 0.320 0.066 0.233 -0.552 0.044 YW + WW 470.76 264.64 0.407 0.069 0.228 -0.486 0.044 * Abbreviations as in Table 2. Table 5. Estimates of across-trait correlations from pairwise bivariate analysis of BW, WW and YW Trait combinations Parameters BW + WW BW + YW WW + YW fr'd 0.373 0.445 0.750 rm 0.082 -0.033 0.214 ^dlm2 -0.042 -0.276 -0.024 Td2ml 0.239 0.309 0.052 rc -- 0.995 re 0.174 0.069 0.027 rP 0.236 0.188 0.377 Vd, rm, rc, rt and rp: direct genetic, maternal genetic, permanent environmental, residual and phenotypic correlations, respectively; rdlm2 and rd2ml: genetic correlation between direct effect of one trait and maternal of the other, and vice- Trivariate analysis Phenotypic variances, heritabilities and correlation matrices obtained from the trivariate analysis are presented in Tables 6 and 7. Generally, heritability-like estimates from trivariate analysis lie within the range of those from bivariate analysis (Table 4). The exceptions were estimates for YW. As compared with the pairwise analysis, the trivariate analysis showed higher rm and lower rd values with the exception of those of WW and YW, which were virtually the same. Table 6. Heritability (direct or maternal, on diagonal), genetic correlation (above diagonal) and genetic covariance (below diagonal) estimates from trivariate analysis Traits BWmt BWdt WWm WWd YWm YWd Bvym 0.093 -0.547 0.383 0.217 0.249 0.259 BWd -0.378 0.267 -0.048 0.180 -0.316 0.253 w w m 1.399 -0.295 0.072 -0.334 0.229 0.047 w w d 1.678 2.357 -23.16 0.320 -0.029 0.750 YWm 0.889 -1.912 7.332 -1.949 0.072 -0.499 YWd 2.311 3.833 3.768 126.89 -39.04 0.448 + Direct or maternal genetic effects. i Table 7. Phenotypic variance, c2-values (on diagonal) and correlations between traits (phenotypic above diagonal and environmental below diagonal) based on estimates from trivariate analysis Traits BW WW YW BW 4.384 0.220 0.163 WW 0.193 457.12 0.391 YW 0.082 0.031 436.03 WW, 0. 121* YW, 0.888* 0.054+ t= (^-values; %= correlation between permanent environment of WW and YW. Discussion Univariate analysis All the variance component estimates for BW are lower than most other recent esti mates for beef cattle (Meyer, 1992a; Johnson et al., 1992; Waldron et al., 1993). The h / value was also low compared with estimates by the above authors, while the hm2 was close to estimates by Meyer (1992a) and Waldron et al. (1993). The hd , hm2 and the genetic correlation between direct and maternal effect of BW in the present study were close to estimates for Brangus cattle in the USA but were in general disagreement with those reported for Brahman, Beefmaster and Santa Gertrudis (Kriese et al., 1991). A recent review by Mohiuddin (1993) gave average literature estimates of hd, h 2, rdm and c2-values of 0.3, 0.1, -0.35 and 0.03, respectively. A work on a subset of this data assuming a genetic correlation between direct and maternal effects of -0.44 (literature average) and Henderson's method 3 gave hd , hm2 and c2 of 0.11, 0.02 and 0.05, respectively (Arnason and Kassa-Mersha, 1987). The disparity between our estimates of zero c2-value and those of Arnason and Kassa-Mersha (1987) cannot be explained. In their analysis, however, parity divided into two classes had no effect on BW, while in the present study parity divided into six classes had a significant effect. The variance component estimates for weaning weight were close to the estimates for Zebu Cross and were lower than those estimates for Hereford and Angus cattle in Australia (Meyer, 1992a) but were higher than estimates for Hereford and Angus cattle in New Zealand (Waldron et al., 1993). Similarly the hd2 and hm2 values were generally lower than estimates by Meyer (1992a) regardless of the model used but they were close to estimates reported by Waldron et al. (1993). The estimates according to model 4 in the present study were similar to those for Gudali and Wakwa cattle in Cameroon (Tawah et al., 1993). Eler et al. (1992) estimated 0.26, 0.28, and -0.91 for direct and maternal h2 and the genetic correlation between direct and maternal effects for 205-day weaning weight of Nelore cattle. Arnason and Kassa-Mersha (1987) gave hd , h 2 and lv a lu e s of 0.22, 0.11 and 0.12 for Boran cattle on the assumption of a genetic correlation of -0.55 between direct and maternal effect. The literature averages as reported by M ohiuddin (1993) were 0.22, 0.13, -0.15 and 0.07 for hd , hm2 and rdm and c2-value, respectively The variance component and genetic parameter estimates for YW were fairly close to the literature averages reported by Mohiuddin (1993), viz. 0.31, 0.11, -0.26 and 0.03 for h 2, h 2, rdm and c^-value, respectively. Meyer (1992a) estimated a higher hm2 and a smaller negative correlation but lower hd and (^-values for Zebu Cross cattle in Australia, than our estimates. General The genetic antagonism between maternal and direct effects obtained in the present study for all traits was higher than the literature averages (Mohiuddin, 1993). However, they were less strong when compared with estimates by Tawah et al. (1993) for Zebu cattle in Cameroon where they applied model 4. In the case of BW, recent estimates of genetic correlations obtained using an animal model were either positive or close to zero, while earlier estimates where sire-maternal grandsire models were used (Meyer,1992a; Waldron et al., 1993) were negative. The fact that the estimates in the present study were highly negative when animal model was used might be related to the need to maintain an intermediate weight in an unfavourable tropical environment. In the case of WW and YW, other factors might also be involved. One reason for the strong antagonism between direct and maternal genetic effects is assumed to be omission of accounting for possible negative correlation between the permanent environment of the dam and her daughter (Baker, 1980). In other words, an overfed daughter, as a dam, would underfeed her own offspring. The role of such an effect in a harsh environment such as, Abemossa might seem limited. However, there is a possibility that such heifers might grow faster and calve earlier and might exhibit a poorer maternal ability than those with normal growth rale. The probability that this could happen is high in Abemossa where the environment is more variable. Another reason could be the fact that the antagonism increases with delay in weaning age (Hill, 1965 cited by Robison, 1981). Weaning occurred at a later age in the present study than in most other studies referred to above. Another possible reason for the high negative correlation ofcjserved in the present study and others in tropical cattle (Eler et al., 1992; Meyer, 1993c; Tawah et al., 1993) might be related to their adaptive mechanisms to harsh environmental conditions. In extreme environments/ keeping the size of the animal within limits genetically might have an adaptive advantage. The importance of this has been discussed by Tawah et al. (1993). The genetic parameter estimates in the present study, apart from not accounting for direct environmental covariance, could be less reliable due to the family structure and the parameter estimates themselves. In a simulation study Gerstmayr (1992) has shown that the h„2 and the genetic correlation between direct and maternal effect had the largest standard deviation when fewer dams and fewer progeny per dam were recorded. In addition, inaccurate estimates were shown when rdm was highly negative while at the same time the maternal variance was equal to or smaller than the direct variance. This was true in the present data and study. In the case of YW, less than a third of all the dams had their own record, whereas for weaning weight, more than half of them had a record (Table 1).; Due to the large sampling error and poor data structure, the estimates for YW appear to be less reliable. One should also noted some other possible sources of bias in the present study. The genetic parameter estimates according to models 4 and 6 - which were considered 'appropriate' - were higher than estimates according to the other models. This might be partly due to a negative sampling correlation between the genetic covariance (between direct and maternal effects) and the direct and m to n a l genetic variances (Meyer, 1992b). Another possible source of bias could 1 0 be the fact that non-additive genetic variances are assumed to be absent. However, the extent of such bias is expected to be small. The fact that the total variance of YW vis~&~vis WW showed a decrease, based particularly on trivariate analysis, is surprising. The difference between them in mean weights was 22.9 kg in favour of YW (Table 1). The fitness of the models measured as R2 (absorbing the random effects) was 0.38 and 0.60 for WW and YW, respectively. Overall the total variance observed in the present study for all traits was low, even compared with breeds with similar mean values. For example, it was about half of that estimated for BW and about a third of that for WW of Gudali and Wakwa cattle (Tawah et al., 1993). The direct genetic correlations between BW and WW (Tables 4 and 6) were lower than literature averages given by Koots et al. (1991). The estimate of maternal genetic correlation between BW and WW from bivariate analysis was lower than that of Koots et al. (1991), but that of trivariate analysis was similar. However, Meyer (1993b,c) reported higher direct and maternal genetic correlations between BW and WW, and BW vs. YW than in the present study. Compared with the current results, Arnason and Kassa-Mersha (1987) reported a similar phenotypic correlation (0.15) and a much lower genetic correlation (0.01) between BW and WW. The low environmental correlations (Tables 5 and 7) between WW and YW were due to variability in the effect of random environmental factors on each trait. The fact that direct genetic correlations among traits - except those of WW and YW - were relatively low indicated that selection for each trait can be effected independently of the other. The relatively weak correlation between BW and later weights is particularly advantageous. Although the rm is probably to be associated with a high standard error, it is much weaker than many other estimates (Meyer, 1993b,c). The generally low correlation between traits estimated in the present study is in agreement with Meyer (1993c) who found a lower correlation between traits for Zebu Cross than for Angus. Consequently she suggested that the genetic determinants of growth at various ages are more diverse in tropical than in temperate environments. It might also be true that genes controlling growth at different ages (different environments) are different in tropical cattle. During times of relative abundance (favourable environmental conditions) that prevailed pre weaning, genes that increase appetite might be important, while during times of scarcity and stress, genes that are associated with efficiency of feed utilization might be more dominant. Conclusions The analysis showed that a model that includes direct and maternal genetic effects and their correlation for BW and a model which also includes the permanent 11 I environmental effect for WW and YW is "appropriate". The genetic parameter estimates obtained for Boran cattle are within the range of those reported for temperate cattle, but the degree of antagonism between direct and maternal genetic effects appears to be stronger, particularly when based on estimates from univariate analysis. Estimates of genetic parameters from bivariate and trivariate analysis were relatively higher than those of univariate analysis. In addition the extent of variation, particularly the phenotypic variation, observed for all traits was fairly limited. Moreover, the across-trait correlation estimates in the present study were relatively lower than in other studies. Acknowledgments The data were collected by staff at the Ministry of Agriculture (Ethiopia) and this study was financed by the Swedish Agency for Research Cooperation with Developing Countries. We thank Dr. K. Meyer for the use of her DFREML package and for supplying us with offprints of her papers. We also thank Dr. J. Philipsson and Dr. Th. Arnason for their fruitful comments. References Amason,Th.; Kassa-Mersha,H., 1987: Genetic parameters of growth of Ethiopian Boran cattle. Anim. Prod. 44:201-208. Baker, R.L. 1980. The role of maternal effects on the efficiency of selection in beef cattle; a review. Proc. N.Z. Soc. Anim. Prod., 40:285-303. Banlaw, K.; Haile-Mariam, M., 1994: Productivity of Boran cattle and their Friesian crosses at Abemossa Ranch, Rift Valley of Ethiopia. II. Growth performance. 