Selection and evaluation of promising indigenous fodder trees and shrubs as supplemental diets for ruminant animals across different agroecological environments

The aim of this study was to select and evaluate promising potential indigenous fodder trees and shrubs (IFTS) as supplemental feeds for ruminant animals. Through interviews with farmers and field inventories, 107 IFTS species were identified as ruminant feeds, from which 37 potential IFTS species were selected. The chemical composition and anti-nutritional factors of the selected IFTS were analysed. Among the 37 selected IFTS species, crude protein (CP) ranged from 83.3 to 230.5 (mean, 134.3) g kg−1 dry matter (DM); CP was greater than 80 g kg−1 DM in all selected species. Condensed tannin (CT) content varied from 2.57 to 210.91 g kg−1 (p < 0.05) among the selected species, of which 72% had < 40 g kg−1 DM, an amount that is tolerable for ruminants. Farmers’ preferences and the nutritive value of the selected fodder species were positively correlated (p < 0.05) for CP (0.61), organic matter digestibility (0.303) and metabolisable energy (0.25). Farmers’ preference ranking and the nutritive value of the plants were considered to select the top 14 IFTS, all of which had CP > 110 g kg−1, neutral detergent fibre (NDF) < 350 g kg−1, and CT < 40 g kg−1. Thus, the selected potential IFTS need further evaluation for their agronomic practice, biomass production and animal performance.


Introduction
Livestock plays a key economic role in African agriculture, and is increasingly viewed as an important means for rural households to escape poverty.Ruminant livestock production in the tropics is limited due to insufficient dietary nitrogen (N) and high content of indigestible fibre from low-quality forages (Selemani et al. 2013).Fodder trees and shrubs offer considerable potential nutrient N to complement the low feeding value of crop residues and natural pastures in which the levels of crude protein (CP) are below the minimal requirement (i.e.80 g kg −1 dry matter) for optimal rumen microbial function (Girma et al. 2015;Melesse et al. 2017).In addition, fodder trees and shrubs are an important component of ruminant livestock diets in diverse farming systems in tropical regions (Babayemi and Bamikole 2006;Derero and Kitaw 2018).
In tropical countries like Ethiopia, the adoption of exotic fodder species by smallholder farmers is limited (Mekoya et al. 2008), thus resulting in the very low contribution of these species to livestock feed (Birhanu et al. 2017).This low contribution consequently encourages expanded use of indigenous fodder species to exploit their advantages in terms of adaptability to the local environment, availability of local planting material, and familiarity to the farmers (Asmare and Mekuriaw, 2019).Owing to a recent policy shift in response to overgrazing, the Ethiopian government has prohibited livestock from entering communal grazing and bush land, and has enclosed these areas in order to rehabilitate and restore the ecosystem (Amare et al. 2017).
A high focus of the rehabilitation plan for these enclosed areas in Ethiopia is the planting of various fodder tree and shrub species, particularly those native to the tropical environment (Mulatu and Kassa 2001;Tefera and Sterk 2010).In these cases, farmers are encouraged to enter the enclosed areas to harvest fodder trees and shrubs and carry them to their animals (Mekuriaw and Aynekulu 2013;Mekuriaw et al. 2017;Shimelse et al. 2019), Consequently, the use of indigenous fodder species as a supplemental livestock feed is becoming increasingly important (Gwaze et al. 2009).This practice benefits smallholder farmers in developing countries, who typically cannot afford industrially manufactured concentrate feeds and thus depend almost entirely on locally available fodder species for supplemental feeding of their animals (Olafadehan and Okunade 2018).For the reasons just stated and because of their multipurpose usefulness, indigenous fodder trees and shrubs (IFTS) are now receiving increased research attention in tropical regions (Murgueitio et al. 2011), and particularly in Ethiopia (Balehegn et al. 2014;Derero and Kitaw 2018).Moreover, recent changes in climate could influence conventional forage production and thus locally available feed ingredients such as indigenous fodders, trees and shrubs could become sustainable supplemental alternative feeds in the dry season (Idamokoro et al. 2016).In addition, these plant materials contain bioactive compounds, such as tannins or other polyphenols, that if not present in very high concentration could improve gut health and thus overall animal health (Huang et al. 2018).
The plant secondary metabolites can also modulate the rumen microbiome to improve protein and fibre degradation that reduce enteric methane emissions in ruminants (Ku-Vera et al. 2020).Thus, the identification and nutritional evaluation of potential quality fodder species is crucial for developing strategic feeding supplementation for sustainable ruminant animal production (Habib et al. 2016;Santos et al. 2017).The subsequent use of fodder species, selected on the basis of farmers' knowledge and scientific findings, would be environmentally friendly, socially acceptable and economically sustainable for increasing ruminant livestock production (Khanal and Subba 2001).Despite the increasing use of indigenous tree and shrub species in Ethiopia, only a few studies have focused on indigenous fodder species (Haile and Tolemariam 2008;Shenkute et al. 2012;Derero and Kitaw 2018), and most of these resources remain unexplored (Derero and Kitaw 2018).In particular, many indigenous fodder plants have been undervalued due to insufficient knowledge about their potential feeding value (Mokoboki et al. 2019).This situation suggests the need to characterise IFTS to sufficiently understand their potential as supplemental feed for ruminants (Assouma et al. 2018).
To this end, we initiated the current study to address the following specific objectives: (1) to identify, prioritise and select IFTS species for feeding ruminant animals; (2) to evaluate the nutritional quality such as chemical composition, anti-nutritional factors, in vitro digestibility and volatile fatty acid production of selected IFTS; and (3) to investigate the relationship between farmers' preference criteria and ranking score with nutritive value in the selection of IFTS.The data obtained regarding these characteristics of IFTS will facilitate their strategic use for supplemental feeding of ruminant animals in Ethiopia and other tropical areas.

