Diet composition and preferences of Bohor reedbuck (Redunca redunca ) in the compound of Alage College, Central Rift Valley of Ethiopia

Abstract Numerous indices have been developed to compare use and availability of foods in field diets of wild ungulates. However, little attention has been given to laboratory analysis for comparing food preferences. To this end, a study aimed at investigating the diet composition and preference of Bohor reedbuck was conducted in the compound of Alage Agricultural College, Central Rift Valley of Ethiopia from 2017 to 2018 encompassing both dry and wet seasons. Bohor reedbuck is a medium sized horned antelope species endemic to Africa. Continuous focal animal observation was used to collect the data on plant species included in the diet of Bohor reedbuck. Focal individuals’ observation was carried out for 30 min in 10 min sampling interval during their active feeding period (early morning and late afternoon) over four different habitat types. The nutrient composition of plants consumed was determined using wet chemistry laboratory analysis. Bohor reedbucks consumed 15 species of plants; herbs comprised 94.3% of the foods they consumed. Digitaria abyssinica was the most preferred plant species with highest crude protein (23.75%) and less fiber (61.8% nitrogen detergent fiber and 27.8% acid detergent fiber). These findings suggest that food preference of Bohor reedbuck is determined by the nutritional content of the plant it consumed, since the area is more or less natural habitat in terms of plant species composition. For sustainable conservation of the species, there is a need to actively promote management of the plant species most preferred by the reedbuck to feed on.

their diet selection. Studies on feeding ecology of Bohor reedbuck in Ethiopia addressed only the diet composition and preferences without addressing the nutritional composition of the plants consumed by the species (Afework et al., 2009;Habtamu et al., 2012).
Feeding ecology is vital ecological phenomenon that describes the interaction between plant communities and herbivores. Foraging behavior is governed by various plant attributes such as plant availability and palatability and plant nutritional composition (Clauss et al., 2010;Tanentzap et al., 2009). Studies have indicated that selective foraging on high-protein and low-fiber content plants boost nutrient and energy intake and also decrease retention time, ultimately increasing intake capacity and fitness (Mysterud et al., 2001;Zweifel-Schielly et al., 2012). It has been reported that for herbivores the level of nutrients and toxins in a plant or its neighbors are key factors that determine the palatability of the plant to be consumed (Provenza et al., 2002). Since, all plants contain some amount of toxin herbivores cannot completely avoid toxins, but have to regulate the intake of toxins (Foley et al., 1999). Ultimately the interaction between plant chemistry and herbivore learning determines the coexistence of plants and animals affecting sustainable biodiversity conservation (Provenza, Villalba, Dziba, Atwood, & Banner, 2003). Furthermore, animal factors such as body size, digestive physiology and experience determine foraging behavior (Clauss & Hummel, 2005;Clauss et al., 2008;Parker & Bernard, 2006).
Studies on feeding behavior are necessary to enable estimates of the coexistence of wild animals in a particular ecosystem (Gutbrodt, 2006). The behavior of individuals and the strategies they adopt to maximize fitness plays an important role when a species' natural behavior can lead to conservation problems in habitats altered by humans (Festa-Bianchet & Marco, 2003). Information on nutritional composition of herbivores is vital for better understanding of resource requirements and offers intuition into herbivore influences on an ecosystem as well as animal populations (Parker & Bernard, 2006;Tanentzap et al., 2009). Clear nutritional composition analysis is crucial for clear understanding of animal normal growth and the maintenance of good health (Clauss et al., 2010). Therefore, this study aimed to determine the nutritional composition and preferences of Bohor reedbuck in the compound of Alage Agricultural College (AAC), Central Rift Valley of Ethiopia.

| Description of the study area
Alage Agricultural College is located at about 217 km south west of Addis Ababa in Central Rift Valley zone of Ethiopia. The college shares its boundaries between two administrative regions, namely, Oromia and South Nations Nationalities and Peoples Regional States ( Figure 1). The study area is geographically situated between 7°35′0′′ to 7°37′30′′N latitude and 38°25′0′′ to 38°27′30′′E longitude ( Figure 1).
The topography of the area is characterized by midland with altitude ranging from 1,580 to 1,650 m above sea level. The area is characterized by bimodal rainfall pattern. It receives most of its rain in the months of June to September, with minor erratic rainy season during the months of March and April. The average annual rainfall is 708-900 mm (Ethiopian National Meteorological Agency, 2017

