Exploring tropical forage options that optimize animal production and reduce methane emissions in mixed crop–livestock systems in Ethiopia

Experimental sites

Field experiments were conducted for two consecutive years (2021/2022–2022/2023) at the Experimental Farm of the School of Plant and Horticultural Sciences of Hawassa University in Hawassa (7°3.02′N, 38°30.31′E, 1712 m asl) in the south and at the Research Farm of Bahir Dar University (11°24.03′N, 37°8.45′E, 1983 m asl) in Bahir Dar in the north of Ethiopia. The Bahir Dar site is within the tepid moist mid-highlands agro-ecological zone in the northern-central highlands, whereas the Hawassa site is within the tepid subhumid mid-highlands agro-ecological zone in the southern highlands (MoA 1998; Elias 2016). Rainfall distribution at the Hawassa site is bimodal, with most (≈80%) occurring in the main rainy season (June–October) and 20% in the short rainy season (April–May). The mean annual rainfall is 961 mm. The average monthly maximum and minimum temperatures in Hawassa were 27.4°C and 12.9°C respectively (Raji and Dörsch 2020). The soil is a loam, with a pH of 6.6, organic C 1.8%, total N 0.12%, and available P 120 mg kg⁻¹ (Mehlich 3 method). The Bahir Dar site receives a unimodal type of rainfall with a mean annual rainfall of 1395 mm that falls between June and September. The average monthly maximum and minimum temperatures of the area were 29.2°C and 9.5°C respectively. The soil is a silty loam, with a pH of 5.6, organic C 1.9%, total N 0.2%, and available P 1.7 mg kg⁻¹.

Evaluation of forage production and crude protein yield

Eight improved tropical grasses and four species of forage legumes were evaluated (Table 1). Species selection was based on their performance at other sites with similar environmental conditions, and international and local experts’ opinion. Of the 12 forage crops, only desho grass (Pennisetum pedicellatum) and lupin (Lupinus angustifolius) are grown locally by farmers. Seeds were acquired from ILRI (Ethiopia). Cuttings of desho grass were obtained from smallholder farms in Central Ethiopia Regional State for the Hawassa site and the planting materials for desho and lupin were sourced from Bahir Dar University for the other site. Three replicate plots (4 × 4 m²) of each of the forage crops were laid in a randomized complete block design. All the plots were established in June 2021 and maintained for the entire study period. The plots were fertilized with 100 kg NPS (19 kg N, 38 kg P and 7 kg S ha⁻¹) at sowing. The annual legumes were resown each year during the long rain, except in 2021 at Hawassa, where they were resown during the short rain as well. The legume Stylosanthes guianensis (cv. Ubon) failed to establish at both sites and was therefore replaced by lupin in 2022. All plots were hand-weeded and received no fertilizer after establishment.

Excluding one cut conducted in the first year at the Bahir Dar site, two to four cuts per season were performed. Sampling was undertaken when most plots reached 50% flowering stage at the first cut (for both annuals and perennials), and with an interval of about 100 days between the subsequent cuts for the latter. Herbage yield was determined at final harvest for the three annual legumes lablab, lupin and sunnhemp. The harvest was done manually at a stubble height of 7 cm. All aboveground forage mass from the central 1 m² quadrats of the plots was weighed fresh, and a subsample of 1 kg was taken for each plot, and then air dried at ambient temperature to constant weight for determination of herbage dry-matter (DM) yield. Additionally, in the 2023 main rainy season, fresh subsamples of 0.5 kg were oven-dried at 65°C for 48 h, then ground through a 1 mm sieve for chemical analyses and in vitro CH₄ determination. Dry-matter yield (Mg DM ha⁻¹) was summed to annual values for the perennial crops involving multiple cuts, whereas it was averaged for the annuals involving resowing (sown in both short and long rains in the same year) of each year. These yields were used to compute mean annual forage mass production across the two years of the study. Crude protein yield (kg ha⁻¹ year⁻¹) for each species was estimated by multiplying the DM yield of each with the corresponding N content (see below) and a conversion factor of 6.25.

Chemical analyses

Plant samples were analysed for DM, ash and crude protein (6.25 × %N) according to AOAC Methods 930.15, 942.05 and 976.05 respectively (AOAC 1990), and for neutral detergent fiber (NDF), acid detergent fiber (ADF) and acid detergent lignin (ADL) according to Van Soest et al. (1991). All samples were analysed in duplicate. Chemical analyses and in vitro studies were conducted at the Animal Nutrition Laboratory of Hawassa University.

