Laboratory Assays and Near-Infrared Reflectance Spectroscopy for Estimates of Feeding Value of Corn Silage

Laboratory Assays and Near-Infrared Reflectance Spectroscopy for Estimates of Feeding Value of Corn Silage

Laboratory Assays and Near-Infrared Reflectance Spectroscopy for Estimates of Feeding Value of Corn Silage A. J. MOE and S. B. CARR Virginia Polytech...

528KB Sizes 2 Downloads 35 Views

Laboratory Assays and Near-Infrared Reflectance Spectroscopy for Estimates of Feeding Value of Corn Silage A. J. MOE and S. B. CARR

Virginia Polytechnic Institute and State University Blacksburg 24061 ABSTRACT


Corn silages samples from the Virginia Tech Forage Testing Program were classified by color, presence or absence of mold, aroma, and fermentation type. In the sample of 142 corn silages, in vitro digestible organic matter averaged 59.8% (range 38.5 to 74.7%) and indicated a range of silage quality. Simple and multiple regression of in vitro digestible organic matter on chemical constituents were derived for the full data set and within fermentation groups. Acid detergent fiber and dry matter explained the largest portion of the variation of digestibility. Simple regression on the full data set had coefficients of determination of .352 and .208 with standard errors of prediction of 5.29 and 5.86 for acid detergent fiber and dry matter. Multiple regression analyses improved coefficient of determination and standard error slightly. Corn silage in vitro digestible organic matter was predicted by near infrared reflectance techniques. A calibration equation resulted in coefficient of determination of .88 and standard error 5.38. The calibration equation was tested for prediction of an additional 90 corn silages. Regression of digestibility predicted by near infrared reflectance resulted in coefficient of determination of .37 and standard error of 4.04. Prediction of corn silage digestibility by near infrared reflectance or laboratory methods was similar.

Silage fermentation is important to preservation of forages with respect to feeding value and animal performance. Chemical and bacteriological changes in the silo during the fermentation process can affect adversely nutrient yield and quality (19, 24). Variations of digestibility of corn silage are small relative to other forages (11), and maximum digestibility corresponds roughly to maximum dry matter yield (13, 14). Highest digestibilities of dry matter and organic matter occurred at the milk, early dough, and dough dent stages of maturity (13), whereas maximum acre yield of corn silage dry matter occurred between dent and glaze stages (14). Prediction of feeding value of corn silage is desirable for least cost ration formulation; however, heterogeneity of corn silage and associative effects of rations complicate such predictions (26). Feeding value of corn silage remains relatively high over a wide range of maturities, because increases of grain content offset decreases of digestibility of plant components until the grain has reached physiological maturity (25). Research does not indicate the degree to which corn silage spoilage is reflected in changes of animal response. No differences in dry matter consumed, milk production, or body weight were observed when normal and late harvested corn silages were fed to dairy cows (10). Total digestible nutrients or digestibility of dry matter components were not different for soft and hard dough stages of maturity (12). However, silage intake and milk yield were increased with increasing maturity (12, 20). The relatively new technology of nearinfrared reflectance spectrophotometry (NIRR) has potential advantages over conventional laboratory analyses for determining chemica] composition and predicting feeding value of forages (6, 8, 9). The rapidity of analysis for several constituents or indicators of feeding value and limited sample preparation (i.e.,


August 13, 1984.

1985 J Dairy Sci 68:2220-2226


NEAR-INFRARED REFLECTANCE FOR CORN SILAGE drying and grinding to fine particle size) make NIRR an attractive alternative to conventional analysis. The accuracy of NIRR can be improved by increasing precision of laboratory procedures as demonstrated by (4). Limited wavelength selection and data processing capabilities reduce applications of some commercial instruments

(9). Numerous chemical and biological methods have been suggested for predicting forage quality. However, evidence is lacking to demonstrate superiority of different methods for predicting animal performance. Objectives were to test several laboratory methods and NIRR for predicting in vitro digestible organic matter (IVDOM) of corn silage. MATERIALS AND METHODS

