Search
Author
Title
Vol.
Issue
Year
1st Page

Journal of Animal Science - Animal Nutrition

In vivo indices for predicting acidosis risk of grains in cattle: Comparison with in vitro methods1

 

This article in JAS

  1. Vol. 91 No. 6, p. 2823-2835
     
    Received: Apr 15, 2012
    Accepted: Feb 27, 2013
    Published: November 25, 2014


    2 Corresponding author(s): ianl@sbscibus.com.au
 View
 Download
 Share

doi:10.2527/jas.2012-5379
  1. I. J. Lean*†,
  2. H. M. Golder 2,
  3. J. L. Black,
  4. R. King§ and
  5. A. R. Rabiee*†
  1. Faculty of Veterinary Science, The University of Sydney, Camden 2570, Australia
    SBScibus, Camden 2570, Australia
    John L Black Consulting, Warrimoo 2774, Australia
    Dairy Australia, Southbank 3006, Australia

Abstract

Our objective was to evaluate a near-infrared reflectance spectroscopy (NIRS) used in the feed industry to estimate the potential for grains to increase the risk of ruminal acidosis. The existing NIRS calibration was developed from in sacco and in vitro measures in cattle and grain chemical composition measurements. To evaluate the existing model, 20 cultivars of 5 grain types were fed to 40 Holstein heifers using a grain challenge protocol and changes in rumen VFA, ammonia, lactic acids, and pH that are associated with acidosis were measured. A method development study was performed to determine a grain feeding rate sufficient to induce non-life threatening but substantial ruminal changes during grain challenge. Feeding grain at a rate of 1.2% of BW met these criteria, lowering rumen pH (P = 0.01) and increasing valerate (P < 0.01) and propionate concentrations (P = 0.01). Valerate was the most discriminatory measure indicating ruminal change during challenge. Heifers were assigned using a row by column design in an in vivo study to 1 of 20 grain cultivars and were reassigned after a 9 d period (n = 4 cattle/treatment). The test grains were dry rolled oats (n = 3), wheat (n = 6), barley (n = 4), triticale (n = 4), and sorghum (n = 3) cultivars. Cattle were adapted to the test grain and had ad libitum access to grass silage 11 d before the challenge. Feed was withheld for 14 h before challenge feeding with 0.3 kg DM of silage followed by the respective test grain fed at 1.2% of BW. A rumen sample was taken by stomach tube 5, 65, 110, 155, and 200 min after grain consumption. The rumen is not homogenous and samples of rumen fluid obtained by stomach tube will differ from those gained by other methods. Rumen pH was measured immediately; individual VFA, ammonia, and D- and L-lactate concentrations were analyzed later. Rumen pH (P = 0.002) and all concentrations of fermentation products differed among grains (P = 0.001). A previously defined discriminant score calculated at 200 min after challenge was used to rank grains for acidosis risk. A significant correlation between the discriminant score and the NIRS ranking (r = 0.731, P = 0.003) demonstrated the potential for using NIRS calibrations for predicting acidosis risk of grains in cattle. The overall rankings of grains for acidosis risk were wheat > triticale > barley > oats > sorghum.



INTRODUCTION

Ruminal acidosis is a costly disorder of ruminants that is expressed over a continuum of severity ranging from mild to acute. It is induced by feeding large amounts of rapidly fermentable substrates, primarily from the nonstructural carbohydrate content of forages and grains (RAGFAR, 2007; Enemark, 2008). The potential to screen grains fed to livestock for acidosis risk is 1 method that may reduce the risk of acidosis in cattle. The intent of this study was to compare animal measures of acidosis with an existing near-infrared reflectance spectroscopy (NIRS) method (Black, 2008) developed from in sacco and in vitro measures from cattle and grain chemical composition analysis to rank grains for risk of acidosis. Results from this method can be compared with rankings based on in vivo rumen fermentation measures of acidosis based on the findings of Bramley et al. (2005, 2008), who demonstrated associations between cattle categorized as “acidotic” with prevalence of lameness, reduced milk fat to protein ratios, and diets greater in nonfiber carbohydrates and lower in NDF. A robust NIRS method could be used to provide rapid and accurate information about the risk of acidosis associated with feeding different grains.

Our first objective was to determine the amount of grain required to create a moderate ruminal challenge in dairy cattle in a method development study. Our second objective was to rank 20 cultivars from 5 grain types for their acidotic risk when fed to cattle using 1) an existing NIRS model developed from in sacco and in vitro measures in cattle and grain chemical composition analysis and 2) in vivo rumen fermentation parameters measured during a grain challenge. The third objective was to compare rankings of 20 grain cultivars derived from the existing NIRS model and the in vivo study. We hypothesized that the NIRS and in vivo estimates of acidosis risk obtained from feeding the same 20 grains would be significantly correlated and support the use of NIRS analysis.


MATERIALS AND METHODS

All experimental procedures were approved by the Bovine Research Australasia Animal Ethics Committee.

Test Grains

A total of 20 different dry rolled grain cultivars from the 2005 to 2006 harvest were used for comparisons between the NIRS and in vivo methods for ranking grains for potential acidotic risk. The test grains were dry rolled oats (n = 3; Dalyup 5817, Dalyup 5818, and Swan), wheat (n = 6; Bellaroi, Sentinal, Ammrock, Kellalac, unknown, and Chara), barley (n = 4; Tantangara, Gairdner 3864, Gairdner 3862, and Binnalong), triticale (n = 4; Maiden, Jackie 6823, Jackie 6824, and Prime 322), and sorghum (n = 3; Liberty, Pacer, and MR43) cultivars. A control was also included consisting of a blend of the 20 grain cultivars in equal proportions.

The 20 grains and mixed control were analyzed for these chemical components: NDF [Royal Australian Chemical Institute (RACI, 1995) method 03-02], CP [Dumas nitrogen with nitrogen value × 6.25; AOAC (1995) method 4.2.04], ADF [AOAC (1995) method 4.6.03], total starch [Megazyme amyloglucosidase/α-amylase method; American Association of Cereal Chemists (AACC, 1976) method 76.13; McCleary et al., 1997; AOAC (2000) method 996.11), total insoluble nonstarch polysaccharides (NSP), total soluble NSP, arabinoxylans, and β-glucans [McCleary method incorporating American Association of Cereal Chemists (AACC, 1976) method 32-23; European Brewery Convention (EBC, 1998) methods 3.11.1, 4.16.1 and 8.11.1; AOAC (2000) method 995.16; Table 1.


