Production Systems and Breeding Practices of Arab and Oromo Goat Keepers in Benishangul Gumuz Region, Northwestern Ethiopia: Implications for the Development of Community-based Breeding Programs

To better understand and describe the production systems and breeding practices of Arab and Oromo goat keepers, we undertook a household survey in semi-arid and sub-humid parts of Benishangul Gumuz region, Northwestern Ethiopia. Multi-stage random and probability proportional to size sampling techniques were employed to select peasant associations and households, respectively.Data were collected from 249 goat keeping households (86 in semi-arid and 163 in sub-humid areas) through personal observations, focus group discussions and structured questionnaires, and analyzed using SPSS. Descriptive statistics and indices were used to present the data. In both areas, goats were kept mainly to generate cash income followed by meat production and saving with overall indices of 0.40, 0.23 and 0.17, respectively. The average ± SD of fl ock sizes per households were 12.5 ± 4.0 in Arab goat keepers and 9.9 ± 3.8 in Oromo. T he mean numbers of does, doe kids, bucks, kids and castrates were significantly higher in Arab goat keepers as compared to Oromo. Body size, twining ability and coat color were the most preferred attributes of both goat keepers in selecting breeding does while breeding bucks were selected based on their body conformation, growth and coat color. Mating was predominantly uncontrolled mainly due to communal grazing lands. Castration of bucks was significantly (p < 0.01) more frequent in Arab goat keepers than Oromo. The average age at first kidding, kidding interval, reproductive lifetime and number of kids born per lifetime of Arab does were 13.9 ± 1.7 months, 7.2 ± 1.8 months, 7.2 ± 2.0 years and 10.7 ± 2.5, respectively. The corresponding values for Oromo does were 14.9 ± 2.4 months, 7.8 ± 1.1 months, 7.9 ± 1.9 years and 9.7 ± 1.6. The difference between Arab and Oromo goat keepers in terms of production objectives, flock sizes, trait preferences and breeding practices should be carefully considered in the development of breed improvement programs in the study areas.

and Bambasi 02, Mutsa 01, Shebora and Womba-selama from Bambasi) were selected. Goat population size (based on the available data from the respective districts' agricultural offices), presence of communal grazing areas, relative significance of goats to the livelihood of the communities, access to market and road were the criteria used in selecting the PAs. In order to clarify the objectives and possible outcomes of the research, meetings were held with the communities at each selected peasant association.
The number of sampled households was determined following the recommended formula (Arsham, 2007): N=0.25/(SE) 2 ; Where: N=sample size and SE=standard error. To make the size of the sample selected from each peasant association proportional to the size of the corresponding peasant association, probability proportional to size (PPS) sampling technique was employed as: W= [A/B] x N; where: W=number of households to be calculated from the selected PA; A=total number of households in the selected PA; B=total number of households in all eight selected PAs and N=the calculated sample size (Reference).
Finally, households who own at least four adult goats with a minimum of one year experience in goat husbandry practice and willing to participate in a community-based breeding program were identified and their lists were prepared in each peasant association with the help of data collectors. Starting from the first household, respondents were selected from the prepared list using systematic random sampling technique until the calculated sample size of each peasant association was maintained. In general, a total of 86 respondents from Homosha and 163 respondents from Bambasi districts were selected for interview.

Data collection
Semi-structured questionnaires and formal interviews were used to gather information from the selected households. Data on socioeconomic characteristics, purpose of keeping goats, flock size and composition, trait preference and selection criteria, mating system and sources of breeding buck, castration and reproductive characteristics were collected by trained enumerators from the respective peasant associations with close supervision by the researcher. The questionnaire was tested before the actual survey to ensure that all questions were of sufficient clarity for the interviewees. To generate additional information and validate the data collected from the individual farmers, participatory focus group discussions with elders, village leaders, women goat owners, youngsters, veterinarians and goat traders were also made in the selected peasant associations.