In vitro studies

The method of Menke et al. (1979) modified by Blümmel and Ørskov (1993) in that feeds were incubated in a thermostatically controlled water bath was employed to determine in vitro CH₄ production. The samples were incubated with buffered rumen fluid in calibrated glass syringes (100 mL) in two separate runs (replicates) and comprised two syringes for each replicate. All syringes were pre-warmed overnight at 40°C. In each syringe, about 200 mg of a ground sample plus 30 mL of the buffered rumen fluid were incubated at 39°C (Menke et al. 1979). The rumen fluid collected from two sheep using oral stomach tube (Muizelaar et al. 2020) fed native grass hay (given ad libitum) plus 400 g concentrate (offered in two meals daily) was strained using gauze and then mixed (1:2, v/v) with anaerobic buffer solution before use. Two blanks were included for correction of CH₄ produced from the inoculum in each run. The syringes were shaken manually every hour for the first 8 h and three times daily at later times. Gas volume readings were taken before incubation (0 h) and at the end of incubation (48 h). Total gas values were corrected for blank incubation (containing only the rumen fluid). However, due to lack of reference standard with known gas values (expected gas production) for calibration in Ethiopia, the gas production values were not compared with a ‘standard feed’. Hence, gas production results should be interpreted with that in mind.

Methane measurement

After 48 h incubation, the gas production of the feed samples was immediately recorded, and it was corrected using the blanks (see the previous section). As an alternate to gas chromatography, the CH₄ at the end of the incubation was measured using the CO₂ trap method as described in Fievez et al. (2005). After measuring the total gas production, 4 mL of NaOH (10 M) was introduced into the vessel via a connector (silicon tube) fitted between a syringe containing NaOH solution and the incubation syringe, which will cause absorption of the CO₂. Because the contribution of other gases to gas production is negligible, the remaining gas was considered to be CH₄ (Pell and Schofield 1993; Mathison et al. 1998; Fievez et al. 2005). The CH₄ production was reported in grams per kilogram of incubated DM, and as a proportion of the total gas recorded at the end of incubation. Methane production was calculated as:

The proportion of CH₄ was determined as:

Statistical analysis

ANOVA was performed on annual forage mass production and crude protein yield, nutrient concentration, and in vitro gas and CH₄ productions with forage species as the main factor (fixed effect; block treated as a random variable) by using a general linear model (GLM) of IBM SPSS Statistics (version 20). Although non-significant, blocks were kept in the model (Frey et al. 2024). Tukey’s test (at P = 0.05) was used to make multiple comparisons among species. Pearson correlation was used to detect the relationship between chemical composition and CH₄ production. Because not all forage crops were grown in both locations, the analysis was performed separately for each site. However, because we were most interested in identifying forage species or cultivars with consistent, high average production, a combined analysis of production data over years at each site was conducted. Xaraes was excluded from the analysis for the Bahir Dar site as there was no yield from any of the replicates. However, because of the limited resources available, the analysis of promising species for forage quality and CH₄ mitigation was restricted to a one-time measurement only, which we assumed a limitation in fully assessing their long-term potential. Candidates for CH₄ abatement considered were those that produced less than or equal amount (i.e. expressed as a percentage of total gas production) of CH₄ to the average CH₄ across all plants minus one standard deviation. All data were analyzed using SPSS ver. 20.

Forage mass production

Forage mass production (averaged across years) showed significant (P < 0.05) differences among the test species at both sites (Table 2). Overall, the native grass desho had the highest forage production compared with the other species tested at both sites, but the most productive forage species varied across the sites. At Hawassa, for example, the production of desho (27.30 Mg ha⁻¹) was similar (P > 0.05) to those of Basilisk and Xaraes, whereas there was even no measurable biomass obtained from Xaraes (as it failed to establish) in the Bahir Dar site. Here, the highest herbage production was obtained for desho (17.58 Mg ha⁻¹) and the lowest for lablab (3.88 Mg ha⁻¹).