Study sites
The study was conducted at three sites, Dibatie, Aba Gerima and Guder, in the tropical Upper Blue Nile basin of Ethiopia (Figure 1).Each site is characterised by specific climate, soil and topographic conditions (Ebabu et al. 2019).The selected sites are livestock producing areas representing three distinct agro-ecologies (Table 1).

Identification of IFTS and farmers' preferences regarding them
The IFTS were identified based on community knowledge gathered through questionnaire surveys, group discussions, and one-on-one interviews across the study sites.A total of 90 farmers (30 from each site) were selected based on their existing experience and knowledge regarding the utilisation of IFTS species as ruminant animal feed.A questionnaire was developed after some initial discussion prior to the actual survey, conducted in 2018.Questions related to the types and quality of plants available, and to farmers' preferences.From the questionnaire survey results, we created a list of available IFTS.In addition, field inventories were performed to identify potential fodder species at each site.Criteria for selection of preferred fodder species were used in response to farmers' stated preferences regarding the important qualities of fodders of choice.It was found that the criteria were almost universal and were similarly identified in other studies in other places (Mekoya et al. 2008;Balehegn et al. 2014).They included characteristics such as: (1) feeding value, palatability and ability to support the animals; (2) year-round fodder availability, in both rainy and dry seasons; (3) local abundance; (4) biomass production; and (5) multipurpose use (Figure 2).Herders were also asked to rank each fodder species according to their preferences (1 = least favoured; 4 = most favoured).For this research there is no established system for ethical approval; however, the protocol for conducting the survey and the focus group discussions with farmers were approved by Bahir Dar University's authorised division and the Offices of Agriculture in the districts of the three study areas.
Initially, a total of 129 indigenous trees and shrubs were collected in both rainy and dry seasons (Supplementary Table S1), and their seasonal availability was recorded over two consecutive years (2018 and 2019).To understand the selected IFTS distribution patterns, a relative abundance vegetation survey was conducted.For this, a total of 105 quadrats were assessed across the three study sites.The quadrat sizes were determined using the minimal area method, 10 × 10 m for trees and 5 × 5 m for shrubs, according to Fisaha et al. (2013).The total numbers of indigenous tree and shrub species in the sampling quadrats were recorded to determine the relative abundance of each species, which was calculated according to Tadesse et al. (2019): Relative abundance = Number of individuals of species × 100% Total number of individuals Specimens of all plant species were transported to the Botany Department of Bahir Dar University for identification at the species level; remaining unknown species went to the National Herbarium at Addis Ababa University for identification.Any specimens still unidentified after evaluation at the National Herbarium were labelled as 'unknown species' (Supplementary Table S1).
To screen the palatability of plant species as potential IFTS (Figure 2), we used four categories of fodder palatability for ruminant livestock: unpalatable, low palatability, moderate palatability and high palatability, according to Hussain and Durrani (2009).The differences in palatability among fodder species was confirmed through knowledge gathered from local herders and field observation.For the palatability of IFTS, the cattle and small ruminant species were considered.Furthermore, to select the top quality IFTS as potential fodder species at each site, we considered the following three parameters: (1) farmers' preference ranking score; (2) chemical composition (CP > 110 g kg −1 DM); and (3) anti-nutritional content (CT < 40 g kg −1 ) (Figure 2).

