The dairy cow requires enough fiber to maintain an efficient and healthy digestive system (2). She has a requirement for both chemical fiber and physical fiber. Chemical fiber, measured as NDF, supplies fermentable carbohydrate that is required to support the microbial growth within the rumen. Physical fiber, measured as particle size, promotes chewing which results in the production of saliva. The buffers contained in saliva help minimize the risk of acidosis. The optimal amount of NDF in barley diets depends on the PI of the barley and the particle size of the forage.

Formulating diets for dietary NDF levels is a way of adjusting the forage-to-concentrate ratio of the diet to account for the variation in fiber content of forages. The NRC (9) recommends a minimum of 28% NDF in dairy cattle diets to maintain 3.5% milk fat and to minimize digestive disorders. It is also recommended that at least 75% of dietary NDF be supplied by forage. These recommendations are based on studies that mainly used corn rather than barley. Although this method of ration formulation is useful, NRC (9) guidelines are not directly applicable for barley-based diets. The NDF content of barley (19 to 25%) is higher than that of corn (7%), making it impossible to meet the NRC criteria for minimum fiber.

A commonly used thumb rule is to formulate the diet to supply 21% NDF from forage sources (NDF-f) without consideration of the NDF from concentrates (i.e., 28% NDF x 0.75 from forage = 21% NDF-f). To examine whether this recommendation is adequate for barley diets, we examined the relationship between NDF-f and milk yield and fat content (Figure 3). Treatment means from 5 experiments in which concentrates based on barley were fed to dairy cows are plotted in Figure 3 (3, 4, 6, 7, 12). Increasing the NDF-f content of the diet decreased milk yield, but increased milk fat content up to a maximum of 32% NDF-f. Beyond this level, milk fat content declined due to lower feed intake and insufficient energy intake. Therefore, within the range of forage fiber normally fed to lactating dairy cows, fat corrected milk was relatively unaffected by dietary NDF-f concentration.

The concentration of NDF-f necessary to maintain 3.5% milk fat in diets based on barley was extremely variable with the average ranging from 21 to 25%, which is higher than NRC (9) recommendations. This difference is attributed to the more extensive ruminal starch digestion of barley compared with corn (Figure 1). If the PI of the barley is low, and/or the particle size of the forage is fine, then the concentration of NDF-f necessary to avoid milk fat depression is at the higher end of the range. In contrast, if forage particle size is coarse and/or if the PI of barley is high, then less NDF-f is needed. Of course, for both barley and corn diets, the concentration of NDF-f necessary to prevent milk fat depression also depends on the genetics of the cows and other feeding practices, such as inclusion of high fiber byproduct feeds, frequency of feeding, and use of total mixed rations versus feeding grain and forage separately. If milk fat depression is not a primary concern, the minimum level of forage fiber necessary in dairy cow diets is that necessary to maintain healthy rumen function. That level is considerably lower than the level necessary to prevent milk fat depression. Thus, the optimum amount of NDF-f in the diet depends on the relative costs of concentrate and forage and the relative price of milk and fat.

The NDF assay assigns the same fiber value to long stemmed hay and ground hay, and does not predict the roughage value or effectiveness of the fiber. The physical components of fiber are often referred to as effective fiber, or effective NDF (eNDF). This pool of long fiber is chewed during eating and must be ruminated before it can leave the rumen. Saliva produced during chewing helps buffer acids produced in the rumen during microbial digestion of feed, thereby minimizing postprandial decreases in ruminal pH.

Figure 4 illustrates the relationship between the level of NDF-f consumed and the time spent eating, ruminating, and total chewing for diets consisting of barley concentrate and either long stemmed alfalfa hay, orchard grass hay, or corn silage (3). Time spent eating and ruminating increased as the amount of forage

fiber in the diet increased. The dairy cow typically spends 3 to 8 h/d eating and 6 to 9 h/d ruminating. The chewing time of cattle fed barley diets is considerable higher than those fed corn indicating a higher eNDF content of barley. In a study in which beef cattle fed 100% grain, rumination was twice as much when barley versus corn was fed (5). The higher rumination for barley is be due to the hulls, as well as a more resistant pericarp, resulting in a greater proportion of ingested intact kernels.

