Conventional production of meat products from ruminant animals in the United States requires inputs including the cultivation and nitrogen fertilization of annual grains such as corn and barley, and transportation of cattle and grain to feedlots. Consumers have concerns about the impact of feedlot conditions on animal health, and about the implications of pharmaceutical inputs such as growth hormones and antibiotics on the environment and human health. These concerns have led to a growing interest in pasturefinished meat production by consumers. Such smaller-scale livestock production systems can be healthier and lower-stress for animals, are integrated into local food systems and are more transparent to consumers, and have higher potential profitability for producers than traditional ruminant production methods. There is a strong market for pasture-finished beef products, and prices for naturally or organically raised beef have remained well above feedlot-produced product prices. There is also concern about the impact of ruminant production on the environment, including air and water pollution from feedlot production and greenhouse gasses that are emitted from ruminant animals during feed digestion. This thesis project explored the potential of a beef production system based on perennial legumes, including the non-bloating legume birdsfoot trefoil (BFT; Lotus corniculatus L.) for producing meat products from cattle while reducing concentrate feeding and methane production. The condensed tannins that are produced by BFT bind proteins in the rumen but allow them to be digested in the abomasum and intestines, which in turn leads to better utilization of forage nutrients during the finishing period and higher gains or milk production. The higher digestibility of legumes compared with grasses reduces methane emissions in cattle both through higher digestibility of the forage and through direct impacts on methanogens operating in the rumen. As reported in this thesis, steers finished on BFT gained significantly more weight per day than steers fed another perennial forage legume, cicer milkvetch, but did not gain as rapidly as feedlot-fed steers. At the end of summer grazing, the blood plasma of pasture-fed steers was lower in saturated and omega-6 fatty acids and higher in transvaccenic and omega-3 fatty acids than the blood plasma of feedlot-fed steers. When beef cows grazed grass and legume pastures, enteric methane emissions were lower on the legume pastures than the grass pasture. These results demonstrate that, compared with other feed sources, perennial legume pastures used for cattle production can improve cattle gains and reduce environmental impacts.

Mathematical modeling in animal agriculture can be applied at various levels including at the tissue, organ, animal, farm, regional and global levels. The purposes of this research were i) to evaluate models used to estimate volatile fatty acid (VFA) and methane (CH4) production and assess their impact on regional enteric CH4 inventory, and ii) to develop a process-based, whole-farm model to estimate net farm GHG emissions. In the first study, four VFA stoichiometric models were evaluated for their prediction accuracy of rumen VFA and enteric CH4 production. Comparison of measured and model predicted values demonstrated that predictive capacity of the VFA models varied with respect to the type of VFA in rumen fluid which impacted estimated enteric CH4 production. Moving to a larger scale assessment, we examined the enteric CH4 inventory from Manitoba beef cattle (from 1990 to 2008) using two mechanistic rumen models that incorporate VFA stoichiometric models: COWPOLL and MOLLY, and two empirical models: Intergovernmental Panel on Climate Change (IPCC) Tier 2 and a nonlinear equation (Ellis). The estimated absolute enteric CH4 production varied among models (7 to 63%) indicating that estimates of GHG inventory depend on model selection. This is an important consideration if the values are to be used for management and/or policy-related decisions. Development of models at the individual farm component level (animal, soil, crop) does not accurately reflect net GHG emissions generated from the whole production system. We developed a process-based, whole-farm model (Integrated Components Model, ICM), using the existing farm component models COWPOLL, manure-DNDC and some aspects of IPCC to integrate farm components and their associated GHG emissions. Estimates of total farm GHG emissions and their relative contribution using the ICM were comparable to estimates using two other whole-farm models (Integrated Farm System Model and Holos model). Variation was observed among models both in estimating whole-farm GHG emissions and the relative contribution of the different sources in the production system. Overall, whole-farm models are required to explore management options that will mitigate GHG emissions and promote best management practices. However, for full assessment of the production system, other benefits of the system (e.g., carbon sequestration, ecosystem services), which are not part of current whole-farm models, must be considered.

Mathematical modeling in animal agriculture can be applied at various levels including at the tissue, organ, animal, farm, regional and global levels. The purposes of this research were i) to evaluate models used to estimate volatile fatty acid (VFA) and methane (CH4) production and assess their impact on regional enteric CH4 inventory, and ii) to develop a process-based, whole-farm model to estimate net farm GHG emissions. In the first study, four VFA stoichiometric models were evaluated for their prediction accuracy of rumen VFA and enteric CH4 production. Comparison of measured and model predicted values demonstrated that predictive capacity of the VFA models varied with respect to the type of VFA in rumen fluid which impacted estimated enteric CH4 production. Moving to a larger scale assessment, we examined the enteric CH4 inventory from Manitoba beef cattle (from 1990 to 2008) using two mechanistic rumen models that incorporate VFA stoichiometric models: COWPOLL and MOLLY, and two empirical models: Intergovernmental Panel on Climate Change (IPCC) Tier 2 and a nonlinear equation (Ellis). The estimated absolute enteric CH4 production varied among models (7 to 63%) indicating that estimates of GHG inventory depend on model selection. This is an important consideration if the values are to be used for management and/or policy-related decisions. Development of models at the individual farm component level (animal, soil, crop) does not accurately reflect net GHG emissions generated from the whole production system. We developed a process-based, whole-farm model (Integrated Components Model, ICM), using the existing farm component models COWPOLL, manure-DNDC and some aspects of IPCC to integrate farm components and their associated GHG emissions. Estimates of total farm GHG emissions and their relative contribution using the ICM were comparable to estimates using two other whole-farm models (Integrated Farm System Model and Holos model). Variation was observed among models both in estimating whole-farm GHG emissions and the relative contribution of the different sources in the production system. Overall, whole-farm models are required to explore management options that will mitigate GHG emissions and promote best management practices. However, for full assessment of the production system, other benefits of the system (e.g., carbon sequestration, ecosystem services), which are not part of current whole-farm models, must be considered.