View Full Table | Close Full ViewTable 1.

Mean chemical composition of the 20 grains and control (mix of equal proportions of all 20 grain cultivars)

 
Grains NDF, % of DM CP, % of DM ADF, % of DM Total starch, % of DM Total insoluble NSP1, % of DM Total soluble NSP1, % of DM Arabinoxylans, % of DM Beta-glucans, % of DM
Control (mixed) 17.6 13.5 6.80 62.7 12.0 1.20 6.70 1.40
Barley (Binnalong) 24.1 19.4 6.20 56.5 10.2 3.50 6.30 3.90
Barley (Gairdner 3862) 19.4 9.40 5.10 66.0 9.01 2.80 5.30 3.60
Barley (Tantangara) 19.5 11.5 4.70 64.7 8.70 3.20 5.10 4.10
Barley (Gairdner 3864) 20.8 15.6 4.50 61.2 8.90 3.50 5.70 4.20
Oats (Swan) 32.6 13.2 16.8 36.1 27.5 2.80 14.4 3.00
Oats (Dalyup 5817) 33.6 11.7 16.1 37.3 26.0 2.60 14.1 3.20
Oats (Dalyup 5818) 26.2 11.2 13.0 42.7 24.9 2.70 15.0 3.40
Sorghum (Liberty) 13.4 13.5 5.40 79.3 6.30 0.00 2.10 0.00
Sorghum (Pacer) 11.4 13.1 5.60 80.1 4.20 0.00 1.90 0.00
Sorghum (MR43) 13.1 13.9 5.20 75.4 5.80 0.00 1.10 0.00
Triticale (Maiden) 15.3 18.2 3.50 65.2 7.70 0.20 4.70 0.10
Triticale (Jackie 6823) 8.60 12.1 3.10 72.3 6.10 0.60 4.80 1.50
Triticale (Jackie 6824) 13.9 14.1 2.70 69.1 7.00 0.60 4.20 0.40
Triticale (Prime 322) 13.3 11.2 3.20 71.2 7.70 0.40 5.30 0.30
Wheat (Bellaroi) 13.6 20.8 3.20 66.3 7.50 0.00 5.50 0.00
Wheat (Sentinal) 14.1 14.3 2.70 70.3 6.40 0.60 4.20 0.60
Wheat (Ammrock) 11.5 15.9 2.30 70.8 6.00 0.40 3.90 0.50
Wheat (Kellalac) 11.0 14.6 2.50 70.5 6.50 0.90 4.10 1.00
Wheat (unknown) 12.0 10.9 3.20 71.7 8.10 0.60 5.30 0.30
Wheat (Chara) 12.1 12.4 2.50 73.0 6.80 0.90 4.70 0.50
1NSP = nonstarch polysaccharide.

Acidosis Indices

The NIRS-derived acidosis index was used to rank the 20 grain cultivars for potential acidotic risk on a scale of 1 to 20 with 1 indicating the greatest risk. The index was calculated using an algorithm consisting of a combination of in sacco starch disappearance, in vitro total acid and lactic acid production, and grain starch and NDF content results reported by Black (2008) as follows:The increase in lactic acid during the in vitro fermentation was described as “net” production because no measurement was made of lactic acid degradation by microbes. The mean values with SE for terms used to calculate acidosis index were 0.55 ± 0.01 for 6 h starch disappearance, 41.2 ± 0.54 for starch content, 15.2 ± 0.20 for total acid production, and 3.70 ± 0.14 for net lactic acid production. The acidosis index was calculated by dividing each value by the greatest value expressed as a percentage.

The in sacco and in vitro fermentation studies were conducted on 92 cereal grains varying in cultivar and/or growing conditions. There were 23 wheat samples, 32 barley samples, 20 oat samples, 9 triticale samples, 7 sorghum samples, and 1 maize sample.

The in sacco assay was based on the methods of Ørskov et al. (1980). In brief, a 5-g (as-fed) sample of rolled grain was put into an artificial fiber bag (45 to 50 µm pore size) and suspended in the rumen of a steer for 6 h. There were 6 replicates for each grain, with the same grain being incubated in the rumen of 3 steers on 2 separate occasions. The steers were fed an amount sufficient to maintain BW on a diet containing 10% rolled oat grain, 10% rolled wheat grain, 10% rolled barley grain, 10% rolled maize grain, 10% rolled sorghum grain, 48% chopped (∼40 mm) wheat straw, and 2% mineral mix. The grains were rolled in a commercial roller mill with a gap between the rollers sufficient to crack all grains without fine crushing. The larger grain particles obtained through roller milling compared with laboratory milling greatly reduce the loss of starch particles from the bags when they are inserted into the rumen. The rate of DM and starch fermented over this period was estimated from the loss of DM and starch from the bag. A “blank” sample bag containing grain soaked in water for 6 h was also tested to determine the amount of material that was washed out of the bag.

The in vitro assay was conducted by incubating at 39°C for 5 h 30 g of finely milled grain sample (0.5- to 1-mm screen) in a 1 L flask containing 125 g rumen fluid and 375 g of McDougall’s buffer (McDougall, 1948), which contained 1.2 g/L urea. The assay was conducted as an incomplete block design with 4 replicates of the 92 treatments arranged in 4 blocks, each with 2 batches. Within each batch there were 2 water baths with 23 flasks per batch. The rumen fluid was collected from the same steers used for the in sacco assay but at a different time. Estimates of starch fermentation over the 5-h period were determined by measuring the difference between the starch content in the original grain and starch content remaining in the flask at 5 h. Similarly, total acid and lactic acid production was measured at the end of the 5-h incubation period by subtracting the initial acid values from the final values.

Starch was measured by the amyloglucosidase-α-amylase method (McCleary et al., 1997). Neutral detergent fiber was determined by the method of Van Soest et al. (1991). Parameters for the acidosis index prediction were derived from simulations with a rumen function model (Nagorcka et al., 2000) and information from Defoor et al. (2002). The NIRS spectra were collected in reflectance mode (log 1/R) from cereal grains using both whole and laboratory milled samples. Whole grain samples were scanned twice using the sample transport module in a model 6500 scanning monochromator (FOSS NIRSystems, Silver Spring, MD) and the mean spectra obtained. Milled grain samples were scanned once in small ring cups using the spinning sample module in a model 5000 scanning monochromator (FOSS). Both instruments had been previously spectrally matched.