Nutritive value analysis
After we screened IFTS and excluded the fodder plants that did not meet the farmers' preferences, and those that had low palatability according to the key informants, 37 potential fodder species underwent further nutritional evaluation.A total of 130 samples of IFTS plant parts edible by the animals were collected for chemical composition analysis from between three and five phonologically similar individuals of each fodder species and pooled to form representative samples.Samples were dried at 105 °C for 24 hr to a constant weight in a forced-air oven, ground in a hammer mill to pass through a 1 mm sieve, and stored in plastic bags at Shimane and Tottori universities in Japan for subsequent determination of nutritive value.Dry matter content was determined by drying feeds at 135 °C for 2 hr (AOAC 2005), and ash was determined by burning dried samples in a muffle furnace set at 550 °C for 3 hr (AOAC 1990).The nitrogen content of the feed was analysed by using a CN coder analyser (Shimadizu, Kyoto, Japan) and CP was calculated as N content × 6.25.The neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin (ADL) contents, which comprised the cell wall components in these samples, were analysed according to Van (1991).Gross energy (GE) of the feed was determined by using an adiabatic bomb calorimeter (model CA-4AJ, Shimadzu) (Santos et al. 2017).To determine the CT content, the sample extract was treated with butanol-HCl in the presence of ferric ammonium sulphate, and CT was expressed in equivalents as absorbance at 550 nm in a spectrophotometer (Shimadzu) (Porter et al. 1985).

In vitro digestibility and fermentation characteristics
The in vitro digestibility was conducted using a glass syringe method and two incubation runs per sample were performed.The amount of substrate in the bottle was 200 mg of each sample.Rumen fluid for evaluation of in vitro digestibility was obtained from two healthy mature female Japanese Corriedale sheep fitted with permanent

Discard
No No

Discard
Step 1: Field inventory and key informant discussion Step 2: Is the IFTS palatable to ruminant livestock?
Step 3: Is the IFTS available year-round (wet and dry season)?
Step 4: Do farmers prefer the IFTS for animal feed?
Recording list of available indigenous trees and shrubs Recording list of available indigenous trees and shrubs

Yes Yes
Yes Yes

Yes Yes
Yes Yes 46 mm rumen cannulas that were fed a daily ration of 800 g timothy hay and 200 g concentrates, divided into two equal meals at 08h00 and 16h00.The sheep were supplemented with minerals and had free access to water.We used duplicate analysis per treatment for incubation.Two incubation runs per sample were performed, totalling 72 runs.The amount of substrate in the bottle was 200 g of each sample.With respect to the volume of fermentation media, rumen fluid was obtained in the morning three hours after the morning feed, flushed with CO 2 , filtered through three layers of cheesecloth, and mixed (1:4, v/v) with an anaerobic mineral buffer solution as described by Makkar et al. (2000).An in vitro mineral buffer media for gas testing was prepared according to Menke and Steingass (1988).A reduced buffer solution was prepared as described by (Mekuriaw et al. 2019).Gas production was recorded before incubation (0 hrs) and after 3, 6, 12, 24, 48, 72 and 96 hrs of incubation.The blank incubation (0 h) gas production value was subtracted from each of those after incubation.The gas production characteristics were estimated by fitting the mean gas volumes to the exponential equation:

