Participatory inventory and nutritional evaluation of local forage resources for smallholder free-range beef production in semi-arid areas of South Africa

Feed scarcity is a major challenge facing free-range beef farming in semi-arid areas. Specifically, low quality and quantity of forage in rangelands and higher feeding costs are the main constraints limiting smallholder free-range beef farmers’ participation in mainstream beef markets. Using farmers’ participatory approaches, this study identified major locally available forage resources (LAFRs) and evaluated their nutritional value. A total of 40 free-ranging commercially orientated smallholder beef farmers were interviewed using semi-structured questionnaires and tasked to identify LAFRs in the Cradock and Middelburg areas of the Eastern Cape, South Africa. Chemical analysis was conducted for the most cited forages such as African sheepbush (Pentzia incana), sweet thorn (Vachellia karroo) leaves and pods, reed (Phragmites australis), lucerne (Medicago sativa) hay, natural pasture grasses (NPGs) and barbary fig, or prickly pear cactus, (Opuntia ficus-indica) cladodes collected from twelve participants’ farms. Feed shortage was ranked by more than 53% respondents as the main constraint to smallholder beef production. Regardless of the farming area, crude protein content of V. karroo leaves and pods averaging 18.8 and 19.5%, respectively, was higher than other LAFRs. However, V. karoo pods had relatively low ash content than other forages in both farming areas. Opuntia ficus-indica attained high in vitro neutral detergent fibre digestibility at 12, 24 and 48 hr incubation periods, due to low neutral detergent fibre, acid detergent fibre and acid detergent lignin. Integration of participatory inventory and chemical analysis proved to be reliable in identifying LAFRs, with V. karroo leaves and O. ficus-indica cladodes being the main potential forage resources for inclusion in beef cattle diets. Further research is recommended to substantiate their supplementary nutritive value and level of inclusion in beef cattle finishing diets.

In developing countries, cattle in the smallholder farming areas are kept for a variety of purposes including improving household food security, generating income, eradicating poverty and to sustain people's livelihoods (Marandure et al. 2016;Mapiye et al. 2020). However, the smallholder beef industry is challenged amongst others by lack of access to land, land use changes, tragedy of the commons, and climate change (Abule et al. 2005;Musemwa et al. 2010). According to Meissner et al. (2014), free-range commercially orientated smallholder beef farming is often practiced extensively under communal set-up where animal feeding relies solely on natural pastures and crop residues with little supplementation. Thus, the profitability in this beef farming sector depends largely on the quality and availability of natural pasture grasses which vary seasonally.
In southern Africa, despite the high number of cattle in the smallholder beef farming sector, overall productivity is quite low compared to the commercial beef farming sector (Bennett et al. 2007). Feed scarcity is among the major challenges, limiting animal performance and productivity (Madzivhandila et al. 2008). The forage grasses are high in fibre and poor in most essential nutrients including crude protein (CP), energy, minerals and vitamins (Baloyi et al. 2009;Tonamo et al. 2016), especially during dry season. Besides fluctuation in grass biomass production and quality, natural pastures remain a vital aspect of extensive beef production (Pell-Stroebel and Kristjanson 2011). Thus, animal nutritionists need to identify and ascertain alternative low cost and locally available forage resources (LAFRs) which are highly nutritious and adapted to arid areas, given recent trends of climate change with highly variable Introduction rainfall and recurrent droughts. This is important not only as a strategy to cope with climate change, but also to ensure that cattle from smallholder beef production meet the standards of the mainstream beef markets. This would in turn ensure food security and help in alleviating poverty in communal areas.
There are plenty of LAFRs that could be harvested and preserved for beef cattle feeding (Mlambo and Mapiye 2015). Arzani et al. (2006) and Fayemi et al. (2011) suggested that preserved forage maintains relatively high nutritional quality than standing grasses and crop residues during the dry season. However, there is limited information on LAFRs that can be used to supplement cattle in free-range smallholder beef cattle production in arid and semi-arid areas of South Africa (Mlambo and Mapiye 2015). In that regard, it is crucial to make an inventory of the LAFRs and evaluate their nutritive value for free-range smallholder beef cattle production in dryland areas to meet the standards of mainstream beef markets (Nyambali et al. 2022). Farmers' participation with their associated indigenous technical knowledge has always been a panacea in understanding livestock feeds and feeding practices because farmers frequently monitor their pastures, animal foraging behaviour and diet selection patterns (Garcia et al. 2019). The objective of this study was, therefore, to identify potential forage resources using indigenous knowledge and evaluate their nutritive value for cattle feeding in free-range commercially orientated smallholder beef production.

