1648
Received: 1 October 2010 Revised: 25 January 2011 Accepted: 31 January 2011 Published online in Wiley Online Library: 28 March 2011
(wileyonlinelibrary.com) DOI 10.1002/jsfa.4361
Colour score as a guide for estimating the protein value of corn gluten feed
Ana Rita J Cabrita, a ∗ Margarida RG Maia, b Marisa Freitas, a Jos ´e Manuel F Abreu a and Ant ´onio J Mira Fonseca b
Abstract
BACKGROUND: Dry corn gluten feed (CGF) is a raw material commonly included in compound feeds, particularly for ruminant animals. Its colour can vary from yellow–light brown to dark brown. A general assumption is that darker CGF has a low nutritive value for animals due to heat-damaged protein. This study aims to evaluate the use of colour as a practical guide for estimating CGF protein value.
RESULTS: Results indicate great variability in the physical and nutritional properties among 28 sources of CGF. Principal component factor analysis was used to reduce a large number of variables into fewer numbers of factors. First factor aggregated 1/Zand acid detergent insoluble N content in opposition toX,Y,eL, andL∗2. Second factor aggregateda,a∗, neutral detergent fibre and soluble crude protein (CP) contents in opposition to ash and CP contents, and digestible N andin vitroorganic matter digestibility. Multiple regression analysis also showed that physical appearance (colour) was related to nutritional properties, stronger relationships being found betweena,a∗andL∗2scores and digestible N.
CONCLUSION: Colour could potentially be used to quickly predict the nutritive value of CGF, particularly digestible N, a key parameter when formulating rations.
c 2011 Society of Chemical Industry
Keywords:colour; corn gluten feed; nutritive value; protein
INTRODUCTION
Corn gluten feed (CGF) is a combination of several co-products from the wet-milling manufacture of corn starch or syrup, namely corn bran, corn fibre and corn steep liquor. Dry CGF is generally included in compound feeds, particularly for ruminant animals as a source of rumen-degradable protein and energy. It can vary in colour from yellow–light brown to dark brown, depending on the amount of steep liquor, drying temperature and drying time. An assumption commonly expressed by compound feed manufacturers and field nutritionists is that darker CGF decreases feed acceptance by the animal, and thus voluntary intake and animal production. This is probably due to heat damage during processing of corn that promotes non-enzymatic browning reactions (the so-called Maillard reaction) that are detrimental to protein quality, lysine being known to be the most sensitive amino acid.1 – 4
Traditionally, acid detergent insoluble N (ADIN) has been used as a measure of heat damage in non-forage plant protein sources,5 – 7 but some evidence suggests that this relationship is very poor.8 – 10 Nakamuraet al.11observed that, although ADIN increased in heat- damaged non-forage protein sources, this increase was not a one- to-one relationship with indigestible protein. Indeed, evidence suggests that ADIN is partially digestible8,10,12and overestimates heat damage.9,10,13 However, these studies evaluated heat- damaged protein through N digestibilities that do not necessarily reveal the metabolism of ingested N. The results obtained by Nakamuraet al.14with growing ruminants suggest that some of
the N in heat-damaged protein was absorbed post-ruminally but was not used for growth.
Although more in vivo work would be needed to further understand the relationship between heat-damaged CGF and animal performance, it is also important to find a simple methodology able to be used in the feed industry to quickly assess its nutritive value. It is well known that colour plays an important role in food acceptance in humans, and it is commonly quantitatively measured in studies on this subject,15but as far as we know few studies have related colour to nutritional parameters of animal feeds, and relied only on visual appreciation. Recent studies16 – 18indicated a linkage between distillers wet grain colour and microbial growth, concluding that colour parameters could be used to predict the microbial stability of distillers wet grains.
Therefore, the present study aimed to evaluate the use of colour as a practical guide for estimating CGF protein value.