Samples were selected from corn silages submitted from Virginia farms for analysis in the Virginia Tech Forage Testing Program. A screening procedure employing visual and olfactory characteristics was employed to select corn silage samples exhibiting deterioration and abnormal fermentations. The screening procedure placed silages into three broad classifications associated with known types of fermentation, clostridial, normal (i.e., lactic acid fermentation), and oxidized. Individual samples were rated for color, aroma, and presence or absence of mold. Silages were selected exhibiting black, green-yellow, and brown colors. Five aromas were distinguished including sour, pleasant, heated, no aroma (i.e., absence of a distinctive or discernable aroma), and musty. A fourth group of samples had silages from the Middleburg Forage Research Station and Dairy Center at Virginia Tech. This group of silages served as control for analytical comparisons. From approximately 300 g corn silage, 25 g was macerated with 125 ml distilled water in a Waring blender and filtered through four layers of cheesecloth. Silage pH was determined from this extract by pH meter. Duplicate 5-g samples of fresh silage were taken for each Kjeldahl nitrogen and volatile nitrogen analysis (2). Ammonia nitrogen content of the fresh sample was measured by distillation with carbonate-free magnesium oxide (2). Total nitrogen was determined by macro-Kjeldahl procedure (2). Fresh silage samples were chopped coarsely in a food processor (Moulinex Products Inc.,


Pennsauker, N J). The remainder of the original 300-g sample was dried in a forced air oven at 65°C for 48 h. Samples were ground to pass a 1-mm screen, and 100 ° dry matter (DM) was determined for correction of results to a DM basis. The IVDOM was determined by a method similar to the modification described by (5) of the Tilley and Terry procedure (21). The method differed from (5) in that residues from the incubation tubes were filtered through sintered glass Gooch crucibles with coarse porosity. Neutral detergent'fiber (NDF), acid detergent fiber (ADF), and chemically bound nitrogen (ADFN) were determined by methods of (22). Lactate was determined by the colorimetric procedure of (3). Potassium was measured by emission techniques (1). Digestible organic matter of corn silage was predicted with a Neotec Model FQAS1 Feed Quality Analyzer. The FQA51 is equipped with filters that permit measurements in selected regions of the near-infrared spectrum from 1.5 to 2.4 /a. Instrument capabilities are described by (7). Differences in optical density were determined from 89 pairs of wavelengths for six corn silages selected to represent the three fermentation groups. Areas of absorption are determined from portions of the filters as they rotate through the light source. For convenience, the area of absorption are termed pulse points and are defined as radii in the rotation of the wheel-mounted filters. Change of optical density (AOD) for each filter was used to determine pulse points where changes of reflectance occurred. Pulse points separating areas of the NIRR spectrum indicating minimum and maximum reflectance were selected and examined as possible areas of the spectrum where constituents related to digestibility could be measured. Ten corn silages from 38.5 to 74.7% IVDOM were employed as a calibration set to test various combinations of pulse points. Forward stepwise multiple regression of optical data (AOD) on IVDOM indicated reflectance related to IVDOM. The FQA51 can employ as many as six areas of the spectrum for regression analysis. Comparison of correlation coefficients and standard error for the prediction (Sy-x) for the 10 silage set determined pulse points to be used for further analysis. Prediction of IVDOM was Journal of Dairy Science Vol. 68, No. 9, 1985



further tested with a 26 sample calibration set and final regression equation determined from this calibration. The regression equation developed was tested by predicting IVDOM of 90 unknown corn silages. The 90 silages did not include any samples used in either calibration set. Procedures described in Statistical Analysis System, were employed for analysis of wet laboratory data. Packages included forward stepwise regression, all possible regressions, and general linear models. RESULTS A N D DISCUSSION