Site selection and description
The study was conducted in the Cradock and Middelburg areas under Inxuba Yethemba Local Municipality, Chris Hani District of the Eastern Cape province (EC) of South Africa. Figure 1 shows the location of the study areas in the EC, South Africa. Cradock lies at 32°09′51.19″ S, 25°37′9.05″ E, in the upper valley of the Great Fish River, 896 m above mean sea level. Middelburg lies at 31°29′22.79″ S, 25°01′1.20″ E, in the Great Karoo, 1 279 m above mean sea level. The vegetation of the areas is classified as mixed veld dominated by Vachellia karoo and spineless cactus pear among other browse trees and shrubs, with mixed grasses and Karoid species dominating the herbaceous layer. The combination of climate, topography and geology limits crop production in the study areas, hence livestock production is a major farming enterprise (CSIR 2004). The annual rainfall at Cradock and Middelburg is 341 and 372 mm, respectively (du Toit 2019; Choruma et al. 2019Choruma et al. , 2022. Rainfall occurs mainly late summer in Cradock (Choruma et al. 2019) and mid-summer in Middelburg (du Toit 2019). While March is the wettest month in both areas, rainfall distribution patterns differ between the two areas, with 27% of rain falling from April to September in Middelburg (du Toit 2019), whereas Cradock is relatively dry at this time of the year. The average midday temperatures in Cradock and Middelburg range from 16.8°C and 15.0°C in June to 29.5°C and 30.0°C in January, respectively. The region is coldest in July when temperatures drop to 2.3°C on average during the night (CSIR 2004;World Bank 2016). Although day length is similar in these study areas, minimum temperatures tend to vary widely, with Middelburg generally being cooler than Cradock (Jordaan et al. 2017).

Baseline Survey: inventory of forage resources
The baseline survey was conducted to identify the locally available forage resources (LAFRs) used by the free-range commercially orientated smallholder beef farmers. The survey was conducted within a 250 km radius of Cradock, which encompasses Middelburg. Inxuba Yethemba Local Municipality of the Chris Hani District was selected purposively on the bases that the Agricultural Research Council (ARC), the local abattoir and local fresh produce retailers are currently working with the free-range commercially orientated smallholder beef farmers to promote their access to free-range beef markets.

Farmer selection and data collection
The target population for the study were free-range commercially orientated smallholder beef farmers from Cradock and Middelburg farming areas who owned at least five cattle. A total of forty free-range commercially orientated smallholder beef farmers were selected randomly in each study area, totalling 80 respondents. Thirty-five percent (35%) of the farmers were individual farm owners and the rest were commonage farmers. Commonage farmers were defined according to Van Rensburg et al. (2009), as a group of farmers who share equal rights of grazing due to common ownership of the grazing land. Farmers were interviewed by trained enumerators using a pre-tested structured questionnaire. The questionnaire was administered in the local languages (IsiXhosa and Afrikaans) and later translated to English. One-on-one interviews were conducted since some farmers were individual farm owners. A farm manager and or leader of the commonage farming group were used as key informants. A guide with definition of concepts was prepared and presented using local languages prior to commencement of the interviews to familiarise respondents with questionnaire. The questionnaire captured data on constraints facing livestock production and types of LAFRs [woody (browse shrubs and trees) and herbaceous plants (legumes and grasses)]. Answers were coded and if respondents were asked to give various factors or attributes, they were instructed to rank them numerically by order of importance.
Thereafter, 12 farms (n = 6 per area) were selected randomly for identification of LAFRs. In each selected farm, three 300-m line transects were laid in three dominant vegetation ecotypes and the key informants were tasked to identify LAFRs within 20 m along the transects. The following LAFRs were identified: reed (Phragmites australis), African sheep bush (Pentzia incana), sweet thorn tree (Vachellia karroo) leaves and pods, lucerne hay (Medicago sativa), natural pasture grass and prickly pear cactus (Opuntia ficus-indica).