Data management and analysis
Data collected from the districts were encoded and fed into MS-Excel (2010) for further analysis using Statistical Package for Social Sciences version 22 (SPSS, 2013). Preliminary data analysis like homogeneity test, normality test and screening of outliers were employed before conducting the main data analysis. Chi-square or t-test was employed when required to test the independence of categories or to assess the statistical significance. Indices were calculated for ranked variables in reference to its formula: Index = Sum of (3 X number of households ranked first + 2 X number of households ranked second + 1 X number of households ranked third) given for an individual reason divided by the sum of (3 X number of households ranked first + 2 X number of households ranked second + 1 X number of households ranked third) for overall reasons.

Results and Discussion General household characteristics
Out of the total of 249 interviewed households, the overall majority (90%) of the respondents were males. The proportion of male respondents was significantly (p<0.01) higher in Oromo goat keepers (93.9%) than Arab (82.6%). The mean (SD) age of the respondents was 47.1 (13.0) and 45.1 (9.7) years in Arab and Oromo goat keepers, respectively. This indicates that people of middle age were more likely engaged in goat farming and its implication in accepting new approaches such as community-based breeding programs could be higher given that young people are generally more receptive to innovations and new technologies than old ones (Adesina et al. 2000). The overall average family size of goat keepers was 7.94 ± 4.2 and it was higher than the national average of 4.80 (CSA, 2010b). Arab goat keepers have significantly (p<0.05) higher family size (8.8 ± 3) than Oromo goat keepers (7.5 ± 4.4). The difference might be due to the observed prevalence of polygamous marriage in the Arabs.
While 38.6% of the Oromo goat keepers attended primary, secondary and adult education, a significantly (p<0.01) lower proportion (22.1%) of respondents in Arab goat keepers were literate. Complete absence of record keeping and identification of goats as observed among the interviewed farmers might be the result of high illiteracy rate in the study areas. This necessitates awareness creation among goat keepers and provision of simple and easy-to-use recording system. Nevertheless, the better literacy rate in Oromo goat keepers would be a good opportunity to implement goat improvement program as it might be helpful in performance and pedigree recording (Kosgey and Okeyo, 2007).
The marital status of Arab and Oromo goat keepers varied significantly (p<0.05). In Arab goat keepers, 98.8% was married and 1.2% was divorced, whereas married and divorced goat keepers represented 89% and 1.6% of the Oromo goat keepers, respectively. The rest of the Oromo goat keepers were widowed (6.1%) and single (3.1%). The average privately owned land per household was 2.3 ± 1.8 and 1.9 ± 1.6 hectare in Arab and Oromo goat keepers, respectively. The private lands were used mainly for crop production and communal grazing areas were available for grazing. Generally, the socio-economic differences of the households should be properly considered in designing of goat improvement programs in both study areas.

Purpose of keeping goats
In the present study, the most frequently reported purposes for keeping goats in both study areas were generation of cash income followed by meat production and saving (using goats as "village bank") with overall index values of 0.40, 0.23 and 0.17, in that order (Table 1). These results clearly show that goats were kept in the rural areas of Benishangul Gumuz region not only for generation of cash income but also to supplement the diets of households through their meat. While the indices for other purposes were relatively small, keeping goats as a sign of wealth status in Arab goat keepers (0.12) and for manure production in Oromo goat keepers (0.10) was of some importance.
None of the interviewed households in this study raised goats for milk production. This is because consumption of goat milk was considered as a cultural taboo in both study areas. Similar findings on goat milk consumption were reported in different parts of Ethiopia like Jimma (Belete, 2009), Metema (Tesfaye, 2009), North Gondar (Alubel, 2015) and Horro Guduru (Ahmed et al., 2015). However, the current result contrasts with reports in Sekota (Muluken, 2006), Dire Dawa (Grum, 2010), west Amhara (Kefyalew et al., 2015), Afar (Feki et al., 2015) and many African countries (Dubeuf, 2007). Lack of awareness of its nutritional and medicinal values is probably the possible explanation for low preference and acceptance of goat milk in Ethiopia in general and the study areas in particular. Overall, given the different purposes and ranks that farmers have for keeping goats, the main breeding goal of Arab and Oromo goat keepers is defined as increasing meat production for marketing and home consumption.