| Sampling design
Four easily recognizable and approachable groups of Bohor reedbuck populations were selected during reconnaissance survey representing the four dominant habitat types namely; Acacia dominated wooded grassland, Riverine forest, Open grass land and Farm land.
Recognizable groups (known group size, natural marks, known home range and activity) of Bohor reedbuck were selected and observed in each of the four different habitat types. Individuals observed up to 50 m far from the group was counted as the member of the same group (Caro, 1999).
The dietary data of Bohor reedbuck were collected through continuous focal animals observation of individuals (ranged from 5 to seven individuals per group, the number of individuals varied based on the habitat type) of the same group that were not reproductively active (all individuals targeted were available during the whole data collection period) (Altmann, 1974). Focal animals' observation was carried out through following up all individuals of one selected easily recognizable and approachable groups of Bohor reedbuck population at a time (Yihune & Bekele, 2012). Observation of focal individuals was carried out for 30 min in 10 min sampling interval during morning hours (7:00 a.m. to 11:00 a.m.) and late afternoon hours (3:00 p.m. to 5:00 p.m.) when the animals were actively foraging, for each recognizable group at each habitat type during both dry and wet seasons. A total of 82 hr in 10 days (both in wet and dry seasons) were spent observing the focal animals. In each season a total of 41 hr in five days were spent observing the focal animals. For each group in each habitat type equal numbers of hours were spent observing the focal animals. The observations were carried out with a distance (between 50 and 200 m) approaching the group from different directions (Afework et al., 2009;Tekalign & Bekele, 2012). All individuals in a group were observed at the same time with a help of trained personnel of the Bohor reedbuck identification techniques, the method of sampling and plant consumed. Field assistants were B.Sc. graduates in wildlife management and related natural resource management disciplines.
The frequency of occurrence of the plant species in the study area that was consumed by Bohor reedbuck and not consumed but available to Bohor reedbuck was estimated using 21 plots (4 m by plots could exhaustively capture the available plant species in the area. The distance between each transect were 0.5 km, and also, the distance between each plot was 0. 5 km and the length of each transect was 2 km.

| Data collection
The data on groups of feeding Bohor reedbuck were recorded particularly; the type of plant species consumed and the type of food item preferred (young leaves, mature leaves, shoots, flowers and fruits) during both wet and dry seasons. To minimize any disturbance effect caused by the observer camouflaging clothes were used. Furthermore, the animals were approached walking gently and quietly against the direction of wind movement to avoid being scented by the animals.
Once the animals were approached to observable distance, before collecting information they were allowed to become acclimated and resume feeding for a minimum of 10 min following Hochman and Kotler The plant species were identified by the guide; Flora of Ethiopia and Eritrea (Phillips, 1995) and by comparison with archived specimens from the collection housed at the National Herbarium in Addis Ababa University.
The untouched fresh (both young and mature leaves, but not old dried ones) plant leaves and shoots, flowers and fruits for only grass species consumed by the Bohor reedbuck from the four habitat types were collected in a plastic bag for the proximate analysis.
All plant parts were mixed together for the proximate analysis, no classifications in to plant parts were made. Forage quality analysis was carried out based on wet season data and across habitat due to logistic constraints. Thus, seasonal or habitat variations were not tested. The proximate analysis of 15 plant species (five from highly preferred, five from moderately preferred and 5 from rarely consumed and avoided) consumed by Bohor reedbuck was used as forage quality test. As part of the proximate analysis Moisture, Dry Matter, Organic Dry Matter, Ash, Nitrogen/Crude protein, Neutral Detergent Fiber (NDF), and Acid Detergent Fiber (ADF) was analyzed following AOAC (1995) analytical procedures. According to AOAC (1995) the following procedures were used to carry out the laboratory analysis for all proximate analysis.
Before proceeding to the analysis the weight of sampled items were weighed using electronic balance and then fresh forage samples were dried for 48 hr at 60°C in an air-circulation oven to obtain air dried samples ready for grinding. The weight of dried sample was recorded with electronic balance and also the dried samples were grounded to 1mm particle size with a Wiley mill. Dried and ground samples were stored in airtight containers away from heat and light to avoid moisture in take. The dry matter content of the sample plant were determined through adding approximately 2 g of ground sample to the crucible and drying it overnight at 105°C with in the oven dry, then the percentage of dry matter (DM) was calculated. The ash content of the sample was determined by igniting dry matter samples overnight at 550°C in muffle furnace, and then percent ash was calculated. Determination of total nitrogen (crude protein) was conducted mixing 0.2g sample with equal amount of catalyst and digesting the sample in sulfuric acid using K 2 SO 4 / CuSO 4 as a catalyst. Then, N was converted into NH 3 , then the distilled solution was trapped in boric acid and titrated with H 2 SO 4 , then Percent of Crude Protein (CP) was calculated. Determination of the Neutral Detergent Fiber (NDF) was carried out through adding 0.5 g sample (Ws) in 600 ml Berzelius beaker and adding 100 ml of neutral detergent solution and 0.5 g sodium sulfite (Na 2 SO 3 ). Then the mixture was boiled for one hour in refluxing apparatus and was poured through glass crucibles and vacuum was admitted, residues were ashed for three hours at 550°C and cooled to room temperature in desiccators. Determination of Acid Detergent Fiber (ADF) was conducted through adding 0.5 g sample (Ws) and 100 ml of acid detergent in 600 ml Berzelius beaker solution. Then the mixture was boiled for one hour in refluxing apparatus and poured through glass crucibles and admitted vacuum, ash residues for three hours at 550°C and cool to room temperature in desiccators.
The relative density and frequency of occurrence of the plant species that was consumed by the Bohor reedbuck and not consumed, but available to Bohor reedbuck were recorded through estimating their density and occurrence in different plots and extrapolating the average to the size of the total study area.