∗ Correspondence to: Ana Rita J Cabrita, REQUIMTE, Departamento de Geociˆencias, Ambiente e Ordenamento do Territ´orio, Faculdade de Ciˆencias, Universidade do Porto, Campus Agr´ario de Vair˜ao, Rua Padre Armando Quintas, 4485-661 Vair˜ao VC, Portugal. E-mail: rita.cabrita@mail.icav.up.pt
a REQUIMTE, Departamento de Geociˆencias, Ambiente e Ordenamento do Territ´orio, Faculdade de Ciˆencias, Universidade do Porto, 4485-661 Vair˜ao VC, Portugal
b REQUIMTE, ICBAS, Instituto de Ciˆencias Biom´edicas de Abel Salazar, Universi- dade do Porto, 4485-661 Vair˜ao VC, Portugal
1649
Table 1. Description of chemical composition (g kg−1 dry matter, DM), digestible N (kg kg−1total N),in vitroorganic matter digestibility (IVOMD, kg kg−1) and colour parameters of corn gluten feed samples (n=28) used in the study
Mean SD Minimum Maximum
Ash 72.7 4.88 64.3 85.7
Starch 121.2 22.43 90.5 193.1
NDF 400.4 17.27 375.4 447.8
CP 245.7 19.41 212.3 287.7
Soluble CP (g kg−1CP) 410.7 56.36 299.2 519.3 ADIN (g kg−1total N) 41.4 20.06 10.9 80.1
Digestible N 0.78 0.035 0.71 0.86
IVOMD 0.76 0.026 0.71 0.83
X 23.4 3.44 14.4 31.0
Y 22.2 3.48 13.2 30.0
Z 13.9 2.66 9.3 20.7
L∗ 53.8 4.25 43.0 61.7
a∗ 7.7 0.91 5.6 9.3
b∗ 23.0 2.65 15.7 27.3
L 46.9 3.77 36.3 54.8
a 6.5 0.69 4.9 7.9
b 15.4 1.69 10.3 17.8
EXPERIMENTAL
Twenty-eight sources of CGF were obtained from Portuguese feed compound manufacturers. Ground CGF samples (1 mm) were analysed for ash (method 942.05)19 and Kjeldahl N (method 954.01).19 Crude protein (CP) was calculated as Kjeldahl N × 6.25. Neutral detergent fibre (NDF) and acid detergent fibre (ADF) were determined by the procedures of Van Soest et al.20 and Robertson and Van Soest,21withα-amylase being added during NDF extraction; sodium sulfite was not added. Neutral detergent fibre (NDF) was expressed without residual ash. Acid detergent insoluble N (ADIN) was determined on ADF residues using the Kjeldahl method. Starch was analysed on finely ground samples (0.5 mm screen) using the method described by Salomonsson et al.22 Soluble protein was determined according to Hovell
et al.23Nitrogen digestibility was determined using the enzymatic method described by Mathiset al.24In vitroorganic matter (OM) digestibility (IVOMD) was determined by the method of Tilley and Terry,25 as modified by Van Soestet al.26 Colour analyses were performed in 0.5 mm milled samples using a colorimeter (model CR 400, Minolta, Osaka, Japan). The instrument was calibrated with a white plate before measurements were made.
All original variables were tested for normality according to the Shapiro–Wilk test, and when needed (starch,Z,L,b,L∗andb∗) transformations were made to reach normality (inverse of starch, exponential ofL, and power of 2 ofb,L∗, andb∗). Correlations were tested between various physical and nutritional characteristics of CGF. Relationships between physical and nutritional characteristics of CGF were estimated by linear regression analysis. Relationships obtained were assessed based on the proportion of variance accounted for by the model (R2), residual mean square error (RMSE) and stepwise regression. That is, the best linear model was selected by minimizing the RMSE, maximizingR2, and stepwise regression by which variables were added one at a time to the model, as long as theFstatisticp-value was below 0.05; after adding a variable, all others already included in the model were re-evaluated, those having an insignificantF statisticp-value being removed. Only after this check was made and the identified variables removed could another variable be added to the model. The stepwise process ended when none of the variables excluded from the model had a significantFstatistic and every variable included in the model was significant. Physical and chemical parameters were submitted to principal component factor analysis. Factor analysis is a technique that is used to reduce a large number of variables into a smaller number of factors, extracting maximum common variance from all variables and putting them into a common score. Eigenvalues were the criteria used for determining the number of factors, a factor being considered if Eigenvalues were greater than one. Eigenvalues show variance explained by that particular factor out of the total variance. The varimax rotation method was used to make the output more understandable. The varimax rotation criterion maximizes the sum of the variances of the squared coefficients within each eigenvector, and the rotated axes remain orthogonal. All computations were done using SAS software (2002, SAS Institute Inc., Cary, NC, USA).