WinISI software, version 3 (FOSS), was used to pretreat the spectral data using “standard normal variate” and “detrend” options. Calibrations for acidosis index were derived using modified partial least squares regression and a second derivative math treatment. The spectral range used for the whole samples was from 700 to 2,498 nm and for the milled samples 1,100 to 2,498 nm. In both cases, 3 outlier samples were omitted.

The statistics for the NIRS calibrations for acidosis index based on whole grain and laboratory milled grain scans are given in Table 2. The whole grain scan provided the better calibration with a 1-variance ratio of 0.86 (Table 2), a SE of cross-validation (SECV) of ± 8.46, an accuracy of prediction with 95% confidence of ± 16.6% units, and a ratio of deviation to prediction (RPD) of 2.66. The RPD value (SD/SECV) is an indicator of the reliability and robustness of the calibration and number between 2.5 and 3.0 suggest that the calibration is generally good. The relationship between the NIRS predicted based on whole grain scans and calculated acidosis index values is shown in Fig. 1.


View Full Table | Close Full ViewTable 2.

Statistics for near-infrared reflectance spectroscopy (NIRS) calibrations for acidosis index developed from whole and milled grain scans1

 
Calibration n Mean SD RSQ SEC 1-VR SECV RPD
Whole grain scans 86 44.8 22.5 0.93 7.02 0.86 8.46 2.66
Milled grain scans 86 44.8 22.5 0.86 8.55 0.82 9.62 2.34
1Mean = the mean predicted acidosis index value (% units); RSQ = R2 values – fraction of the variance accounted for by the NIRS calibration when all accepted observations are included in the relationship; SEC = SE of the calibration; 1-VR = 1-variance ratio – fraction of variance accounted for in NIRS prediction when some observations are used for “cross-validation” of the calibration as determined by the NIRS software; SECV = SE of cross-validation – SE of the calibration when some observations are used for “cross-validation” of the calibration as determined by the NIRS software; RPD = ratio of prediction to deviation = SD/SECV, an indication of the value of the calibration.
Figure 1.
Figure 1.

Relationship between the calculated acidosis index and values predicted using the first near-infrared reflectance (NIR) spectroscopy calibration, which was based on NIR spectroscopy of whole grain scans. The fine solid line represents the line of equivalence and the dotted lines represent the SE of cross-validation ± 8.46% units. The heavier solid line is the linear regression (y = 0.96x + 1.83).

 

Method Development Study

Eight Holstein heifers, 18 mo of age (410 to 650 kg), and 8 Holstein nonlactating cows (650 to 800 kg) were randomly selected for the trial. All grains were processed using a roller mill to provide an effective crush to industry standard and screen sizes were recorded (Table 3). The cattle were allocated into 1 of 2 dietary challenge groups, mixed grain (control) or triticale cultivar Jackie (n = 8 animals/group). Cattle were fed 1 kg of mixed grain daily with ad libitum ryegrass silage (Table 4) for a 7-d preadaptation period followed by a 5-d period when either 1 kg of rolled mixed grains (n = 8) or 1 kg of rolled triticale (n = 8) were fed daily with ad libitum access to ryegrass silage (Table 4). Cattle were then withheld from all feed for a period of 14 h before challenge. On the challenge day all cattle were fed first with 1 kg of ryegrass silage to reduce the saliva contamination during the rumen sampling and immediately after their allocated challenge diets of the test grains. Each grain cultivar was fed at 0, 0.4, 0.8, or 1.2% of BW (n = 4 heifers/rate). The control group received a blend of the 20 grain cultivars in equal proportions consisting of a mixture of oat, wheat, barley, triticale, and sorghum cultivars. Approximately 5 min after ingestion of test ration an initial rumen sample was collected using a stomach tube and custom-designed stomach pump. A second rumen sample was collected 1 h later and every 45 min for a subsequent 3 samples. The stomach tube was approximately 4 m in length and 19 mm in diameter with an aluminum multiholed probe inserted at 1 end to act as a filter. All rumen fluid samples were tested for saliva contamination as described by Bramley et al. (2008) and any contaminated samples were discarded. Rumen pH was measured immediately in the unprocessed rumen fluid using a pocket pH meter (pHTestr 30; Oakton Instruments, Vernon Hills, IL). Rumen fluid was then centrifuged at 1,512 × g for 15 min at 15°C and the supernatant was removed and stored at –20°C for VFA, ammonia, and lactic acid analyses.


View Full Table | Close Full ViewTable 3.

Proportion of different particle sizes of rolled grains

 
Grains 3.5-mm sieve, % 2.0-mm sieve, % <2.0-mm sieve, %
Oats (Dalyup 5817) 99.2 0.4 0.4
Oats (Dalyup 5818) 99.2 0.4 0.4
Oats (Swan) 99 0.5 0.5
Barley (Tantangara) 97 2 1
Wheat (Bellaroi) 30 60 10
Wheat (Sentinal) 60 35 5
Wheat (Ammrock) 65 30 5
Wheat (Kellalac) 77 20 3
Wheat (unknown) 65 32 3
Wheat (Chara) 82 15 3
Barley (Gairdner 3864) 98 1 1
Barley (Gairdner 3862) 98 1 1
Barley (Binnalong) 97 2 1
Triticale (Maiden) 92 7 1
Triticale (Jackie 6823) 91 8 1
Triticale (Jackie 6824) 89 10 1
Triticale (Prime 322) 84 15 1
Sorghum (Liberty) 4 94 2
Sorghum (Pacer) 19 80 1
Sorghum (MR43) 10 80 10
Control mixed grain 86 11 3

View Full Table | Close Full ViewTable 4.