Dibatie Aba Gerima Guder
where G is the gas production (ml 0.2g −1 DM) at time t, a is the gas production from the immediately soluble fraction, b is the gas production from the insoluble but degradable fraction, and c is the rate constant of gas production (h −1 ) (Ørskov and McDonald, 1979).
The in vitro gas production data were fitted by using the 'Neway' curve-fitting program (Macaulay Land Use Research Institute 2004, cited in Ichinohe and Fujihara 2009) to estimate the rumen degradation parameters according to the model of McDonald (1981).Rumen degradability of fodder plant species was estimated by using the following equation (Kamalak et al. 2005): where EDDM is effective degradability of dry matter, a is the gas production from the immediately soluble fraction, b is the gas production from the insoluble fraction, (a + b) is the total potential gas production, c is the gas production rate, R 2 is the coefficient of determination, and RSD is relative standard deviation.
For analysis of volatile fatty acids (VFA), 5 ml of the supernatant in the syringe was collected at the end of the 96 h gas production incubation and centrifuged at 25 000 × g for 15 min at 4 °C.The supernatant was transferred to a 15 ml micro-centrifuge tube (capacity) and stored at −20 °C until analysis.The content of crotonic acid (internal standard) is 10 mmol l −1 , which is the same as the content of external standard.The external standard solution comprised a mixture of acetic, propionic, n-butyric, isobutyric and crotonic acids, and was treated in the same manner as the sample.The VFA contents of test samples were analysed using gas chromatography (model GC 14A, Shimadzu).The VFA content was decided by comparing the proportion of peak area for target VFA to that for crotonic acid between the samples and the external standard solution as described in Mekuriaw et al. (2019).

Statistical analysis
Due to the differences in the types of IFTS available at each site, means were calculated, and all data were analysed using a general linear model procedure (SAS 2001) with a completely randomised design within site.The normality of data distribution was tested with the Shapiro-Wilk test.When normality could not be assumed, data were log-transformed.The data were analysed separately by location.The experimental design was completely randomised and the treatments are individual IFTS.The effect of IFTS species type on nutrient composition was analysed using the model: where Y ij is the dependent variable of chemical composition parameters, μ is the overall mean, IFTS i the fixed effect of IFTS species (1-15 for each site), and e ij the random error.When the F-test was significant (p < 0.05) Tukey's test was used to compare significant differences (p < 0.05) among the nutritive values of IFTS species.Samples of IFTS species were collected during both the wet and dry seasons and analysed as pooled data.Pearson correlation coefficients between chemical composition and farmers' preference ranking of IFTS were determined using SPSS (SPSS 2016).A regression analysis was performed to test the complementarity between farmers' preference ranking scores and nutritive values of IFTS, such as CP (Figure 2) (Mekoya et al. 2008).The CP content was defined as the fodder plant nutritional quality parameter to relate to the farmers' ranking score to nutritional composition (Nunes et al. 2016).The preference ranking scores were treated as quantities measured on a continuous scale (Kuntashula and Mafongoya 2005).Cross tabulations were used to identify the proportion of respondents for each IFTS ranking score.
The correlation and complementarity between farmers' preference and the nutritive value of Dodonaea angustifolia L.f. was performed using PCA (principal component analyses).The PCA was computed in R Studio using the autopilot function of ggplot2 (R Core Team 2022).