Table 2. Coefficients of linear correlation (r) between chemical composition, digestible N andin vitroorganic matter digestibility (IVOMD) (for each parameter, probability is presented in the second row)
Ash 1/Starch NDF CP SolCP ADIN Dig. N IVOMD
Ash 1.000
1/Starch 0.330 1.000
0.086
NDF −0.380 0.073 1.000
0.046 0.711
CP 0.525 0.576 −0.227 1.000
0.004 0.001 0.245
SolCP −0.478 −0.732 0.189 −0.606 1.000
0.010 <0.001 0.335 <0.001
ADIN 0.133 0.286 0.043 0.029 −0.409 1.000
0.499 0.140 0.828 0.882 0.031
Dig. N 0.555 0.369 −0.365 0.793 −0.433 −0.067 1.000
0.002 0.053 0.056 <0.001 0.021 0.736
IVOMD 0.399 0.412 −0.302 0.651 −0.596 0.332 0.558 1.000
0.035 0.029 0.118 <0.001 <0.001 0.085 0.002
1650
Table 3. Coefficients of linear correlation (r) between colour parameters (for each parameter, probability is presented in the second row)
X Y 1/Z L∗2 a∗ b∗2 eL a b2
X 1.000
Y 0.999 1.000
<0.001
1/Z −0.948 −0.947 1.000
<0.001 <0.001
L∗2 0.958 0.955 −0.899 1.000
<0.001 <0.001 <0.001
a∗ −0.678 0.711 0.657 −0.608 1.000
<0.001 <0.001 <0.001 <0.001
b∗2 0.348 0.336 0.100 0.344 −0.116 1.000
0.070 0.081 0.614 0.073 0.556
eL 0.999 0.999 0.949 0.954 −0.702 0.355 1.000
<0.001 <0.001 <0.001 <0.001 <0.001 0.064
a −0.529 −0.568 0.515 −0.462 0.981 −0.020 −0.555 1.000
0.004 0.002 0.005 0.013 <0.001 0.920 0.002
b2 0.582 0.573 −0.341 0.569 −0.321 0.959 0.587 −0.203 1.000
0.001 0.001 0.076 0.002 0.096 <0.001 0.001 0.301
Table 4. Coefficients of linear correlation (r) between chemical composition, digestible N,in vitroorganic matter digestibility (IVOMD) and colour parameters (for each parameter, probability is presented in the second row)
Ash 1/Starch NDF CP SolCP ADIN Dig. N IVOMD
X −0.072 0.097 −0.227 0.406 −0.083 −0.407 0.330 0.190
0.715 0.624 0.245 0.032 0.673 0.032 0.087 0.333
Y −0.046 0.110 −0.246 0.426 −0.106 −0.401 0.362 0.217
0.817 0.579 0.207 0.024 0.591 0.035 0.058 0.268
1/Z 0.053 −0.031 0.221 −0.372 0.042 0.479 −0.291 −0.115
0.790 0.875 0.258 0.051 0.830 0.010 0.133 0.560
L∗2 −0.123 0.125 −0.159 0.412 −0.119 −0.326 0.270 0.208
0.531 0.527 0.420 0.029 0.546 0.090 0.164 0.289
a∗ −0.426 −0.277 0.480 −0.658 0.449 0.208 −0.778 −0.546
0.024 0.153 0.010 <0.001 0.017 0.289 <0.001 0.003
b∗2 0.023 0.154 −0.037 0.288 −0.150 −0.010 0.164 0.087
0.908 0.435 0.850 0.137 0.446 0.958 0.404 0.661
eL −0.044 0.104 −0.241 0.431 −0.103 −0.408 0.360 0.206
0.825 0.597 0.216 0.022 0.604 0.031 0.060 0.293
a −0.491 −0.294 0.492 −0.637 0.487 0.126 −0.803 −0.587
0.008 0.129 0.008 <0.001 0.009 0.522 <0.001 0.001
b2 0.011 0.183 −0.107 0.366 −0.188 −0.105 0.249 0.170
0.957 0.351 0.588 0.055 0.337 0.596 0.202 0.387
RESULTS AND DISCUSSION
The results of this research indicate that there is a high degree of variability in the physical and nutritional properties among sources of CGF (Table 1) that are available to the feed industry, clearly illustrating the importance of its analysis, as this variation can cause misformulation and thus affect animal productivity.
