Livestock Research for Rural Development 22 (2) 2010 Guide for preparation of papers LRRD News

Citation of this paper

Characterizing and predicting chemical composition and in vitro digestibility of crop residue using near infrared reflectance spectroscopy (NIRS)

Dereje Fekadu, Seyoum Bediye and Zinash Sileshi

Ethiopian Institute of Agricultural Research (EIAR), Holetta Research Center, P.O.Box 31, Holetta, Ethiopia
drjfekadu@gmail.com   or   derejefekadu@ymail.com

Abstract

Livestock production is an integral part of the farming system in Ethiopia. Ruminants in Ethiopia feed mainly on poor-quality plant material, such as crop residues. Crop residues provide large amount of dry matter that can back up the feed shortage, natural pasture, in the dry season of a year. 45 to 131 crop residue samples from crop types were collected and analyzed for the chemical entities (dry matter, ash, crude protein, neutral detergent fiber, acid detergent fiber, lignin) and in vitro dry matter digestibility. NIRS was evaluated by the coefficient of determination in calibration (R2), standard error of calibration (SEC), and standard error of cross validation (SECV).

 

The results showed NIRS is a method of choice for prediction of chemical composition including in vitro digestibility of crop residues.

Keywords: Chemical composition, crop residue, dry season feed, NIRS, nutritive value


Introduction

A wide variety of arable crops is grown on subsistence farm holdings. Many of these crops have residues which can form an important source of livestock feed, following the harvesting of grain. Livestock in mixed crop-livestock farming systems two to three months into a dry season feed on cereal straws, stubble or other leftovers such as maize stover. The potential and abundance of crop residues that could be used for livestock feeding in Ethiopia in most cases, drown from grain yield, using multiplier (FAO 1987, Nordbloom 1988, Kossila 1988) is 13.7 million ton (13.6 million ton in the rural area and 136 thousands ton in urban areas) from cereals having CP value ranging from 3.1 - 6.7% with digestibility level about 40.7 - 54.1% and 1.32 million ton (1.31 million ton in the rural area and 6 thousands ton in urban areas) from pulses having CP value ranging from 5.7 - 14% with digestibility level about 34.4 - 52.3%. They are suited for all classes of livestock in the country according to their nutritional characteristics.

 

Techniques of feed evaluation has been modified and refined since the mid 1980s when Weende method was proposed. Since then various chemical, biological and physical methods have been proposed and applied for feed resource characterization. Near infrared spectroscopy (NIRS) is one of the recent techniques being applied for feed resource characterization. Boever et al 1997 have assessed application of NIRS for predicting in situ degradability feeds in Belgium and reported good prediction potential in favor of NIRS. The predictive accuracy of NIRS in general relies heavily upon obtaining a calibration set which represents the variation in the main population, accurate laboratory analyses and the application of the best mathematical procedures (Park et al 1998). Although the reliability of NIRS has been investigated well for temperate feeds little work has been done for tropical feeds. Moreover, the variation in ecological set up, the biological diversity in feed resources in the country requires quite robust and cost effective method for characterization. This research result meant to fill these gaps with an objective of assessing the potential of this technique in characterizing nutritive value and predicting the chemical composition of crop residues: study reliability of NIRS technique for characterizing and predicting nutritional status of crop residues.

 

Materials and methods 

Major crop residues (tef straw, wheat straw, barley straw, maize stover, sorghum stover, pea straw, chick pea straw, lentil straw, bean straw, cowpea straw) were considered.

 

The procedures for determination of parameters undertaken are chemical entities (DM, Ash, CP, NDF, ADF and lignin) and bio-availability (IVOMD). Chemical analysis (DM, Ash, and CP) determined conventionally using procedures of AOAC (1990) and determination of NDF, ADF, and Lignin using procedure (Goering and Van Soest 1970). In vitro digestibility determined using the two-stage rumen fluid–pepsin technique (Tilley and Terry 1963).

 

Samples of the crop residues were taken for NIRS analysis of each feed types that are analyzed using conventional analysis. NIRS spectroscopy performed on 3 g of ground sample (1mm sieve) using Foss NIRS 5000 in the 1108-2492 ranges with an 8 nm step. Before scanning the samples pre-dried at 60oC overnight in an oven to standardize moisture conditions. The spectra of each sample taken by scanning (Win Scan version 1.5, 2000, Iintrasoft international, L.L.C). Mathematical and statistical treatments of the NIRS spectra were first treated using ISI. Average spectra for each sample were obtained from the scanning. Calibration equations were calculated by step-wise multiple linear regressions on the samples and 30 samples from each type of crop residue used for validation purposes. The samples for calibrations and validation were selected systematically to cover the range and to fairly represent the population for each feed they are drawn from.

 

The correlation of predicted and conventionally determined values used to assess the reliability of NIRS and residual behavior predicted. Regression analysis of the predicted values and conventionally determined values also undertaken to assess the precision of NIRS.