Experimental design
The experiment was arranged as a randomised complete block design, with the farming area considered as a block, whereas LAFRs (n = 7), farms (n = 6 per area) and the 300 m × 20 m belt transects (n = 3 per farm) were considered as the main treatment, replicates and experimental units, respectively. However, due to differences in some climatic attributes (e.g. rainfall) between the two farming areas, we assessed the interaction between LAFRs and farming area to establish if chemical composition of LAFRs varied by agro-ecological zone.

Sampling of potential forages
Leaves, pods, cladodes and fresh grass herbage were hand-clipped and placed into brown paper bags during the rainy season (January) from different farms (n = 12). Grasses were cut as bulk at a stubble height of 5 cm in each quadrat randomly placed every 50 m along the 300-m transects. All non-grassy material was removed from the grass biomass to avoid contamination. Cattle were intended to be the bulk grazers in this study (Kimuyu et al. 2017), therefore grasses were sampled and analysed in bulk rather than by species. The species composition of grasses sampled for chemical analysis is presented in Table 1. For browse species, V. karoo and P. incana particularly, the foliage was clipped on the lower, middle and top parts of the canopy of six individuals randomly selected along the transect. At least three cladodes of the prickly pear cactus were clipped per randomly selected individual encountered along the transects. The samples for M. sativa were collected from bales purchased locally or produced directly at the survey farms, assuming minor variation in chemical composition which might be caused by differences in growing conditions (fertilisation, irrigation etc.) applied in different farms. Samples (n = 6 per LAFR) were oven-dried to constant weight at 65°C for grasses, legumes and shrubs (Sanson and Kercher 1996) and at 105°C for cactus cladodes (Akanni et al. 2015). The dried samples were milled to pass a 0.8 mm sieve for analysis of chemical composition and in vitro NDF digestibility.

Chemical analyses
Subsamples (n = 6) of 0.5g were pooled for each LAFR and analysed using Dumas method (Leco FP-528; Leco Corporation, St. Joseph, MI, USA) to determine nitrogen (N) content (Wiles et al. 1998). Prior to each session, ethylenediamine tetraacetic acid (EDTA) was used as a standard. The forage samples were combusted in the presence of oxygen at 1 000 °C to release nitrogen oxide (NO x ), carbon dioxide (CO 2 ) and H 2 O. Combustion products were equilibrated, after which a gas mixture was passed over hot copper to remove O 2 for conversion of NO x to N 2 (Raffrenato et al. 2019). The CO 2 and H 2 O were trapped and the N 2 was detected using a Thermal Conductivity Detector (Muller 2017  The neutral detergent fibre (NDF) and acid detergent fibre (ADF) were analysed using the semi-automated equipment for fibre analysis (ANKOM Technology, ANKOM 200/220 fibre analyser, using alfa-amylase). The acid detergent lignin (ADL) was determined as described by Van Soest fiber analysis (Van Soest et al. 1991). The fibre values were expressed exclusive of residual ash content. Dietary starch was analysed according to Hall (2009) method. To determine in-vitro digestibility of NDF, ruminal fluid was extracted from two cannulated Holstein Friesian cows (body weight = 449.4 ± 4.23 kg), feeding solely on a kikuyu cultivated pasture. The 0.5 g LAFR samples were inoculated with ruminal fluid with a buffer solution (Raffrenato et al. 2019). The jars were placed in the rotation racks inside the Ankom Daisy II incubator containing perforated agitator (Tassone et al. 2020). The in vitro digestibility of NDF was determined at 12, 24 and 48 hr incubation periods at 39°C (Ammar et al. 1999). Dry matter loss was determined according to Tilley and Terry (1963), as a difference between pre-incubation sample weight and post-incubation weight for each period (Tassone et al. 2020; Gobena et al. 2022).

Statistical analyses
The data, largely on farmers' knowledge was analysed using descriptive statistics (frequencies and means). The Chi-square test was used to assess the differences on the perceptions between farm type and farming areas using Statistical Package for Social Sciences version 20.0 (SPSS 2011). The Wilcoxon test (NPAR1WAY procedure) was used to derive mean ranks. The General Linear Model (SAS 2009) was used to ascertain the interaction between LAFRs and farming area and the effect on the chemical composition and in vitro digestibility. The linear model was as follows: where Y ijk = response variable (forage chemical attributes); µ = overall mean; A j = jth effect of the farming area (Cradock and Middelburg); A k = kth effect of the LAFRs; Aβ jk = farming area and LAFRs interactions, and Ԑ ijk = random error.
Mean separation was conducted using Tukey test, with means considered significantly different at p < 0.05.