Three colour scales were analysed in this study. In the CIE (1931) system,27X,YandZcorrespond respectively to red, green and blue.
BecauseXYZvalues are not easily understood in terms of object colour, other colour spaces have been developed. Nowadays, there are two popularL,a,bcolour scales in use: the HunterL,a,b and CIEL∗,a∗,b∗(CIELAB). Despite being similar in organization, a colour will have different numerical values in these colour spaces.
The Hunter colour procedure gives a score to a sample based on its
lightness, redness and yellowness. The HunterLscore ranges from 0 (black) to 100 (white). Positiveavalues are red, negative are green and 0 neutral. Positivebvalues are yellow, negative blue and 0 neutral. Hunter and CIELAB scales are both mathematically derived from theXYZvalues, but neither scale is visually uniform: Hunter L,a,bis over-expanded in the blue region of the colour space and CIELAB is over-expanded in the yellow region. The current CIE recommendation is to use CIELAB. In the present study, theL,L∗, X andYscores ranked the CGF samples essentially in the same order, as well asaanda∗scores, but theb,b∗andZscores did not (data not shown).
Correlations between chemical composition, digestible N, IVOMD and colour parameters are presented in Tables 2, 3 and 4. Soluble protein negatively correlated with CP and 1/starch
1651
Figure 1.Plot of factor 1versusfactor 2 from principal component factor analysis of chemical composition, digestible N,in vitroorganic matter digestibility and colour parameters (orthogonal rotation, varimax).
Table 5. Predictions of nutritive value of corn gluten feed according to linear fixed-effects regression models based on colour score as independent regression variables (n=28)
Independent variable Intercept X1 X2
Dependent variable X1 X2 Value SE P-value Value SE P-value Value SE P-value RMSEa r2
Ash (g kg−1DM) eL a 168.2 30.78 <0.001 −38.0 15.64 0.023 −5.4 1.36 <0.001 16.366 0.386 L∗2 a∗ 126.1 12.98 <0.001 −0.01 0.002 0.004 −4.3 1.05 <0.001 15.63 0.414
NDF (g kg−1DM) a∗ 328.8 25.84 <0.001 9.3 3.32 0.010 247.1 0.230
a 319.3 28.31 <0.001 12.5 4.36 0.008 243.5 0.242
CP (g kg−1DM) X Y 271.6 35.97 <0.001 −52.5 19.56 0.013 54.3 19.36 0.010 268.3 0.365
a∗ 356.1 24.93 <0.001 −14.3 3.20 <0.001 230.1 0.433
a 363.8 28.18 <0.001 −18.3 4.34 <0.001 241.2 0.406
Sol. CP (g kg−1CP) a∗ 192.2 85.90 0.034 28.3 11.0 0.017 2731.3 0.201
a 148.6 92.68 0.121 40.6 14.26 0.009 2608.8 0.237
ADIN (g kg−1N) 1/Z −13.2 19.91 0.513 736.8 264.65 0.010 333.7 0.230
eL 264.4 97.81 0.012 −139.4 61.08 0.031 361.0 0.167
Dig. N (kg kg−1N) L∗2 a∗ 1.15 0.070 <0.001 −0.001 0.001 0.035 −0.04 0.006 <0.001 0.001 0.670
a 1.04 0.039 <0.001 −0.04 0.006 <0.001 0.001 0.645
IVOMD (kg kg−1OM) a∗ 0.88 0.037 <0.001 −0.016 0.005 0.003 0.001 0.298
a 0.91 0.040 <0.001 −0.02 0.006 0.001 0.001 0.344
aRMSE, residual mean square error.
contents. Digestible N was positively correlated with ash and CP contents and negatively correlated with soluble protein, not being related to ADIN content.In vitroOM digestibility was positively correlated with ash, 1/starch and CP contents, and digestible N, and negatively correlated with soluble protein. The absence of a significant correlation between digestible N or IVOMD and ADIN agrees with previous results11and suggests that ADIN could be
partially digestible, as pointed by other authors.8,10,12In general, colour coordinates significantly correlated with each other.