'ijrop. Anim. Hlth. and Prod. 26:49-57. Eleij, J.P.; Ferraz, J.B.S.; Lobo, R.B.; Josakian, L.A., 1992: Genetic antagonism bjetween growth and maternal ability in Nelore cattle. J. Anim. Sci. 70 Suppl. 1:138 (Abstr.). Gerstmayr, S., 1992: Impact of the data structure on the reliability of estimated genetic parameters in an animal model with maternal effects. J. Anim. Breed, cjenet. 109:321-336. Haile-Mariam, M.; Kassa-Mersha, H v 1994: Genetic and environmental effects on age at first calving and calving interval of naturally breeding Boran (zebu) cows ( in Ethiopia. Anim. Prod. 58:329-334. Johnson, Z.B.; Wright, D.W.; Brown, C.J.; Bertrand, J.K.; Brown, A.H., 1992: Effect of including relationship in the estimation of genetic parameters of beef calves. jJ Anim. Sci. 70:78-88. Kocits, K.R.; Gibson, J. P.; Wilton, J.W., 1991: Analysis of genetic parameters for beef cattle. J. Anim. Sci. 69 Suppl. 1: 205-206 (Abstr.). 12 Kriese, L.A.; Bertrand, J.K.; Benyshek, L.L., 1991: Genetic and environmental growth trait parameter estimates for Brahman and Brahman-derivative cattle. J. Anim. Sci. 69:2362-2370. Meyer, K., 1992a: Variance components due to direct and maternal effects for growth traits of Australian beef cattle. Livest. Prod. Sci. 31:179-204. Meyer, Kv 1992b: Bias and sampling covariances of estimates of variance components due to maternal effects. Genet. Sel. Evol. 24:487-509. Meyer, Kv 1993a: DFREML version 2.1. Programs to estimate variance components by restricted maximum likelihood using a derivative-free algorithm. User notes. Animal Genetics and Breeding Unit, University of New England, Armidale, NSW. Mimeo. Meyer, K., 1993b: Covariance matrices of growth trai of Australian Polled Hereford cattle. Anim. Prod. 57:37-45. Meyer, K., 1993c: Estimates of direct and maternal correlations among growth traits in Australian beef cattle. Livest. Prod. Sci. 38:91-105. Mohiuddin, G., 1993: Estimates of genetic and phenotypic parameters of some performance traits in beef cattle. Anim. Breed. Abstr. 61:495-522. Quaas, R.L., 1976: Computing the diagonal elements and inverse of a large numerator relationship matrix. Biometrics 32:949-953. Robison, O.W., 1981: The influence of maternal effects on the efficiency of selection; A review. Livest. Prod. Sci, 8:121-137. Tawah, C.L.; Mbah, D.A.; Rege, J.E.O.; Oumate, H.; 1993: Genetic evaluation of birth and weaning weight of Gudali and two-breed synthetic Wakwa beef cattle populations under selection in Cameroon: genetic and phenotypic parameters. Anim. Prod. 57:73-79. Waldron, D.F.; Morris, R.L.; Baker,R.L.; Johnson, D.L. 1993: Maternal effects for growth traits in beef cattle. Livest. Prod. Sci. 34:57-70. Willham, R.L., 1972: The role of maternal effects in animal breeding: III. Biometrical aspects of maternal effects in animals. J. Anim. Sd. 35:1288-1293. 13 IV Estimates of genetic and environmental trends of growth traits in Boran cattle. Mekonnen Haile-Mariam and Jan Philipsson Dept, of Animal Breeding and Genetics, Swedish Univ. of Agric. Sciences, S-750 07 Uppsala, Sweden. Summary Data from a Boran cattle breeding scheme run by the Ministry of Agriculture (Ethiopia) were analysed by separating the genetic and environmental trends in growth traits. The data used included weight records at birth (BW) and weaning (WW) and as yearlings (YW) for animals born between 1959 and 1985. The mean generation interval as calculated was 6.75 years and 3.29 generations of selection occurred in a 24-year actual selection period, during which inbreeding increased to an average of 1.7% in animals bom in 1985. The regressions of estimated direct and maternal breeding value on year of birth were -0.002 and 0.003 kg for BW, 0.32 and 0.02 kg for WW and 0.22 and -0.005 kg for YW, respectively. The aggregate breeding value, which is the sum of the maternal and direct breeding values, showed an increase of 0.34 and 0.21 kg per year for WW and YW, respectively. Whereas the maternal environmental trends were slight, the variation in the direct environmental effects was the greatest. Generally speaking, a less marked annual genetic trend was discerned, when breeding values were regressed on generation coefficients than on year of birth. Despite the good opportunity for selection as demonstrated by the more than 50 kg difference between the maximum and minimum breeding values for WW and YW, the maximum genetic gain in aggregate breeding value was only 0.2% of the mean WW per year. This limited response was due to the high level of genetic antagonism between direct and maternal effects and also problems in running the selection scheme, such as selection on phenotypic performance alone and relatively long generation intervals. Introduction The main objective of animal breeding is to obtain genetic improvement in economically important traits by means of selection. The progress achieved in any selection work can be assessed by estimating the actual and expected genetic gain. One such method by which breeding schemes can be evaluated is by separating the genetic from the environmental portions of the total phenotypic trend. Monitoring of genetic and environmental trends is crucial in tropical areas where the annual variation in climatic conditions, particularly rainfall, is high. Despite 1 I this, there seems to be little published information on the status of selection schemes in tropical areas (Burrow et alv 1991; Tawah et al., 1994). An analysis of data collected from a Boran Cattle Breeding programme at Abemossa ranch in Ethiopia, which was established to improve beef production through selection, showed that birth weight increased consistently from about 20 to 26 kg (Kassa-Mersha and Arnason, 1986). At the same time, weaning weight showed an increase in the first 10 years and a decline in the subsequent 14 years. Similarly an analysis of data of animals bom at the ranch between 1977 and 1985 showed that birth weight increased by 0.24 kg per annum, whereas weaning weight showed a declining trend (Banjaw and Haile-Mariam, 1994). However, in neither study was the genetic trend separated from the phenotypic trend. This paper attempts to assess the status of the breeding work which was conducted for 26 years, by estimating genetic and environmental trends for direct and maternal effects, assuming variance and covariance estimates reported by Hai e-Mariam and Kassa-Mersha (1994) for growth traits in this herd. In addition, the level of inbreeding and its influence - as well as the effects of certain other fixed factors on growth traits - was considered. Generation intervals for different paths were also estimated. Mat erials and Methods Environment and management The data came from the Abemossa/Adami-Tulu Boran Cattle Breeding and Improvement Ranch in Ethiopia which was established in 1959 by the Ministry of Agriculture. A description of the study area and herd management is given elsewhere (Banjaw and Haile-Mariam, 1994; Haile-Mariam and Kassa-Mersha, 1994; Kassa-Mersha and Arnason, 1986). Although the objective of the ranch was to improve the productivity of Boran cattle through selection, the criteria for selepion were not clearly defined. The general policy at the ranch was to select replacement animals on the basis of their own and parents' weaning weight (phenotypic performance), conformation and colour. In addition, mating of closely related animals was avoided. Data and analysis The j data used for this study were the same as those of Haile-Mariam and Kassa- Mersha (1994) where (co)variance estimates were obtained for birth weight (BW), weaning weight (WW) and yearling weight (YW). The numbers of records available were 5,256 for BW, 4,082 for WW and 2/417 for YW. 2 In Haile-Mariam and Kassa-Mersha (1994) six models were applied to analyse each trait and the most 'appropriate' model was ultimately selected. In the present study the variance and covariance estimates of the selected models were considered as 'true' values and were used to predict breeding values for both maternal and direct genetic effects. For BW a model which included direct and maternal genetic effects and their covariance was used, whereas for WW and YW the model used included a permanent environmental effect of the dam as well. The fixed effects included were year-season of birth or weighing, sex and parity for all traits. In addition, animal's age at weighing and its inbreeding coefficient were included as a covariate for YW and WW. In the case of BW the coefficient of inbreeding and permanent environmental effects were found to be unimportant in a preliminary analysis and were therefore not included in the final analysis. The coefficient of inbreeding was calculated, considering all pedigrees, assuming that animals were not inbred at the time they were purchased from their pastoralist owners. However, 94 animals born at the ranch lacked a sire record. Breeding values and solutions for fixed effects were obtained using the Derivative-Free Restricted Maximum Likelihood (DFREML) programme by Meyer (1993). Aggregate breeding values, the sum of the direct and maternal breeding values according to Azzam and Nielsen (1987), were also calculated for each trait. Then the aggregate breeding value as well as the maternal and direct breeding values of animals bom in the same year were averaged. These averages were plotted and/or regressed on their birth year in order to illustrate trends. The aggregate, maternal and direct breeding values of animals were also averaged within their generation coefficients which were calculated as outlined below. The averages were then regressed on generation coefficients (GC) to assess the effectiveness of the selection programme. The reason for calculating the regression of breeding values on GC, which measures one more than the number of generations of selection (Mrode, 1988), as well, was the fact that some animals were kept for a long time and produced offspring over several years, due to which the generations of selection were in some cases markedly different from their birth year. The GC of the animals was calculated as, GCc=l + (GQ + GCd)/2, where GCa GCS and GCd are generation coefficients of the animal, sire and dam, respectively (Brinks et al., 1961). Foundation animals were assigned a GC of zero. Generation interval was calculated as an average age of parents when their reproducing offspring were bom, considering the four pathways. Cumulative selection differentials (CSD) were also calculated for WW (considered as primary trait) and for BW and YW (considered as secondary traits) for each sex. First the selection differential for each individual was calculated as a deviation in performance from the contemporary group mean (year-sex) after adjusting for 3 parity, age of calf and inbreeding, when applicable. To this deviation, the average CSD of the parents of all progeny bom in the same group was added, as described by Newman et al. (1973). Foundation animals were assigned a selection differential of zero. The mean selection differential per year was estimated from the regression of mean annual CSD (pooled over sexes) on year of birth. Then the expected annual genetic response for each trait was calculated from the total fe2-value, which measures the regression of an animal's total genotype (direct and maternal) on its phenotype (according to Willham, 1972), times mean selection differentials (Sharma et al., 1985). The total h2 was obtained from Haile-Mariam and Kassa- Mersha (1994). Maternal environmental trends were calculated by taking averages of the permanent environmental solutions for cows bom in a given year. Environmental trends for the direct effects were estimated by averaging the solutions for year- season of animals bom (weighed) in a given year. The environmental trends were also plotted and regressed against their year of birth or weighing. Results Generation interval(Gl) and coefficients(GC) The overall average GI was 6.75 years with very little variation between the four paths (Table 1). The average GC for 213 animals bom in the final year (after 24 yeajrs of actual selection) was 4.29 (s.e.