Constraints to beef cattle production
The main constraints to free-range smallholder beef farming sector are reported in Table 2. Feed shortage was perceived by 53% and 62% respondents at Cradock and Middelburg, respectively, as the most important constraint to beef production. In both farming areas, poor productive and reproductive performance of cattle was ranked as a second most important constraint by on average 18% respondents. The lack of knowledge on cattle and pasture management was ranked as the third most important  The lower the ranking index of a resource, the greater is its importance, * Mean rank index of the different resources are significantly different at p < 0.05, Sig 1 = Significance; NS = not significant; * significant, Cereal grains: maize corn and oats, Browse species: Acacia karroo and Pentzia incana, Herbaceous legumes: Medicago sativa, Natural pasture grasses are as described in Table 1.

Feed resources
Farmers across both the farming areas and farm types identified locally available forage feed resources and they were ranked according to their level of utilisation (Table 3). Regardless of farming area and farm type, most farmers ranked natural pasture grasses as the main feed resource for cattle feeding. The crop residues were ranked the second most important feed source utilised in both areas, but the perception differed with the farm type, with browse species regarded as the second most utilised feed source in commonage farms. Similarly, cereal grains and browse legumes were third and fourth most utilised feed resources, respectively, in both farming areas.

Chemical composition and in-vitro NDF digestibility of selected forage resources
The chemical composition and in vitro NDF digestibility of the locally available forage resources in the Cradock and Middelburg areas are presented in Tables 4 and 5. LAFRs and farming area interacted significantly (p < 0.05) on DM, CP, starch, NDF and ADL content. Natural pasture grasses, P. incana and V. karoo leaves had significantly (p < 0.05) higher DM at Middelburg relative to other species at Cradock. Regardless of the farming area, V. karroo pods had significantly (p < 0.05) higher OM (94% on average) and CP contents (19.5% on average), and relatively low total ash content (6.25% on average) than other species. Similarly, leguminous forage of V. karroo and M. sativa contained relatively high CP than other species regardless of the farming area, with their CP averaging 18.8 and 16.8%, respectively. Likewise, V. karroo and M. sativa had significantly (p < 0.05) more ash content than other species. Natural pasture grasses and P. australis contained higher NDF contents than other species regardless of the farming area. Acid detergent fibre content was highest in V. karroo leaves and natural pasture grasses compared to other species. Likewise, V. karroo leaves and pods had higher ADL contents compared to other species, with leaves having on average 6% higher ADL than pods.
As depicted in Table 5, farming area by LAFR interaction was significant (p < 0.05) for in vitro NDF digestibility after 12 and 24 hrs. Opuntia ficus-indica had a 4.0-fold and 1.6-fold higher in vitro NDF digestibility at Middelburg than at Cradock after 12 and 24 hrs, respectively. The in vitro NDF digestibility of O. ficus-indica increased with incubation period from 75% to 83% from 12 to 48 hrs. The in vitro NDF digestibility was relatively low for P. australis and V. karroo pods after 12 and 24 hrs; but, from 24 to 48 hrs, the NDF digestibility for P. australis increased 3.0-fold higher.