Chemical composition of ryegrass silage

 
Method development study
In vivo study
Item, % of DM Adaptation Challenge Adaptation Challenge
DM 41.8 39.0 57.9 52.9
NDF 57.4 54.1 52.8 52.3
CP 14.2 15.9 17.5 16.0
Soluble protein, % of CP 56.8 63.0 49.8 52.5
Crude fat 3.90 4.60 3.95 4.33
Ash 8.95 10.7 13.5 14.1
Ca 0.58 0.80 0.77 0.79
P 0.32 0.40 0.26 0.29
Mg 0.22 0.20 0.29 0.25
K 2.96 3.40 2.42 3.05
S 0.21 0.20 0.27 0.25
Cl 1.16 1.40 1.33 1.82
ADF 37.8 36.1 36.2 36.0
Acid detergent insoluble CP 1.03 1.00 1.30 1.53
Neutral detergent insoluble CP 2.77 2.60 4.10 3.35
Lignin 5.03 4.60 5.63 5.00
Nonfiber carbohydrates 18.4 17.3 16.4 16.7
Nonstructural carbohydrates 11.1 8.50 10.2 12.3
ME, MJ/kg 7.77 9.10 9.15 9.08
Available protein 13.1 14.9 16.2 14.5
Adjusted CP 14.2 15.9 17.5 15.5
Degradable protein, % of CP 70.3 70.0 66.0 67.8
Starch 2.27 2.80 1.63 2.28
Sugar 8.90 5.60 8.60 9.95
Lys 0.46 0.50 0.58 0.53
Met 0.17 0.20 0.21 0.20

In vivo Study

Forty Holstein heifers (230 to 500 kg) were assigned to 1 of 20 grain cultivars in a randomized controlled clinical trial (run 1). The study was formulated as row–column design with 40 rows (cow) and 2 columns (run). This design maximized the number of treatments in each day and allowed for good estimation of the between grain treatment effects. Each individual heifer had equal chance to be selected in each run. Cattle were then randomly reallocated after a 9 d wash-out period to a different test grain (run 2; n = 4 heifers/test grain). Study personnel were blinded to the NIRS ranking of the 20 grains. Cattle were challenged in 4 groups of 10 on different days for ease of management in both run 1 and 2. During the 9 d wash-out period cattle were fed ryegrass (Lolium multiflorum) silage and 1 kg of triticale cultivar Jackie to prevent confounding of the results in run 2 from previous test grains fed in run 1.

Cattle were accustomed to the feeding system using a preadaptation period of 7 d when cattle were fed 0.5 kg of triticale cultivar Jackie and 0.5 kg of the test grain with ad libitum ryegrass silage (Table 4) access. For a subsequent 4 d period 1 kg/d of test grain and ryegrass silage with a mean NDF % of DM of 52.75 was fed. Cattle were withheld from all feed for 14 h before the day of challenge. On the challenge day cattle were fed 0.3 kg of ryegrass silage with a mean NDF % of DM of 52.33 and then 1.2% of BW of the test grain. Rumen samples were collected according to the procedures used in the method development study. Grain refusal was minimal with the exception of 1 animal in each of these groups: sorghum (Pacer), sorghum (MR43), oats (Dalyup 5817), and barley (Binnalong). Refusals from sorghum (Pacer) and sorghum (MR43) groups were administered by drenching the grain with 0.5 to 2 L of water.

Laboratory Analysis

A ryegrass silage sample was collected from 3 different silage bales as they were opened during the adaptation period of the method development study and results were pooled (Table 4). One ryegrass silage sample was taken during the challenge period of the method development study (Table 4). A ryegrass silage sample was collected from each freshly opened bale during the adaptation period for both run 1 (2 samples) and run 2 (2 samples) of the in vivo study, totaling 4 samples. A ryegrass silage sample was taken the day before each challenge day and on each challenge day and pooled providing 4 pooled samples for each run of the study. All ryegrass silage samples were frozen at –20°C before NIRS (AOAC, 2000) and wet chemistry analysis at George Weston Technologies (Sydney, NSW, Australia). Wet chemistry techniques were as follows: DM [AOAC (2000) method 930.15], NDF (Van Soest et al., 1991), CP [AOAC (2000) method 990.03], soluble protein (Cornell sodium borate-sodium phosphate buffer procedure), crude fat [AOAC (2000) method 2003.05], ash [AOAC (2000) method 942.05], lignin [AOAC (2000) method 973.18], ADF [AOAC (2000) method 973.18], acid and neutral detergent insoluble crude protein (ADICP and NDICP; Leco Tru-Mac N Macro Determinator; Leco Corp., St. Joseph, MI), starch (YSI 2700 SELECT Biochemistry Analyzer; YSI Inc., Yellow Springs, OH), water soluble carbohydrates (Hoover and Miller-Webster, 1998), and ethanol soluble carbohydrates (Hall et al., 1999). The nonfiber carbohydrate (NFC) was calculated as NFC = 100 – [(NDF – NDICP) + CP + crude fat + ash]. The minerals were analyzed by inductively coupled plasma–optical emission spectroscopy (George Weston Technologies). Results from the adaptation periods during the in vivo study were pooled as were results from the challenge periods (Table 4).

Rumen VFA concentrations were analyzed by gas chromatography using an Agilent series gas chromatograph with HD6890 injection, 30 by 0.53 mm by 1.0 μm capillary column (Agilent Technologies Inc., Wilmington, DE), and Chemstation software (Agilent Technologies Inc.). The interassay CV for propionate, acetate, isobutyrate, butyrate, isovalerate, valerate, and caproate were 3.8, 5.1, 3.2, 4.0, 3.4, 3.7, and 3.2%, respectively. Ammonia concentrations were analyzed using a Boehringer Mannheim kit (catalog number 11112732; Arrow Scientific, Lane Cove, Australia). D- and L-lactate concentrations were analyzed using a Boehringer Manheim kit (catalog number 11112821; Arrow Scientific).

Statistical Analysis

Method Development Study.

A GLM with repeated measures (SPSS, version 12.0; Apache Software, Forest Hill, MD) was used to analyze the effects of time, grain rate, and grain type and their interactions on rumen pH and concentrations of VFA, D- and L-lactate, and ammonia (Table 5). Estimated marginal means ± SE were calculated using the GLM and a Tukey’s honestly significant difference test was performed to distinguish between the treatment means (Table 6).


View Full Table | Close Full ViewTable 5.