-0.02) and it varied between 3.0 and 4.8 among individuals. A similar GI of 6.79 years was calculated as the inverse of GC on the animals' year of birth (1959-85). Tatle 1. Estimated Generation Interval (GI in years) for Boran cattle Pat' iway No. of observed pairs GI±s.e. Sire =>son 31 6.02±0.23 Dam^-son 27 6.25±0.46 Sire^wiaughter 660 6.62±0.06 Dam=>-daughter 591 6.96±0.12 Weighted overall mean 1309 6.75 Inbreeding coefficients The estimated level of inbreeding in the herd increased consistently until 1971 due ma: nly to increased pedigree information, and then fluctuated between 1.3 and 2.3% (Fig. 1). The most inbred animal had an inbreeding coefficient of 26.6%. Of all animals, 26.5% were found to be inbred, with an average inbreeding coefficient of 4.23%. The estimated mean inbreeding for all animals was 1.01%. The major reason for inbreeding was parent-progeny mating. Selection applied Table 2 shows the number of animals selected (selection here defined as having progeny) and the selection applied. For YW, CSD could be calculated for only 60% of the males and 35% of the females. However, females were selected from progeny bom every year except the last 2 years, while males were not selected from those bom in some years. Table 2. Number of selected parents and estimated selection differentials (sd) for birth weight (BW), weaning weight (WW) and yearling weight (YW) in Boran cattle Item Trait BW WW YW No. of animals selected Males 45 45(42)a 45(27) Females 899 899(642) 899(317) Years of selection Males 19 19 11 Females 24 24 17 Mean annual SD (kg)b -0.01 1.34 0.367 Expected annual response (kg)b -0.002 0.281 0.088 aFigures in parentheses for WW and YW are the actual numbers of animals having a record and for which SD was calculated. Tooled over sexes. Genetic trends The differences between the maximum and minimum breeding values are large, especially for WW and YW direct (Table 3). In all traits the direct breeding value estimated was more variable, showing the opportunity for selection. Looking at the estimated breeding values of most widely used non-foundation sires (7 sires with 1,567 progeny) and cows (18 cows with 187 progeny) showed that the presumably selected animals were not necessarily the best according to their breeding value. 5 Table 3. Mean, maximum and minimum estimated breeding values (kg) for growth traits in Boran cattle Trait No. of records* Mean±s.e. Max. Min. BW 5256 Maternal 0.009±0.003 0.802 -0.852 Direct 0.090±0.006 1.943 -2.343 Aggregate 0.099±0.005 1.636 -1.638 WW 4082 Maternal 0.296±0.028 8.671 -9.140 Direct 3.876±0.082 32.189 -19.093 Aggregate 4.172±0.067 23.979 -13.711 YW 2417 Maternal| -0.299±0.023 7.456 -7.889 Direct 2.622±0.082 29.632 -27.65 Aggregate 2.323±0.065 22.448 -20.194 "Number of animals with record of their own in the data. Table 4 shows the regression of estimated breeding values (direct/ maternal and aggregate) on year of birth and GC. The aggregate breeding value was similar to the direct breeding value, for which reason it was not plotted in Figs. 2, 3 and 4. The maternal genetic trend was only significant for BW, where it increased by about 0.003 kg per year, or by 0.029 kg per generation. For both WW and YW the direct breeding value showed a significant increase. However, the extent of the increase was not consistent when breeding values were regressed against year and GC. Dividing the regression of breeding values on GC by 6.75 (the GI) gave a lov er annual genetic gain than that predicted from the annual trends in the case of direct and aggregate breeding value for WW and YW. On the other hand the regression on year of birth showed a lower trend for maternal breeding value of BW and WW. The reason for this discrepancy could be, inconsistency in the gei ietic improvement over the years. The direct breeding value for WW and YW of animals born between 1971 and 1973 and again 1983 and 1985 showed a decline (Figs. 3 and 4). The direct breeding value for YW before 1966 was close to zero, while that of WW showed a fairly consistent increase from 1961 to 1970. 6 Table 4. Genetic and environmental trends (b±s.e.) calculated by regressing estimated breeding values and environmental solutions (kg) on year of birth and generation coefficients (GC) Trait Regressed on year Regressed on GCa BW Maternal genetic 0.003±0.001* 0.029±0.013* Direct genetic -0.002±0.003NS -0.049±0.031NS Aggregate genetic 0.001 ±0.002NS -0.020±0.021NS Direct env.b 0.180±0.020** WW Maternal genetic 0.021 ±0.015NS 0.279±0.106** Direct genetic 0.319±0.040** 1.080±0.329** Aggregate genetic 0.340±0.032** 1.359±0.284** Maternal env. -0.030±0.029ns Direct env. -0.905±0.406* YW Maternal genetic -0.005±0.010NS 0.079±0.085NS Direct genetic 0.220±0.033** 0.758±0.31 O* Aggregate genetic 0.215±0.026** 0.837±0.248** Maternal env. -0.002±0.010ns Direct env. -1.084±0.736NS *=P<0.05; **=P<0.01; NS=not significant; aOnly breeding values are regressed on GC. bDirect environmental trends are regressed on year of birth or weighing. Environmental trends and fixed effects The regression of the maternal and direct environmental trends on year of birth or weighing is illustrated in Table 4. Fig. 2 shows the direct environmental trend for BW and Fig. 5 for WW and YW. The maternal environmental trends which were close to zero are shown in Figs. 3 and 4 for WW and YW, respectively. The direct environmental trend for BW showed a consistent increase throughout the study period, with an average annual increase of 0.18 kg. The direct environmental trend for WW and YW showed an inconsistent but generally declining trend. The worst years were between 1974 and 1978 and between 1984 and 1986. The maternal environmental trend was less important for all traits. Apart from the effect of year-season, the other fixed effects were also found to have a significant effect (Table 5). Female calves weighed 0.9,13.9 and 19.9 kg less than males at birth, weaning and at 1 year of age. The relation of parity with weight was rather curvilinear, whereby calves of first and second parity and of seventh and above were lighter than those of third to sixth parity. Weights decreased by 0.34 and 0.56 kg for every 1% increase in the inbreeding coefficient of the calf for WW and YW, respectively (Table 5). 7 Table 5. Solutions for fixed effects (except year-season) for BW, WW and YW Effect BW(kg) WW(kg) YW(kg) solution±s.e. solution±s.e. solution ±s Mean 23.6 168.6 191-5 Parity 1 0.0 0.0 0.0 2 0.25±0.10 4.72±1.13 0.57±1.56 3-4 0.45±0.10 7.45±1.11 3.12±1.50 5-6 0.47±0.10 5.84±1.24 2.24±1.67 7 0.45±0.12 0.73±1.39 -2.41 ±1.87 8+ 0.26±0.15 -5.61 ±1.84 -5.99±2.59 Sex Male 0.0 0.0 0.0 FeiW e -0.86±0.G5 -13.92±0.64 -19.86±0.94 Regression on age at weighing Linear - 0.26±0.G2 0.17±0.01 Quadratic - -1*5'3±3.0"10 25A± 3 S n Regression on inbreedine(%) Linear -0.34±0.1! -0.56±0.14 Discussion Generation interval(GI) and coejfidents(GC) The weighted average GI (Table 1) is consistent with the 6.6 year estimated by co:risidering the calf mortality rate, age at first calving and calving interval of the s.amie herd (Haile-Mariam, 1987). Longer GI than that in the present study was reported for Gir cattle (8.02 years) in Brasil (Queiroz and Lobo, 1993) and for Wakwa cattle (8.67 years) in Cameroon (Tawah et al., 1994). However, this is still relatively long when compared with the overall average of 4.36 years (Mrode, 1988) obtained in several beef cattle selection experiments, though his estimate was based largely on temperate cattle breeds. The reason for the long GI observed in the present study apart from delayed age at first breeding is the continuous use of bulls or cows for several years and that no male animals were selected from among those bom in some years (Table 2). Four sires which had altogether more than 1114 progeny in the data were used for over 8 years each. Similarly some cows had more than 10 progeny which implied that they were used as breeding cows for over 10 years. Inbreeding coefficients Although the level of inbreeding may be regarded as low, the fact that some 94 animals had an unknown sire suggests that the coefficient calculated is a minimum value. As the variation in the level of inbreeding among animals was also high, measures to avoid inbreeding should be considered. First, breeding animals should be used for a limited period of time, which will also help to reduce GI. Secondly, it is necessary to avoid parent-progeny matings as inspection of the pedigrees showed such matings to be the main cause of inbreeding. Selection applied Expressed in standard deviation units, the selection differentials observed for WW and YW are only 0.06 and 0.02 (average over sex), respectively, compared with an average of 0.23 for YW estimated from several selection experiments by Baker et al. (1991). The designation of WW as a primary trait of selection is not strictly valid as 3 sires and 257 cows were used as parents without having records of their own for WW. One possibility is that the records on these animals might have been used during selection, though they could not be traced in the data used for this study. The most probable reason, however, is that they were selected on the basis of pedigree, colour or conformation. Genetic trends The difference in estimated breeding values for all traits between the individuals shows the possibility for selection. The differences between the maximum and minimum for WW are similar to those reported for 205-day weaning weight of Angus cattle in the USA (Johnson et al., 1992). When using mixed-model methodology for separating genetic trend from environmental trends, three conditions should be fulfilled (Sorensen and Kennedy, 1984). First, genetic and phenotypic (co)variances of the populations before selection should be known. Second, selection should be based on a linear function of the records. Third, the complete relationship matrix must be used in the analysis. Almost all three assumptions were met in the case of BW and WW. However, due to the fact that there was some selection based on WW and conformation or colour after weaning, the genetic trend for YW might have been biased. This was also shown in the higher direct genetic variance estimates for YW from multitrait vis~ 9 The 0.21 kg annual genetic trend (aggregate breeding values) observed in YW was therefore mainly correlated response due to selection for WW. This is to be expected, as the direct genetic correlation between the two traits was 0.75 (Haile- Mariam and Kassa-Mersha, 1994). The reason for the lower trend in YW than in WW could also be explained by this. Estimation of breeding values in a multitrait analysis using genetic parameter estimates of trivariate analysis could have eliminated this bias and would perhaps have given more accurate estimates. The accuracy of the estimated breeding values was not computed, though it is expected to be poorer due to the fact that maternal effects are an embedded trait and the two breeding values (maternal and direct) are estimated from the same phenotypic record. They are also likely to be more closely correlated than would be the case for two distinct traits. Roeche and Kennedy (1993) concluded from a siniulation study that the absence of disparity in the correlation between the true breeding values of direct and maternal traits on the one hand and the estimated breeding values of the same traits on the other hand could be considered as an indication of the estimation of the one, independent of the other. In the present study the correlation of the estimated breeding values between maternal and direct effects were -0.54, -0.62 and -0.80 for BW, WW and YW, respectively. They are similar to the assumed 'true' genetic correlations of -0.55, -0.57 and -0.68 (Haile-Mariam and Kassa-Mersha, 1994) for the corresponding traits. The fact that the disparity between the two correlations was slightly higher in the case of YW is «n indication