Identification of and farmers' preferences regarding indigenous fodder species
Overall, 129 indigenous trees and shrubs were identified across the three study sites (Table S1).According to the farmers' interviews, 22 of the indigenous trees and shrubs were unpalatable to ruminant livestock and 40 were poorly or rarely palatable, so they were excluded from further consideration.From the 67 fodder species that were moderately or highly palatable, 37 potential indigenous fodder species were selected in line with the ranking order of local farmers' preferences and their use as feed for ruminant livestock (Table 2).In the final stage of the selection procedure (Step 5 in Figure 2), which was based on farmers' preference-ranking scores and nutritive value (CP > 110 g kg −1 , NDF < 350 g kg −1 , CT < 40 g kg −1 ), the top five IFTS at each site were selected as supplementary feeds for ruminant livestock: Albizia schimperiana Oliv.,

Relative abundance of selected IFTS
The distribution of the selected potential fodder species revealed variations in abundance even within a site (
The NDF, ADF and ADL values of the selected IFTS ranged from 209 g kg −1 to 513.2 g kg −1 DM, 145.3 to 450.2 g kg −1 DM, and 42.5 to 367.7 g kg −1 DM, respectively.The lowest fibre content was recorded in C. aurea in Aba Gerima whereas the highest was in O. abyssinica in Dibatie.In addition, the ash content of IFTS varied significantly (p < 0.001) among species, from 48.2 g kg −1 (C.africana) to 186.0 g kg −1 (Dombeya torrida(J.F.Gmel.)Bamps).Furthermore, although the CT content differed significantly (p < 0.05) among species from 2.5 g kg −1 (D. penninervium) to 210.9 g kg −1 (C.macrostachyus), 72% of the selected IFTS had less than 40 g kg −1 DM of CT which is within the accepted value for ruminant animal feed.
The highest content of GE was recorded in C. macrostachyus (19.6 MJ kg −1 DM), and 65% of the selected IFTS had GE contents that were greater than the mean value (15.6 MJ kg −1 ).The largest and smallest values for in vitro OMD were observed for M. lanceolata (83.9%) in Guder and C. tomentosa (21.6%) in Aba Gerima, respectively.The ME values of D. penninervium (10 MJ kg −1 ), M. lanceolata (12.8 MJ kg −1 ), and U. hypselodendron (10.8 MJ kg −1 ) in Guder; F. sur (10.2 MJ kg −1 ) in Dibatie; and Vernonia amygdalina Delile (10.6 MJ kg −1 ) in Aba Gerima were higher (p < 0.05) than the mean value (6.01 MJ kg −1 ) for all selected IFTS.The effective degradability of dry matter (EDDM) ranged from 27.6% to 56.9% in the selected IFTS.The content of total VFA ranged from 72.1 mmol L −1 to 161.6 mmol L −1 and was greater than the mean value (124.73 mmol L −1 ) in 68% of the selected IFTS.

Principal component analysis (PCA)
Principal component analysis (PCA) provides an overview of the similarities and differences between the measured properties and was used in this study to visualise the variations in chemical components and nutritional properties of the indigenous fodder species (Figure 3).The results show that the proportions of eigenvalues of the correlation matrix for the first two principal components were 23.43% and 16.83%.This supports the result for the correlation and complementarity between IFST nutritive values.

Correlation and complementarity between farmers' preference ranking and plant nutritive value
Complementarities between the farmers' preference ranking score and nutritive values (specifically CP) of the 37 selected IFTS are shown in Figure 4. Farmers in Dibatie and Guder used their preferences and acquired knowledge to effectively discriminate potential quality IFTS species (with the exception of Z. spina-christi in Dibatie) (Figure 4).Farmers' preference-ranking scores were positively correlated with CP, OMD, ME and TVFA (Table 4).
In contrast, in Aba Gerima, farmers' preference ranking was unrelated to CP and could not be used to discriminate potential individual IFTS.The combined regression equation for the relationship of farmers ranking score (Y) and CP (X) was: Y = 0.0169x + 0.314 (R² = 0.58) (Figure 5).

