Evaluation of forage quantity and quality in the semi-arid Borana Lowlands, Southern Oromia, Ethiopia Evaluación de la cantidad y calidad del forraje natural en la zona semi- árida de Borana Lowlands, Southern Oromia, Etiopía

This study was conducted with the aim of determining herbaceous biomass during different seasons, plus nutritive value of herbaceous species and forage on selected woody plants and documenting pastoralists’ perceptions of the value of various forage species in Borana Zone, Oromia, Ethiopia. Data were collected from a total of 92 main plots of 500 m during rainy and dry seasons located across different functional Land Use Units called Kalo (enclosed areas), Worra (grazed by lactating stock) and Foora (more remote and grazed by dry and non-lactating stock). Total herbage and leaves of woody plants were analyzed for concentrations of crude protein (CP), organic matter (OM), neutral detergent fiber (NDF), acid detergent fiber (ADF), acid detergent lignin (ADL) and ash. Perceptions of farmers were determined through group discussions. Herbage biomass plus chemical composition of both herbaceous and woody forage species varied significantly across seasons and Land Use Units. Mean herbaceous biomass in all Land Use Units was poor (876‒1,469 kg DM/ha). Mean CP, NDF and ADF concentrations of the herbaceous samples were 62, 749 and 444 g/kg DM, respectively. Mean CP% of leaves from woody plants was higher (11%) than from herbage (6%). In both groups, crude protein concentrations were highest during the wet season and lowest during the dry season, whereas fiber concentrations were highest in the dry season. Mean CP% of herbaceous forage species was below the critical level recommended for both beef cattle (7%) and small ruminants (9%) but forage from woody species should provide a reliable supply of supplementary nitrogen. Management of rangelands should be designed to ensure that desirable herbaceous species are preserved, while desirable woody species are also a valuable asset. Determination of management strategies to ensure that the desirable mix of species is maintained is imperative if sustainable production is to continue.


Introduction
The Borana Lowlands occupy 95,000 km 2 (Alemayehu Mengistu 2004) in Ethiopia and are populated by pastoralists who represent a vital part of Ethiopia's population, contributing significantly to the nation's GDP. Review of different studies, e.g. Shapiro et al. (2017), estimated direct contribution of livestock production in lowland pastoral systems of Ethiopia to agricultural GDP and national GDP to be 39 and 17%, respectively. The area supports 480,000 families with an annual population growth rate of 2.5-3% (Homan et al. 2004). Livestock production dominated by the Boran breed has been the major source of livelihood for Borana pastoralists. According to CSA (2008), in 2007 there were 1,771,589 cattle, 1,991,196 goats, 699,887 camels and 52,578 donkeys in the Borana zone. The Boran breed remains one of the most productive breeds as it is fastgrowing and fertile with good milk production compared with other indigenous cattle breeds in Ethiopia (Aynalem Haile et al. 2011).
Livestock play a crucial role in the subsistence economy, culture and religion of pastoralists in Ethiopia, and represent both social capital and an insurance against disaster (Herlocker 1999). Borana pastoralists are known for their strong tradition of livestock production through using their indigenous rangeland and water management strategies. The herbage on offer in the rangelands, however, is highly variable, both in quantity and quality. Vázquez-de-Aldana et al. (2000) reported that the botanical composition of available forage was highly variable as was the nutritional quality, which was further exacerbated by topographic relief, soil characteristics, climate, season and management. The semi-sedentary Borana pastoralists have developed strategies to exploit this highly variable resource, and are known for sustainably using the Borana land in southern Ethiopia for livestock production.
Rangeland management markedly affects botanical composition and, consequently, herbage quantity and quality. In order for the grazing system to be sustainable, better understanding of the characteristics of the forage available is needed. However, little or no data are currently available on the quality of plant resources in the study area.
Therefore, this study was conducted in the Borana Lowlands to determine both quality and quantity of forage resources in this semi-arid pastoral production system throughout the year. The specific objectives were to determine herbaceous biomass and nutritive value of forage species and document pastoralists' perceptions on forage species.