Significance (P-values) for effect of time1, grain rate2, type of grain (mixed vs. triticale), and their interactions on rumen fermentation products in the method development study (n of heifers = 16; n of samples = 80)

 
Item Time (T) Rate (R) Grain (G) T × R T × G R × G T × R × G
Rumen pH 0.77 0.01 0.35 0.28 0.50 0.40 0.84
Total VFA 0.80 0.18 0.20 0.02 0.78 0.74 0.39
Acetate 0.43 0.32 0.20 0.06 0.74 0.75 0.33
Propionate 0.98 0.01 0.07 <0.01 0.92 0.84 0.51
Butyrate 0.32 0.37 0.46 0.03 0.89 0.60 0.64
Isobutyrate 0.91 0.14 0.94 0.56 0.57 0.78 0.29
Valerate <0.01 <0.01 0.05 <0.01 0.05 0.89 0.66
Isovalerate 0.95 0.29 0.87 <0.01 0.41 0.70 0.70
Caproate 0.86 0.66 0.74 0.21 0.76 0.67 0.80
Acetate:propionate 0.24 0.06 0.62 0.09 0.84 0.95 0.48
D-lactate 0.08 0.40 0.83 0.35 0.77 0.64 0.95
L-lactate 0.06 0.66 0.66 0.72 0.37 0.42 0.52
Ammonia 0.04 0.20 0.27 <0.01 <0.01 0.79 0.23
1Rumen samples were collected 5, 65, 110, 155, and 200 min after diet consumption.
2Grain was fed at 0, 0.4, 0.8, or 1.2% of BW.

View Full Table | Close Full ViewTable 6.

Estimated marginal means ± SEM of rumen fermentation products over 5 time periods (5, 65, 110, 155, and 200 min after diet consumption) for cattle fed 2 grains (mixed vs. triticale) at 4 rates in the method development study

 
Rate
Grain
Item Control (0 kg) 0.4% of BW (2 to 3 kg) 0.8% of BW (4 to 6 kg) 1.2% of BW (6 to 8 kg) SEM Control (mixed grain) Triticale Jackie SEM
No. of heifers 4 4 4 4 8 8
Rumen pH 7.21a 7.12ab 7.03b 6.99b 0.03 7.07 7.08 0.02
Variable, mM
Total VFA 59.1 62.0 74.7 76.1 5.45 66.2 71.6 3.85
Acetate 45.1 45.6 54.0 54.5 4.15 48.0 52.3 2.93
Propionate 7.99a 8.85a 11.6b 12.8b 0.74 10.1 11.0 0.53
Butyrate 4.24 5.67 6.52 6.20 0.75 5.78 5.92 0.53
Isobutyrate 0.53 0.49 0.62 0.63 0.04 0.58 0.57 0.03
Valerate 0.31a 0.46ab 0.81bc 0.88c 0.07 0.61A 0.69B 0.05
Isovalerate 0.76 0.78 0.98 1.05 0.10 0.93 0.90 0.07
Caproate 0.08 0.12 0.17 0.12 0.04 0.13 0.13 0.03
Acetate:propionate 5.67 5.22 4.64 4.28 0.26 4.80 4.89 0.19
L-lactate 0.16 0.08 0.07 0.01 0.12 0.06 0.07 0.09
D-lactate 0.13 0.12 0.16 0.08 0.07 0.13 0.11 0.05
Ammonia 3.96 4.47 6.75 6.09 0.85 5.15 5.78 0.60
a−cMeans of rates with different superscripts are significantly different P = 0.05.
A,BThe mean of the control is significantly different from the mean of triticale cultivar Jackie.

The estimated marginal means from all time periods were used to rank the 4 different grain rates and 2 types of grain for each rumen measure from 1 to 4 (on a 4 point scale) to determine the most effective grain type and dose rate for use in the in vivo study. For example, groups with the greatest concentrations of VFA, D- and L-lactates, and ammonia were ranked 1 and those with the lowest concentrations of these variables were ranked 4. Cattle with the lowest rumen pH were ranked 1 and those with the highest rumen pH were ranked 4. A weighted ranking was estimated by combining the ranking scores of the individual rumen VFA and ammonia. The grain type and grain rate that had the greatest number of variables ranked 1 were considered the most effective for use in the in vivo study.

To provide a cross-validation of the rankings derived from the estimated means, a discriminant score was estimated for each animal using K-means cluster and discriminant analysis methods (SPSS; Apache Software) and algorithms defined by Bramley et al. (2008). The Bramley et al. (2008) methods are based on a model that categorized dairy cows into 1 of 3 groups: acidotic, suboptimal rumen function, and normal based on K-means cluster analysis and discriminant scores of rumen fermentation parameters measured from 800 cows from 100 commercial dairy herds across southern Australia. The index developed was strongly influenced by rumen VFA and ammonia concentrations, with valerate and propionate being the most discriminatory measures whereas lactic acid and pH were less important predictors. The index was not influenced by herd and can be best understood as reflecting a continuum of ruminal conditions from clinical acidosis through to normal rumen function. Bramley et al. (2008) also characterized milk production, diet characteristics, and lameness for each of the 3 acidotic categories. The mean of the discriminant scores for each grain type and grain rate in this current study were then ranked on a score of 1 to 2 or 1 to 4, respectively, with 1 indicating most effective for use in the in vivo challenge study.

In vivo Study.

A mixed model linear regression using nlme package (Pinheiro et al., 2012) in R (R 1.14.2; R Development Core Team, 2012) was used to analyze the effects of time, day of sampling, and grain type and their interactions on rumen pH and concentrations of VFA, D- and L-lactate, and ammonia (Table 7). The linear mixed effects function in the nlme library (Pinheiro et al., 2012) uses the Laird-Ware form of the linear mixed model (Laird and Ware, 1982):in which yij is the response variable (e.g., rumen pH) for the jth of ni observations, β1,…, βp are the fixed-effect coefficients, x1ij,…, xpij are the fixed-effect regressors (grain type, time, day of sampling, and interaction terms), bi1 is the random-effect coefficient, z1ij is the random-effect regressor (cow identification number), and εij is the error for observation j in group i.


View Full Table | Close Full ViewTable 7.