Flock size and composition
The average flock size of a household was 12.5 ± 4.0 heads (ranging from 3 to 27goats) in Arab goat keepers and 9.9 ± 3.8 heads (ranging from 3 to 27goats) in Oromo ( Table 2). The current results are comparable with 10.8 goats per household in western lowland goats, but lower than 48.5 in Abergelle goats (Abegaz et al., 2013). On the other hand, Ahmed et al. (2015) reported a relatively small flock size (7.6) in western Ethiopia. The variations in flock sizes of goats might be due to the differences in production system and production environment, the role of livestock as major source of livelihood and availability of land and feed (Solomon et al., 2014). The mean numbers of does (females above 1 year) and doe kids (females between 6-12 months) were significantly (p < 0.01) higher in Arab goat keepers than in Oromo. Similarly, the mean numbers of bucks (males above 1 year), kids (both males and females less than 6 months) and castrates were significantly (p<0.05) higher in Arabs. Breeding does constituted the highest proportion (37.5%) followed by kids (22.6%), doe kids(20.3%), buck kids (8.6%), bucks (7.2%) and castrates (3.9%) in Arab goat flocks. A similar pattern was observed in Oromo goat flocks by which breeding does were dominant (35.9%) followed by kids (24.9%), doe kids (20.6%), buck kids (8.4%), bucks (7.3%) and castrates (3%). Larger proportion of breeding does obtained in the present study could imply the practice of retaining does for breeding purpose and the production of more number of kids. This in turn might be advantageous to implement village-based selection as it increases selection intensity and effectiveness of selective breeding. On the other hand, bucks and buck kids had lower proportion as compared to their female counter parts. This might be due to the removal of males for sale or home consumption. The ratio of breeding buck to breeding does was about 1:5 in Arab goats and the same in Oromo. The result is lower than the recommended ratio of 1:25 for tropical traditional production system (Wilson and Durkin, 1988). The higher ratio obtained in this study revealed that sufficient number of male goats were kept in the flock for breeding purpose. Castrated goats represented the lowest share of the flock per household in both goat keepers and the practice of goat castration might be helpful to avoid mating of unwanted bucks.  Table 3 summarizes the relative importance of different traits as ranked by the goat keepers. Irrespective of keepers, results from the trait preference of households indicated that body size, twining ability, coat color and kid growth were the most preferred traits in selecting breeding does with overall index values of 0.34, 0.22, 0.17 and 0.10, respectively. Likewise, body conformation (wide chest, long body and up right standing) followed by fast growth, coat color and libido were found to be the most important selection criteria of breeding bucks with indices of 0.30, 0.24, 0.20 and 0.08, respectively. Traits such as kidding frequency, age at first maturity, kid/s survival and mothering ability in does and adaptability, ability to walk long distance, horns, age at first maturity, character and pedigree in bucks were also mentioned as selection criteria, but with lower ranking.

Trait Preference and selection criteria
High preference of twining ability in the present study indicates the genetic potential of Arab and Oromo goats for multiple births and the farmers' general objective which is to increase flock size and overall flock productivity. Similarly, high emphasis on body size/conformation, fast growth and coat color in both sexes might be due to the observed market condition. Larger sized goats fetch premium market price and fast growing goats reached market weight sooner. Due to these conditions, goat keepers in the study areas tend to sale off good looking and fast-growing goats, particularly bucks that are potentially "best" breeding at early age to fetch better market price. This implies that bucks with unwanted traits are replicating which resulted in negative selection that may compromise the livelihoods of the producers. Awareness creation on Arab and Oromo goat keepers about negative selection is therefore highly recommended. Moreover, community level selection of bucks and using them effectively before they are sold off is suggested.