Discussion
Smallholder farmers identified feed shortage as their main constraint to free-range beef production in that they rely largely on natural pasture grasses for cattle feeding. In particular, farmers perceived low forage quantity and quality during the dry season as the main cause of poor animal performance. Consequently, during the dry season, farmers depend more on maize stover and cereal grains for feeding (Table 3). This feeding regime is common in free-range beef cattle farming, but deficiency of essential minerals and CP calls for use of alternative feeding sources. Amongst others, farmers at Cradock indicated that they also use browse as cut and feed to feed their beef cattle. The use of browse for cattle feeding has been suggested also by Arzani et al. (2006) because browse species are rich in protein, energy and minerals. Nutritional analysis of forage resources in the current study indicated that natural pasture grasses were deficient in CP, with very high accumulation of fibre content (Table 4). These findings concur with farmers' perception that natural pasture grasses have low quality. The commonage farmers ranked browse foliage as the second most important source of feed (Table 3). The commonage farmers' perception concurs with the nutritional analysis that indicated that V. karoo leaf foliage was rich in CP and minerals as indicated by higher ash content. Current results indicated that indigenous farmers' knowledge on cattle feeding is reliable.
The nutritional analysis indicated that regardless of the study area, CP content of leguminous species (V. karroo and M. sativa) was higher than that of other species and surpassed the CP requirements for all physiological stages of a beef cow, including lactation stage (Raseigh et al. 2014). These results are fundamental given that the cattle herds consist largely of cows in both farming areas (Supplemental Table S1). Higher CP content in legumes is not surprising since they depend not only on soil N, but also on symbiotic N fixation, which increases CP content (Nyamukanza and Sebata 2020). This finding suggests that V. karroo can be used as a low-cost source of protein for beef cattle feeding. Nevertheless, all LAFRs (except O. ficus-indica) met the CP requirements for a beef cow, e.g. dry period ( 8%) and late gestation (10%), but not the lactation stage (Raseigh et al. 2014).
The NDF of all LAFRs fall within ranges reported by Raseigh et al. (2014) for all physiological stages of a beef cow, except for P. australis (which had higher NDF and low in vitro NDF digestibility). These results indicated that other LAFRs (except P. australis) have potential for use as alternative feeds for beef cattle, particularly during dry periods and for gestating cows. However, the low ash content of V. karroo pods in the Cradock (6%) and Middelburg (6.5%) areas is indicative of low mineral contents (Hoffman 2005). This renders pods less important in cattle diets. By contrast, the ash content of M. sativa hay, P. australis and P. incana (Table 4) surpassed the threshold (9%) reported by Hoffman (2005), indicating that these forages might be contaminated. The wind, rain splash and floods are the main sources of soil contamination, which increase mineral content of forages (Weiss 2019).
Forage contamination reduces forage quality, thus proper timing and mode of harvesting and storage of forage are necessary. The in vitro NDF digestibility of V. karroo forages tended to be lower than other species, whereas O. ficus-indica attained higher in vitro NDF digestibility than other species at all incubation periods. This could be ascribed to higher ADF and ADL contents in the leaves and pods of V. karroo relative to cladodes of O. ficus-indica (  Means with different subscripts within a column are significantly different (p < 0.05). NS = not significant; * significant at p < 0.05; ** significant at p < 0.01; *** significant at p < 0.001 Table 4: Means (± SD) of chemical attributes of locally available forage resources (LAFRs) for cattle feeding at Cradock and Middelburg areas of the Eastern Cape, South Africa. DM = Dry matter; OM = Organic matter; CP = Crud Protein; NDF = Neutral detergent fibre; ADF = acid detergent fibre; ADL = acid detergent lignin contains high condensed tannins and phenolics that reduce the ruminal microbial populations, thereby inhibiting fermentation and degradability of ADF, NDF and ADL (Theart 2015). The high ADF and ADL contents in V. karroo leaves were likely due to inclusion in the samples of more fibrous and ligneous rachis and petioles. Thus, it is essential to derive foliage harvesting methods that will maximise harvesting of more vegetative leaflets and less fibrous and ligneous material. The O. ficus-indica cladodes had relatively low CP, more so at Middelburg than at Cradock (Table 4). The cactus CP at Middelburg is comparable to CP values reported by Mciteka (2008) for different cactus varieties. Low levels of NDF, ADF and ADL and higher total ash and in vitro ADF digestibility warrants the use of cacti as beef cattle feed, but low CP content suggests the necessity for supplementation with protein sources.

Conclusions
Integration of participatory inventory and chemical analysis was a reliable approach to identify the potential forages for cattle feeding in free-range beef farming. The LAFRs differed in terms of nutritive value, but in conjunction with the perception of farmers, grasses had low nutritional quality and browse species, specifically Acacia karroo foliage, were of high quality as indicated by the high CP and ash contents. *** *** *** Farming area × LAFRs *** * NS NS = not significant; * significant at p < 0.05; ** significant at p < 0.01; *** significant at p < 0.001. abcd Means with different subscripts within a column are significantly different (p < 0.05). The comparisons were not conducted across different incubation periods.