Study area
The study was conducted in Arero and Yaballo Districts of Borana Zone, Oromia, Ethiopia ( Figure 1). This study was part of a larger project of the Borana Lowland Development Program (BLPDP)/Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ) aimed at developing a pastoraloriented self-help concept for sustainable natural resource management under changing ecological and socio-economic conditions. Field data were collected from 2001 to 2003 in different seasons.
The main study sites were Dida Hara Pastoral Association (PA) in Yaballo and Web PA in Arero. In addition, a government ranch called Dida Tuyura and Foora (an area used for dry or non-lactating livestock) were selected randomly for the forage resource assessment. The government ranch was reputed to be in relatively good rangeland condition and was included as a benchmark for comparing the other Land Use Units. Yaballo town is 570 km south of Addis Ababa (9°0'19'' N, 38°45'49'' E; 2,355 masl). Dida Hara and Web are located about 30 km northeast and 85 km southeast of Yaballo town, respectively. Foora is located in Dida Hara PA, about 48 km southeast of Yaballo town. Dida Tuyura Ranch is in Dida Yaballo PA (Yaballo district), about 25 km northeast of Yaballo town.
Livestock populations in the Borana Lowlands are predominantly cattle, while small ruminants and camels are also important in the production system. Rearing dromedaries has expanded since the 1990s. Estimates have shown that herd composition in Tropical Livestock Units (TLU) was 90% cattle, 5% small ruminants and 4% dromedaries (Homann 2004). Borana cattle are a Bos indicus breed that belong to the Large East African Zebu breed group (Homann 2004).

Climatic characteristics
The elevation of the study area ranges from 750 to about 2,000 m above sea level. Rainfall is bimodal, with the long rains during March-May and short rains during October-November (Haugen 1992;Coppock 1994). Mean annual rainfall is 412 mm in Web (Web weather station; data from Southern Range Development Unit) and 566 mm in Dida Hara (Yaballo town as the nearest station; data from the National Meteorological Services Agency of Ethiopia). While mean annual temperature varies from 19 to 24 °C (Alemayehu Mengistu 1998), the mean maximum temperatures for Yaballo stations ranged from 24.4 to 26.4 °C and minimums from 13.8 to 14.8 °C (1989-2001 raw data from the National Meteorological Services Agency of Ethiopia). In general, December-February is the hot dry season, March-May is the long rainy season, June-August is the cool dry season and September-November is the short rainy season. The difference between the long rains and short rains is the amount of rain that the area receives.

Soil characteristics
The soils in the study area are granitic and volcanic soils and their mixtures (Coppock 1994). Valley bottomlands of the Borana rangelands are dominated by vertisols. Review of studies that described upland rangeland soils in the study area showed that the soils vary in color (yellow, brown, grey or red) and have almost equal proportions of sand, silt and clay (Alemayehu Mengistu 2004). In general, Dida Hara soils are the lightest, containing the highest proportion of sand, whereas Web has soils with higher levels of available P, Ca, Mg, CEC and pH. Mean available P ranged from 2.0 ppm in Foora to 30 ppm in Web Worra. Concentrations of P and Ca and CEC are highly variable in both Dida Hara and Web (Gemedo Dalle 2004).