that the breeding values for BW and WW were more accurately esti mated than that of YW. This is likely to be true as there were more BW and WW records than YW in the data. Another reason for the disparity could be the fac ithat all the assumptions for using mixed-model methodology to separate ger etic from environmental trends were not fully met in the case of YW. The estimates of genetic trend in the present study are close to the genetic trends of 3102 and 0.40 kg per year for BW and WW reported for a Hereford herd maintained in a nutritionally stressful environment and selected by visual appraisal in USA (Zhang et al., 1991). Cantet et al. (1993) reported 0.92 kg per year as an average regression of direct breeding value, but a near-zero maternal brefeding value for WW of Angus cattle in Argentina. The trends in breeding values (both direct and aggregate) for WW and YW are moldest when compared with an average of 1.15 kg and 2.65 kg per year for the corresponding traits as reported by Mrode (1988) after reviewing the results of selection experiments. The majority of the estimates of genetic trend reported by Mrode (1988) are from cattle breeds in temperate environments. Few studies on rate of improvement have been reported from tropical environments based on zetju (Packer et al., 1986) or Bos indicus-taurus crosses (Burrow et al., 1991). In both casBs a higher level of gain was reported in a single generation of selection than in i;he present study. Tawah et al. (1994) also reported annual changes in sire estimated transmitting ability of 0.67 and 1.69 kg per year for WW in Gudali (African zebu) and Wakwa (a synthetic breed between Gudali and Brahman), respectively, in Cameroon. In this herd, theoretically, considering a total h2 of 0.21 and 0.24 and phenotypic standard deviations of 22.2 kg and 22.4 kg for WW and YW (Haile-Mariam and Kassa-Mersha, 1994), respectively and assuming an estimated selection intensity of 1.54 (Haile-Mariam, 1987) would give a genetic gain of about 1.0 kg per year in both traits, if single-trait individual selection was practised. The genetic gain observed (Table 4) is less than about a third of what seems possible even if one considers the regression of breeding value on year of birth. The reason for the low genetic gain is that animals were selected on the basis of their own and their parents' unadjusted WW as well as on colour and conformation. In addition, male animals in particular were not selected at all among those bom in some years (Table 2) and also those selected were not necessarily superior to their contemporaries. Moreover, the genetic antagonism between direct and maternal effects was strong suggestive of the difficulty of improving them simultaneously. Weighing the estimated direct and maternal breeding values by equal values, as in the present study, might be reasonable when evaluating the program in retrospect. However, when planning future selection strategies the weighing factor should probably depend on production cost. General The estimates of expected genetic responses (Table 2) lie within the range of estimates from the regression of aggregate breeding value on year and GC (Table 4) in the case of BW and WW but are close to that of regression on GC in the case of YW. The decline in estimated breeding values in 1971 and 1977 coincided with the increase in the level of inbreeding of animals bom in these years. This was the same when inbreeding was not included as a covariate as well. The fall in estimated breeding values in 1983-85 might be related to the emphasis placed on measures which reduce inbreeding rather than on those which increase performance. The decline could also be associated with the occurrence of drought during these years, which might have reduced expression of genetic difference. The fact that there was a decrease in the direct breeding value of BW and that the direct environmental trend increased markedly is difficult to speculate about. However, the reasons for such an increase might be associated with an inaccurate separation of the trend into its components or the introduction of some kind of systematic error. On the other hand the near-zero selection differential of BW (Table 2) is in agreement with the absence of any genetic trend. The decreases in 11 direct environmental trend observed for BW during years of drought were consistent with those of WW and YW. The importance of environmental effects (particularly year-season) in affecting performance in a tropical environments is clearly shown in Figs 2 and 5. The decline in direct environmental effects observed in 1974-78 is consistent with the drought and years of instability in the country. The drought of 1974 was followed by years of political instability, due to which the grazing area allocated for animals was drastically reduced. Again, 1984 saw the outset of another cycle of drought which seriously marred the performance of the animals. The regression of WW and YW on percentage inbreeding is similar to average reported by Burrow (1993) who estimated a decline of 0.44 and 0.69 kg, respectively, per 1% increase in inbreeding. The absence of any decline in the case of BW is inconsistent with Burrow's (1993) estimate of 0.06 kg decline. However, Burrow also concluded that the effect of inbreeding on BW is marginal compared with that on growth from birth to maturity. In the present analysis, inbreeding of the animal was included as a covariate, while that of the dam was disregarded, for two reasons: first, the estimated level of inbreeding in the dams was less than 1% which might suggest that its effect is probably low. Secondly, the inbreeding level of the calf and its dam may be correlated, in which case the inclusion of both mi^ht cancel each other out. Conclusion The genetic trend estimates indicate the possibility of improving growth traits in Boran cattle. The maximum annual genetic gain observed for the aggregate breeding value of WW was 0.2% of the mean (168.9 kg). This limited response is to be expected, due to the fact that selection was based on phenotypic performance, colour and conformation. Also the relatively high level of genetic antagonism between direct and maternal effects in Boran cattle reduced the possibility of genetic gain. The marked variation between years in direct environmental trend suggests that improvement in environmental conditions shctild be part of the genetic improvement program if the advantages of the selection work are to be realized. Moreover, animals should be selected on the basis of their estimated breeding value rather than on phenotypic performance. In addition, measures to reduce both coefficient of inbreeding and generation interval need to be considered. Acknowledgments i We thank the staff at the Ministry of Agriculture (Ethiopia) for data collection and SA'^EC for financially supporting this study. We also thank Dr. K. Meyer for her DF SEML Program and Dr. Th. Arnason for useful comments. R eferences Azzam, S.M.; Nielsen, M.K., 1987: Expected responses to index selection for direct and maternal additive genetic effects of gestation length or birth date in beef cattle. J. Anim. Sci. 64:357-365. Baker, R.L.; Morris, D.L., C.A.; Johnson, D.L.; Hunter, J.C.; Hickey, S.M. 1991: Results of selection for yearling or 18-month weight in Angus and Hereford cattle. Livest. Prod. Sci. 29:277-296. Banjaw, K.; Haile-Mariam,M.,1994: Productivity of Boran cattle and their Friesian crosses at Abernossa Ranch, Rift Valley of Ethiopia. II. Growth performance. Trop. Anim. Hlth. and Prod. 26:49-57. Brinks, J.S.; Clark, R.T.; Rice, F.J., 1961: Estimation of genetic trends in beef cattle. J. Anim. Sci. 20:903 (Abstr.). Burrow, H.M., 1993: The effect of inbreeding in beef cattle. Anim. Breed. Abstr. 61:737-751. Burrow, H.M.; Seifert, G.W.; Hetzel, D.J.S., 1991: Consequences of selection for weaning weight in Zebu, Bos taurus and Zebu x Bos taurus cattle in the tropics. Aust. J. Agric. Res. 42:295-307. Cantet, R.J.C.; Gianola, D.; Misztal, I.; Fernando, R.L., 1993: Estimates of dispersion parameters and of genetic and environmental trends for weaning weight in Angus cattle using a maternal animal model with genetic grouping. Livest. Prod. Sci. 34:203-212. Haile-Mariam, M., 1987: Evaluation of reproductive and growth performance of Boran cattle and their crosses with Friesians at Abernossa, Ethiopia. M.Sc. Thesis, Alemaya University of Agriculture, Dire Dawa, Ethiopia. Haile-Mariam, M.; Kassa-Mersha, H., 1994: Estimates of direct and maternal (co)variance components of growth traits in Boran cattle. [J. Anim. Breed. Genetics. Accepted]. Johnson, Z.B.; Wright, D.W.; Brown, C.J.; Bertrand, J.K.; Brown, A.H., 1992: Effect of including relationship in the estimation of genetic parameters of beef calves. J. Anim. Sci. 70:78-88. Kassa-Mersha, H.; Arnason, Th., 1986: Non-genetic factors affecting growth of Ethiopian Boran cattle. World Rev. Anim. Prod. 22(2):45-55. Meyer, K., 1993: DFREML version 2.1. Programs to estimate variance components by restricted maximum likelihood using a derivative-free algorithm. User's notes. Animal Genetics and Breeding Unit, University of New England, Armidale, NSW. Mimeo. Mrode, R.A., 1988: Selection Experiments in Beef Cattle. Part 2: A review of responses and correlated responses. Anim. Breed. Abstr. 56:155-167. Newman, J. A.; Rahnefeld, G.W.; Fredeen, H.T., 1973: Selection intensity and response to selection for yearling weight in beef cattle. Can. J. Anim. Sci. 53:1- 12. 13 Packer, I.U.; Razook, A.G.; Trovo, Bonilha, L.M.; Figueiredo/ L.A.; Nascimento, J.; Facola, I.; Candido, J.G.; Campos, B.E.S.; Machado, W.B., 1986: Selection for yearling weight in Nelore and Guzera zebu breeds: selection applied and response. Proc., 3rd World Congress on Genetics Applied to Livestock Production. 11:419-423. Queiroz, S. A.; Lobo, R.B., 1993: Genetic relationship, inbreeding and generation interval in registered Gir cattle in Brazil. J. Anim. Breed. Genet. 110:228-233. Roeche, R.; Kennedy, B.W., 1993: The influence of maternal effects on accuracy of evaluation of litter size in swine. J. Anim. Sci. 71:2353-2364. Sharma, A.K.; Willms, L.; Hardin, R.T.; Berg, R.T., 1985: Selection response in a purebred Hereford and multibreed synthetic population of beef cattle. Can. J. Anim. Sci. 65:1-9. Sorensen, D. A.; Kennedy, B.W., 1984: Estimation of response to selection using least-squares and mixed model methodology. J. Anim. Sci. 58:1097-1106. Tawah, C.L.; Rege, J.E.O.; Mbah, D.E.; Oumate, H, 1994: Genetic evaluation of birth and weaning weight of Gudali and two-breed synthetic Wakwa beef cattle populations under selection in Cameroon: genetic and phenotypic trends. Anim. Prod. 58:25-34. Willham, R.L., 1972: The role of maternal effects in animal breeding: III. Biometrical aspects of maternal effects in animals. J. Anim. Sd. 35:1288-1293. Zhang, M.H.; DeNise, S.K.; Golden, B.L., 1991: Comparison of genetic trends of pte and post weaning traits estimated by single- and multiple-trait animal models. J. Anim. Sd. 69 Suppl. 1:215 (Abstr.). -o- direct genetic - - maternal genetic — direct environmental environmental direct — genetic maternal - - genetic -o- direct Fig. 2. Genetic and environmental trends for birth weight. birth for trends environmental 2.Fig.and Genetic Fig. 1. Trend in inbreeding coefficient. inbreeding in 1.Trend Fig. inbreeding (% ) 2.5 f—2.5 4 6 8 0 2 4 6 8 0 2 84 82 80 78 76 74 72 70 68 66 64 ero birth of year year of birth of year 15 year of birth —- direct genetic - - maternal genetic - mat. environmental Rg. 3. Genetic and maternal environmental trends for weaning weight. year of birth direct genetic - - maternal genetic •«- mat. environmental Fig. 4. Genetic and maternal environmental trends for yearling weight. 16 wetght-kg -— -wwyw Fig. 5. Direct environmental trends for weaning (ww) and yearling (yw)weight. yearling and (ww) weaning for trends Directenvironmental Fig. 5. 