In vitro digestible organic matter was chosen as the index of feeding value to be predicted by laboratory analysis and NIRR. The IVDOM was preferred over in vitro dry matter digestibility because the former corrects for variation of insoluble ash content. In these comparisons, IVDMD (mean 67.7%) was closely correlated (r = .94) with IVDOM (mean 59.8%). Means for the chemical constituents measured for the entire data set (142 silages) are in Table 1. They agree with results for corn silage. However, lactate was considerably lower than in (10, 15, 17, 20). Lactate was low for most samples and not detected in 17 silages. It is possible that lactate was degraded to other metabolites during secondary fermentation while being shipped to the laboratory, Kempton and Clemente (16) have reported lactate is replaced by butyrate in the silo even after an original proliferation of lactic acid bacteria. In silages frozen shortly after sampling (standard fermentation group), mean lactate of 6.97% was substantially higher than in samples from the forage testing program. The wide range for all measurements was expected because of the screening procedure of sample selection. Means by fermentation group are in Table 1. The standard group provided a comparison of normal silages with other fermentation groups. The standard group comprised silages from the Virginia Tech Dairy Center and the Virginia Forage Research Station at Middleburg. These silages were considered normal in that they supported adequate growth rates of beef heifers and above average milk production of dairy cattle. Higher quality silage was indicated in the standard group by significantly higher IVDOM than selected silages. The selected groups Journal of Dairy Science Vot. 68, No. 9, 1985

(normal, clostridial, oxidized) were not significantly different from one another for IVDOM. Higher digestibility for the standard group indicates adequately preserved silage from well-eared corn and proper sampling procedures. Standard silages were significantly lower in ADF, which is consistent with digestibility results. The ADF is negatively correlated with digestibility, and lower ADF is expected for high quality silages (18). However, ADFN was significantly higher for the standard group. This response was an artifact associated with filtration of samples from the standard group. Slow filtration inflated ADFN. Substantial heat damage was indicated in the oxidized and even the normal fermentation groups. Within the normal and oxidized groups, ADFN were as high as 30.0% and 42.2%. Heat damage usually is considered a minor problem for corn silage. However, increased bound nitrogen could diminish feeding value. Lactate was significantly higher in the standard group. Mean lactate of 1.33% for the normal fermentation group, although lower than expected, was significantly higher than clostridial and oxidized groups. Standard and normal groups were not different in pH but were significantly lower than clostridial and oxidized silages, which is consistant with lactate. There were no differences among fermentation groups for NDF and crude protein. Ammonia nitrogen (NHaN) was significantly higher for the clostridial group, which could correspond to greater amino acid catabolism. Potassium was significantly higher for standard silages, which suggests either lower grain content or harvest of silage of an earlier maturity. Also, potassium is related to fertilization, which may explain part of this variation. The IVDOM was regressed on the 11 chemical constituents to test these laboratory methods for predicting feeding value. Results of simple linear regression are in Table 2. The Sy..x is defined as the mean square error of regression. The R 2 were not large; however, ranking of individual measurements by R 2 was similar to results of (18). The best predictor of IVDOM was ADF, which explained 35.2% of the variation with Sy.x of 5.29. Marten et al. (18) found ADF to be the best predictor of in vivo DM digestibility, accounting for 61% of the

TABLE 1. Mean c o m p o s i t i o n and standard deviations of corn silages by selected f e r m e n t a t i o n groups. Measurement ~


N u m b e r o f observations



Acid detergent fiber Acid d eterg ent fiber nitrogen (N), % of N Crude protein Dry matter, % Lactic acid Neutral detergent fiber pH Potassium A m m o n i a nitrogen, % of N In vitro digestible organic m a t t e r