There is considerable difference among forages in the time spent chewing, even when compared at the same level of fiber intake (Figure 4). Eating time was much shorter for the diet containing corn silage due to the smaller particle size of corn silage compared with long hay. When feeds are chopped or ground, rumination and eating times are reduced (2). Grinding and pelleting feed results in a drastic decrease in chewing time, whereas the effect of chopping depends upon the fineness of chop.

Effective NDF measurements. The Cornell Net Protein and Carbohydrate model (11) assigns eNDF values to individual feeds. For example, the eNDF content (% of NDF) of alfalfa hay is 92%, alfalfa silage is 82%, barley silage is 65%, and barley grain is 34% (Table 3). The values represent a combination of particle size measurements (i.e., measured as the proportion of NDF retained on a 1.18 mm square mesh screen after dry sieving) and subjective adjustments based on the feed digestion characteristics and their effects on ruminal pH. However, it is not clear how the eNDF values should be adjusted to account for on-farm variation in chop length of silage. Likewise, it is not specified how the adjustments for digestion characteristics are calculated.

The Cornell model (11) predicts ruminal pH from eNDF (% of DMI). However, the equation used is based on studies using dairy cows, sheep, and steers. To verify this relationship for lactating dairy cows, we compiled data for mean ruminal pH reported in 22 publications representing 95 diets fed to 619 cows. Ruminal pH was positively affected by level of NDF-f in the diet, but the variation in ruminal pH at a particular level of NDF-f was tremendous (Figure 7). Mean ruminal pH ranged from 5.6 to 6.7 within the NDF-f levels commonly fed to high producing dairy cows. This confirms that the minimum NDF-f level needed in the diet to minimize milk fat depression and ensure healthy rumen function is variable.

Using eNDF content (or eNDF intake) rather than NDF-f content did not improve the prediction of ruminal pH. It is possible that the eNDF values of feeds were not estimated correctly, thereby under or overestimating eNDF intake. However, it is likely that most of the variation is caused by differences in fermentability of the carbohydrate sources which is not accounted using the current system. This issue is extremely important if the model is to be used for diets containing barley concentrates because barley is extensively digested in the rumen causing low ruminal pH. Predictions of ruminal pH must account for the effective fiber content of the diet (saliva production) as well as the ruminal digestion of carbohydrates (acid production and absorption).

The Cornell model assumes that rumen pH is directly related to the efficiency of microbial yield (11). Microbial yield is reduced by 2.5% for each percentage unit reduction in eNDF < 20% of DMI, thus feeding low fiber diets increases the need for supplemental feed protein and ration costs. Accurate estimates of the efficiency of microbial protein synthesis are critical in predicting the amount of supplemental protein required and the cost of the diet. An example of this is given in Figure 8 where a diet was balanced for 48 kg of milk (3.5% fat; 3.12% protein) using the new Cornell Penn Miner (beta version) model. By increasing microbial protein synthesis by 50%, potential allowable milk yield increased from 48 to 92 kg/d in this theoretical application of the model. Clearly, the potential for altering the eNDF content of the diet to reduce the need for expensive protein supplements is large.

We do not have data for cows fed barley diets to verify the relationship between fiber and microbial protein synthesis for western Canadian dairy diets. However, we have shown that for diets containing the same proportion of forage (22% NDF-f), microbial protein synthesis is higher for barley diets than corn diets (12). Higher microbial protein synthesis was due to the greater proportion of ruminal fermented organic matter, rather than improved efficiency (Table 4).

Table 4. Digestibility of nitrogen by lactating dairy cows fed diets based on barley or corn grain (12)¹