Discussion
In this study, the results of the interviews with local farmers and the field inventory indicated the availability of diverse indigenous fodder resources at all three study sites, and the number of species used as fodder is remarkable.Depending on plant availability and the preferences of particular animal species, most (83%) of the 129 fodder plant species were palatable to ruminant livestock (Table S1).Similarly, Roothaert and Franzel (2001) reported that farmers in three zones of Kenya used more than 90% of the 160 IFTS evaluated for feeding their livestock.Some of the selected IFST in the current study were previously documented as ruminant livestock feed in various parts of Ethiopia, for example, A. abyssinica as the most commonly utilised fodder species in the Rift Valley of Ethiopia (Gedda, 2003); V. amygdalina, Buddleja polystachya Fresen.and M. lanceolata in southern Ethiopia (Haile and Tolemariam, 2008); and Y. alpina (Mekuriaw et al. 2011) and Ficus thonningii Blume (Asmare and Mekuriaw 2019) in northern Ethiopia.In contrast, several species were previously unregistered and were recognised as ruminant livestock feed in the current study.Atypically, a few of the species identified as fodder are toxic to livestock when consumed in large quantities, such as the leaves of the Prunus africana (Hook.f.) Kalkman trees mentioned by farmers in Guder adversely affect sheep and cattle, potentially leading to death (personal communication and discussion with key informants).In support of this finding, Abera et al. (2015) reported that P. africana was one of the most frequently ingested toxic plants among livestock in the Horro Gudurru Wollega Zone, western Ethiopia.This issue of toxicity needs further detailed investigation so that farmers can minimise adverse effects of IFTS on ruminant livestock (Kiptot, 2007).
The different relative abundances of the 37 selected IFTS species between and within study sites might be explained by local management practices, characteristics of the plants' reproductive mechanisms and their ability to regenerate (Aregawi et al. 2008).The high relative abundance of most of the preferred species indicates that these species are common in the study sites.However, the sparse distribution of preferred fodder species that show low relative abundance, such as Z. spina-christi in Dibatie, calls for increased attention to multiplying these species, to make them more available for ruminant livestock feed.
In our current study, the chemical composition of the selected IFTS varied widely, indicating fodder quality differs remarkably among species (Pal et al. 2015).The nutritive values of the IFTS species in the current study ranged within those reported previously for native fodder species in Ethiopia and tropical regions (Tolera et al. 1997;Mengistu et al. 2017;Derero and Kitaw 2018).In the current study, the nutritive value of the selected IFTS such as crude protein was higher for A. abyssinica, C. africana, M. lanceolata and V. amygdalina than reported by Haile and Tolemariam (2008) and Ndagurwa and Dube (2013).Further, the CP contents of these IFTS species are comparable to those of the much-publicised accessions of exotic species, such as Leucaena and Sesbania (Melaku et al. 2010, Melesse et al. 2017).This similarity implies that the nutritive quality, in terms of CP, of the 37 selected IFTS species that we evaluated could provide from 83.3 to 230.5 g kg −1 DM, which is sufficient to supply adequate protein for maintenance (80 g kg −1 DM), moderate growth (113 g kg −1 DM), and milk yield (120 g kg −1 DM for lactating dairy cows) in ruminant animals (NRC 2001).Therefore, the 37 selected IFTS could serve as potential protein supplements that are comparable in quality to improved fodder species, to enhance the utilisation ) EDDM = effective degradability of dry matter (%) ME = metabolisable energy (MJ kg −1 ) ADL = acid detergent lignin (g kg −1 ) GE = gross energy (MJ kg −1 ) NDF = neutral detergent fibre (g kg −1 ) CP = crude protein (g kg −1 DM) IVOMD = in vitro organic matter digestibility (%) TVFA = total volatile fatty acid (mmol l −1 ) CT = condensed tannins (g kg −1 ) SEM = standard error of the mean.Means followed by the same letter in the column do not differ (p > 0.05) statistically by the Tukey test of IFTS species within each site of low-quality feed resources such as fibrous crop residues for ruminant diets in tropical regions (Habib et al. 2016).In support of our results, Santos et al. (2017) also reported that indigenous fodder plants, in addition to their adaptation to the local environment, have greater potential nitrogen supply for animal metabolism than tropical grasses.In addition to their high CP content, the 37 IFTS species selected in this study may be advantageous to rumen microbes because most of these plants have moderate tannin content, which can be used as a dietary source of nitrogen (Mokoboki et al. 2019).The variations in the chemical composition of the fodder trees and shrubs in the present study, compared The group refers to the location-based classification of the fodder species: colours represent Dibatie (blue), Aba Gerima (red) and Guder (green), using the following abbreviations: with the works reported earlier, could be due to differences in their ability to accumulate proteins at the stage of their leaf sampling; growth potential of the plants and possible differences in the amounts of nutrients in the soil; season of harvest and harvesting stages between sites (Geta et al. 2014;Derero and Kitaw 2018).
The NDF, ADF and ADL contents that we measured seem to concur with values reported for other native fodder   4: Complementarities between farmers' preference and crude protein (CP) content of selected fodder species in Dibatie, Aba Gerima and Guder species (Mengistu et al. 2017;Melesse et al. 2017).In our study, the high NDF value for O. abyssinica in Dibatie may reflect differences in the stage of maturity of the plants at harvest (Ndagurwa and Dube 2013).As reported in Table 3, there are no differences among species for NDF, ADF and ADL at any of the locations that certain species (such as C. aurea) which had the lowest fibre content when the statistical analysis reflects no effect.In addition, the CT content of 65% of the selected IFTS were below the level (40 g kg −1 DM) suggested to be acceptable in the diets of ruminant animals.A further usefulness of these shrubs or trees is that they contain bioactive compounds such as moderate tannin or other polyphenols, which, if not present in very high concentration, could improve gut health and thus overall animal health and hence performance (Alonso-Díaz et al. 2010).However, the effect of CT on protein degradation also depends on the composition of the tannin (Yayneshet et al. 2009).
The in vitro OMD values for some of the fodder species that we evaluated were within ranges reported for other multipurpose tree species in northern Ethiopia (Melaku et al. 2004).The ME values of D. penninervium, M. lanceolata, U. hypselodendron, F. sur and V. amygdalina are higher than those in other selected IFTS that we studied and may be due to the high digestibility of their cell wall carbohydrates, as indicated by their very low ADL contents.In addition, the GE of 67% of the 37 selected IFTS species was greater than the mean value (15.6 MJ kg −1 ), thus indicating that they can produce relatively high volumes of in vitro gas and suggesting that their leaves are highly nutritious for ruminant animals (Melesse et al. 2017).Furthermore, the total VFA contents that we obtained are in line with other reports involving tropical fodder species, and the high average value (124.7 mmol L −1 ) in our study indicates better substrate utilisation by ruminal microbiota for the support of ruminant livestock (Gemeda andHassen 2014, Mengistu et al. 2017).
In our study, farmers' preferences regarding fodders were good indicators of the relative quality of IFTS for ruminant livestock, even though the farmers' preference criteria for multipurpose fodder species are complex (Ayenew et al. 2021).This result is in agreement with those of Roothaert and Franzel (2001) in Kenya and Mekoya et al. (2008) in Ethiopia, who likewise reported that farmers' appreciation of fodder species reflected their nutritive value.By extension, the complementarity pairwise comparisons in our study showed that farmers effectively discriminated between IFTS species with low protein content as compared with high content.This finding is in line with Mekoya ) EDDM = effective degradability of dry matter (%) NDF = neutral detergent fibre (g kg −1 ) ADL = acid detergent lignin (g kg −1 ) GE = gross energy (MJ kg −1 ) OMD = organic matter digestibility (%) CP = crude protein (g kg −1 DM) ME = metabolisable energy (MJ kg −1 ) TVFA = total volatile fatty acid (mmol l −1 ) CT = condensed tannins (g kg  (2008) and Talore (2015) who reported significant complementarity between farmers' assessment of fodder quality and laboratory data.strong relationship between farmers' preferences and the nutritive value (particularly for CP, OMD, ME and VFA) of IFTS (except in Aba Gerima) confirms that farmers, through their accumulated experiential knowledge, effectively select good-quality fodder species for supplementing their ruminant livestock.Moreover, in this study, new selected IFTS used for animal feed (which have not been reported in earlier studies) are reported, as summarised in Supplementary Table S2