| Data analysis
Microsoft Excel 2010 was used for data summarization and to organize results into tables and figures. All statistical testes were analyzed using SPSS version 23 computer software program. Mann-Whitney rank test analysis was used to examine the relationship between relative frequency of consumption of species and its availability (i.e., relative frequency of occurrence) following Johnson (1980).
The food preference indices (FPI) for each plant species consumed by Bohor reedbuck was calculated and ranked to determine the most preferred plants. The feeding preference index (FPI) of Bohor reedbuck for each plant species was calculated by the The frequency of occurrence (FO) of the plant species in the area also was calculated using equation (3) below (1) FPI = frequencyofusagetheplantspecies frequencyofoccurrence (2) Frequency of usage of the plant species = number of record the plant species consumed Total number of sample records The absolute density (density) of the plant species in the area is calculated using Equation (4) below The Relative Density (RD) of the plant species in the area was calculated by Equation (5); Percentage preference of Bohor reedbuck to plant parts (%PPP) was computed by Equation (6); Sorenson's similarity (Ss) index between the plant species consumed by Bohor reedbuck during wet and dry season was made fol- where a = number of species consumed in both seasons, b = number of species unique to wet season c = number of species unique to dry season.
Chi-square test was used to examine frequency of each plant species consumed among habitat type.
According to AOAC (1995) the percentage values of all the proximate analysis were calculated by the Equations (8)-(14); where %DM = Percentage of dry matter, W 0 = Weight oven-dry crucible + sample, W t = Weight oven-dry crucible, and W s = record weight.
where Percentage of Acid detergent fiber (ADF), W 0 = Weigh sample and crucible, W t = Weigh oven-dry glass crucible and W s = weight of sample.
Since nutrient requirement data for Bohor reedbuck is not available, the nutrient requirements of common bushbuck (Macleod et al., 1996) were used to determine an ideal or optimal nutritional profile.

