**4. Is it possible to predict the rumen digestibility (feeding value) of unknown and underutilised forages?**

#### **4.1. Prediction of degradation of forages in the rumen using feed and animal properties**

#### *4.1.1. Materials and methods*

**Table 5.** Nylon bag degradability of urea treated and untreated forage grasses (roughages) in cows fed kikuyu pasture.

*curvula*, ECB: *Eragrostis curvula* at bloom stage, KG: kikuyu grass, SE: *Schizachyrium exile*, VGHD: veld grass hay from Dundee, VGHC: veld grass hay Camperdown, VGHP<sup>1</sup>

2, CPH: cowpea husks, CRP: cassava root peels, GNH: groundnut haulms, UTCPH: ureatreated cowpea husks, UTDH: urea-treated *Diheteropogon hagerupii*, UTET: urea-treated *Eragrostis tremula*, UTSE: urea-treated *Schizachyrium exile*, UTMIS: urea-treated maize stover, SS: sorghum stover, UTSS: urea-treated sorghum stover, SSLS: sorghum stover leaves and sheath, SSS: sorghum stover stems, MB: millet bran, WB: wheat bran, and CSC: cot-

CMLB: *Colophospermum mopane* leaves—brown, CMLG: *Colophospermum mopane* leaves green, CMPG: *Colophospermum mopane* pods, CPH: cowpea husks, CRP: cassava root peels, GNH: groundnut haulms, MPL: *Mucuna pruriens* leaves, AQLP: *Afzelia quanzensis* legume pods, BOAL: *Brassica oleraceae var. acephala* leaves, UTCPH: urea-treated cowpea husks, MB: millet bran, WB: wheat bran, CSC: cottonseed cake, a: rapidly degradable fraction, b: slowly degradable fraction, c: rate of degradation, PD: potential degradability, and ED: effective

MS: maize stover, ML: maize leaves, MT: maize stalks, WS: wheat straw, EC: *Eragrostis curvula*, ECB: *Eragrostis curvula* at bloom stage, KG: kikuyu grass, VGHD: veld grass hay from Dundee, VGHC: veld grass hay Camperdown, VGHP1: veld grass hay Pietermaritzburg area 1, VGHP2: veld grass hay from the Pietermaritzburg area 2, kp: rate of passage of particles in the rumen, a: rapidly degradable fraction, b: slowly degradable fraction, c: rate of degrada-

MS: maize stover, ML: maize leaves, MT: maize stalks, WS: wheat straw, EC: *Eragrostis curvula*, ECB: *Eragrostis curvula* at bloom stage, KG: kikuyu grass, VGHD: veld grass hay.

tion, PD: potential degradability, and ED: effective degradability.

grass hay Pietermaritzburg area 1, VGHP<sup>2</sup>

tonseed cake.

96 Forage Groups

degradability.

: veld

: veld grass hay from the Pietermaritzburg area

Data were collected from studies that reported at least average values for in sacco (nylon bag technique) degradability parameters (a, soluble fraction; b, slowly degradable fraction and c, rate of degradation) of roughages and stated the diet, feeds and feed supplements given to animals. A dataset was created bearing degradability parameters from wild and domesticated ruminants from 40 studies. Factors affecting degradability were identified in each of these studies and were categorised into two main groups: (1) diet properties (i.e. fed to the animal) and (2) feed sample properties (i.e. incubated in the rumen). Diet properties were used to account for the effects of rumen ecology on fermentation and included neutral detergent fibre (NDF), starch (STA) and crude protein (CP) contents of entire diet (all in g/kg), level of concentrate supplementation (%) and provision of a urea supplement in the form of a lick (presence = 1, absence = 0). Feed sample properties included urea treatment (%) of sample and feed compositional attributes (DM, dry matter; CP, crude protein; NDF, neutral detergent fibre, ADF, acid detergent fibre; HEM, hemicellulose and ash all in g/kg). Starch content of the diet fed to animals was calculated using the formula: STA = 1000–(NDF + CP). Potential degradability (PD) and hemicellulose (HEM) content were calculated in studies that did not report them using the formulae: PD = a + b; and HEM = NDF—ADF, respectively. Studies that did not report dietary composition of feeds but mentioned names of feeds used had their composition looked up in studies that reported them. These factors were used as input parameters to develop regression models for predicting degradability of feeds in the rumen.