Conclusions
In the current study, we identified 107 IFTS among three sites in Ethiopia, of which 37 species were considered to be good-quality fodder in light of farmers' preferences and CP values exceeding 80 g kg −1 .Moreover, using nutritive values and considering farmers' preference rankings, we selected the top 5 IFTS species for each location; A. schimperiana, C. acrostachyus, D. abyssinica, F. sur and F. vasta for Dibatie; A. abyssinica, D. angustifolia, F. sur, P. schimperi and T. catappa for Aba Gerima; and D. penninervium, E. abvssinica, E. schimperi, G. gnemon and R. apetalus for Guder as potential supplements for poor-quality roughage in ruminant livestock feed in tropical regions.Moreover, this study -as it was conducted in Ethiopia in a comprehensive way across wide agro-ecological environments, mainly in the high land, mid land and low land -can be a basis for studies scaled-up to the broader tropical regions.Studies like this will improve our understanding of available feed resources and would ultimately help design sustainable feeds and feeding systems.Thus, the identification of indigenous fodder trees and shrubs with moderate to high nutritive quality would improve the sustainability of livestock production in tropical regions and help small-scale farmers alleviate poverty.Extension bodies, policymakers and farmers should be aware of the importance of these IFTS, and should consider including IFTS in feeding formulations as supplemental diet for ruminant livestock feed.Further research is needed to evaluate animal performance following the inclusion rates of these IFTS, and to determine whether reliance on a combination of different species is better than use of a single species.In addition, in relation to the species selected in this study, it is necessary to review agronomic practices, biomass production and suitability for silvopastoral agriculture.Further research also needs to be done on mechanisms for preventing toxicity due to animal consumption of P. africana.