| Diet composition
this study is a study conducted in Dinder National Park of Sudan indicated that Bohor reedbucks observed to consume a wide variety (25) of plant species in their diets (more than 85% of available plants) (Ahmed, 2005). Similarly, the generalist feeder common warthog has been observed to forage on about 83% of available plants in Tanzanian savanna (Treydte et al., 2006). The reduced variety of plant consumed in the present area could be attributed to the reduced floral diversity and poor nutritional quality of plants available in the present study area (WWDSE, 2013). The present study area is characterized by infertile sandy soil and erratic rainfall that promote sparse vegetation diversity and less nutritional quality feed. Furthermore, since the area is less protected as compared to Dinder National Park disturbances such as human activities, livestock grazing could reduce both the diversity of plant species and foraging opportunity of the species. The majority (50%-55%) of plant species avoided by Bohor reedbuck in the study area was lowest in nutritional composition (low crude protein, which is likely accompanied by increased carbohydrate) as compared to the benchmarked species common bushbuck (Macleod et al., 1996) and rarely occurs during dry season (WWDSE, 2013). Furthermore, as the dry season progresses the nutritional content (protein content, which is likely accompanied by increased carbohydrate) of the forage species decline and hence, become less preferred and attractive (Omphile et al., 2004). Several studies have indicated that floral diversity of a particular area may influence feeding composition of ungulates and shape their selective feeding behavior (Clauss et al., 2007;Codron et al., 2007;Owen-Smith, 1994). Bohor reedbucks were both grazers and browsers but they were more grazers than browsers. Grasses comprised 73% of the plant species consumed by Bohor reedbuck and it contributed 94.3% of dry matter of their feed (Table 1). The predominant selective grazing of grass species over other plant species available by Bohor reedbuck was an indicator of their grazing behavior (Cerling et al., 2003;Estes, 1991;Kingdon, 1997). Similarly, Ahmed (2005) and Gutbrodt (2006)  A total of 11 and 13 species of plants were consumed during the wet and dry seasons respectively (Table 1). Despite the fact that plant species Panicum infestum, Pennisetum ciliare, Acacia seyal, and Balanites aegyptiaca were available in the study area during both dry and wet seasons, they were only consumed during dry season (Table 1). Cenchrus cilaries and Hyparrhinia hirta were available as forage only during wet season and consequently consumed only during wet season (Table 1) The significantly high diet overlap between dry and wet seasons could be explained by the year round availability of most consumed species in the study area. The slight increase in plant species consumed during dry season could be due to the fact that during dry season, the forage availability and quality of most plant species consumed decline and, hence the species had to include few more species to meet up the compromised forage quality and availability. Since   (Gutbrodt, 2006), it may include alternative foods during the dry season to meet their nutritional requirements.
The findings of feeding preference and proximate analysis indicated that crude protein and moisture content were the potential driving forces for selective grazing/browsing (Tables 2 and 3). Bohor reedbucks were selecting for a consistent nutrient profile (mainly crude protein and moisture) each season. Species highly preferred were those with relatively higher crude protein and high moisture contents during both dry and wet seasons. However, during dry season moisture content was found to highly determine selective foraging. For instance, Acacia seyal was exclusively browsed in dry season due to its better moisture content potential, despite its lowest crude protein content. Acacia seyal is typical arid tree species that conserve moisture in its leaves due small   (Table 3). Several authors have reported that selective grazing is highly governed by nutritional content such as CP, moisture, NDF and ADF (e.g., Chinomona et al., 2018;Gutbrodt, 2006;Kingdon & Hoffmann, 2013;Omphile et al., 2004).

| Feeding preference
The plants species such as Cynodon dactylon (0.89 frequency of usage) and Chloris gayana (0.54 frequency of usage) were frequently consumed plant species during both seasons (Table 2) reported that diet composition of many ungulate species varies substantially among seasons (Ego et al., 2003;Halsdorf, 2011).
There was also statistically significant variations in the frequency of each plant species consumed among habitat types during both dry (χ 2 = 24.23, df = 3, p = .01) and wet (χ 2 = 17.27, df = 3, p = .01) seasons. This is mainly attributed to the change in availability of a particular species and change in nutrition content, a general decline of nutrient content as dry season continues (Ombabi et al., 2001).
Particularly, all those species frequently consumed during the dry seasons were available during both seasons. Both Pennisetum ciliare and Panicum infestum were frequently consumed during the dry seasons mainly due to their highest moisture (11.97 and 11.54% respectively) content (Table 3).
The final partition of the cluster group analysis clustered all the observations at 66.7% similarity in to three clusters. Cluster 3 has the highest number of observations (8), while cluster 1 has the least number of observations (2) (Table 4, Figure 2). The three clusters are highlighted in different color in Figure 2.
The 2 species clustered in cluster one had higher crude protein, moderate NDF and ADF, the 5 species under cluster two had moderate moisture, crude protein highest, NDF and ADF proportion. On the other hand, species under cluster three had the lowest moisture, crude protein, NDF and ADF proportions ( Table 5).
The most preferred species (with relatively highest protein content); Digitaria abyssinica and Cynodon dactylon followed by Panicum maximum and Chloris gayana contribute the highest proportion (60.57%) of the diet of Bohor reedbuck. The report from Sudan at Dinder National Park (Bordering Ethiopia) by Ahmed (2005) shows that Cynodon dactylon is the first ranked species to be consumed by Bohor reedbuck and also this plant species contributed the highest proportion to the diet of nyala (Tragelaphus angasi) in southern Africa (Pienaar, 2013). For herbivores, foraging preferences are shaped by multiple constraints, such as the availability of the food, nutritional quality of the food and the requirements of the animals (Sinclair et al., 2006). The feeding preference of Bohor reedbucks might be influenced by the nutritional quality, especially crude protein and moisture content and frequency of occurrence of the plant species. The first rank preferred plant species by Bohor reedbuck (Digitaria abyssinica and Cynodon dactylon) illustrate the above statement. Digitaria abyssinica was preferred by its high nutritional quality (high crude protein and the lowest ADF) as evidenced by proximate Note: The clusters are indicated by different colors in the Dendrogram (Figure 2).