Sampling strategy
Borana pastoralists classify their grazing lands into enclosed grazing lands for calves (Kalo), grazing lands for lactating livestock (Worra) and grazing lands for dry livestock (Foora). Based on suitability for different classes of livestock (i.e. availability of forage and watering points), the pastoralists establish their villages (pastoral camps) locally called Olla. Classification and demarcation of the grazing land into Kalo, Worra and Foora is based on distance from the villages and accessibility of watering points: Kalo is adjacent to the villages, Worra the next removed and Foora the most remote. Similarly, Kalo and Worra are located within walking distance (distance from water covered by grazing livestock in a single day, which is about 12 km) from watering points, whereas Foora are remote from the watering points (having no permanent watering point within the grazing area) and dry livestock utilizing this area depend on surface rainwater or must walk long distances to access watering points. Kalo was fenced and protected from grazing from early wet season to hot dry season, and was accessible for grazing only during the hot dry season. Worra and Foora were open to livestock throughout the year. A stratification sampling technique was used to collect samples from these functional Land Use Units. Within each Land Use Unit, the initial sampling point was established randomly, but subsequent units were established at 200 m intervals on a linear transect.
Samples of both herbaceous and woody forage species were gathered from different Land Use Units. Herbaceous samples were collected during cool dry (June-August), short rains (September-November) and long rains (March-May) seasons, whereas woody samples were taken during short rains, hot dry (December-February) and long rains seasons. Within the various districts, an effort was made to sample from different sites in all seasons, in an attempt to ensure that the samples were representative for the specific study sites.

Forage sampling
Forage samples were collected from a total of 92 main plots of 50 × 10 m (500 m 2 ) each (Table 1). Within each 500 m 2 plot, 5 subplots of 0.5 × 0.5 m (0.25 m 2 ), 4 at the corners and 1 in the center, were established and herbaceous samples were collected for both biomass and forage nutritive value determination. Samples from the 5 subplots were pooled and assumed to represent the main plot. To demarcate these subplots, a 3-sided frame of welded metal (0.5 × 0.5 m), left open at one side as recommended by Whalley and Hardy (2000), was used. All grasses, herbaceous forbs and sedges rooted within the marked area of 0.25 m 2 were cut at 2 cm above ground following the method of Vázquez- de-Aldana et al. (2000). Immediately after harvesting, the material sampled in each plot was sorted manually into species and weighed in the field using a portable digital balance to determine contributions of individual species to total fresh biomass. Because of logistical issues sorted samples could not be dried individually and were pooled again, dried at 60 °C for about 48 hours in a well-ventilated oven (Adesogan et al. 2000) and weighed to determine both total herbaceous biomass and the contribution of individual species to total dry biomass.
For woody plants, samples of green leaves (including young and old) of each plant and each species were collected at random. For Vachellia tortilis (syn. Acacia tortilis), fruits (pods) were also collected as they were preferred by animals. A total of 75 samples (25 samples for each of the 3 seasons) were collected and analyzed. An effort was made to sample from the same species in all 3 seasons. However, during the hot dry season, some species had shed leaves, so samples were taken from other drought-resistant forage species as indicated by the pastoralists. In each season, only 1 sample per species was taken from the first site where the identified woody plant was encountered. In other words, only a single sample of each browse species was collected and only from a single tree/shrub. Nomenclature of plant species follows published volumes of Flora of Ethiopia and Eritrea (Hedberg and Edwards 1995;Edwards et al. 1995Edwards et al. , 1997 and was updated according to the taxonomy of the Genetic Resources Information Network GRIN (npgsweb.ars-grin.gov/ gringlobal/taxon/taxonomysearch.aspx).

Chemical analyses
After oven-drying of samples at 105 °C, dry matter (DM), organic matter (OM), crude protein (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF), acid detergent lignin (ADL), ash, ADF-Ash and in vitro digestibility of dry matter (IVDMD) were determined in the laboratory of the International Livestock Research Institute (ILRI), Addis Ababa, Ethiopia. IVDMD was analyzed only during the cool dry season.
CP was determined using the Kjeldahl method (N × 6.25), IVDMD by the in vitro rumen digestibility procedure (Van Soest and Robertson 1985) and NDF, ADF and ADL by the detergent system of analysis (Van Soest and Robertson 1985). Ash was determined by igniting samples at 500 °C (AOAC 1990).