Mating system and sources of breeding bucks
The majority of the respondents in Arab (69.8%) and Oromo (92%) goat keepers practiced uncontrolled mating ( Table 4). Predominance of uncontrolled mating within households' flocks was also reported in Ethiopia (Tekleyohannes, 2012;Abegaz et al., 2013;Hulunim;Netsanet et al., 2014). In contrast to these, controlled mating through selection, culling, castration and physical restriction was practiced in Afar goats (Feki et al., 2015). The main reasons for the absence of controlled mating in the present study were communal grazing areas (81.4% in Arab and 92.6% in Oromo) and lack of farmers' awareness about the deleterious effects of inbreeding (18.6% in Arab and 7.4% in Oromo). An advantage of natural uncontrolled mating is that it allows for all year round parturition (Kosgey, 2004). Most of the Arab (95.3%) and Oromo (80.4%) goat keepers practiced mixing of their flocks during grazing, on average with 5 other flocks in Arab and 3 other flocks in Oromo. This would minimize the problem of inbreeding by increasing the probability of mating of unrelated animals (Jaitner et al., 2001). Nearly two-third (65.1%) of the respondents in Arab goat keepers and the majority (81%) in Oromo used breeding bucks born within the flock whereas 27.9% of Arab and 10.4% of Oromo used neighbors'/communal bucks and the rest purchased breeding bucks from the nearby market. The main reasons of keeping bucks were mating, fattening and both. The average service year of a buck in the flock was 3.8 ± 1.54 years in Arab and 4.1 ± 1.31 years in Oromo; after which it will be usually castrated or disposed through sale or slaughter. Apparently, the average age at which the breeding buck is changed, revealed in Arab community, could be relatively better in the prevention of inbreeding. Generally, uncontrolled mating together with small flock sizes, utilization of breeding bucks born within the flock and poor record keeping on the pedigree would result in severe inbreeding and decreased genetic diversity (Saico and Abul, 2007). Equally important to note is that bucks were kept up to 3-5 years in the study areas. The implication of these results is that community-based breeding scheme would be appropriate for both Arab and Oromo goat keepers.

Castration
Castration of bucks was a common phenomenon in both study areas, but it was significantly (p < 0.01) more frequent in Arab goat keepers (97.7%) than Oromo (60.1%) ( Table 5). The low rate of castration in Oromo goat keepers might be due to high demand of intact male goats in the area. Irrespective of keepers, fattening was found to be the most common reason of castration followed by control mating and better temperament. The frequency of castration methods differed significantly (p<0.05) between the keepers in that 63.8% of Oromo and 54.7% of Arab goat keepers used traditional castration method, by crushing the vas deferens using smooth and round river stone locally known as 'alello' and banded the testes until they wither due to lack of blood flow. The rest of keepers took their male goats to a nearby veterinary clinic to use burdizzo.
The mean age of castration in Arab goat keepers (1.9 ± 0.6 years) was significantly (P<0.05) lower than that of Oromo goats (2.2 ± 1.1 years). These results are close to the findings of Hulunim (2014) and Abegaz et al. (2013) who documented 2.2 ± 0.11 and 2.10 ± 0.68 years for Borena and Western lowland goats, respectively. However, the current findings contrast with what has been recommended by Alemu (2009). The same author suggested castration of male kids as soon as the testicles descend into the scrotum, from a few days of age to three weeks. In general, the practice of castration in both study areas might be advantageous to design goat genetic improvement program through control of inbreeding and unwanted mating. However, castration of bucks before sexual maturity is recommended to avoid unwanted mating.