17 Y Anim. Prod. 1994, 58:329-334 0003-3561 /94/32320329$02«0 © 1994 British Society of Animal Production Genetic and environmental effects on age at first calving and calving interval of naturally bred Boran (zebu) cows in Ethiopia M. Haile-Mariam and H. Kassa-Mersha Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, S-750 07 Uppsala, Sweden f t 'S Abstract Twenty-four years of data from a Boran cattle breeding and improvement ranch in the Rift Valley of Ethiopia were used to study the influence of genetic and environmental factors on age at first calving (AFC) and calving interval (Cl) using an individual animal model. The mean AFC and Cl were 41-8 months and 442 days, respectively. The h- values for AFC were 0-062 and 0-075 when estimated on the original and selected data, from which coivs that did not calve after 5 years of age were deleted, respectively. The h2 values for Cl were 0-037 and 0-043 when estimated on the original and selected data, from xvhich cows loith AFC higher than 5 years and Cl longer than 2 years were deleted, respectively. The corresponding c2 values (ratio of permanent environmental variance to total) were 0 031 and 0-028 for the respective data sets. Parameter estimates of the original data from bivariate analysis ivere 0-012 {TrJ, 0-065 (h2), -0-054 and -0-176 for first Cl, AFC, their genetic and environmental correlation, respectively. The h2 values estimated in a bivariate analysis of the first three CIs varied between 0-002 and 0-093. Year and season had significant effects on both traits. The annual genetic change for both traits was not significant while the solutions for year effects were highly variable. The regression of the solutions of year effects for AFC on year showed that it increased by about 10 days while the variation in Cl was due to random year to year fluctuation. K ey w o rd s: Boran, environmental factors, fertility, genetic parameters. Introduction Abernossa ranch in the Rift Valley of Ethiopia was The main reason for the relatively low reproductive reported by Haile-Mariam, Banjaw, Gebre-Meskel efficiency of cattle in the tropics is recognized to be and Ketema (1993). This study was conducted to environmental. However, the limited evidence determine genetic parameters for age at first calving available from tropical regions suggest that there is (AFC) and calving interval (Cl) and estimate the also substantial genetic variation (Seebeck, 1973; environmental effects and genetic trends using 24 Thorpe, Cruickshank and Thompson, 1981; years of data. In addition the influence of Mackinnon, Taylor and Hetzel, 1990). The majority retrospective culling of cows, on the basis of AFC of past studies on reproductive performance from and Cl, on genetic parameter estimates, was also tropical areas have been largely limited to the assessed. assessment of effects of non-genetic factors and breed differences (e.g. review by Galina and Arthur, 1989). Genetic parameter estimates and information Material and methods on the extent of within breed variation is scarce. Environment, animals and management Even of those few estimates available from tropical The climatic conditions and environmental areas, the majority are of temperate breeds or their description of the Adami Tulu and Abernossa crosses kept in tropical or subtropical environments ranches is given elsewhere (Kassa-Mersha and (Meyer, Hammond, Parnell, Mackinnon and Arnason, 1986; Haile-Mariam et al., 1993). The Sivarajasingam, 1990; Mackinnon et al., 1990). animals were maintained at Adami Tulu (165 km Generally, estimates of genetic parameters for Bos south of Addis Ababa) until 1974 and were indicus (zebu) cattle, with the exception of Brahman transferred to Abernossa (15 km further south) in or their derivatives, are rare. 1975. The grazing area allocated was about 1500 and 1100 ha at Adami Tulu and Abernossa ranches, The influence of fixed factors on fertility of Boran respectively. The climatic conditions were the same cattle and their crosses using data (1977 to 1983) from at both ranches but the grazing area per animal was 329 330 Haile-Mariam and Kassa-Mersha less in the latter. The average annual rainfall is about vector of unknown fixed effects which included 700 mm with most precipitation between July and season and year of calving, sex of calf and parity (For September. Short wet seasons occur between March AFC it included year and season of birth, season of and June. From November to February rainfall is rare conception and parity of birth. A season included 4 and natural vegetation is poor in quality and months defined on the basis of rainfall and pasture quantity. The mean minimum and maximum conditions as indicated in the previous section.); u, c temperatures are 12-0°C and 27-1 °C, respectively, and e are vectors of unknown random animal and with the highest maximum temperatures occurring permanent environmental effects and random in the short wet Season. residual error, respectively; W is the incidence matrix relating records to cows with observations (fitted for Animals grazed on natural pasture composed mainly Cl only). of Hyperheiiin species scattered at varying densities w ith Acacin trees. During the 1973, 1980 and 1984 Estimation of variance components was carried out drought animals were supplemented in the dry with the derivative-free restricted maximum season with hay and concentrate made of wheat bran likelihood (DFREML) programs of Meyer (1989) and oil-seed cake. using an individual animal model. Convergence was considered reached when the variance of function Theifoundation stock were 323 female and 10 male values (-21og L) in the Simplex was less than 10-8. aninftals purchased in 1959 from the Borana area in Sampling errors as well as likelihood ratio tests (LRT) southern Ethiopia. Cows were divided into five to were obtained for h2 and c2 values. Genetic parameter seven breeding groups each composed of 40 to 50 estimates were also obtained from bivariate analysis cows and a bull. Occasionally cows and bulls were of the first Cl with AFC and the first three CIs of the mov'ed from group to group. On a few occasions original data. breeding bulls with disease or fertility problems wer£ removed and replaced after a minimum of 25 Finally, variance component estimates, obtained dayj by other unrelated bulls. Until 1980, cows were from the analysis of the original data, were assumed exposed continuously to bulls. In 1981, a move to be the true values to predict breeding values. towards a seasonal (September to December) mating Prediction of breeding values and hypothesis test of scheme was started. Weaning occurred at about 8 fixed effects was carried out by the p e s t package months of age in a batch system. Replacement (Groeneveld, 1990) using the model above. Although animals were selected, based on their birth and predicted breeding values (PBV) were obtained for weaning weight and their own conformation, as well all animals through their relationships, only PBVs of as jon their pedigree, to avoid inbreeding. animals with their own observation were averaged Replacement heifers were considered sufficiently within birth year to obtain genetic trend. The matjire to be mated after 24 months of age and at solutions for year effects were considered as the about 250 kg weight. However, this policy was not estimate of environmental trend. The solutions for strictly followed and thus the age of heifers at first year and the averaged breeding values were plotted exposure varied from 24 to 36 months. against year. Data and analysis AFQ data included animals bom between 1959 and Results 1981 and Cl data included calvings which occurred Genetic parameters betvyeen 1959 and 1983. Then a data set, from which The means and measures of variations along with the AFC higher than 6 years and Cl longer than 3 years number of records are shown in Table 1. Table 2 were deleted, was considered as the original data. shows variance components and parameter Selected data of AFC were formed after deleting AFC estimates. The h2 values were slightly higher when higl er than 5 years. In the case of selected Cl data, the analyses were carried out on the selected data. AFu higher than 5 years and Cl longer than 2 years, The h1 values of AFC from univariate analysis were and all their records thereafter, were deleted. Both not significantly different (at a 5% level of error the selected and original data were analysed probability) from zero, contrasting LRT criteria of assuming the following model shown in matrix 2-60 and 3-13 for original and selected data, notation: respectively, with a x2 value of 3-84 for 1 degree of freedom . | y = Xb + Zu + (Wc) + e The c2 value (the ratio of the permanent where, y is the vector of observations; X and Z are environmental variance to the total) was slightly knoivn incidence matrices relating records to fixed lower when estimated on the selected data. Given the and1 random animal effects, respectively; b is the h2 valu e, the LRT criteria (3*60) for c2 sh ow ed that it Calving age and interval in zebu cows 331 Table 1 Da ta structu re, u nadjusted meansand measuresofvariation Table 3 Estimate of parameters from bivariate analysis of age at of the original and selected data first calving (AFC) and first calving interval (Cl) and the first three Cfet Age at first calving Calving interval AFC cnt CI1 CI2 CI3 Item Original Selected Original Selected AFC 0-065 -0-054 No. of records 875 857 4094 3493 41946 ^ 1 No. of cows 875 857 939 878 cut -0-176 0-012 Mean (days) 1271-4 1257-7 4417 422 0 -4632 17355 s.d. (days) 223-8 204-7 129-2 95-3 cn 0-015 0-999 0-653 CV 0-176 0-163 0-292 0-225 17509 143 211 Minimum (days) 727 727 300 300 CI2 -0-006 0-002 0-997 Maximum (days) 2092 1816 1091 730 -95 16140 41 C13 0-008 -0-011 0093 107 -140 11374 Table 2 Genetic variance and parameter estimates for age at first calving and calving interualf t /j2and phenotypic variances (pooled values) on thediagonal, genetic correlation and covariances above diagonals and Age at first calving Calving interval environmental correlation and covariance below diagonal. C llt, CI1, CI2 and CD are first Cl when analysed with AFC, Parameter Original Selected Original Selected first, second and third CIs, respectively. Table 4 2602 2487 547 353 Solutions and significance levels for season of conception a,2 462 229 and birth and parittf for age at first calving a,.2 38609 30892 13962 7615 No. Mean s.e. 41211 33379 14972 8197 /l2V 0-063 0075 0037 0-043 Overall 875 1271 (0-049) (0-051) (0-015) (0-019) Season of conception *» c2 0-031 0-028 November to February 288 0 (0-017) (0-020) March to June 283 -74 18 July to October 304 -18 17 t o„2, of2 and a,.2, are the phenotypic, additive genetic, a * , Season of birth ♦ permanent environmental and error variances, respectively. November to February 370 0 March to June 276 - -30 19 was not different from zero at a 5% level of July to October 229 -18 19 probability while the estimate on the original data Parity of birth was (LRT = 5-95). Ignoring permanent environmental 1 158 0 effect the h2 value of the selected data was 0 056 (s.e. 2 183 31 26 118 1 28 0014). 