Improving Nutrient Management on a 100-Cow Free-Stall Dairy Farm

Nutrient balance--Miner Institute has been involved in nutrient balance investigations since 1983. The 1992-93 balances found that we had improved P and K efficiency since 1983, but were less efficient for N. We were also less efficient in nutrient utilization than the typical NY dairy farm. Adjustments were then made in dairy rations, the crop rotation adjusted to better utilize dairy manure, and forage was sold to bring inventories more in line with existing herd size. Nutrient inputs were greater in 1994, but outputs were considerably greater, resulting in an improvement in nutrient efficiency. A comparison between a typical 100 cow dairy farm and the 4-yr. Miner Institute mean is in Table 1.

Compared to the typical NY dairy farm, the average (1992-95) mass balance for N, P, and K on the Miner Institute farm was 13%, 11%, and 23% higher, respectively. The typical farm uses 63% less fertilizer N, 75% less fertilizer P, and has 75% less N from nitrogen fixation than does Miner Institute. However, during 1992-93 Miner Institute was pursuing an aggressive fertilization program to correct low soil potassium levels on a number of fields. Miner Institute also has a higher proportion of cropland in alfalfa, as can be seen from the considerable difference in N fixation.

Nutrient modeling--Three models were investigated: DAFOSYM, Penn State Farm Balance Model, and the Cornell Net Carbohydrate Protein Model, UPenn version (NCPS). DAFOSYM and the Penn State model are whole farm models, while NCPS is a single cow model. DAFOSYM integrates all aspects of the farm into one model. In exploring DAFOSYM’s capabilities it was found to be too restrictive to use in this project. Although considerable time was spent attempting to refine the inputs to approximate farm conditions (including discussions with the developer of the model), constraints within the model would not allow accomplishment of this goal. The Penn State model is an accounting model with little predictive capability. Since the objective is to improve the predictive ability of existing models, the Penn State model was found to be unsuitable for this purpose. As an accounting tool, however, this model is user-friendly and easy to interpret.

Efforts therefore concentrated on the NCPS model. Elements within the model were extracted to form a report based on predicted nutrient output. The model is partially mechanistic and predicts the potential based on inputs, so it is easy to compile nutrients not used for production or maintenance into a predicted output report. For parameters not predicted in the model, the literature was reviewed and predictive equations were used which are generally accepted by the scientific community.

Once the report page was compiled, we conducted a study to determine the quantity of manure excreted from four classes and two breeds of dairy cattle, comparing observed outputs with those predicted by the model. The model adequately predicted the excretion of dry matter, N, and P for all groups. However, as the milk production of the animals increased, the predictive ability of the model was reduced.

Substitution of manure for fertilizer--A 1992 research trial at Miner Institute found that topdressing 3000 gallons/acre of slurry dairy manure had no adverse effect on alfalfa or groundwater quality. This was confirmed by a 1994 research trial in Massachusetts. In 1994 applications of manure to established alfalfa at Miner Institute increased from less than 20% of total manure (average of 1987-93) to over 40%. At the same time the amount of commercial fertilizer applied to alfalfa decreased by 40%.

Applications of dairy manure at rates which meet the K needs of the crop result in considerably more P than needed by alfalfa. Analyses of the dairy manure at Miner Institute shows a P2O5:K2O of 1:2, while alfalfa removes these nutrients from the soil at a ratio of 1:4. In 1994 a soil P stratification study on four fields characterized the fate of the P applied as manure. The fields had widely varying histories of manure application, ranging from 0 to 175 tons/acre since 1981. Samples were taken from six depths in the top 24” of soil. Soil analysis determined that essentially all P remained in the plow layer. Plant available P was medium and high in the two fields which had received heavy applications of manure, but P levels on all fields were low at depths below 8”. Water-soluble P was high in the top 1” of soil in one heavily manured field, indicating the importance of practices which prevent soil erosion and limit water runoff.

Presidedress nitrate tests (PSNT)--Five Miner Institute corn fields were sampled in 1994, four in 1995, and five in 1996. Nitrate levels over 21-25 ppm (the range of sufficiency levels recommended by universities in the Northeast), indicate a need for supplemental N. In 1994 nitrate levels on the Miner Institute fields ranged from 3-13 ppm, in 1995 from 7-23 ppm, and in 1996 from 3-11 ppm. Since soil nitrate levels were generally low, supplemental N was applied to all fields tested, and this test therefore did not result in a decrease in fertilizer N use. However, evaluation of soil nitrate levels soon after corn harvest in these fields found low to medium levels of nitrate, indicating that most of the supplemental N had been used by the crop.

One of the problems with PSNT is the necessity for the test to be done very soon before nitrogen application. It can take a week or more to send the sample to the laboratory and get the results back. For this reason, in 1996 we evaluated the N-Check test which is intended for on-farm use. We used the N-Check on the same five sample which were submitted to the University of Vermont Agricultural Testing Laboratory. The results are in Table 2.

Since any reading below 21-25 ppm indicates a need for supplemental N, the N-Check performed acceptably for these samples. One problem is that the test is designed for nitrate levels in water samples, so it’s calibrated for very low nitrate levels and lacks the refinement between 20 and 30 ppm, where the greatest accuracy is needed. It may be possible to dilute samples, although the efffect of this on accuracy isn’t known. The test kit costs about $50 for 25 samples, and would take someone not familiar with laboratory procedure about 1 1/2 hours to run five tests.