Standard 142










23.1 a


29.8 b


31.1 b


31.4 b


15.4 8.2 40.9 1.7

6.6 1.2 9.6 2.3

20.1 a 8.6 40.3 a 7.0 a

9.10 1.05 4.48 1.89

12.6 b 8.0 38.2 a 1.3 b

4.43 .92 9.99 .98

14.4 bc 8.6 34.4 a .7 c

3.02 1.03 6.31 .78

16.7 c 8.2 45.6 b .5 c

6.90 1.54 9.42 .67

60.3 4.5 1.4

7.7 1.1 .5

59.5 3.9 a 1.8 a

6.28 .17 .70

57.5 4.2 a 1.3 b

7.11 .44 .28

60.0 5.0 b 1.4 b

5.44 1.30 .34

63,1 4.9 b 1.3 b

8.39 1.28 .55



7.4 a



10.3 b




64.0 a



59.1 b



9.4 ab 60.1 b

8.4 ab 58.2 b

4.63 7.22

,q Z O O Z e"

a'b'CMeans within rows not bearing a c o m m o n superscript differ (P<.05). E x c e p t where noted, percentage of dry matter. < o o, 0o

Z o xo 0o vl

b~ t~



TABLE 2. Simple linear regression of in vitro digestible organic matter (IVDOM) on laboratory measurements for 142 corn silages. Constituent~


Acid detergent fiber Acid detergent fiber nitrogen (N) Acid detergent fiber % of N Crude protein Dry matter, % Lactic acid Neutral detergent fiber pH Potassium Ammonia N Ammonia N % of N





-.802 a




-.180 a



62.3 61.0 47.1 59.0

-12.152 a - . 156 .311a .986

.031 .001 .208 .016

6.$ 6.6 5.9 6.5

68.0 58.5 62.7 60.5 60.6

- . 136 .282 -2.157 --.083 7.842

.025 .002 .026 .003 .007

6.5 6.6 6.5 6.6 6.6

aRegression coefficient is significant (P<.05). ExcePt where noted, percentage of dry matter.

variation. The next best predictor was DM, which explained 20.8% of the variation with Sy.x of 5.86. Correlation between ADF and DM was significant r = --.28. All other constituents explained less than 5% of variation. Fiber represents a relatively indigestible portion of forages (23), and ADF would be expected to have a significant negative correlation with digestibility. The relationship of DM to IVDOM is not as readily apparent. Corn silage DM increases with maturity and with increasing ratio of grain:plant (10, 13). Digestibility of corn silage tends to increase and then plateau during the milk, early dough, and dough dent stages of maturity (13), and increased DM could indicate later maturity with increased ratio of grain:plant and higher digestibility. These data demonstrate correlation between DM and digestibility but do not indicate cause and effect. Results of multiple regression (forward stepwise regression) are in Table 3: The two variable model, which included ADF and DM, explained 44.2% of the variation of IVDOM with a Sy.x of 4.93. There was little improvement of R 2 by more variables added and only ADF and DM had significant regression coefficients for all 11 models. The full model explained 47.6% of the variation of IVDOM with Sy.x of 4.94. These data indicate little Journal of Dairy Science Vol. 68, No. 9, 1985

advantage of .additional variables other than ADF and DM. The screening process to select abnormal corn silage samples resulted in a wide range of silage qualities and fermentations that can occur in corn silage. Under practical forage testing conditions, assays for predicting silage quality should be confined to relatively stable chemical constituents. Readily degradable or volatile compounds would be lost or vary with exposure rather than relate to fermentation in the silo. Results show ADF and DM were the most important estimators of corn silage IVDOM. Additional variables did not improve the prediction enough to warrant their inclusion. Near-Infrared Reflectance

An optimum correlation search employing 10 silages resulted in R 2 .943 and Sy.x of 4.36. Pulse points yielding these results were 265 to 285 and 590 to 600, which correspond to wavelengths 2.0472 to 2.0713 and 2.2360 to 2.2397/~. The former pair of pulse points are in the area where protein and starch absorb, and the latter are in the area of fiber and sugar absorption. A calibration set of 26 silages was selected to test the pulse points obtained. The pair 410 t o 420, which corresponds to wavelengths 2.1452

TABLE 3. Multiple regression of in vitro digestible organic m a t t e r (IVDOM) on l a b o r a t o r y m e a s u r e m e n t s for 142 corn silages. Regression coefficients by n u m b e r of variables in m o d e l Parameter 1