 

Results and discussion

The feed samples used for calibration and validation  varied in their chemical entities (DM, Ash, CP, NDF, ADF and lignin) and bio-availability (IVOMD) as is seen in Table. The mean and range of each entity were seen being similar to previously observed values (Seyoum et al 2007). There were significant differences among the samples for all entities of the feed types which suggest the presence of sufficient variation among the samples of the feed types to develop NIRS equation separately.


Table 1.  Performance of NIRS calibration

Feed type

Parameters/traits

N

Range

R2

SEC

1-VR

SECV

Barley straw

DM

131

87.4 – 93.6

0.86

0.38

0.82

0.43

Ash

130

3.01 – 11.8

0.74

0.75

0.66

0.85

CP

130

1.00 – 3.40

0.90

0.19

0.86

0.23

NDF

131

65.8 – 75.6

0.66

0.96

0.56

1.09

ADF

131

39.9 – 45.3

0.64

0.88

0.43

0.94

Lignin

131

3.33 – 8.93

0.63

0.92

0.55

0.96

In vitro (DOMD)

131

51.5 – 58.1

0.72

1.06

0.22

1.09

Bean straw

DM

78

91.9 – 92.2

0.90

0.02

0.77

0.02

Ash

74

8.97 – 9.66

0.96

0.02

0.77

0.06

CP

75

7.47 – 7.81

0.95

0.01

0.71

0.03

NDF

74

61.6 – 63.4

0.95

0.06

0.80

0.13

ADF

78

45.0 – 45.9

0.96

0.03

0.54

0.11

Lignin

76

8.41 – 8.61

0.76

0.02

0.41

0.03

In vitro (DOMD)

77

51.3 – 52.7

0.94

0.05

0.78

0.10

Chickpea straw

DM

79

91.5 – 92.0

0.87

0.03

0.79

0.03

Ash

79

8.67 – 9.00

0.83

0.02

0.46

0.04

CP

84

6.19 – 6.37

0.80

0.02

0.35

0.02

NDF

79

55.1 – 57.5

0.98

0.06

0.93

0.11

ADF

77

40.5 – 41.4

0.86

0.06

0.62

0.10

Lignin

75

8.04 – 8.52

0.98

0.01

0.96

0.02

In vitro (DOMD)

80

56.0 – 58.4

0.92

0.12

0.80

0.18

Cowpea straw

DM

47

88.5 – 95.9

0.99

0.03

0.99

0.04

Ash

51

9.49 – 19.4

0.72

1.35

0.17

1.55

CP

51

7.45 – 28.3

0.84

2.31

0.34

2.88

NDF

45

32.0 – 52.5

0.99

0.17

0.99

0.21

ADF

47

17.8 – 23.5

0.95

0.21

0.94

0.23

Lignin

50

2.97 – 3.95

0.99

0.01

0.99

0.01

In vitro (DOMD)

51

60.6 – 60.7

0.65

0.02

0.97

0.02

Lentil straw

DM

67

86.2 – 98.5

0.68

2.05

0.01

2.05

Ash

67

2.89 – 13.6

0.61

1.79

0.02

1.81

CP

67

5.08 – 11.0

0.64

0.99

0.01

1.00

NDF

67

35.5 – 79.6

0.67

7.33

0.01

7.46

ADF

67

12.5 – 68.6

0.63

9.34

0.01

9.38

Lignin

67

4.41 – 12.6

0.66

1.37

0.02

1.39

In vitro(DOMD)

67

39.2 – 70.2

0.65

5.10

0.02

5.11

Maize stover

DM

72

86.0 – 96.6

0.92

0.51

0.89

0.58

Ash

71

2.89 – 14.1

0.84

0.91

0.81

1.01

CP

72

1.11 – 4.73

0.71

0.42

0.41

0.46

NDF

72

65.7 – 91.5

0.86

1.61

0.65

2.53

ADF

72

40.2 – 59.1

0.76

2.09

0.43

2.38

Lignin

72

0.85 – 11.0

0.60

1.72

0.72

1.77

In vitro(DOMD)

72

44.3 – 66.8

0.76

2.50

0.41

2.88

Pea straw

DM

106

92.3 – 92.7

0.85

0.03

0.75

0.04

Ash

111

7.44 – 8.36

0.95

0.04

0.80

0.07

CP

102

6.82 – 7.41

0.96

0.02

0.87

0.04

NDF

106

64.4 – 66.4

0.74

0.17

0.34

0.28

ADF

107

54.3 – 55.1

0.61

0.08

0.24

0.12

Lignin

105

10.0 – 10.4

0.88

0.02

0.59

0.04

In vitro (DOMD)

109

53.8 – 54.6

0.93

0.04

0.81

0.06

Sorghum stover

DM

117

84.4 – 96.0

0.92

0.55

0.89

0.65

Ash

118

2.73 – 11.6

0.71

0.79

0.58

0.95

CP

118

1.04 - 7.81

0.61

0.85

0.49

0.97

NDF

118

57.3 – 83.2

0.46

3.17

0.43

3.25

ADF

118

39.0 – 60.4

0.57

2.34

0.44

2.68

Lignin

117

4.04 – 9.32

0.47

0.64

0.35

0.71

In vitro (DOMD)