Chemical composition, in vitro gas and methane production

Chemical composition, gas and CH₄ yield (g kg⁻¹ DM incubated) of forage species used in the study varied somewhat at each site. In Hawassa, the highest crude protein (CP) content was obtained for lablab and Greenleaf desmodium/sunnhemp (21.12% and 16.98%/16.53% of DM respectively (with no significant difference between Greenleaf desmodium and sunnhemp; P > 0.05, Table 3)) and the lowest for desho (4.87%) (but this value was similar (P > 0.05) to those for Basilisk, Mombasa and Cayman). The CP in two-thirds of the test species were above the recommended minimum level of 7% for ruminant feeds, below which intake is usually decreased and rumen function affected (Van Soest 1994). At Bahir Dar, the CP ranged from 4.95% to 21.22% of DM and was highest in sunnhemp. Here, the concentrations of CP in five species were above the threshold level of 7%. In Hawassa, ash content ranged from 8.5% to 14.24% of DM and was lowest in Greenleaf desmodium/sunnhemp (sunnhemp was not significantly different from Greenleaf desmodium), whereas the ash content at the Bahir Dar site ranged between 9.42% and 18.2% of DM respectively, and was lowest (P < 0.05) in sunnhemp. At Hawassa, the NDF values for most of the species (8 of 10) were below or equal to 60% and in Bahir Dar there were four species that had NDF concentrations below this threshold.

The total gas yield in Hawassa varied widely, from 712.82 to 1234.1 g kg⁻¹ DM, with the lowest being for the grass Cayman and the highest for desho (Table 3). The lowest CH₄ yield was recorded in Mulato II (143.75 g kg⁻¹ DM; this value was similar to those for Cayman, Basilisk, lablab or Greenleaf desmodium) and the highest value in desho (190.18 g kg⁻¹ DM). Methane as a proportion of the total gas yield varied among plants from 13.97 (in lablab) to 20.55% (in Cayman). Most of the other species were clustered, with a proportion CH₄ to total gas production of 15.43–17.07%. At this site, across all plants, the average CH₄ proportion (% CH₄ in total gas yield) was 16.70 (s.d. ± 2.16)%. Therefore, the promising plants (for their CH₄ production reduction potential) were those that produced CH₄ of ≤14.54%. These plants were lablab (13.97%) and Mombasa (14.90%; this value was similar to that recorded for lablab). In Bahir Dar, the lowest total gas yield was recorded for Greenleaf desmodium (648.70) and the highest value for desho (1072.09 g kg⁻¹ DM). Both CH₄ yield (P ≥ 0.695) and the proportion of CH₄ emission to total gas production did not differ among test species at this site (P ≥ 0.302; Table 3). At this site, across all plants, the average amount of CH₄ produced as a percentage of total gas production was 21.99 (s.d. ± 6.12)%.

Crude protein yield

The CP yield (kg ha⁻¹ year⁻¹) showed variations among species in both sites, but it was overall higher for the legumes than grasses (Fig. 1). In Hawassa, the mean CP yield ranged from 177.70 to 1197 kg ha⁻¹ year⁻¹ and was highest in lablab, whereas the CP yield at the Bahir Dar site ranged between 188 to 1300 kg ha⁻¹ year⁻¹ and was highest in sunnhemp.

Fig. 1.

Mean crude protein yield (kg ha⁻¹ year⁻¹). Bars respresent standard error of the mean. Species/cultivars with different or no letters within a site are significantly different (Tukey’s test, P < 0.05). GD, Greenleaf desmodium.

Relationships among forage mass production, chemical composition, and in vitro gas and methane production

In Hawassa, ash was negatively correlated with CP content (r = −0.67; P < 0.01) and forage mass production (r = −0.66; P < 0.01), and positively correlated with NDF content (r = 0.47; P < 0.05) (Table 4). The CP content had a strong negative relationship with NDF (r = −0.87; P < 0.01), but was positively related to forage mass production (r = 0.63; P < 0.01). Only NDF was correlated (r = 0.54; P < 0.05) with CH₄ yield, and that was a positive correlation. Total gas yield was positively correlated with CH₄ yield (r = 0.73; P < 0.01), but negatively correlated with percentage of CH₄ to total gas yield (r = −0.87; P < 0.01). In Bahir Dar, the ash content was negatively correlated with CP (r = −0.85; P < 0.01), and positively correlated with NDF content (r = 0.80; P < 0.01) but not with forage mass production. The correlation between CP and NDF content was also negative (r = −0.90; P < 0.01). Although the correlation between total gas and CH₄ yield was negative (r = −0.77; P < 0.01), that of CH₄ to total gas ratio was positive (r = 0.90; P < 0.01).