Figure 1 :
Figure 1: Map showing the location of the study sites

Figure 2 :
Figure 2: The methodological framework for selection and evaluation of indigenous fodder trees and shrubs (IFTS).CP = crude protein, NDF = neutral detergent factor, CT = condensed tannin

Figure 3 :
Figure 3: Principal component analysis (PCA) results showing indigenous fodder trees and shrubs (IFTS) data in relation to the chemical composition parameters.The Y-axis (Component 1) explains 23.43% of the variance, while the X-axis (Component 2) explains 16.83%.The length of vectors indicates the loading of each parameter and the angle between vectors indicates the relationship among the parameters: ADL = acid detergent lignin (g kg −1 ) NDF = neutral detergent fibre (g kg −1 ) OMD = organic matter digestibility (%) CP = crude protein (g kg −1 DM) EDDM = effective degradability of dry matter (%) TVFA = total volatile fatty acid (mmol l −1 ) GE = gross energy (MJ kg −1 )

Figure 5 :
Figure 5: A linear and quadratic regression line depicting the relationship of farmers' ranking score and CP

Table 2 :
Ranking score (RS) and relative abundance (RA) of the indigenous fodder trees and shrubs (IFTS) most preferred by farmers at each study site

Table 3 :
Nutritive value of selected indigenous fodder trees and shrubs (IFTS) species for ruminant animal feed

Table 4 :
Rank correlations of farmers' preference score with nutritive value , i.e. C. aurea,  C. tomentosa, C. lanceolata, D. angustifolia, G.gnemon, Polyscias fulva (Hiern) Harms, P. schimperi, R. apetalus, Rytigynia neglecta, U. hypselodendron and Ximenia caffra Sond. in north western part of Ethiopia.Studies like this will improve our understanding of available feed resources and would ultimately help design sustainable feeds and feeding systems.