Pastoralists' perceptions
Eight community-level group discussions were held in 4 places: Dikale (DIK), Dambala Abba Chana (DAC) (both in Dida Hara), Tesso Qallo (TSQ) and Dhibu Kolocho (DBK) (both in Web PA). One hundred and eight pastoralists (52 men and 56 women) participated in the group discussion. Pair-wise preference ranking was used to identify the most preferred forage species. According to their palatability to livestock, grass species were classified as highly desirable (decreasers), intermediate (increasers) and least desirable (pioneers), based on pastoralists' perceptions and field observations. Decreasers were defined as desirable grass species that are likely to decrease with heavy grazing pressure (Baars et al. 1996).

Data analysis
Descriptive statistics were used in organizing, summarizing and describing the data. Comparison of mean values was performed using two-tailed t-test following Fowler and Cohen (1996). ANOVA was applied to investigate variability across Land Use Units and seasons. Effects of season × land use interactions were determined using GLM (General Linear Model).

Rainfall
Rainfall data for Dida Hara were taken from Yaballo town, which was the nearest station to Dida Hara and that of Web was from Web station, which was collected by Southern Range Development Unit. The mean annual rainfall in Dida Hara (Yaballo) and Web is presented in Table 2. While mean monthly rainfall ranged from 5.9 to 144.1 mm in Dida Hara and 2.0 to 113.9 mm in Web (Table 2), variation within individual months was great as observed from the CV%.
Two rainfall peaks are conspicuous, demonstrating the bimodal nature of rainfall in the Borana Lowlands. Annual totals were highly variable, ranging from 188 to 803 mm with mean of 545 mm in Dida Hara, and from 211 to 638 mm with mean of 412 mm in Web. This difference in annual rainfall between the two sites was statistically significant (t = 2.196, df = 22 at P=0.05), indicating that Web is more arid than Dida Hara.

Herbaceous biomass
The above-ground herbaceous presentation yields (standing crop) were highly variable both spatially and temporally (Tables 3 and 4), with significant differences across seasons (P = 0.026) and Land Use Units (P = 0.000). Presentation yields for Dida Hara Kalo and Dida Tuyura Ranch, which were relatively protected sites, were higher than those in other Land Use Units, which were communally grazed, during the main rainy season.
Mean herbage biomass over all Land Use Units ranged from 921 kg DM/ha in the early wet season to 1,241 kg DM/ha in the cool dry season following the long rains. Available herbage biomass during the cool dry season was higher than in the other seasons.
In the early wet season (Long rains; Table 4), C. aucheri contributed nearly half (46%) to available herbaceous biomass at Dida Hara, and about 58% at Foora sites. C. mezianus made a large contribution (31%) at Foora site. In Web, Cenchrus ciliaris contributed the highest proportion (34%), while at Dida Tuyura Ranch, the dominant contributor was H. contortus (35%). Overall, C. aucheri was the main contributor to herbaceous biomass at most sites during both seasons. At Web sites, the dominant species were C. mezianus in November (short rains) and C. ciliaris in April (long rains).
Estimation of contributions to available biomass by highly desirable, intermediate and least desirable forage grasses showed that intermediate grasses were most prominent in Dida Hara, Dida Tuyura and Foora areas but not in Web zones (Table 5).

Nutritive quality of herbaceous species
Chemical composition of herbaceous samples was compared across both Land Use Units and seasons (Table  6). In general, there was a strong trend for an increase in CP% from the cool dry season to the long rainy season. Mean CP% was 48 g/kg DM in cool dry season, 62 g/kg DM after short rains and 76 g/kg DM during the long rainy season. There was variation in CP% across the seasons (P = 0.000) and Land Use Units (P = 0.013).
However, there were no differences in CP levels across various Land Use Units during the cool dry season. During the short rains, Dida Hara and Foora sites had higher CP than Web sites and Dida Tuyura Ranch. During the long rains, Web sites showed higher CP levels than Dida Hara sites with Dida Tuyura Ranch showing lowest levels (Table 6).