Forage production

This study was designed to evaluate the efficacy of selected improved tropical grasses and legumes for increasing forage mass production in the Ethiopian highlands. Regardless of observable numeric differences (Table 2) between the two sites (although these differences were not statistically tested, see Statistical analysis), most of the forage crops (except desho planted with cuttings, which often leads to higher initial dry-matter production than when planted with seeds) evaluated in this study demonstrated faster crop establishment and early plant growth and vigor as reflected by early biomass production (data not shown), suggesting that most of the improved materials are likely to be suitable for the study regions. Overall, averaged over the 2 years, the native grass desho performed well in terms of forage mass production (27 Mg ha⁻¹ at Hawassa; ≈18 Mg ha⁻¹ at Bahir Dar), but the annual legumes lupin at Hawassa (≈4 Mg ha⁻¹) and lablab at Bahir Dar (2.1 Mg ha⁻¹) performed the least. These wide variations in herbage yields, especially among the grass species, imply that the species under study exhibit different adaptation potentials to the study area conditions. In addition, the variability observed in top-performing forage crops across the sites highlighted that yields could be improved with location-specific species/cultivar selection. The significantly higher forage mass of the grasses desho, Basilisk and Xaraes indicated that they are suitable for enhancing forage production in the highland mixed-farming systems of southern Ethiopia, whereas the native desho is likely to be suitable for the highland mixed systems in the north. It is also important to note that these species can grow for many years (CIAT 2004; Feedipedia 2024), and even under rain-fed condition can provide several harvests per year (we note that seven to eight growth cycles for grasses were considered for data collection in this study), which would likely lead to added production benefits for risk-averse, resources-poor farmers and therefore may increase the desirability of these grasses. Although the value of early growth is a key in influencing on-farm adoption of improved forages (Philp et al. 2019; Balehegn et al. 2020), long-term productivity is essential for the success of forage crops (Vanlauwe et al. 2014; Garcia et al. 2019). Forage mass yields of most of the tested crops were generally within the expected range reported for tropical regions (Leta et al. 2013; Nguku 2015; Feedipedia 2024). However, there is little information on the B. hybrid (Cayman) with which to compare our data. The relatively low forage mass production of the crops, with even Xaraes not establishing as expected, at the Bahir Dar site is possibly due to inadequate rainfall conditions (data not shown) during establishment of the experiment and low pH and P status of the soil prevailing at the site. Poor crop establishment, especially in the early phase of establishment, under unpredictable rainfall or limited soil moisture and poor soil conditions are major obstacles to obtaining reasonable yields in many farming systems in sub-Saharan Africa (Reynolds et al. 2015). The poor establishment of Xaraes at the Bahir Dar site suggests that the cultivar may not have been well suited to the environment.

Relationship between chemical composition and methane production

The nature and amount of cell wall components are known to influence ruminal degradability and, hence, nutritive value of forages (Van Soest 1994). Even though most of the species (8 of the total examined) possessed reasonably low NDF contents (≤60%), the significant positive correlation of this fiber with CH₄ yield (r = 0.54) from forages grown in Hawassa site corresponds to the commonly expected positive association between NDF and in vitro organic matter digestibility in forage plants (Jayanegara et al. 2012; Chen et al. 2016) and, consequently, CH₄ production. The positive correlation of NDF with CH₄ production is likely to be the result of a higher digestibility and passage rate of the NDF (Beauchemin et al. 2020) present in the forage plants tested here. Forages with a higher digestibility are known to cause more intensive ruminal fermentation and thereby increase in CH₄ yields (Beauchemin et al. 2020; Weiby et al. 2022) because more H₂ derived from feed will be used to create CH₄.

The lack of a significant correlation between other chemical constituents (even the relation of NDF to CH₄ production was non-significant in forages grown in Bahir Dar) and CH₄ yield in this and other studies (Banik et al. 2013; Berhanu et al. 2019; Ansah et al. 2021) highlighted the importance of alternative factors to methanogenesis in vitro. For example, factors affecting the extent or duration of digestion in the rumen (i.e. residence time), such as the degree of lignification of plant cell walls or the concentration of condensed tannins and other secondary metabolites in the forages, might have more important effects on CH₄ production than nutrient composition (Archimède et al. 2011; Beauchemin et al. 2020).

