The following answer was provided by Mike Shearing and Rick Brown, of Venture Milling-Precision Dairy Nutrition in Charles City, Iowa.

Q: There are several new and revised ration balancing software programs available to nutritionists today. What are the most important features you look for when evaluating the programs you will use in your practice?

A: Today, many nutritionists are using or considering models to balance dairy rations. Models present challenges and opportunities, and can require more time and effort to use than conventional ration software. That being said, many have found the benefits to be worth the cost. Understanding what a model is, and the reasons to use one, can help to determine if one is right for you.

Much of what we feed to a dairy cow is not used directly by the cow, but rather is fermented in the rumen. The end products of rumen fermentation, such as volatile fatty acids (which provide energy), and microbial crude protein (which provide amino acids), are used by the cow once they reach the small intestine. Feeds which are not fermented in the rumen reach the small intestine where they can be absorbed — or pass through and be excreted as waste.

Many factors affect the yield of ruminal microbial crude protein and volatile fatty acids; content of physically effective NDF, content and potential rate of fermentation of NDF, NFC, and protein. A dynamic rumen sub-model will take these factors into consideration to generate a performance prediction for a specific ration fed to a defined cow at a defined inclusion of dry matter intake.

When used as intended, a well-designed model can make very accurate predictions – potentially more accurate than is possible when using a conventional or linear program. Models can also be used to troubleshoot situations where performance is below expectations, such as milk fat depression. Models also allow more detailed evaluation of specific areas, such as the fatty acid profile of the diet.

An effective model will be based on sound science and ideally will be the collaborative work by a group of unbiased researchers. A model should have a complete and accurate feed library which is updated regularly, and will contain a biological system that is updated and improved as new information becomes available.

A key factor to consider when choosing a model is the biological system used — it must be proven and have the ability to generate accurate predictions from accurate information. A first class model will be easy to use, provide flexibility, allow customizations (reports, user-defined nutrients), and be supported by a responsive organization. In many cases, least-cost optimization is desired in order to provide equal nutrition at a better cost. Some models contain optimizers, and some optimizers work better than others. An effective non-linear optimizer can reduce the cost to supply non-linear nutrients such as metabolizable energy, metabolizable protein, metabolizable methionine, and metabolizable lysine.
Finally, it must be remembered that a model is simply another tool to be used by the nutritionist.

Models are not perfect, and all have some weaknesses and biases. It remains the responsibility of the nutritionist to accurately assess each situation, and integrate all available information into a sound feeding program that provides the best results possible for the dairyman.