Reproductive performance
The results obtained for reproductive performance of Arab and Oromo goats are presented in Table 6. There was significant (P<0.01) differences between the two goat populations, all reproductive performance traits being higher in Oromo goats than Arab goats. The average age at sexual maturity of Arab and Oromo male goats were found to be 7.0 ± 1.0 and 7.6 ± 0.9 months, respectively, in the meantime female goats were mated at an average age of 7.9 ± 0.9 and 8.3 ± 0.7 months, in that order. The average age at first kidding of Arab and Oromo goats were estimated to be 13.9 ± 1.7 and 14.9 ± 2.4 months, respectively. The present findings are within the range of 7-15 months of age at sexual maturity in both sexes and 12-20 months of age at first kidding for indigenous goats (Solomon et al., 2014). Similarly, the kidding interval obtained in the current study, 7.2 ± 1.8 months in Arab goats and 7.8 ± 1.1 months in Oromo goats, falls within the range of 180-300 days reported for tropical goat breeds (Banerjee et al., 2000).
A breeding doe can serve in the flock for an average of 7.2 ± 2.0 years in Arab and 7.9 ± 1.9 years in Oromo. For the corresponding areas, 10.7 ± 2.5 and 9.7 ± 1.6 mean numbers of kids were born per lifetime of does. The inverse relation of average reproductive lifetime of a doe and number of kids born per lifetime of a doe indicated that Arab goats have better reproductive performance than Oromo goats. This could be attributed to their genetic superiority or the environment. With regard to the environment, Arab goats predominate in semiarid zones whereas Oromo goats are found in sub-humid zones of Benishangul Gumuz region (Getinet et al., 2005). As elucidated by Solomon et al. (2014), the highest values for most reproductive traits, except kidding interval and average number of kids born, in the Ethiopian indigenous goat breeds were observed in goats from arid and semi-arid areas. Regarding the genetic superiority, molecular analysis of the two goat populations is recommended. In general, the reproductive performance of both Arab and Oromo goats is fairly good and in the short-term, this parameter may not be a priority for intervention and efforts at genetic improvement of goats in the study areas should be directed towards production traits. 7.0 ± 1.0 6-12 7.6 ± 0.9 6-9 ** Age at 1 st sexual maturity of female goats (months) 7.9 ± 0.9 6-12 8.3 ± 0.7 7-10 ** Age at 1 st kidding (months) 13.9 ± 1.7 12-18 14.9 ± 2.4 11-24 ** Kidding interval (months) 7.2 ± 1.8 5-12 7.8 ± 1.1 6-12 ** Reproductive life time of does (years) 7.2 ± 2.0 4-15 7.9 ± 1.9 4-15 ** Number of kids born/life time/doe (number) 10.7 ± 2.5 5-20 9.7 ± 1.6 6-16 ** N=Number of respondents, SD= standard deviation, **p<0.01

Conclusions and recommendations
The present study provided insights into the production systems and breeding practices of Arab and Oromo goat keepers in Benishangul Gumuz region, Northwestern Ethiopia. Goats were kept in both study areas for multiple purposes and keepers have an established practice of selecting breeding does and bucks. Increasing meat production (improving conformation and growth) for marketing and home consumption, twining ability and coat color were found to be the most preferred traits in defining breeding goals. Low level of literacy especially in the Arab goat keepers, negative selection, small flock size, poor recording, utilization of breeding bucks born within the flock are the characteristics of the study areas. Therefore, larger effort should be put on awareness creation on goat keepers of the study areas about inbreeding and negative selection. Strengthening the existed practice of mixing different goat flocks within the village based on communal grazing land is also suggested in order to make selection within village rather than within each flock of a household. Simplified and easy-to-use recording methods should be introduced to make selection based on recorded data. Overall, introduction of communitybased breeding programs, considering the difference between Arab and Oromo goat keepers' production systems and breeding practices, are highly recommended to improve the livelihood of the goat keepers.

Declarations Ethics approval and consent to participate
The review paper meets all applicable standards with regard to ethics and integrity. As a researcher and educator in animal breeding and genetics and along with the co-author, the paper has been submitted with full responsibility, following due ethical procedure, and there is no duplicate publication, fraud or plagiarism.

Consent for publication
Not applicable