Physical appearance (colour) was related to nutritional proper- ties, with the exception ofb2andb∗2scores.X,Y,L∗2andeLscores were positively related to CP content and negatively related to ADIN content. 1/Zwas positively correlated with ADIN. Cromwell et al.28have also observed, in dry distillers grain samples, a signifi-
1652
cant relationship between colour and ADIN content. In the present study,aanda∗scores were positively related to NDF and soluble protein contents, and negatively related to ash and CP contents, digestible N and IVOMD.
The principal component factor analysis appeared to provide a reasonable description of the data (Fig. 1). The first factor, explaining 44.3% of total variation, aggregated 1/Z and ADIN content in opposition to colour scores negatively related to ADIN content (X,Y,eLandL∗2). The second factor explained 21.6% of variation and aggregateda,a∗, NDF and soluble CP contents in opposition to ash and CP contents, and digestible N and IVOMD.
Table 5 presents the predictions of nutritive value of CGF according to linear fixed effects regression models based on colour score as independent regression variables. Relationships between colour scores and NDF, ADIN, soluble protein contents and IVOMD were much weaker than those obtained between a,a∗ andL∗2 and digestible N. The results suggest that colour could be effectively used to quickly predict the nutritive value of CGF, particularly digestible N – a key parameter when formulating rations.
CONCLUSIONS
A high variability in the physical and nutritional properties among 28 sources of CGF were found. TheL,L∗,XandYscores ranked the CGF samples essentially in the same order. The principal component factor analysis appeared to provide a reasonable description of the data. The first factor aggregated 1/Zand ADIN content in opposition to X, Y, eL and L∗2. The second factor aggregated a,a∗, NDF and soluble CP contents in opposition to ash and CP contents, and digestible N and IVOMD. Multiple regression analysis also showed that physical appearance (colour) was related to nutritional properties, stronger relationships being found betweena,a∗andL∗2scores and digestible N. The results suggest that colour could be effectively used as a guide for protein quality of CGF by the feed industry.
ACKNOWLEDGEMENT
The authors gratefully acknowledge Dr LM Cunha for his help with colour measurement.
REFERENCES
1 Hurrell RF and Carpenter KJ, The estimation of available lysine in foodstuffs after Maillard reactions.Prog Food Nutr Sci5:159–176 (1981).
2 Weiss WP, Conrad HR and Shockey WL, Amino acid profiles of heat- damaged grasses.J Dairy Sci69:1824–1836 (1986).
3 Weiss WP, Conrad HR and Shockey WL, Digestibility of nitrogen in heat-damaged alfalfa.J Dairy Sci69:2658–2670 (1986).
4 Mao L-C, Lee K-H and Erbersdobler HF, Effects of heat treatment on lysine in soya protein.J Sci Food Agric62:307–309 (1993).
5 Van Soest PJ, Fox DG, Mertens DR and Sniffen CJ, Discounts for net energy and protein, inProceedings of the Cornell Nutrition Conference for Feed Manufacturers(4th edn). Cornell University, Ithaca, NY, pp. 123–136 (1984).
6 Poos-Floyd M, Klopfenstein T and Britton RA, Evaluations of laboratory techniques for predicting ruminal protein degradation.J Dairy Sci 68:829–839 (1985).
7 Schroeder GE, Erasmus LJ, Leeuw K-J and Meissner HH, The use of acid detergent insoluble nitrogen to predict digestibility of rumen undegradable protein of heat processed plant proteins.S Afr J Anim Sci26:49–52 (1996).
8 Rogers JA, Conrad HR, Dehority BA and Grubb JA, Microbial numbers, rumen fermentation, and nitrogen utilization of steers fed wet or dried brewer’s grains.J Dairy Sci69:745–753 (1986).