Significance (P-values) of effect of time1, day2, grain3, and their interactions on rumen fermentation products in the in vivo study (n of heifers = 80; n of samples = 400)

 
Parameter, mM Time (T) Day (D) Grain (G) T × D T × G
Rumen pH, no units 0.001 0.001 0.002 0.009 0.83
Total VFA 0.001 0.001 0.001 0.98 0.51
Acetate 0.07 0.001 0.001 0.001 0.001
Propionate 0.001 0.001 0.001 0.99 0.32
Butyrate 0.001 0.001 0.001 1.00 1.00
Isobutyrate 0.033 0.001 0.001 0.68 0.33
Valerate 0.001 0.001 0.001 0.84 0.001
Isovalerate 0.001 0.001 0.001 0.92 0.001
Caproate 0.001 0.001 0.001 0.71 0.84
Acetate:propionate 0.001 0.001 0.001 0.84 0.17
L-lactate 0.025 0.001 0.001 0.99 0.89
D-lactate 0.010 0.001 0.001 1.00 0.98
Ammonia 0.001 0.001 0.001 1.00 0.049
1Rumen samples were collected 5, 65, 110, 155, and 200 min after challenge ration consumption.
2Heifers were sampled on 1 of 8 d.
3The 20 grain cultivars were oats (n = 3; Dalyup 5817, Dalyup 5818, and Swan), wheat (n = 6; Bellaroi, Sentinal, Ammrock, Kellalac, unknown, and Chara), barley (n = 4; Tantangara, Gairdner 3864, Gairdner 3862, and Binnalong), triticale (n = 4; Maiden, Jackie 6823, Jackie 6824, and Prime 322), and sorghum (n = 3; Liberty, Pacer, and MR43).

To enable comparison with the NIRS-derived acidosis index, 2 separate indices were used to assign an estimated acidosis ranking to each of the 20 grain cultivars tested in the in vivo study on a scale of 1 to 20 with 1 indicating the greatest acidotic risk (Table 8). The mixed grain control was also ranked in both indices. The first index was based on a discriminant score determined from individual VFA, ammonia, lactate, and pH measures from the final time period (200 min) using the same methodology (Bramley et al., 2008) that ranked the grain cultivars.


View Full Table | Close Full ViewTable 8.

Acidosis rankings of 20 grain cultivars estimated from near-infrared reflectance spectroscopy (NIRS), discriminant scores, and coefficients of rumen valerate

 
Grain NIRS Discriminant Valerate2
Barley (Tantangara) 15 15 11
Barley (Gairdner 3864) 5 13 13
Barley (Gairdner 3862) 4 11 14
Barley (Binnalong) 12 16 12
Oats (Dalyup 5817) 17 17 15
Oats (Dalyup 5818) 18 14 19
Oats (Swan) 16 12 16
Sorghum (Liberty) 14 18 20
Sorghum (Pacer) 19 19 17
Sorghum (MR43) 20 20 18
Triticale (Maiden) 13 6 7
Triticale (Jackie 6823) 6 1 6
Triticale (Jackie 6824) 7 5 2
Triticale (Prime 322) 9 9 9
Wheat (Bellaroi) 11 8 1
Wheat (Sentinal) 10 2 3
Wheat (Ammrock) 8 10 5
Wheat (Kellalac) 2 3 4
Wheat (unknown) 3 7 10
Wheat (Chara) 1 4 8
Control mixed grain NA1 10 12
1Not available.
2Ranking based on valerate concentrations from the 200 min rumen sampling on challenge day

The second index was based on the coefficients of rumen valerate concentrations at the final sampling period (200 min). The coefficients of valerate were chosen as a second method of ranking because with the exception of propionate (P = 0.01) and pH (P = 0.01), valerate (P < 0.01) was the only variable that was significantly affected by the rates of grain fed in the method development study and the only variable that significantly differed between the 2 grain types fed, triticale and the mixed grain control, in that study. Valerate was also shown to be a critical indicator of acidosis by Bramley et al. (2008). The coefficients of rumen concentrations of valerate at the final sampling (200 min) were estimated using a mixed effect model with heifer as a random effect and day as a covariate in STATA 9 (StatCorp LP, College Station, TX). Grains with the greatest coefficients for valerate were ranked 1, and those with the lowest coefficients were ranked 20, and the control group was used as a reference group. Values at 200 min only were selected for both indices as this sample time showed the largest difference between test grains.

Correlations between the acidosis index rankings estimated by 1) NIRS measures, 2) discriminant scores, or 3) valerate coefficients were determined using a Spearman’s coefficient of rank from the actual index scores.


RESULTS

Method Development Study

No animals showed clinical signs of ruminal acidosis during the experiment. Cattle fed grains at 0.8 and 1.2% of BW regardless of grain type had a decreased rumen pH (P = 0.01) compared with control (0% of BW) fed cattle (Tables 5 and 6). Concentrations of rumen propionate (P = 0.01) were increased in cattle fed at 0.8 or 1.2% of BW in comparison with control or 0.4% of BW fed cattle (Tables 5 and 6) and increased over time (Fig. 2B). Concentrations of propionate in the control and 0.8% of BW grain fed cattle decreased over time (P = 0.01; Fig. 2B). Valerate concentrations increased more than 200% in cattle fed grain at 1.2% of BW compared with the control fed cattle and increased in comparison with those fed grain at 0.4% of BW (P < 0.01; Tables 5 and 6; Fig. 2A). Concentrations of valerate also increased (P < 0.05) over time in the 0.8 and 1.2% of BW grain fed cattle up to 155 min; the opposite occurred for the control and 0.4% of BW grain fed cattle (Table 5; Fig. 2A). The time × rate interaction was also significant for valerate and concentrations were greater (P = 0.05) in cattle fed with triticale cultivar Jackie compared with the control cattle fed the mixed grains (0.69 ± 0.05 and 0.61 ± 0.06 mM, respectively; Table 6). Rumen ammonia concentrations decreased over time for all rates with the exception of 1.2% of BW grain fed cattle (Table 5; Fig. 1C).

Figure 2.
Figure 2.

Mean rumen A) valerate concentration, B) propionate concentration, and C) ammonia concentration from cattle (n = 16) fed 0, 0.4, 0.8 or 1.2% of BW of mixed grain or triticale at 5 time periods after diet consumption in the method development study.

 

Both the acidosis index rankings, derived from the estimated marginal means and discriminant scores, showed that cattle fed grain at the rate of 1.2% of BW had the greatest acidosis rank, and cattle in the control (0%) and 0.4% groups had the lowest rank (data not shown).