A step-wise regression procedure on the Statistical Analysis System 9.3 (SAS Institute Inc., Cary, NC, USA) was used to select parameters that qualified to develop regression equations to predict (1) rapidly degradable fraction of fibre (a), (2) potential degradability (PD), (3) time lag for fermentation to occur (tL), and (4) rate of degradation (c) in the rumen. One parameter from a pair of correlated parameters was dropped in model development when both correlated parameters significantly influence degradation parameters. Those parameters that qualified for model development were CP and NDF content of feed sample (model for soluble fraction of fibre); ADF content of feed sample and STA content of diet (model for potential degradability); ADF, CP and ash content of feed sample, and STA content of diet (model for time-lag); NDF and CP content of feed sample, and, STA and DNDF content of diet (model for degradation rate).

Regression models were used to simulate the rumen degradability of *Colophospermum mopane* leaves and pods, *Diheteropogon hagerupii*, *Eragrostis tremula*, *Mucuna pruriens* leaves, Marula oil cake, *Afzelia quanzensis* legume pods, *Brassica oleraceae var. acephala* leaves, maize stover, leaves and stalks, millet stover, wheat straw, *Eragrostis curvula*, Kikuyu grass, *Schizachyrium exile*, veld grass hay, cowpea husks, cassava root peels, groundnut haulms, *Eragrostis tremula*, sorghum stover, leaves and sheath, and stems, millet bran, wheat bran, and cottonseed cake. The effective degradability of these forages was calculated using the model of McDonald [47].

#### *4.1.2. Statistical analyses*

For all evaluations, regression analyses of observed against predicted degradability were carried out using the linear regression procedure. Coefficients of determination (R<sup>2</sup> ) were used to evaluate the precision of regression lines in approximating real data points of models and standard error of the mean (SEM) was used to determine the accuracy of prediction.

#### **4.2. Results**

#### *4.2.1. Model development*

From the step-wise regression procedure for all prediction models, level of concentrate supplementation, provision of a urea supplement in the form of a lick and urea treatment of feed sample were rejected in model development.

The regression model for predicting the soluble fraction (a) was a = 558.12(±62.45) + 0.27 (±0.133) CP–0.57(±0.07) NDF (n = 113, SEM = 6.86), accounting for 59% of the variation in development.

The regression model for predicting the potential degradability (PD) was PD = 1025.96(±66.64) –0.91(±0.10) ADF + 0.32(±0.08) STA (n = 113, SEM = 9.27), accounting for 65% of the variation in development.

The regression model for predicting the time-lag (tL) was tL = −11.33(±1.89) + 0.030(±0.002) ADF + 0.01(±0.003) CP–0.006(±0.001) STA + 0.02(±0.007) ASH (n = 113, SEM = 0.17), accounting for 77% of the variation in development.

The regression model for predicting the rate of degradation (c) was c = 0.12(±0.05) + 0.00013 (±0.00002) CP–0.00012(±0.00006) STA–0.00002(±0.00001) NDF–0.00008(±0.00005) DNDF (n = 113, SEM = 0.0009), accounting for 55% of the variation in development.

#### *4.2.2. Model predictions*

The regression model for predicting the soluble fraction of feeds accounted for 70% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates (**Figure 1**).

The regression model for predicting the effective degradability of feeds accounted for 57% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concen-

Evaluation and Prediction of the Nutritive Value of Underutilised Forages as Potential Feeds…

http://dx.doi.org/10.5772/intechopen.83643

99

**Figure 1.** Relationship between observed and predicted degradability of soluble fraction.

**Figure 2.** Relationship between observed and predicted potential degradability.

Among the forage legumes, trees and shrubs, *Brassica oleracea var. acephala* leaves had a superior crude protein content and the lowest neutral and acid detergent fibre contents. The CP content of *Brassica oleracea var. acephala* is slightly higher than those reported by McDonald et al. [21] and Barry et al. [22]. The rate of degradation of *Colophospermum mopane* pods was similar to that of *Brassica oleracea var. acephala*. High levels of degradability of these feeds were partly due to

trates (**Figure 5**).

**4.3. Discussion**

The regression model for predicting the potential degradability accounted for 24% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates (**Figure 2**).