3 4 115 12 29 54- 301 3 25 The genetic correlation between AFC and first Cl, though close to zero, indicates possible antagonism Table 5 Solutions and significance levels for season of calving, sex between the two traits (Table 3). The It2 of the first Cl of calf and parity for calving interval estimated from the analysis with AFC is similar to that obtained from the analysis with second and No. Mean s.e. third Cl. However, the sampling errors of all the Overall 4094 442 parameters, though not estimated, are expected to be * high. Season of conception November to February 1487 0 March to June 1276 12 5 Fixed effects and genetic trend July to October 1331 5 5 The solutions for fixed effects are shown in Table 4 Sex of birth (AFC) and Table 5 (Cl). In both traits the solutions Male 1978 0 for year effects showed high variation (Figure 1). In Female 2116 -A 3 the case of AFC season of birth and conception were Parity of birth *» also significant (P < 0 05) but parity of birth was not 1 922 0 significant (P > 0-05). Similarly the effect of sex of calf 2 752 -9 6 was not significant (P > 0 05) in the case of Cl. 3 624 -35 7 4 493 -21 7 5 411 -21 8 The genetic and environmental trends are shown in 6+ 892 -29 8 Figure 1. The average breeding values of AFC and Cl 332 Haile-Mariam and Kassa-Mersha The fact that removing AFCs longer than 5 years decreased the error variance without marked effect on the additive genetic variance suggests that non additive genetic factors that were unaccounted for by the model fit had affected the performance of the late calving animals. This was more so in the case of Cl, where the elimination of extreme CIs decreased the error variance by about half with small effect on the additive genetic variance. Similarly Milagres, Dillard and Robison (1979) reported that deleting records of barren cows (that did not calve for 2 years) increased It2 of calving rate from 0 01 to 0-21 as a result of a small increase in genetic variation and a substantial decrease in the environmental variance. However, Duarte-Grtuno, Thorpe and Tewolde (1988) reported variable changes in the h2 estimates after deleting Year of birth (calving) higher AFCs and longer CIs. Figure 1 Genetic and environmental trends for calving interval (CO ?nd age at first calving (AFC) of the analysis based on the The h2 values estimated for Cl are close to those original data: (—▲— AFC-environmental and —Q— genetic reported for beef cows (zebu) in the Tropics of trends; —■— Cl-genetic and —0— environmental trends). Mexico (Duarte-Ortuno el al., 1988) but are lower than those for Sahiwal cattle in Kenya (Rege, Lomole and Wakhungu, 1992). Heritability estimates of Cl on birth year did not show any trend and were close for beef cattle are scarce in the literature due to the to zero. However, the solutions for year of AFC fact that either calving rate or calving data (also tended to increase by about 10 (s.e. 3) days every called days to calving) are considered to be yean (P < 0-01). The lowest AFC was in 1962 (142 appropriate measures of fertility in seasonal mating days below the mean) while the highest AFC was in schemes. In the present study, where year-round 1973 (259 days above the mean). In the case of Cl the mating took place for the majority of the period, Cl shortest interval (37 days below the mean) was in appears to be a comparable measure of reproduction. 1969 and the longest (121 days above the mean) was With this assumption, the h2 values for Cl lie within in 1975. As heifers bom in 1973 are expected to calve the range of the estimate for calving rate in tropical sometime after 1975, the reason for the longest Cl Australia of 0-07, 0-02 and 0-171 for Hereford, Angus and highest AFC were likely associated with and zebu crosses respectively (Meyer et al., 1990). managerial problems which happened during the Thorpe et dl. (1981) reported h2 value of 0-08 for transfer of the animals from Adami Tulu to calving rate of Boran in Zambia. Abernossa in 1975. The repeatability (r) values, the sum of c2 and I12, are close to estimates by Duarte-Ortuno et al. (1988) but Discussion are lower than estimates for days to calving by The mean AFC and Cl is higher than estimates for M eyer et al. (1990) for zebu crosses. The slightly the same breed under ranch conditions in Kenya lower c2 value for the analysis of the selected data (Trail;and Gregory, 1981; Trail, Gregory, Durkin and might be related to the decrease in the number of Sanafjord, 1984). However, the mean for Cl after records per covy from 4-36 to 3:98. The low r values records longer than 2 years were deleted was similar are in agreement with the negative environmental to tne mean of 421 days reported by Trail et al. (1984). correlations between adjacent CIs. This indicates that The means in the present study are in agreement cows with first short CIs were requiring a rest period with the averages (41-5 months for AFC and 448-4 and thereby prolonging their second interval, days for Cl) reported by Kassa-Mersha and Arnason perhaps due to a higher degree of environmental (1986) but were lower than the least squares means stress. Although genetic correlations between (454 months for AFC and 465 days for CD reported adjacent CIs were unity, the sampling errors by Haile-Mariam et al. (1993), both working on data associated with them were likely to have been high, from the same ranch, given the size of the data. The low h1 value for the first and second Cl might be related to the Thes/i2 value of AFC is similar to that estimated by vulnerability of young cows to environmental stress. Boufdon and Brinks (1982) for temperate beef breeds in USA. Trail and Gregory (1981) observed no The significance of the fixed effects in the present add tive genetic variation for AFC in Kenyan Boran. study is in general agreement with the literature Calving age and interval in zebu cows 333 (Duarte-Ortuno et al. 1988; Trail et al., 1984). different Cl varied considerably, improvement of Including weaning age of previous calf and adjusted these traits through selection is limited. Variation in weaning weight as a covariate for Cl and AFC AFC could arise from differences in growth rate, age respectively, did not have any statistically significant at onset of puberty and the ability to conceive effect. Information on lactation status at joining for (Galina and Arthur, 1989), which means that the h2 Cl (M eyer et al., 1990) and age at first exposure observed reflects variations in a number of (Bourdon and Brinks, 1982) for AFC might have been interrelated traits. Therefore, considering alternative important had it been available. However, the former traits, such as scrotal circumference, which is can be recorded in a seasonal mating scheme only favourably correlated with semen characteristic and and the effect of the latter might have been removed a good indicator of age at puberty in both sexes by year effect as animals born at the same year were (M eyer et al., 1990), might be important. The other probably exposed for breeding at the same time. limitation of both AFC and Cl as measures of fertility is that they exclude animals which are not calving For Cl the absence of any environmental trend and which might justify a shift to recording other (P > 0-05), despite year itself defining considerable traits like calving rate or days to calving. In the short variation, suggests that the significance of year term, however, improvement in fertility in this herd effects is mainly due to random annual fluctuation in has to come from improved management, including rainfall and thus food availability. The somewhat better data recording, timely culling of cows with consistent increase in AFC over the years suggests long Cl and high AFC and using bulls with known that this might have been due to a change in age at semen characteristics. first exposure for breeding. Heifers were mated at about 2 years of age in the earlier years but this policy seems to have changed to mating at about 3 Acknowledgements years. 'The fact that this effect was removed as The data were collected by the staff of the Ministry of Agriculture, Ethiopia. This study was financially supported environmental trend suggests that it is unlikely to by the Swedish Agency for Reseach Cooperation with have influenced the genetic parameters estimated. Developing Countries (SAREC). The assistance of Dr ]. That the regression of predicted breeding value on Philipsson in the preparation of the manuscript and Mr A. year of birth was close to zero for both traits is in Sigurdsson in data analysis is acknowledged. We thank Dr agreement with the fact that there was no selection K. Meyer and Dr E. Groeneveld for the use of their for fertility. computer packages. The performances in terms of Cl and AFC in this References herd were low, particularly in view of the high level Bourdon, R. M. and Brinks, J. S. 1982. Genetic, of fertility of the breed (Trail and Gregory, 1981; Trail environmental and phenotypic relationships among el al.. 1984; Maule, 1990). One explanation for this gestation length, birth weight, growth traits and age at first could be that breeding bulls were not evaluated for calving in beef cattle, journal of Animal Science 55: 543-553. their semen quality or quantity and that cow to bull Duarte-Ortuno, A., Thorpe, W. and Tewolde, A. 1988. ratio was on the higher side when compared with Reproductive performance of purebred and crossbred beef other herds (e.g. Mackinnon et al., 1990). The effect of cattle in tropics of Mexico. Animal Production 47:11-20. these factors could not be assessed because Galina, C. S. and Arthur, G. H. 1989. Review of cattle information on calving or conception rate for each reproduction in the tropics. Parts 1 and 2. Animal Breeding bull could not be derived. Another reason for the low Abstracts 57: 583-590; 679-686. level of fertility could have been inbreeding. G roeneveld, E. 1990. PEST user's manual. Department of However, the level of inbreeding which was 1-1% Animal Science, University of Illinois, Urbana. and 4-2% for the whole herd and the inbred animals Haile-Mariam, M., Banjaw, K., Gebre-Meskel, T. and respectively, was found to be unimportant in a Ketema, H. 1993. Productivity of Boran cattle and their preliminary analysis. The levels of inbreeding were Friesian crosses at Abemossa Ranch, Rift-valley of Ethiopia. 0-7 and 0-9% for cows with record of Cl and AFC, I. Reproductive performance and preweaning mortality. respectively. Tropical Animal Health and Production 25: 239-248. Kassa-Mersha, H. and Arnason, T. 1986. Non-genetic Overall, the genetic parameter estimates in this herd factors affecting growth of Ethiopian Boran cattle. World are within the range of those reported for tropical Revieiv of Animal Production 22:45-55. cattle, though there were high sampling errors. M ackinnon, M. J., Taylor, J. F. and H etzel, D. J. S. 1990. Deleting AFCs higher than 5 years and CIs longer Genetic variation and covariation in beef cow and bull than 2 years increased the estimates of the genetic fertility. Journal of Animal Science 68:1208-1214. parameters by decreasing the error variance. Given M aule, J. P. 1990. The cattle of the tropics. University of that the It2 of AFC is close to zero and the possibility Edinburgh, Centre for Tropical Veterinary Medicine, of a negative correlation with Cl and that the h1 of Edinburgh. 334 Haile-Mariam and Kassa-Mersha Meyer, K. 1989. Restricted maximum likelihood to estimate Australian. Journal of Agricultural Science, Cambridge 81: variance components for animal models with several 253-262. random effects using a derivative-free algorithm. Gcnctics Thorpe, W., Cruickshank, D. K. R. and Thompson, R. Selection and Evolution 21:317-340. 1981. Genetic and environmental influences on beef cattle Meyer, K., Hammond, K., Parnell, P. F., Mackinnon, M. J. production in Zambia. 4. Weaner production from and Sivarajasingam, S. 19%. Estimates of heritability and purebred and reciprocally crossbred dams. Animal repeatability for reproductive traits in Australian beef Production 33:165-177. cattle. Livestock Production Science 25:15-30. Trail, Jf. C M, and Gregory, K. E. 1981. Characterisation of Milagres, J. C., Dillard, E, U. and Robison, O. W. 1979. the Boran and Sahiwal breeds of cattle for economic Heritability estimates for some measures of reproduction in characters. Journal of Animal Science 52:1286-1293. Hereford heifers. Journal of Animal Science 49:668-674. Trail, J. C M., Gregory, K. E., Durkin, J. and Sandford, J. Rege, J. E. O., Lomole, M. A. and Wakhungu, J. W. 1992. 