In vivo Study

No cattle developed clinical signs of ruminal acidosis during the experiment. Rumen pH decreased over time in the majority of groups with the lowest rumen pH reported from the barley (Gairdner 3864), oats (Dalyup 5817), oats (Swan), triticale (Maiden), triticale (Jackie 6824), and wheat (Chara) fed cattle (P = 0.001; Tables 7; Fig. 3A). Figures 3A to 3D show marked differences in change in concentrations of pH and VFA over the 5 sample times among grains. Overall rumen concentrations of total VFA (P = 0.001), propionate (P = 0.001), butyrate (P = 0.001), isobutyrate (P = 0.033), valerate (P = 0.001), isovalerate (P = 0.001), and caproate (P = 0.001) increased over the experimental period (Table 7). D-lactate concentrations decreased over time (P = 0.01; Table 7). Rumen pH, ammonia, and concentrations of total and individual VFA were significantly influenced by the day of sampling. Rumen concentrations of propionate (P = 0.001), valerate (P = 0.001), isovalerate (P = 0.001), and ammonia (P = 0.001) were generally greater in the wheat and triticale cultivars than the control, barley, oat, and sorghum cultivars (Table 7; Fig. 3B to 3D). The interactions between the time of sampling and the day of sampling were only significant for rumen pH (P = 0.009) and acetate (P =0.001) whereas the interactions between the time of sampling and types of grain were significant acetate for (P = 0.001), valerate (P = 0.001; Fig. 3C), isovalerate (P = 0.001), and ammonia (P = 0.049; Fig. 3D) with increasing concentrations over time in the wheat and triticale cultivars (Table 7).

Figure 3.
Figure 3.

Mean rumen A) pH, B) propionate concentration, C) valerate concentration, and D) ammonia concentration for 20 grain cultivars and control over 5 time periods, 5, 65, 110, 155, and 200 min, after diet consumption in the in vivo study.Barley1 (Tantangara), Barley2 (Gairdner 3864), Barley3 (Gairdner 3862), Barley4 (Binnalong); Oats1 (Dalyup 5817), Oats2 (Dalyup 5818), Oats3 (Swan); Sorghum1 (Liberty), Sorghum2 (Pacer), Sorghum3 (MR43); Triticale1 (Maiden), Triticale2 (Jackie 6823), Triticale3 (Jackie 6824), Triticale3 (Prime 322); Wheat1 (Bellaroi), Wheat2 (Sentinal), Wheat3 (Ammrock), Wheat4 (Kellalac), Wheat5 (Unknown), and Wheat6 (Chara). The control consisted of a blend of the 20 grain cultivars in equal proportions. See online version for figure in color.

 

The correlation between the discriminant score ranking at the final time period for each of the 20 different grain cultivars and the NIRS derived ranking was highly significant (Spearman’s rank correlation coefficient = 0.731, P = 0.0003). The correlation between the ranking derived from the coefficients of valerate and the NIRS-derived ranking was also significant (Spearman’s rank correlation coefficient = 0.586, P = 0.008). The discriminant score ranking and valerate ranking were also highly correlated (Spearman’s rank correlation coefficient = 0.842, P = 0.0001). Despite the high correlation between the different indices, the NIRS acidosis index values were low for the oat cultivars (Table 8). Based on all 3 indices the ranking of acidosis risk in descending order is wheat > triticale > barley > oats > sorghum. Subsequently, the data obtained from this study was used as the basis for guiding adjustments to the parameters in the calculation of acidosis index from the in sacco and in vitro results for the original 92 grains. The statistics for the NIRS calibrations developed using the recalculated acidosis index values are given in Table 9. The relationship between the newly calculated acidosis index and revised NIRS calibration predicted values for whole grain is shown in Fig. 4 for whole grain scans.


View Full Table | Close Full ViewTable 9.

Statistics for revised near-infrared reflectance spectroscopy calibrations for acidosis index developed from whole and milled grain scans1

 
Calibration n Mean SD RSQ SEC 1-VR SECV RPD
Whole grain scans 86 48.0 19.8 0.90 6.15 0.88 6.93 2.86
Milled grain scans 86 48.1 19.9 0.87 7.12 0.81 8.59 2.32
1Mean = the mean predicted acidosis index value (% units); RSQ = R2 values – fraction of the variance accounted for by the NIRS calibration when all accepted observations are included in the relationship; SEC = SE of the calibration; 1-VR = 1-variance ratio – fraction of variance accounted for in NIRS prediction when some observations are used for “cross-validation” of the calibration as determined by the NIRS software; SECV = SE of cross-validation – SE of the calibration when some observations are used for “cross-validation” of the calibration as determined by the NIRS software; RPD = ratio of prediction to deviation = SD/SECV, an indication of the value of the calibration.
Figure 4.
Figure 4.

Relationship between the calculated discriminant score and revised predicted acidosis index for the experimental grain cultivars. The line represents the linear regression (y = 0.0157 × x – 1.731; R2 = 0.69). NIR = near-infrared reflectance.

 


DISCUSSION

In vivo testing confirmed the ability of an existing NIRS equation to estimate the acidotic risk of grains as hypothesized. The method development study showed 1.2% of BW of grain is a sufficient dose rate to induce moderate rumen changes. Grains differ in acidosis risk between and within species with wheat and triticale posing the greatest risks based on 3 acidosis indices. The NIRS estimates of acidosis risk of 20 grain cultivars derived from in sacco, in vitro, and grain chemical composition measures were successfully confirmed by high correlations with 2 individual acidosis indices as hypothesized; however, discriminant scores were used to recalibrate the NIRS equation. The characteristics for the new equation were similar to the original, with the best WinISI-generated mean global H and neighborhood H calibration values being 1.095 and 0.456, respectively. Figure 4 shows that there was a good fit for the revised equation to the in vivo data.

The method development study showed that feeding grain at a rate of 1.2% of BW created a significant change in rumen pH, valerate, and propionate, particularly in comparison with the control animals. Data evaluation, consistency with previous studies in North America (Nagaraja and Titgemeyer, 2007), and lack of evidence of severe acidosis lead us to conclude that a 1.2% of BW dose was satisfactory for a challenge model that could discriminate between grains in the in vivo study.