The regression model for predicting the slowly degradable fraction of feeds for forage legumes, trees and shrubs, forage grasses and concentrates (**Figure 3**).

The regression model for predicting the rate of degradation accounted for 4% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates (**Figure 4**).

Evaluation and Prediction of the Nutritive Value of Underutilised Forages as Potential Feeds… http://dx.doi.org/10.5772/intechopen.83643 99

**Figure 1.** Relationship between observed and predicted degradability of soluble fraction.

**Figure 2.** Relationship between observed and predicted potential degradability.

The regression model for predicting the effective degradability of feeds accounted for 57% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates (**Figure 5**).

#### **4.3. Discussion**

sorghum stover, leaves and sheath, and stems, millet bran, wheat bran, and cottonseed cake. The effective degradability of these forages was calculated using the model of McDonald [47].

For all evaluations, regression analyses of observed against predicted degradability were car-

to evaluate the precision of regression lines in approximating real data points of models and

From the step-wise regression procedure for all prediction models, level of concentrate supplementation, provision of a urea supplement in the form of a lick and urea treatment of feed

The regression model for predicting the soluble fraction (a) was a = 558.12(±62.45) + 0.27 (±0.133) CP–0.57(±0.07) NDF (n = 113, SEM = 6.86), accounting for 59% of the variation in

The regression model for predicting the potential degradability (PD) was PD = 1025.96(±66.64) –0.91(±0.10) ADF + 0.32(±0.08) STA (n = 113, SEM = 9.27), accounting for 65% of the variation

The regression model for predicting the time-lag (tL) was tL = −11.33(±1.89) + 0.030(±0.002) ADF + 0.01(±0.003) CP–0.006(±0.001) STA + 0.02(±0.007) ASH (n = 113, SEM = 0.17), accounting

The regression model for predicting the rate of degradation (c) was c = 0.12(±0.05) + 0.00013 (±0.00002) CP–0.00012(±0.00006) STA–0.00002(±0.00001) NDF–0.00008(±0.00005) DNDF

The regression model for predicting the soluble fraction of feeds accounted for 70% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates

The regression model for predicting the potential degradability accounted for 24% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates

The regression model for predicting the slowly degradable fraction of feeds for forage

The regression model for predicting the rate of degradation accounted for 4% of the variation in prediction for forage legumes, trees and shrubs, forage grasses and concentrates (**Figure 4**).

(n = 113, SEM = 0.0009), accounting for 55% of the variation in development.

legumes, trees and shrubs, forage grasses and concentrates (**Figure 3**).

) were used

ried out using the linear regression procedure. Coefficients of determination (R<sup>2</sup>

standard error of the mean (SEM) was used to determine the accuracy of prediction.

*4.1.2. Statistical analyses*

*4.2.1. Model development*

sample were rejected in model development.

for 77% of the variation in development.

**4.2. Results**

98 Forage Groups

development.

in development.

*4.2.2. Model predictions*

(**Figure 1**).

(**Figure 2**).

Among the forage legumes, trees and shrubs, *Brassica oleracea var. acephala* leaves had a superior crude protein content and the lowest neutral and acid detergent fibre contents. The CP content of *Brassica oleracea var. acephala* is slightly higher than those reported by McDonald et al. [21] and Barry et al. [22]. The rate of degradation of *Colophospermum mopane* pods was similar to that of *Brassica oleracea var. acephala*. High levels of degradability of these feeds were partly due to

the brans tended to have faster degradation rates than cotton seed cake and *Brassica oleracea var. acephala* leaves. *Colophospermum mopane* leaves and pods had comparable CP and NDF levels compared to maize and wheat brans, suggesting that *Brassica oleracea var. acephala* and, *Colophospermum mopane* can be used as good sources of supplementary protein to ruminants. Relationships between two variables are said to be ideal when the coefficient of determination

Evaluation and Prediction of the Nutritive Value of Underutilised Forages as Potential Feeds…