TA B L E 4
Summary of statistics of cluster group analysis for the proximate analysis of 15 plant species consumed by Bohor reedbuck, cluster observations using centroids' distance analysis, even though it had the lowest occurrence. However, Cynodon dactylon had moderate nutritional quality (18.56% CP and 11.71% moisture) but its high occurrence could make it be the more frequently consumed.
Furthermore, the optimal foraging theory states that animals maximize fitness through a foraging strategy that incur low cost and with maximum energy gain, maximizing the net energy gain.
Particularly, Cynodon dactylon (0.79 frequency of occurrence) followed by Panicum maximum (0.79 frequency of occurrence) and Chloris gayana (0.81 frequency of occurrence) are the most available forage species in the area and Chloris gayana is the second highest in terms of CP content (Table 4). Hence, a feed with second highest nutritionally quality (Chloris gayana, 21.17% CP) and that is readily available (low energy cost) maximize net energy gain and will have high potential for selective grazing. But, frequency of occurrence by itself might not affect the feeding preference of Bohor reedbuck; Cynodon aethiopicus demonstrates this. It was the least preferred species despite its highest frequency of occurrence (0.90) in the area. This is possibly due to the lower digestibility and food quality of the plant species; nutritionally this species had the lowest crude protein (10.41%) and the highest ADF (43.2%) that makes it more indigestible and less palatable. Furthermore, the presence of anti-nutritional factor such as secondary metabolites (flavonoids, tannin etc) determines selective grazing (Reed, 1995). Particularly, Cynodon aethiopicus could be least preferred due to its high concentration of toxic substance. It has been reported that C. aethiopicus has the potential to produce high levels of prussic acid (HCN) any time during the growing season up to 250 ppm (Harlan et al., 1970). Furthermore, Rhus quartiniana had the highest CP content (21.85%), but among the least preferred. This is probably attributed to its high concentration of tannin. It was reported that R. quartiniana contains high concentration of tannin in the leaves and barks (Miller et al., 2001).
Bohor reedbucks preferred to consume young leaves and shoots (comprised over 95% of their consumption) of the plants during the study period, while flowers (1.33%) (few grass species) and fruits (2.90%) (few grass species) of the plant species were the least preferred ( Figure 3). Seasonally, Bohor reedbuck's preference to consume young leaves was similar during both wet and dry seasons, but a slight difference was observed with the consumption of other plant parts. There was higher consumption of shoots of plants during the wet season than the dry season whereas leaves consumption increased during the dry season (Figure 3). Note: % M = percent moisture, % CP = percent crude protein, % FP = Percent food preference. The centroid represents the "average observation" within a cluster across all the variables (% M, % CP and % FP) in the analysis.

F I G U R E 2
Since the majority (74%) of plants selectively grazed by the Bohor reedbuck was grasses, the nutritional analysis constituted mainly leaves shoots and rarely flowers and fruits. For the remaining nongrass species, only leaves were considered for nutritional analysis.
The Bohor reedbuck preference to feed on young leaves and shoots could be due to the fact that green parts of the plant are more nutritious with high moisture content and easily digested due to low-fiber content as opposed to the dry parts (Woie, 1984). The proximate analysis also indicated that high concentration of nutrients (e.g., CP) in the leaves and shoots for those that were highly preferred. It was reported that the standard range of crude protein ranges from 8% to 30%

| CON CLUS ION
Both plant species availability and nutritional quality determine food preference of Bohor reedbuck. The feeding preference of Bohor reedbucks to graze grass species having high-quality nutritional profile implies the need for promoting sustainable conservation and growth of those preferred plant species as an input for sustainable conservation of the Bohor reedbucks. Further in-depth research needs to be done on ecological importance of Bohor reedbuck.

ACK N OWLED G M ENTS
We would like to acknowledge Technical, Vocational and Educational Training Coordination Office of the Ethiopian Ministry of Agriculture and Natural Resources for financing the data collection. We are highly indebted to Tezazu Animaw for his kind support during laboratory work.

CO N FLI C T O F I NTE R E S T S
We the authors declare that we have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
I will here confirm that I will avail the data that support the findings of this study up on after publication of the manuscript. Particularly, the data that support the findings will be available in Bohor reedbuck diet analysis file name at Zenodo data repository following the date of publication.