Interestingly, despite the ability to discriminate between the test grains, the 1.2% of BW dose rate did not produce declines in rumen pH as pronounced as those of Opatpatanakit et al. (1994) or Wales et al. (2001), who challenged heifers on short-chopped alfalfa hay (Opatpatanakit et al., 1994) or cows 49 ± 14.1 d in milk on perennial ryegrass and clover (Wales et al., 2001), respectively, with similar amounts of grain. We hypothesize the fiber available from the ad libitum access to ryegrass silage before the 14-h feed withholding period may have increased the buffering capacity of the rumen (Erdman, 1988), preventing large declines in rumen pH. The sugar content in alfalfa hay in the study by Opatpatanakit et al. (1994) may have influenced rumen fermentation resulting in a lower pH. Furthermore, rumen pH was obtained from fistulated cattle in the studies of Opatpatanakit et al. (1994) and Wales et al. (2001). Opatpatanakit et al. (1994) obtained their pH samples from the anterior reticulum. It is possible that samples obtained using a stomach tube, typically from the dorsal sac of the rumen, have a higher pH than those obtained from the central rumen (Shen et al., 2012).

The method development study showed that rumen pH, propionate, and valerate were the main indicators of change in rumen function. This is consistent with the work by Bramley et al. (2008) that showed propionate, valerate, and ammonia are critical indicators of acidosis and rumen pH and lactate concentrations play only minor roles. Of these, only valerate also significantly differed between grain types, time × rate, and time × grain type effects. These data strongly indicate that changes in the rumen concentration of valerate after the challenge were more discriminatory than other single measures such as rumen pH or total VFA. Hence the coefficients of valerate concentration were used to rank the 20 grain cultivars.

Both propionate and valerate are synthesized from lactic acid (Hungate, 1966; Stewart et al., 1997). Although the pathways for propionate production are well described, those for valerate are not as easily found but have been described by Ladd (1957). Removal of lactic acid (pKa = 3.1) from the rumen is important as it is a 10 times stronger acid compared with VFA (average pKa = 4.8; Schwartzkopf-Genswein et al., 2003). The conversion of lactic acid to propionate and valerate is a safe method of sequestering hydrogen and maintaining rumen pH. Valerate is used by bacteria and is efficiently metabolized by the rumen wall into acetyl-CoA and propionyl-CoA by β-oxidation before absorption into the portal vein (Kristensen, 2005) and use by bacteria. Therefore, hydrogen sequestration during the synthesis of valerate from lactic acid may be a reason for its formation during grain challenge. Associations between increased valerate concentrations and subacute ruminal acidosis (Morgante et al., 2007) and ruminal acidosis (Bramley et al., 2008) have been demonstrated.

There were marked differences in fermentation characteristics among grains over time. The ranking of acidotic risk in descending order, wheat > triticale > barley > oats > sorghum, based on all 3 indices is consistent with and a reflection of rankings determined by gas production estimations of starch degradation (Opatpatanakit et al., 1994; Lanzas et al., 2007) and rumen digestibilities measured in vivo (Moran, 1986; Galloway et al., 1993). Oats have the lowest concentration of starch; however, oat starch is rapidly fermented (Herrera-Saldana et al., 1990; Cone, 1991). The differences in rumen fermentability between and within grain species reflect variations in the physiochemical properties and structure of grains, which can be influenced by both genetic and environmental factors (Opatpatanakit et al., 1994). In particular, slower fermentations such as that from sorghum can be associated with the presence of protein and nonstarch carbohydrate matrices, high amylose to amylopectin ratios, small starch grain size, and antinutritional factors (Rooney and Pflugfelder, 1986; Opatpatanakit et al., 1994). Differences in particle size after processing may have influenced the ranking and alternative processing methods may affect the rate and site of digestion and change the acidosis index ranking of a grain (Lanzas et al., 2007). Table 3 shows the effects of the roller-mill processing commonly used to crush grain on the particle size profile of grains.

These results provide validation for the in sacco, in vitro, and grain chemical composition derived NIRS model as it does significantly correlate with the coefficients of valerate and discriminant analysis scores. The valerate coefficients and discriminant analysis scores were strongly correlated; however, the only measures that relate to biological outcomes of acidosis, milk production, milk fat content, lameness, and diet are those produced from the discriminant analysis (Bramley et al., 2008). These findings suggest that the current NIRS model is likely to predict acidosis risk; however, the model would benefit from further refinement. Subsequently, the data obtained from this study was used to recalibrate the NIRS model.

Despite ongoing research, there remains confusion and inconsistencies in the definitions of acute and subacute ruminal acidosis. Although studies based on area under the curve estimates of rumen pH are very likely sound, these have been largely based on fistulated cattle. The advantage of a study using repeated measures and stomach tubing of the type undertaken is that the cattle are not surgically prepared and larger groups of animals can be more readily studied. The limitation is that the rumen pH obtained by stomach tube is higher than samples obtained by rumenocentesis (Bramley et al., 2008). This higher pH may lead to an erroneous assumption that cattle are not adequately challenged by diets.

Although no clinical signs of either acute or subacute ruminal acidosis were observed in this study, the shifts in rumen pH, VFA, and ammonia observed indicate acidosis risk for the 20 grains tested. We hypothesize that acidosis occurs along a continuum of ruminal conditions; therefore, at the 1.2% of BW feed rate, cattle were at the more subacute end of the continuum but would likely move up the continuum at a rate relative to their ranking as feed rate increased. The effects of the underlying physiochemical properties and structure of the grains responsible for their fermentability at the current feed rate are likely to be magnified with increased rate. A more severe challenge resulting in clinical signs of acidosis would be required to support this hypothesis.


Conclusion

Grain type and cultivar should be considered when formulating rations to reduce the risk of acidosis. Wheat and triticale are likely to pose the greatest risk of acidosis to cattle and sorghum the least. Feeding 1.2% of BW of grain is sufficient to induce significant decreases in rumen pH and increases in propionate and valerate without clinical signs of acidosis. Valerate was a key indicator of ruminal change during grain challenge. The NIRS equation derived from in sacco, in vitro, and grain chemical composition analysis provides an estimate of the risk of acidosis for grains fed to cattle.

 

References

Footnotes


Comments
Be the first to comment.



Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.