) is in unity; any deviation from the unity degree indicates the degree of imperfection. The above parameters were used to determine the effective degradability (ED): (ED = a + (PD−a) × c/ (c + kp); where 'a' is a soluble fraction, PD is the potential degradability, 'c' is the rate of degradation and kp is the rate of passage of particles through the rumen. Effective degradability is equivalent to digestibility in the rumen. The predicted effective degradability indicated in **Figure 5** followed the expected trends, suggesting that these models (for predicting 'a', PD, and 'c') in the meantime can be used for this purpose. The overall trend between the observed and the predicted digestibility is positive, though accounting for just 36–52% of the total varia-

the simulation model to temperate roughages [43] and those from this study. The amount of variation accounted for in observed against predicted digestibility for simulations by Nsahlai and Apaloo [49, 50] was comparably higher than those reported in empirical studies by Shem

The rather low precision in predicting the rate of degradation (mainly for concentrates, legume forages, trees and shrubs) and the potential degradability (concentrates) of feeds in this study may have been due to the fact that the studies that were used in model development reported data on degradation of roughages grasses only, which are generally of low quality, and did not use data on concentrates, legume forages, trees and shrubs. Despite this, simulations of solubility and effective degradability were good, suggesting that slight modification of model parameters may give better prediction of all degradability (nutritive value) of a large number and classes of forage crops. Generally, there is a poor simulation of

of 70% obtained with the application of

http://dx.doi.org/10.5772/intechopen.83643

101

tion [49], which does not compare favourably with R2

et al. [10], Kibon and Orskov [51] and Umunna et al. [52].

**Figure 5.** Relationship between observed and predicted effective degradability.

(R2

**Figure 3.** Relationship between observed and predicted degradability of slowly degradable fraction.

**Figure 4.** Relationship between observed and predicted rates of degradation .

high levels of crude protein, which could help in the proliferation of microbial populations in the rumen, increasing ED and rate of degradation of these forages. Faster rates of degradation may suggest faster rates of passage of these feeds in the rumen, which could increase microbial protein supply for host animals in the hindgut, improving animal's nutritional status. The CP level in *Colophospermum mopane* leaves was comparable to results of Halimani et al. [14], while NDF contents tended to be comparably higher than those reported by other authors [14, 17].

Compared to concentrates used in the study, *Brassica oleracea var. acephala* leaves tended to have superior crude protein levels than the 'brans' and cotton seed cake. Despite this trend, the brans tended to have faster degradation rates than cotton seed cake and *Brassica oleracea var. acephala* leaves. *Colophospermum mopane* leaves and pods had comparable CP and NDF levels compared to maize and wheat brans, suggesting that *Brassica oleracea var. acephala* and, *Colophospermum mopane* can be used as good sources of supplementary protein to ruminants.

Relationships between two variables are said to be ideal when the coefficient of determination (R2 ) is in unity; any deviation from the unity degree indicates the degree of imperfection. The above parameters were used to determine the effective degradability (ED): (ED = a + (PD−a) × c/ (c + kp); where 'a' is a soluble fraction, PD is the potential degradability, 'c' is the rate of degradation and kp is the rate of passage of particles through the rumen. Effective degradability is equivalent to digestibility in the rumen. The predicted effective degradability indicated in **Figure 5** followed the expected trends, suggesting that these models (for predicting 'a', PD, and 'c') in the meantime can be used for this purpose. The overall trend between the observed and the predicted digestibility is positive, though accounting for just 36–52% of the total variation [49], which does not compare favourably with R2 of 70% obtained with the application of the simulation model to temperate roughages [43] and those from this study. The amount of variation accounted for in observed against predicted digestibility for simulations by Nsahlai and Apaloo [49, 50] was comparably higher than those reported in empirical studies by Shem et al. [10], Kibon and Orskov [51] and Umunna et al. [52].

The rather low precision in predicting the rate of degradation (mainly for concentrates, legume forages, trees and shrubs) and the potential degradability (concentrates) of feeds in this study may have been due to the fact that the studies that were used in model development reported data on degradation of roughages grasses only, which are generally of low quality, and did not use data on concentrates, legume forages, trees and shrubs. Despite this, simulations of solubility and effective degradability were good, suggesting that slight modification of model parameters may give better prediction of all degradability (nutritive value) of a large number and classes of forage crops. Generally, there is a poor simulation of

**Figure 5.** Relationship between observed and predicted effective degradability.

high levels of crude protein, which could help in the proliferation of microbial populations in the rumen, increasing ED and rate of degradation of these forages. Faster rates of degradation may suggest faster rates of passage of these feeds in the rumen, which could increase microbial protein supply for host animals in the hindgut, improving animal's nutritional status. The CP level in *Colophospermum mopane* leaves was comparable to results of Halimani et al. [14], while NDF contents tended to be comparably higher than those reported by other authors [14, 17].