> Z


In tercep t Y 83.585 Acid detergent fiber -.802 a Acid detergent fiber nitrogen Acid detergent fiber nitrogen, % of N Crude protein Dry matter, % Lactic acid Neutral detergent fiber pH Potassium A m m o n i a nitrogen A m m o n i a nitrogen, % of N R2 .352 Sv. x 5.29

-.682 a

-.662 a

-.665 a

-.651 a

-4.467 .213 a

.224 a .898

.211 a .896

.216 a .822

-.632 a

-5.236 .220 a .678 .048

.442 4.93

.455 4.89

-.648 a

-.650 a

-.645 a

-.645 a

-.648 a






-.426 .211 a .777

- . 351 .211 a .795

4.368 -.480 .208 a .795

4.364 -.480 .208 a .795

.059 .393 -.645

.059 .382 -.665

.058 .383 -.749

.475 4.89

-3.160 .475 4.91

-3.256 .476 4.92

-.531 .217 a .763




.065 .424 -.424

.463 4.87

.466 4.88

.469 4.88

.473 4.88

.058 .383 -.749 .001 -3.290 .476 4.94

t~ r~

Z "n Q

Z E ~3

aRegression coefficient is significant (P<.05). t~ < o ro~ o0

1E x c e p t where noted, percentage of dry matter.

z b~ t~ b~



to 2.1566 /a, was also significant for this calib r a t i o n . T h e c a l i b r a t i o n w i t h t h r e e pairs o f pulse p o i n t s r e s u l t e d in c o r r e l a t i o n .879 and Sy. x 5.38, T h e r e f o r e , t h a t calibration was c h o s e n and t e s t e d for t h e p r e d i c t i o n o f I V D O M f o r 90 u n k n o w n corn silages. L a b o r a t o r y I V D O M was regressed o n N I R R p r e d i c t e d I V D O M w i t h an R 2 o f .367 and S y . x o f 4.04. This c o m p a r e s w i t h t h e p r e d i c t i o n b y A D F , R 2 o f .352, a n d S y . x o f 5 . 2 9 . Paired c o m p a r i s o n s showed NIRR underestimated laboratory IVDOM (t = 4.29, P < . 0 5 ) . The N I R R p r e d i c t i o n o f I V D O M was c o m parable t o t h e p r e d i c t i o n b y A D F . These d a t a suggest t h a t N I R R has p o t e n t i a l m e r i t f o r r a p i d screening o f c o r n silages. The r a p i d i t y o f determinations makes the NIRR prediction attractive.


1 Analytical Methods for Atomic Absorption Spectrophotometry. 1976. Perkin-Elmer. Norwalk, CT. 2 Association of Official Agricultural Chemists. 1975. Methods of analysis. 12th ed. Washington, DC. 3 Barker, S., and W. H. Summerson. 1941. The colorimetric determination of lactic acid in biological material. J. Biol. Chem. 138:535. 4 Barton, F. E., II, W. R. Windham, S. W. Coleman, and R. Meyer. 1982. An interlaboratory experiment comparing conventional and semi-automated fiber analyses for use in the calibration of NIR reflectance spectrometers. Page 3 in Proe. Am. Soc. Anim. Sci., So. Sect. Orlando, FL. 5 Barnes, R. F. 1970. In vitro dry matter and organic matter digestion. Page 5051 in Nutrition research techniques for domestic and wild-animals. Vol. 1. L. Harris ed. Publ. L. E. Harris, Logan, UT. 6 Barnes, R. F. 1973. Laboratory methods of evaluating feeding value of herbage, Page 179 in Chemistry and biochemistry of herbage. G. W. Butler and R. W. Bailey, ed. Academic Press, New York, NY. 7 Burdick, D., F. E. Barton, II, and B. D. Nelson. 1979. Rapid determination of forage quality with near-infrared filter spectrometer. Page 81 in Proc. 36th So. Pasture Forage Crop Improve. Conf., Beltsville, MD. 8 Dexter, S. T. 1966. Gas evolution from silage as a diagnostic tool for chemical reaction and energy losses. Agron. J. 58:383. 9 Gibson, T., A. C. Stifling, R. M. Keddie, and R. F. Rosenberger. 1958. Bacteriological changes in silage made at controlled temperatures. J. Gen. Microbiol. 19:112. 10 Gordon, C. H., J. C. Derbyshire, and P. J. Van