1984. Crossbreeding cattle in beef production programmes An evaluation of a long-term breeding programme in a in Kenya. II. Comparison of purebred Boran and Boran closed Sahiwal herd in Kenya. I. Effect of non-genetic crossed with the Red Poll and Santa Gertrudis breeds. factors on performance and genetic parameters estimates. Tropical Animal Health and Production 16:191-200. Journal of Animal Breeding and Gaieties 109:364-373. Seebeck, R. M. 1973. Sources of variation in the fertility of a herd ;of Zebu, British and Zebu X British cattle in Northern (Received 23 April 1993—Accepted 22 November 1993) VI Genetic and environmental effects on performance of dairy cattle of Friesian origin and their Boran (zebu) crosses at Alemaya, Ethiopia M. Haile-Mariam Dept, of Animal Breeding and Genetics Swedish University of Agricultural Sciences S-750 07 Uppsala, Sweden Abstract Data collected from Alemaya University of Agriculture dairy farm in Ethiopia (1965-87) were analysed to study the performance of dairy cattle of Friesian origin and their crosses with Boran (Zebu). The overall unadjusted mean lactation milk yield, lactation length, calving interval and abortion rate were 4,058 kg, 324 days, 437 days and 7.3%, respectively. The deviation of the F1 crosses from those of Friesian origin was -2,365 kg, -75 days and -29 days in lactation yield and length and Cl, respectively. The lactation yield of grade Friesians was similar to that of Friesians but that of the 3/4 crosses was 897 kg below that of the Friesians. The incidence of abortion was relatively low (5.5%) in the F7 crosses compared with the overall least-squares mean of 8.3%. Abortion rate was influenced by season, with the bulk of the incidence occurring in the dry period. Heritability and c2 (the ratio of permanent environmental variance to total) values were 0.08 and 0.24 for lactation milk yield, 0.0 and 0.20 for lactation length and 0.0 and 0.02 for calving interval, respectively. The inbreeding level of all animals (except the F1) in the herd increased to a mean of 3.84%. Although the productivity of the cattle of Friesian origin and their grades was significantly better than that of their F7 and three-quarter crosses, the fact that milk yield decreased by an average of 66 kg per annum while there was no negative genetic trend indicates that the feeding level and management at the farm has deteriorated over the years. Keywords: genetic or breed group effects, environmental effects, dairy cattle, Ethiopia. Introduction Ethiopia has about 30 million cattle which are predominantly zebu. Of the total cattle population only 10% are milked and the average milk yield per cow per year is about 206 kg (FAO, 1992). One of the reasons for the low milk production is the unsuitability of the indigenous cattle for dairying. To meet the pressing need for increased milk production in the country, importations of exotic cattle as well as crossing of indigenous cattle with temperate breeds has been going on for the past three decades. 1 Although the proportion of imported temperate animals to the total cattle population is small in Ethiopia, their contribution to the development of dairying could be significant as they can be used in their 'pure form' in intensive dairy production systems in the highlands and for crossbreeding with indigenous animals. Their proper use for both purposes requires having parameter estimates as expressed under the local environmental conditions. Studies on comparative performance of temperate cattle and their crosses with local breeds are also important for planning appropriate breeding programmes. This report, therefore, presents the results of breed group effects and genetic parameter estimates for milk yield traits of dairy cattle of Friesian origin and their crosses with Boran at Alemaya, Eastern Ethiopia, based on 23 years of data. In addition, environmental and breed group effects on traits related to survival, such as abortion rate are, also considered. Ms terial and Methods Emnronmenial features Alemaya is located 500 km east of the capital Addis Ababa in Harrerghe Region of Ethiopia. Due to its high altitude of 1,980 metres it is characterized by a mild subtropical climate with mean maximum and minimum temperatures ranging from 17°C to 25°C and from 2°C to 13°C respectively. Alemaya lies at a latitude of 9° 20"N and a longitude of 42° 3'E. Although the pattern and amount fluctuates, the annual average rainfall is 800 mm and it is bimodal. The main rains which fall between July and October account for 50% of the total annual mean. The small rains occur from February through June, while the rest of the year is dry. The vegetation consists of annual legumes and pei ennial grass spedes such as Hyparrhenia, Andropogott, and Cynodon. Breeds and management The foundation stock used were 26 Friesian heifers imported from Kenya in 1963 and their crosses with Boran (zebu) cows produced at the farm. The F1 cows were later up-graded using Friesian sires. As a result the herd at the farm was composed of cows with different levels of cattle of Friesian origin. The Boran cows used for crossbreeding were bought locally. Normally, all heifers were first mated after they had attained a minimum body weight of 300 kg. Initially artificial insemination using semen imported from USA was practised but after 1969 Friesian bulls bom at the farm and selected on the basis of their dam's milk yield were used. Generally, all cows were mated at first observed heat 45 days after calving. However, there were irregularities in the breeding policy as well as in the herd management over the years. Cows were kept in a loose housing system and were not allowed to graze, except in 1978, when first-calving cows grazed in paddocks. Feeding and management practises, which were followed in the earlier years until 1974 were generally similar to those reported by Wells, Wagner, Holland, Stringer and Wondafrash (1969). However, after 1974 there were changes in the management of the herd due to budget constraints and transfer of professional staff. Especially between 1976 and 1977, a critical shortage of feed and poor management were encountered. Generally, animals were fed hay ad. lib. and were supplemented with green alfalfa when available. During milking, each cow was provided with 3 to 9 kg of concentrates consisting of com, wheat-bran, oil-seed cake, bonemeal and salt, depending on milk yield. After 1984, breweries grain was partly used instead of concentrate. Com or sorghum silage was also fed until 1983. All cows regardless of their breed composition were managed similarly. Cows were machine milked, starting 1 to 3 days after calving. Calves were separated from their dams at about 2 days of age and were fed milk at the rate of 3 to 4 kg per day until weaning at about 3 months of age. After weaning, young animals were group-managed with their main feed being grass hay. Animals were normally vaccinated against the notifiable diseases in the area and in the earlier years calfhood vaccination for brucellosis was also provided. Animals were also sprayed once or twice a month for ectoparasites. However, there were outbreaks of foot-and-mouth disease, heartwater and lumpy skin disease on the farm. The other problems which persisted over the years included calving difficulty, abortion and mastitis. Data recording and amlysis The data used in this study were compiled from the farm records which included information on milk yield, breeding, calving and pedigree as well as health records. All lactations except those shorter than 50 days were used to study lactation milk yield and lactation length. After appropriate edits concerning calving interval and other inconsistency, there were 742 lactation milk yield and lactation length records and 587 calving interval records of 219 and 194 cows, respectively. The number record for abortion rate (including stillbirth) analysis was 876 of 252 cows. The data on lactation yield, lactation length and calving interval were analysed using the DFREML (Meyer, 1989) programme. The model used included the fixed effect of breed group(cattle of Friesian origin, their F, with Boran cows and their 3/4 crosses, their Grades (crosses with 7/8 and 15/16 Friesian inheritance)), the 3 additive genetic effect of cow (~0,Aa/), the permanent environmental effect of cow (rQJaf), the fixed effect of year of calving (1965-87), the fixed effect of season of calving (July-October, Nov-Feb., March-June)and the fixed effect of parity, the assignment of cows into parities considered abortions as well, (1,2,3,4,5+). The relationship matrix A was based on 935 animals including 35 sires and 252 dams. For lactation yield, the genetic trends of all cows and sires with the exception of F: and 3/4 crosses were estimated by regressing their estimated breeding value on year of birth. Although breeding values were estimated for all animals, based on their relationship, only cows with at least one lactation yield record were included in the estimation of genetic trend. The solutions for year of calving were used to estimate environmental trends. With regard to classification of the animals into breed groups the following points should be noted. First, the Kenyan bom Friesians were put together with the locally bom Holstein sired animals out of Friesian dams after the preliminary analysis had indicated that their performance levels were similar. Furthermore, designating this group as Friesian breed is not strictly true but they may appropriately be described as temperate dairy cattle of Friesian origin. Secondly, the | crossbreeds (Fv 3/4 and those designated as grade Friesian) were not in production throughout the study period but were available through much of the period. The data on abortion rate were analysed by the least-squares procedure of Harvey (1985). In addition, the available farm records were used to calculate the proportion of cows disposed of for different reasons and the number of lactations covered by cows of each breed. Furthermore, the coefficient of inbreeding was calculated on the assumption that all animals were non-inbred at the start of the programme. The coefficient of inbreeding was included as a covariate using the same model above to look into its effect on the traits considered. Results Mill; yield traits The overall mean lactation yield, lactation length, calving interval and abortion rate are given shown in Table 1. Table 2 shows the variance component and the genetic parameter estimates for the milk yield traits. The /z2-value for all traits was low> though the genetic variance for lactation yield was modest. The total phenotypic variance (Table 2) for lactation yield was also considerable even when adjusted for lactation length where it was reduced to 1.2 million kg2. 4 Table 1. Means and standard deviations of traits for dairy cattle of Friesian origin and their crosses with Boran (zebu) Trait Mean s.d. Lactation milk yield (kg) 4057.8 1663.4 Lactation length (days) 324.0 102.0 Calving interval (days) 437.0 115.0 Abortion rate (%) 7.3 26.0 Table 2. Variance components and genetic parameter estimates of milk yield traits for dairy cattle of Friesian origin and their crosses with Boran(zebu) ► Parameter Lact. yield Lact. length Calving interval 188536 0+ 0f • 567084 1897 226 i 1615143 7637 11316 2370763 9534 11542 h! 0.080(0.068) 0.0+(0.106) 0.0+(0.054) c2 0.239(0.075) 0.199(0.10) 0.020(0.069) 1 Approximated to zero. The solutions for breed group, season of calving and parity for total lactation , yield, lactation length and calving interval are shown in Table 3. The environmental trend of lactation milk yield over the years shows (Figure 1) that it decreased almost consistently. This represented an average decrease of 66.2 kg (s.e. 15.5) per annum. On the other hand the genetic trend, though small, showed an annual increase of 7.2 kg (s.e. 2.3). Overall, the decline in lactation length was similar to that shown for milk yield, which means that cows responded to nutritional stress by terminating their lactation early rather than by decreasing their daily milk yield. ► b r ► > 5 Table 3. Solutions for breed group, parity and season of calving for total lactation milk yield (kg), lactation length (days) and calving interval (days) Variable Lact. yield Lact. length Cl No. Mean(s.e.) Mean(s.e.) Mean(s.e.) Overall 742 4058 324 437 Breed group Friesian 513 0 0 0 F1 Friesian-Boran 42 -2365(427) -75(23) -29(20) 3/4 Friesian-Boran 39 -897(363) -31(21) -3(21) Grade Friesian 148 -127(228) -24(13) -32(13) Parity number i 204 0 0 0 2 175 -316(144) -38(9) -21(13) 3 138 -228(156) -37(10) 1(14) 4 95 -327(181) -30(12) -8(13) 5+ 130 -438(184) -45«11) - Season of calving March-Jun. 256 0 0 0 Nov.