**Figure 4.** Relationship between observed and predicted rates of degradation .

**Figure 3.** Relationship between observed and predicted degradability of slowly degradable fraction.

100 Forage Groups

Compared to concentrates used in the study, *Brassica oleracea var. acephala* leaves tended to have superior crude protein levels than the 'brans' and cotton seed cake. Despite this trend, digestibility for low quality roughages, which are commonly grazed and fed to ruminants in the tropics. Ambient temperature grossly affects the digestibility of plant material through its influence on lignin deposition in plants. Studies should focus on development of digestibility models that account for variability in diet quality as brought about by ambient temperature. Future studies may need to account for the type of model used in computation of degradation parameters.

**References**

Research. 2011;**6**(15):3461-3469

Production. 2015;**47**:1131-1137

cal foods. Animal Science. 1995;**60**:65-74

[1] Orskov ER, DeB Hovell FD, Mould F. The use of the nylon bag technique for the evalua-

Evaluation and Prediction of the Nutritive Value of Underutilised Forages as Potential Feeds…

http://dx.doi.org/10.5772/intechopen.83643

103

[2] Delve RJ, Cadisch G, Tanner JC, Thorpe W, Thorne PJ, Giller KE. Implications of livestock feeding management on soil fertility in the smallholder farming systems of sub-

[3] Njarui DMG, Gatheru M, Wambua JM, Nguluu SN, Mwangi DM, Keya GA. Feeding management for dairy cattle in smallholder farming systems of semi-arid tropical Kenya. Livestock Research for Rural Development. 2011;**23**, Article #111. Retrieved July 30, 2013

[4] Mutimura M, Everson TM. Assessment of livestock feed resource-use patterns in low rainfall and aluminium toxicity prone areas of Rwanda. African Journal of Agricultural

[5] Mutimura M, Ebong C, Rao IM, Nsahlai IV. Nutritional values of available ruminant feed resources in smallholder dairy farms in Rwanda. Tropical Animal Health and

[6] McDonald P, Edwards RA, Greenhalgh JFD, Morgan CA, Sinclair LA, Wilkinson RG. Animal Nutrition. 7th ed. Harlow, England: Pearson Education Limited; 2011. 692 p [7] Amata IA. The use of non-conventional feed resources (NCFR) for livestock feeding in

[8] Okafor JC. Myrianthus arboreus. P. Beauv. In: Grubben GJH, Denton OA, editors. PROTA 2: Vegetables/Legumes. Wageningen, Netherlands: PROTA Foundation; 2004 [9] Amata IA, Lebari T. Comparative evaluation of the nutrient profile of four selected browse plants in the tropics, recommended for use as non-conventional livestock feed-

[10] Shem MN, Orskov ER, Kimambo AE. Prediction of voluntary dry matter intake, digestible dry-matter intake and growth of cattle from the degradation characteristics of tropi-

[11] Nefzaoui A, Salem HB. Forage fodder and animal nutrition. In: Nobel PS, editor. Cacti: Biology and Uses. Berkeley, CA, USA: University of California Press; 2002. pp. 190-210

[12] Salem HB, Smith T. Feeding strategies to increase small ruminant production in dry

[13] Hall-Martin AJ. Food selection by Transvaal lowveld giraffe as determined by analysis of stomach contents. South African Journal of Wildlife Research. 1974;**4**(3):191-202 [14] Halimani TE, Ndlovu LR, Dzama K, Chimonyo M, Miller BG. Metabolic response of pigs supplemented with incremental levels of leguminous *Acacia karroo*, *Acacia nilotica*

and *Colophospermum mopane* leaf meals. Animal Science. 2005;**81**(1):39-45

the tropics: A review. Journal of Global Biosciences. 2014;**3**(2):604-613

ing materials. African Journal of Biotechnology. 2011;**10**(64):14230-14233

environments. Small Ruminant Research. 2008;**77**(2):174-194

tion of feedstuffs. Tropical Animal Health and Production. 1980;**5**:195-213

Saharan Africa. Agriculture, Ecosystems and Environment. 2001;**84**:227-243