Nutritive value of woody plants
Mean chemical composition of leaves from woody plants is summarized in Table 7. CP concentration ranged from 47 to 168 g/kg DM. Senegalia brevispica (syn. Acacia brevispica), Balanites aegyptiaca, Chionothrix latifolia, Combretum hereroense, Cordia sinensis (syn. Cordia gharaf), Dalbergia microphylla, Ficus sycomorus, Maerua triphylla and Ziziphus sp. were the top species with higher CP%. CP% for most species reported in Table 6 was an average for 3 seasons. However, C. hereroense, C. sinensis, D. microphylla, F. sycomorus and Ziziphus sp. were sampled only during the hot dry season when other highly ranked species had shed their leaves. Salvadora persica, Kirkia burgeri, F. sycomorus, Commiphora kataf (syn. Commiphora erythraea) and Terminalia prunioides had high amounts of ADF, NDF and ADL. CP% in woody plants was highest in the wet season (169 g/kg DM) and lowest in the dry season (139 g/kg DM; Table 8).

Pastoralists' perceptions
Ranking of the forage value of plant species was performed through group discussions with pastoralists. According to the perceptions of Borana pastoralists, Cenchrus ciliaris and Digitaria milanjiana were grasses with the highest nutritive value (Tables 4 and 9). Similarly, Senegalia brevispica, Grewia tembensis and Maeroa triphylla were the most highly appreciated woody species (Table 8). Some forbs, such as Commelina africana, have high nutritional quality. All sedges (Cyperus spp.) were perceived to have intermediate nutritional value (Table 4). During the cool dry season, samples were not collected from DTR. OM = organic matter; NDF = neutral detergent fiber; ADF = acid detergent fiber; ADL = acid detergent lignin; CP = crude protein; IVDMD = in vitro dry matter digestibility.    - Empty cell (-) means the species was ranked below rank 5.

3
Species not listed in Table 7 (Table 7 contains species that were sampled, whereas this table contains species mentioned by local communities during free listing).

Herbaceous standing crop
The greater availability of forage on the government ranch and Kalos (enclosures) than on open-grazed areas such as the Worra and Foora showed the importance of exclosures for conserving and sustainably using forage resources during the dry season. This was in agreement with previous reports (Oba et al. 2001;Ayana Angassa et al. 2010). The pattern for highest mean herbaceous biomass to occur in the cool dry season (Jun-Jul) and lowest in the main rainy season (Mar-Apr), after the hot dry season, was according to expectations. In the cool dry season growth of vegetation has been stimulated by the long rains and herbage is mature. It is traditional for Borana pastoralists to protect Kalos from grazing during the wet season and open them for general grazing in the hot dry season (Dec-Jan), when there is relatively high accumulation of herbage mass. Grazing land management practices may be the main reasons for the significant differences in presentation yields of herbaceous biomass across the Land Use Units in this study rather than the actual productive ability of these areas. The different presentation yields of forage on the government ranch from those on the communal grazing lands may reflect grazing management imposed. There were only few Borana cattle on the government ranch during the study years and it was therefore subjected to only low stocking pressure during all seasons. Old growth accumulated and rank grass dominated the new growth of desirable species. Furthermore, the ranch was highly encroached by woody species that might have contributed to low forage quality.
Although the author of this manuscript agrees with protection of the ranch area, proper management following standard range management techniques should be followed. It should serve as a demonstration farm and learning laboratory for surrounding pastoralists. Ayana Angassa et al. (2010) also reported that there was variation as a result of rangeland management that affected biomass of most herbaceous species, plus grass basal cover and herbaceous species richness and diversity.
Quantifying the contributions of various species to forage mass allows useful comparisons of the productivity of different species and different management practices, and provides a basis for appropriate stocking rates to be developed (Sollenberger and Cherney 1995). Rainfall (precipitation), soil moisture, radiation, temperature, soil nitrogen and phosphorus are important environmental factors that affect herbage production (Gutman et al. 1990). The maximum presentation yield of herbage recorded in this study was 1,840 kg DM/ha during the short rains, while studies in similar semi-arid ecosystems reported much higher values. In similar arid areas of northern Kenya, mean herbaceous biomass inside exclosures was 4,180 kg DM/ha and that of continuously grazed open rangelands 1,802 kg DM/ha (Oba et al. 2001). Our results for herbaceous biomass yields from all sites in the Borana Lowlands fall within the category of poor, and such low herbaceous yields would directly affect livestock production and ecosystem stability.
Unlike the study of Ayana Angassa and Baars (2000), very low percentages of highly palatable grasses were found. Grasses considered of intermediate nutritive value by pastoralists were dominant at most sites, while highly palatable species were present at much lower levels, in agreement with the rangeland condition assessment report by Gemedo Dalle et al. (2006). This aspect is discussed further in a subsequent section.