118

39.2 – 66.6

0.23

4.00

0.12

4.28

Tef straw

DM

130

88.5 – 95.0

0.67

0.62

0.54

0.73

Ash

130

4.56 – 10.8

0.71

0.87

0.12

0.98

CP

130

0.41 – 6.95

0.66

0.64

0.57

0.71

NDF

130

70.0 – 86.0

0.70

1.93

0.40

2.05

ADF

130

35.1 – 54.3

0.64

2.97

0.06

3.12

Lignin

130

2.15 – 9.42

0.74

1.17

0.05

1.20

In vitro (DOMD)

130

45.4 – 61.3

0.61

2.42

0.04

2.59

Wheat straw

DM

95

89.5 – 94.1

0.86

0.28

0.81

0.33

Ash

96

6.20 – 17.4

0.67

1.06

0.57

1.23

CP

93

0.94 - 6.60

0.79

0.62

0.71

0.73

NDF

96

40.5 – 86.2

0.89

3.09

0.85

3.60

ADF

88

26.8 – 69.0

0.92

2.01

0.88

2.39

Lignin

91

2.37 – 10.4

0.78

0.63

0.72

0.71

In vitro (DOMD)

93

42.2 – 77.0

0.83

2.37

0.78

2.72

DM = Dry matter, CP = Crude protein, NDF = Neutral detergent fiber, ADF = Acid detergent fiber, DOMD = Digestible organic matter in the dry matter, SEC = standard errors of calibration, 1-VR = coefficient of determination of cross validation, SECV = standard errors of cross-validation


The table shows the calibration and external validation statistics for the various classes of feed resources and their traits considered. The calibration equations for DM, Ash, CP, NDF, ADF, Lignin and in vitro show relatively high determination coefficient, and low standard errors of calibration (SEC) and standard errors of cross-validation (SECV)and hence, these traits could be predicted with good precision. Moreover, the predicted means for each trait were similar to the means based on conventional chemical analyses. Higher SEC value was recorded for the feed types of each class may be due to the broader range of variation for the trait.

 

SECV is a basic statistics to measure accuracy for a calibration equation (Shenk and Westerhaus 1993). Accordingly, the calibration error should be comparable to the sampling error and this value is similar to SEP (standard error of performance). Thus, the best performance in calibration equations for individual traits corresponded to those traits for which the variability in the calibration set was wider (see table), indicating that successful calibration equations using NIRS depend on the variability of constituents under investigation.

 

Conclusion 

 

References 

AOAC 1990 Official Method of Analysis, 15th edition. Association of Official Analytical Chemists, Arlington, USA. 

 

Boever J L, de Waes J Van, Cottyn B J and Bouque C V 1997 Potential of solubility, enzymatic methods and NIRS to predict to in situ escape protein. Netherlands Journal of Agricultural Science 45 (2):291-306 http://library.wur.nl/ojs/index.php/njas/article/view/519/233

 

FAO 1987 Assistance to Land Use Planning in Ethiopia: Economic Analysis of Land Use. UNDP/FAO Technical Report No. 8, Food and Agriculture Organization of the United Nations, Rome, Italy.

 

Goering H K and van Soest P J 1970 Forage Fibre Analysis. USDA, ARS Agric. Handbook, No. 379, pp. 1 – 12.

 

Kossila V 1988 The availability of crop residues in developing countries in relation to livestock population. In: J D Reed, B S Capper and P J H Neate (editors). Plant breeding and the nutritive value of crop residues. Proceedings of a workshop held at ILCA, Addis Ababa, Ethiopia, 7-10 December 1987. ILCA, Addis Ababa. http://www.ilri.org/InfoServ/Webpub/Fulldocs/X5495e/x5495e03.htm

 

Nordbloom T 1988 The importance of crop residues as feed resources in West Asia and North Africa. In J D Reed, B S Capper and P J H Neate (editors): Proceedings of a workshop on plant breeding and the nutritive value of crop residues, Addis Ababa, 7-10 December 1987, ILCA, Addis Ababa, Ethiopia. pp. 41-63. http://www.ilri.org/InfoServ/Webpub/Fulldocs/X5495e/x5495e04.htm

 

Park R S, Agnew R B, Gorden F J and Steen R W J 1998 The use mear infrared spectroscopy (NIRS) on undried samples of grass silage to predict chemical composition and digestibility parameters. Animal Feed Science and Technology, 72(1/2):155-167.

 

Seyoum Bediye, Zinash Sileshi and Dereje Fekadu 2007 Chemical composition and nutritive values of Ethiopian feeds. Research report No. 73. Ethiopian Institute of Agricultural Research, pp. 24.

 

Shenk J S and Westerhaus M O 1993 Analysis of agriculture and food products by Near Infrared Reflectance Spectroscopy. ISI Monograph. 116 pp.

 

Tilley J A and Tery R A 1963 A two-stage technique for the in vitro digestion of forage crop. Journal of British Grassland Society 18: 104 – 111.



Received 23 November 2009; Accepted 21 December 2009; Published 7 February 2010

Go to top