9 Cleale RM IV, Klopfenstein TJ, Britton RA, Saterlee LD and Lowry SR, Induced non-enzymatic browning of soybean meal. III. Digestibility and efficiency of protein utilization by ruminants of soybean meal treated with xylose or glucose.J Anim Sci65:1327–1335 (1987).
10 Weiss WP, Erickson DO, Erickson GM and Fisher GR, Barley distillers grains as a protein supplement for dairy cows.JDairy Sci72:980–987 (1989).
11 Nakamura T, Klopfenstein TJ and Britton RA, Evaluation of acid detergent insoluble nitrogen as an indicator of protein quality in nonforage proteins.J Anim Sci72:1043–1048 (1994).
12 Britton RA, Klopfenstein TJ, Cleale R, Goedeken F and Wilkerson V, Methods of estimating heat damage in protein sources, in Proceedings of the 41st Distillers Feed Conference, p. 67 (1986).
13 Britton RA, Klopfenstein TJ, Cleale R, Goedeken F and Wilkerson V, ADIN as a measure of N digestibility in protein supplements (abstract).J Anim Sci65:465 (1987).
14 Nakamura T, Klopfenstein TJ, Gibb DJ and Britton RA, Growth efficiency and digestibility of heated protein fed to growing ruminants.J Anim Sci72:774–782 (1994).
15 DeSantos FA, Ramamoorthi L, Bechtel P, Smiley S and Brewer MS, Effect of salmon type and presence/absence of bone on color, sensory characteristics, and consumer acceptability of pureed and chunked infant food products.J Food Sci75:S279–S285 (2010).
16 Lehman RM and Rosentrater KA, Microbial development in distillers wet grains produced during fuel ethanol production from corn (Zea mays).Can J Microbiol53:1046–1052 (2007).
17 Rosentrater KA and Lehman RM, Physical and chemical properties of corn distillers wet grains.Appl Eng Agric24:57–62 (2008).
18 Rosentrater KA and Lehman RM, Predicting stability of distiller’s wet grains (DWG) with color analysis.Food Bioprocess Technol3:204–212 (2010).
19 AOAC,Official Methods of Analysis(15th edn). Association of Official Analytical Chemists, Arlington, VA (1990).
20 Van Soest PJ, Robertson JB and Lewis BA, Methods for dietary fiber, neutral detergent fiber and non-starch polysaccharides in relation to animal nutrition.J Dairy Sci74:3583–3597 (1991).
21 Robertson JB and Van Soest PJ, The detergent system of analysis and its application in human foods, inThe Analysis of Dietary Fiber in Foods, ed. by James WPT, Theander O. Marcel Dekker, New York, pp. 123–154 (1981).
22 Salomonsson A, Theander A and Westerlund E, Chemical character- ization of some Swedish cereal whole meal and bran fractions.
Swedish J Agric Res14:111–117 (1984).
23 Hovell FD, Ngambi JW, Barber WP and Kyle DJ, The voluntary intake of hay by sheep in relation to its degradability in the rumen as measured in nylon bags.Anim Prod42:111–118 (1986).
24 Mathis CP, Cochran RC, Vanzant ES, Abdelgadir, IEO, Heldt JS, Olson KC,et al, A collaborative study comparing anin situprotocol with single time-point enzyme assays for estimating ruminal protein degradability of different forages.Anim Feed Sci Technol93:31–42 (2001).
25 Tilley JMA and Terry RA, A two-stage technique for the in vitro digestion of forage crops.J Br Grassland Soc18:104–111 (1963).
26 Van Soest PJ, Wine RH and Moore JA, Estimation of the true digestibility of forages by thein vitro digestion of cell walls, in Proceedings of the Xth International Grassland Congress, University of Helsinki, Finland, pp. 438–441 (1966).
27 CIE, Commission Internationale de l’ ´Eclairage Proceedings, 1931.
Cambridge University Press, Cambridge, UK (1932).
28 Cromwell GL, Herkelman KL and Stahly TS, Physical, chemical, and nutritional characteristics of distillers dried grains with soluble for chicks and pigs.J Anim Sci71:679–686 (1993).