Seasonal changes in forage value
Both herbage biomass and chemical composition of the herbaceous forage samples varied significantly across the seasons. Physiological age of forage species, time of grazing, species and botanical fraction are some of the factors that cause variability in chemical composition of forages (Adesogan et al. 2000). The decline in CP% in herbaceous forage from 7.6% in the wet season to 4.8% in the dry season, and that of woody plants from 16.9 to 13.9%, respectively, is in agreement with previous findings (Pérez Corona et al. 1998;Hussain and Durrani 2009;Habtamu Teka et al. 2013;Zhai et al. 2018). According to the report by Habtamu Teka et al. (2013), chemical composition of all grass species showed significant (P<0.05) variation between sites, seasons and species in agreement with our results.
In contrast, nutrient concentrations of browse from woody plants is subject to relatively less seasonal variation (Crowder and Chheda 1982) and this particularly enhances their value as dry season feeds for livestock (Dicko and Sikena 1992). About three-quarters of the woody forage plants studied were perceived as drought-resistant species by pastoralists. Several studies from arid ecosystems, e.g. Dicko and Sikena (1992), have shown that perennial shrubs retained high CP% for a longer period than herbaceous species, with a range of 7% in winter to 14% in spring. During the hot dry season when herbaceous species were almost dry and in limited supply for livestock, mean CP% in leaf material on woody plants was 13.9%, which would provide a valuable N supply to rumen microflora provided the nitrogenous components were digestible. Mean in vitro dry matter digestibility of the herbaceous forage was very low (35%) with a range from 32 to 38% indicating its limited potential to contribute energy. Conservation of these drought-resistant species is an important strategy for sustaining livestock production, especially during dry periods.

Comparing pastoralists' perceptions and scientific knowledge on forage nutritive value
The pastoralists identified Cenchrus ciliaris, Digitaria milanjiana, Megathyrsus maximus and Themeda triandra as highly palatable grasses. Similar perceptions of Borana pastoralists were reported by Habtamu Teka et al. (2013) and in an earlier report by Skerman and Riveros (1989). Furthermore, the pastoralists identified Chrysopogon aucheri and Digitaria milanjiana as drought-resistant forage grasses, concurring with the report by Skerman and Riveros (1989). Among the woody species Vachellia tortilis, Boscia mossambicensis, Chionothrix latifolia, Grewia damine and Pappea capensis for grazers (cattle) and Senegalia brevispica, Grewia tembensis and Maerua triphylla for browsers, e.g. camels and goats, were ranked as the top forage species. Le Houérou and Corra (1980) also reported that most of the woody plants identified during this study were considered palatable for animals, being preferred over other species.
Comparing pastoralists' indigenous knowledge with laboratory results showed that Borana pastoralists have an accurate perception of the nutritive value of the various forage species. In general, the pastoralists' knowledge of forage species growth and quality indicated that they know which species should be retained in the pasture and which were least useful to ensure sustainable animal production in the area.