-Feb. 237 49(138) 11(9) 26(11) July-Oct. 249 28(137) 13(9) 14(11) TotM lactation yield decreased as parity number increased. However, when the solutions were adjusted for the effect of lactation length by including lactation length as a covariate, the opposite trend was observed. In addition, cows on the farm calved for the first time when relatively old (mean of 35.2 months). In a separate analysis (not shown here) of the data of 213 heifers (all breeds except Ej), age at first calving did not have a significant effect on first-lactation yield. The leastrsquares mean yield of 66 heifers that calved when they were younger than 2.5 years was 3,922 kg while that of 69 heifers that calved between 2.5 and 3 years was 4,240 kg. The mean for those calving after 3 years decreased to 3,945 kg. Abortion rate The overall mean abortion rate (%) and the least-squares means for breed group, par ty and season of abortion are given in Table 4. Season of calving influenced abortion rate (P<0.05). Cows calving during the dry season had the highest abortion rate. Of the total 63 abortions, 41.5% occurred between November and January. The mean age of the fetus at abortion or stillbirth for 47 cases with recorded conception dates was 228 (s.e. 7) days and ranged between 123 and 274 days. Table 4. Estimated least squares means for abortion (induding stillbirth) rate (%) with their standard errors (s.e.) Variable No. Mean(s.e.) Overall 876 8.3 (1.5) Breed group Friesian 584 6.7 (1.1) F-[ Friesian-Boran 49 5.5 (3.9) 3/4 Friesian-Boran 47 11.8 (3.9) Grade Friesian 196 9.2 (2.1) Parity number 1 253 7.8 (2.1) 2 205 7.4 (2.2) 3 150 9.3 (2.5) 4 111 7.9 (2.8) 5 4 - 157 8.9 (2.3) Season of calving Nov.-Feb. 272 12.8b (2.0) March-Jun. 307 6.3a (1.9) July-Oct. 297 5.5a (1.8) Within variable groups, means with the same letter do not differ significantly (P>0.05). Number of calvings The average herd-life, calculated as the number of lactations per cow, was 3.3,5.1, 3.3 and 3.4 for dairy cattle of Friesian origin and their F-, and 3/4 crosses and their Grades, respectively. The farm records show that only 7.6% of the 105 cows with recorded disposal reason were culled for their low production and more than a quarter of the culled cows were disposed of due to infertility and reproductive problems, while mastitis, old age, physical injury and brucellosis and other diseases accounted for the removal of 10.5, 4.8, 4.8 and 30% of the cows culled, respectively. Inbreeding The increase in coefficient of inbreeding for different categories of animals is shown in Table 5. The maximum level of inbreeding among milch cows was 17.9% and that among all animals was 36.7%. Indusion of inbreeding coeffident as a covariate for the analysis of all traits had no effect. 7 Table 5. The increase in coefficient of inbreeding (%) of dairy cattle of Friesian origin and their crosses (except F1) Category No. Mean(s.e.) All animals* Milch cows 192 2.15(0.20) Including calves 716 3.84(0.15) Inbred animals Mildi cows 112 4.29(0.30) Including calves 378 7.20(0.61) Excluding those with unknown sire. Discussion The overall mean lactation yield (Table 1) was higher than that reported by Syrstad (1988) but the calving interval was similar to his estimate. The h2-value of lactation yield is low compared with the 0.32 reported by Rege (1991) for Friesian cattle in Kenya. All the /^-values for all traits estimated are also lower than other estimates for temperate dairy cattle in the tropics (Khattab and Sultan, 1991; Salman and Yousif, 1990). The repeatability values (the sum of h2 and c2-values) were also lower than those estimated for 305-day milk yield and calving interval reported by Rege (1991). The low genetic parameter estimates might be attributable to the variable and stressful nature of the environment whereby ind vidual animals responded differently to environmental fluctuation and their ind yidual conditions; suppressed the effects of genes derived from their parents. Her vever, the genetic parameter estimates in the present study are less reliable as they are based on a small dataset. The slight superiority of the F} crosses in fitness trait (abortion rate) and their inferiority in milk yield traits are consistent with the literature (McDowell, 1985; Syrstad, 1988). However, the advantage for milk yield of up-grading to Friesian in the present study was higher than in several reports, indicating a rather good management standard. After re-analysing crossbreeding data from the tropics, Syrstad (1988) showed that such increase was only slight. McDowell (1985) noted that the superiority of a 3/4 improved breed over the first-cross was only 6%. The reason for the difference between the estimate in the present study and the others might be the fact that cows in the present study were managed indoors and given concentrate supplement and that the climate was somewhat subtropical. Moreover, all c ows were fed concentrate while being milked and those which produced more milx had a chance to consume more. Thus, Friesian cows by virtue of their 8 relatively higher milk yield, may have eaten more concentrate than their crosses. The environmental variation in lactation yield (Figure 1) observed is much the same in the tropics, with annual effects accounting for over 30% of the total variation in yield (McDowell, 1985). The lowest lactation yield for 1977 was due to lack of feed and poor management. The decrease in lactation yield in 1984-85 coincided with the drought of 1984-85 throughout the country. The reason for the insignificant effect of parity on lactation yield was due mainly to the longer lactation length of first-calvers as compared to second lactation (Table 3). This longer lactation, may have been due to the persistency of first lactation. In addition, the fact that the animals calved at about 3 years of age suggests that they were fully mature at first calving. The insignificant effect of season of crlving on milk yield traits was in agreement with that observed in Friesian cattle in Kenya (Rege, 1991). This is to be expected, as the cows in the present study were usually kept indoors and as there was little variation in feeding regime between seasons. A significant effect of season on abortion rate was also reported from the Central Highlands of Ethiopia (Bekele, Kasali and Alemu, 1991) and in the hot-humid tropics of Orissa, India (Bhuyan and Mishra, 1985). Bekele et al. (1991) observed the highest abortion rate during March-Sept. In their study, most of the abortions occurred at 150 to 260 days of gestation. Although 30% of the 105 cows culled at Alemaya were removed due to Brucellosis and related infectious diseases, the seasonal nature of the abortions as well as the stage at which they occurred may suggest that Brucella was probably not the real cause. The fact that the abortions were occurring during the dry season when cows were fed on stored hay indicates rather that the cause was probably mouldy hay. Vandeplassche (1982) reported that mycotic (fungal) abortions are seasonal and occur at about 7 to 8 months of gestation. The smaller number of lactations by cattle of Friesian origin and their grades, as compared with Fl7 is consistent with that reported by Vaccaro (1990). The abortion rate was slightly lower than the mean given by Vaccaro (1990) who observed 6.7% and 6% losses due to abortion and stillbirth, respectively, in temperate cattle bom in the tropics. The reason for the absence of any effect of inbreeding on performance might be the fact that the inbreeding level was not high enough, particularly in milking cows. This would agree with Miglior, Szkotnicki and Burnside (1992) who concluded that an inbreeding level of less than 12.5% does not cause inbreeding depression. However, measures to reduce inbreeding should be considered immediately. 9 In theory, some improvement in lactation yield could be obtained by practising selection among cows, though any improvement would be limited by the low ft2- value and the lack of adequate replacement heifers. In this herd the large proportion of abortions (8.3%) and high age at first calving (35.2 months) limited thelscope for selection within the herd. Moreover, the relatively extended calving interval (437 days) and the fact that many cows were culled for reasons other than production, further limits the possibility of any improvement by selection. To avoid inbreeding and to utilize genetic improvement gained elsewhere, sires or semen should be imported from other sources from time to time. The superiority of the cattle of Friesian origin regarding milk yield traits and their slightly inferior performance in calving interval, abortion rate and number of lactations per cow, as compared to the FT crosses, is in agreement with the literature. The high degree of superiority of the Friesians and their grades in this study may indicate that the better performance was possible under fairly good management conditions, at least in the earlier years and in an area where the clinpate is moderated by high altitude. Total lactation yield and lactation length decreased over the years, showing the difficulty of maintaining higher levels of production particularly of Friesians. The main reason for the decline in productivity seems to be associated with scarcity of feed due to drought and lack of consistent management. On the other hand there was no negative genetic trend in lactation yield. This herd may make a considerable contribution if used as are line for crossbreeding with zebu breeds for;milk production. Ac] enowledgements I wish to express sincere gratitude to the former and present staff members of the Animal Science Department at the Alemaya University of Agriculture who were involved in the management and data collection. In particular my thanks go to Drs Kano Banjaw and Alemu Yami. Also to Dr E. Bums and Mr Solomon Zewdie of ILuA and Drs J. Philipsson, T. Henningsson and A. Dahlin of the Swedish Un iversity of Agricultural Sciences for their help in this study. References Bekele, T., Kasali, & Alemu, T. 1991. Reproductive problems in crossbred cattle in central Ethiopia. Animal Reproduction Science, 26:41-49. Bh lyan, R.N. and Mishra, M. 1985. Performance of imported Jersey cattle in hot- uimid climate of Orissa. Indian Journal of Animal Production and Management, 1:123-127. 1 0 I" FAO (Food and Agriculture Organization), 1992. Production Year-book, 1991. Vol. 45, Rome. Harvey, W.R. 1985. User's guide for LSMLMW. Ohio State University, Columbus, USA. Khattab, A.S. and Sultan, Z.A. 1991. Comparison of different selection indices for genetic improvement of some dairy traits in Friesian cattle in Egypt. Journal of Animal Breeding and Genetics, 108:349-354. McDowell, R.E. 1985. Crossbreeding in tropical areas with emphasis on milk, health and fitness. Journal of Dairy Science, 68:2418-2435. Meyer, K. 1989. Restricted maximum likelihood to estimate variance components for animal models with several random effects using derivative-free algorithm. Genetics, Selection and Evolution, 21:317-340. Miglior, F., Szkotnicki, B. and Burnside, E.B. 1992. Analysis of levels of inbreeding and inbreeding depression in Jersey cattle. Journal of Dairy Science, 75:1112- 1118. Rege, J.E.O. 1991. Genetic analysis of reproductive and productive performance of Friesian cattle in Kenya. I. Genetic and phenotypic parameters. Journal of Animal Breeding and Genetics, 108:412-423. Salman, M.H. and Yousif, L.M. 1990. Heritabilities, genetic and phenotypic correlations of milk production traits in Holstein and Friesian cattle in Iraq. Mesopotamia Journal of Agriculture, 22:93-103. Syrstad, O. 1988. Crossbreeding for increased milk production in the tropics. Norwegian Agricultural Science, 2:179-185. Vaccaro, L.P.de. 1990. Survival of European dairy cattle breeds and their crosses with Zebu in the tropics. Animal Breeding Abstract, 58: 475-494. Vandeplassche, M. 1982. Reproductive efficiency in cattle. A guideline for projects in developing countries. FAO, Rome, No. 25, p. 95. Wells, M.E., Wagner, D.G., Holland, G.L., Stringer, B. and Wondafrash, T. 1969. Production characteristics and management of dairy cattle in Ethiopia. East African Agricultural and Forestry Journal, 24:293-298. 11 year Environmental Genetic Fig. 1. Genetic and environmental trends for lactation milk yield. Institutionen for husdjursforadling Rapport i 13 och sjukdomsgenetik Publication No. 113 Swedish University of Agricultural Sciences Uppsala 1994 Department of Animal Breeding ISSN 0347-9706 and Genetics ISHN SLU^HFS-R-113 -S b DISTRIBUTION: Swedish University of Agricultural Sciences Department of Animal Breeding and Genetics Box 7023 S-750 07 UPPSALA, SWEDEN Phone no.: 46-18-671941