Forage quality and candidate forages for methane abatement

Besides herbage yield, the quality of forage is an essential element to consider on selection of forage varieties for livestock production. The NDF and CP contents are among the commonly used nutritive metrics for ruminant feeds (Van Soest et al. 1991; Waghorn and Clark 2004). The level of NDF in tropical forages beyond which DM intake in ruminants, cattle for example, would be negatively affected is 60% (Meissner et al. 1991), suggesting that 8 of 10 of the forages analysed from the Hawassa site and four of nine from the Bahir Dar site have acceptable NDF values. Crude protein serves as a reliable measure of overall nutritional value, and forages having a CP of 7% or more are considered to be of adequate quality. In this study, a CP value above this threshold was observed in eight (from both sites) species (Table 3). Although CP and NDF contents are the commonly used measures to assess feed quality, it is the CP yield that describes the overall and actual productivity of quality forage (Abera et al. 2022; Iliev et al. 2022). As such, the legumes lablab, Greenleaf desmodium or sunnhemp and the grass Xaraes at the Hawassa site and sunnhemp and lablab or Greenleaf desmodium in Bahir Dar may provide both high protein production per hectare per year and high protein content in biomass. These forage plants also contained lower concentration of NDF. Feeding these plants as supplements could thus enable improved utilization of protein-deficient native forages. Moreover, the considerable amounts of CP in these forage species can also have a suppressing effect on CH₄ production in ruminants. However, only the legumes lablab, Greenleaf desmodium and sunnhemp had CP concentration higher than the 15% required for optimal animal growth (Poppi and McLennan 1995). The NDF and CP values of the forages are within the ranges reported in the literature (Jayasinghe et al. 2022; Feedipedia 2024).

The difference observed in CH₄ production and its proportion to total gas production among species might be attributed to genetic variation among species but also plant phenology, which affects chemical composition (Bezabih et al. 2014; Gemeda and Hassen 2014; Lee 2018). There is little information on CH₄ and gas yield for many introduced (or improved) tropical forage grasses and legumes under tropical African conditions, with which to compare our data. We assume that the higher values in our report than those frequently found in literature could partially be attributed to relatively higher digestibility or passage rate of the NDF (Beauchemin et al. 2020) (see the previous section), and relatively lower CP contents recorded in grasses (Table 3). In contrast, the higher CH₄ production (per unit of incubated DM) for legumes (e.g. sunnhemp) in the current study is possibly linked to higher amounts of rapidly fermentable carbohydrates (Chaves et al. 2006; Pirondini et al. 2012); however, we did not measure the concentrations of these carbohydrate fractions in our analysis. According to Beauchemin et al. (2020), better-quality forages including legumes often contain a higher proportion of non-fibre carbohydrates relative to NDF, which promotes organic matter digestibility and intense fermentation in the rumen and thereby increase in CH₄ production (Pirondini et al. 2012). In an earlier study, Chaves et al. (2006) found a higher CH₄ production (g kg⁻¹ DM) for forage legumes, although they possessed low concentrations of NDF. The CH₄ proportion of total gas production of tested species except Greenleaf desmodium was lower than a value normally expected (i.e. 27%) from forages fermented in the rumen (Mathison et al. 1998).

The ratio or percentage of CH₄ to total gas production has been widely used as an important indicator rather than absolute CH₄ amount to screening of plants for low CH₄ production (Jayanegara et al. 2011; Bezabih et al. 2014; Melesse et al. 2019) because it is correlated with total gas production (microbial fermentation) (Bhatta et al. 2012) and because it can also express feed efficiency (Van Soest 1994; Gemeda and Hassen 2014). A lower percentage indicates reduced methanogenic potential of digestible feed components, resulting in lower CH₄ production per unit net gas volume output (Bhatta et al. 2012; Bekele et al. 2024). In this study, plants that possessed a percentage of CH₄ to total gas production of ≤14.54% were considered promising plants (for their potential to reducing CH₄ emissions in vitro) for the Hawassa site, and those with ≤15.86% being considered promising for the Bahir Dar site. These plants were lablab and Mombasa for the former site and desho for the latter, which can be desirable candidates for further in vivo trials. However, the CH₄ suppressing effect of these candidate species may be more pronounced if they are used as supplements rather than as sole-feed diets.