Comparison of the Land Use Units
The high fluctuation of species composition on Web sites might be due to presence of permanent watering points (Web deep wells or Ela) in the area, which allowed high numbers of livestock to remain in the area resulting in overgrazing and consumption of desirable species. During the long rains, most livestock are taken away from the deep wells as water is more readily available and highly desirable species get the opportunity to regrow during this time.
When comparing Land Use Units, CP% varied across the Land Use Units as reported by Habtamu Teka et al. (2013). Overall, forage on the Dida Tuyura Ranch had the lowest CP levels and the highest concentrations of NDF and ADF reflecting the under-utilization of this area and accumulation of a bulk of mature fibrous material. This suggests that 'over-protection' of rangelands is not necessarily a desirable management strategy and significant defoliation by grazing animals at certain times might be needed to stimulate pastures and ensure a sustainable system.

Status of forage nutritive value in relation to livestock production
Herbaceous species had lower forage quality than the woody browse species in agreement with previous reports (Hussain and Durrani 2009;Gete Zewudu and Gemedo Dalle 2019). CP concentrations in standing forage exceeded the threshold level of 7% (Humphreys 1991;Pérez Corona et al. 1998;Ganskopp and Bohnert 2001) only during the long rains. This and other studies, e.g. Habtamu Teka et al. (2013), showed that quality of standing herbaceous forage in the Borana Lowlands was largely below the threshold level for good livestock production. The minimum recommended CP concentration in diets for small ruminants is even higher (9%) (Araújo Filho et al. 1998). While animals would select a higher quality diet from herbaceous forage than the whole plant data indicate, as the seasons progressed CP concentrations in the selected diet would decline. Access to some browse from woody species would alleviate this N deficiency as time progressed, while feeding of N supplements would also increase intake of the low quality forage. The study highlighted the importance of restoring degraded rangelands and also the need to improve forage quality through focused interventions aimed at increasing CP concentration in the herbaceous forage.
Tree leaves are nutritionally desirable, mainly as a source of CP (Forwood and Owensby 1985). The mean CP concentration in foliage of woody plants determined by this study was 11%. This was significantly higher than the CP concentration in herbaceous forage in agreement with reports from other areas, e.g. Musco et al. (2016). Furthermore, NDF, ADF and ADL concentrations in forage from woody plants were lower than those of herbaceous species. Trees and shrubs represent an integral part of diets for domestic ruminants in Africa and may constitute an important source of proteins, minerals and vitamins, especially during the dry season (Dicko and Sikena 1992). Borana pastoralists recognized the importance of woody species with higher CP concentration in this semi-arid environment. For long-term sustainability of the system, grazing management must be designed to retain the desirable woody plants, while restricting encroachment by undesirable trees and shrubs.
In conclusion, this study provides additional information on quantity (herbaceous) and nutritive value (both herbaceous and woody plants) of available forage in different seasons across different functional Land Use Units in a semi-arid rangeland of the Borana Lowlands, Ethiopia, where the main production system is livestock production, predominantly cattle breeding. Mean nutritive value of available forage from herbaceous species was below the critical level recommended for maintenance of both beef cattle and small ruminants for much of the time, while forage from woody species contained moderate levels of crude protein. This indicates the importance of maintaining a mixture of herbaceous species and desirable woody species in these rangelands. In the Dida Hara rangelands, very few highly desirable herbaceous species were present, even in areas where grazing was restricted. There is an urgent need to restore these rangelands to a condition where desirable species are more prevalent. Surprisingly, in Web areas desirable species were dominant during the long rains. Further studies are warranted to determine why these differences occurred between the Land Use Units and whether strategies can be developed to improve the situation in the Dida Hara area. It seems that 'over-protection' as has been practiced on the Dida Tuyura Farm is not the solution and more intensive study of the Web sites may yield possible solutions for testing. Furthermore, herd diversification for effective utilization of browse species was recommended as a result of this study.