Journal of Dairy Science Vol. 68, No. 9, 1985

Soest. 1968. Normal and late harvesting of corn for silage. J. Dairy Sci. 51:1258. 11 Hillman, D. 1968. Supplementing corn silage. Sympos. Dairy Cattle Feeding. 63rd Annu. Mtg. Am. Dairy Sci. Assoc., Columbus, OH. 12 Huber, J. T., G. C. Graf, and R. W. Engel. 1965. Effect of maturity on nutritive value of corn silage for lactating cows. J. Dairy Sci. 48:1121. 13 Johnson, R. R., and K. E. McClure. 1968. Corn plant maturity IV. Effects on digestibility of corn silage in sheep. J. Anim. Sci. 27:535. 14 Johnson, R. R., K. E. McClure, L. J. Johnson, E. W. Klosterman, and G. B. Triplett. 1966. Corn plant maturity. I. Changes in dry matter and protein distribution in corn plants. Agron. J. 58:151. 15 Jorgensen, N. A., H. J. Larsen, R. E. Roffler, J. M. Sund, G. P. Barrington, and R. P. Niedermeier. 1969. High energy corn silage for lactating dairy cattle. Res. Rep. 42, Coll. Agric, Life Sci. Univ. Wis. 16 Kempton, A. G., and C. L. San Clemente. 1959. Chemistry and microbiology of forage-crop silage. Appl. Microbiol. 7:362. 17 Klosterman, E. W., R. R. Johnson, H. W. Scott, A. L. Moxon, and J. V. Stavern. 1960. Whole plant and ground ear corn silages, their acid content, feeding value, and digestibility. J. Anita. Sci. 19:522. 18 Marten, G. C., R. D. Goodrich, A. R. Schmid, J. C. Meiske, R. M. Jordan, and J. G. Linn. 1975. Evaluation of laboratory methods for determining quality of corn and sorghum silages:II. Chemical methods for predicting in vivo digestibility. Agron. J. 67:247. 19 McDonald, P., and R. Whittenberg. 1973. The ensilage process. Page 33 in Chemistry and biochemistry of herbage. G. W. Butler and R. W. Bailey, ed. Academic Press, Lond. 20 Owens, M. J., M. A. Jorgensen, and H. H. Voetker. 1968. Feeding value of high dry matter corn silage for dairy cattle. J. Dairy S ci. 51:1942. 21 Tilley, J.M.A., and R. A. Terry. 1963. A two-stage technique for the in vitro digestion of forage crops. J. Br. Grassl. Soc. 18:104. 22 Van Soest, P. J. 1965. Use of detergents in analysis of fibrous feeds. III. Study of effects of heating and drying on yield of fiber and lignin in forages. J. Assoc. Offic. Agric. Chem. 48:785. 23 Van Soest, P. J. 1967. Development of a comprehensive system of feed analyses and its application to forages. J. Anim. Sci. 26:119. 24 Vetter, R. L., andK. N. Von Glan. 1978. Abnormal silages and silage related disease problems. Page 281 in Fermentation of silage: A review. M. E. McCullough, ed. 25 Weaver, D. E., C. E. Coppock, G. B. Lake, and R. W. Everett. 1978. Effect of maturation on composition in vitro dry matter digestibility of corn plant parts. J. Dairy Sci. 61:1782. 26 Woody, H. D., D. G. Fox, and J. R. Black. 1983. Predicting net energy value of corn silage varying in grain content. J. Anim. Sci. 57:710.