Working Paper No. 94 CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) Luis Orlindo Tedeschi Mario Herrero Phillip K. Thornton Working Paper An Overview of Dairy Cattle Models for Predicting Milk Production An Overview of Dairy Cattle Models for Predicting Milk Production Their Evolution, Evaluation and Application for the Agricultural Model Intercomparison and Improvement Project (AgMIP) for Livestock Working Paper No. 94 CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) Luis Orlindo Tedeschi Mario Herrero Phillip K. Thornton 1 Correct citation: Tedeschi L.O, Herrero M, Thornton P.K. 2014. An Overview of Dairy Cattle Models for Predicting Milk Production: Their Evolution, Evaluation, and Application for the Agricultural Model Intercomparison and Improvement Project (AgMIP) for Livestock. CCAFS Working Paper no. 94. CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS). Copenhagen, Denmark. Available online at: www.ccafs.cgiar.org Titles in this Working Paper series aim to disseminate interim climate change, agriculture and food security research and practices and stimulate feedback from the scientific community. This document is published by the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), which is a strategic partnership of the CGIAR and the Earth System Science Partnership (ESSP). CCAFS is supported by the CGIAR Fund, the Danish International Development Agency (DANIDA), the Australian Government Overseas Aid Program (AusAid), Irish Aid, Environment Canada, Ministry of Foreign Affairs for the Netherlands, Swiss Agency for Development and Cooperation (SDC), Instituto de Investigação Científica Tropical (IICT), UK Aid, and the European Union (EU). The Program is carried out with technical support from the International Fund for Agricultural Development (IFAD). Contact: CCAFS Coordinating Unit - Faculty of Science, Department of Plant and Environmental Sciences, University of Copenhagen, Rolighedsvej 21, DK-1958 Frederiksberg C, Denmark. Tel: +45 35331046; Email: ccafs@cgiar.org Creative Commons License This Working Paper is licensed under a Creative Commons Attribution – NonCommercial–NoDerivs 3.0 Unported License. Articles appearing in this publication may be freely quoted and reproduced provided the source is acknowledged. No use of this publication may be made for resale or other commercial purposes. © 2014 CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS). CCAFS Working Paper no. 94 DISCLAIMER: This Working Paper has been prepared as an output for the Data and Tools theme under the CCAFS program and has not been peer reviewed. Any opinions stated herein are those of the author(s) and do not necessarily reflect the policies or opinions of CCAFS, donor agencies, or partners. All images remain the sole property of their source and may not be used for any purpose without written permission of the source. 2 Abstract The contemporary concern about anthropogenic release of greenhouse gas (GHG) into the environment and the contribution of livestock to this phenomenon have sparked animal scientists’ interest in predicting methane (CH4) emissions by ruminants. Focusing on milk production, we address six basic nutrition models or feeding standards (mostly empirical systems) and five complex nutrition models (mostly mechanistic systems), describe their key characteristics, and highlight their similarities and differences. Four models were selected to predict milk production in lactating dairy cows, and the adequacy of their predictions was measured against the observed milk production from a database that was compiled from 37 published studies from six regions of the world, totalling 173 data points. We concluded that not all models were suitable for predicting predict milk production and that simpler systems might be more resilient to variations in studies and production conditions around the world. Improving the predictability of milk production by mathematical nutrition models is a prerequisite to further development of systems that can effectively and correctly estimate the contribution of ruminants to GHG emissions and their true share of the global warming event. Keywords Adequacy; Comparison; Modelling; Nutrition; Simulation; Testing. 3 About the authors Luis O. Tedeschi is an Associate Professor in the Department of Animal Science at Texas A&M University and Texas A&M AgriLife Research, College Station, TX. Contact: luis.tedeschi@tamu.edu Mario Herrero works as Chief Research Scientist, Food Systems and the Environment, at The Commonwealth Scientific and Industrial Research Organisation (CSIRO), Brisbane, Australia. Contact: Mario.Herrero@csiro.au Philip Thornton leads Theme 4 at CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) and is based at the International Livestock Research Institute (ILRI). Contact: P.Thornton@cgiar.org 4 Acknowledgements Authors express their sincere appreciation to those who assisted them by providing accurate information on the history of the development of nutrition models and who improved the discussion of this manuscript. Special thanks to André Bannink (Wageningen University), Antonello Cannas (University of Sassari), Danny Fox (Cornell University), James France (University of Guelph), Pekka Huhtanen (Swedish University of Agricultural Sciences), Ermias Kebreab (University of California-Davis), and John McNamara (Washington State University). Some of this work was supported by funding through the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) from the CGIAR Fund, AusAid, Danish International Development Agency, Environment Canada, Instituto de Investigacão Científica Tropical, Irish Aid, Netherlands Ministry of Foreign Affairs, 20 Swiss Agency for Development and Cooperation, Government of Russia, UK Aid, and the European Union, with technical support from the International Fund for Agricultural Development. 5 Contents Introduction .................................................................................................................... 9 Evolution of Nutrition Models ..................................................................................... 11 Description of empirical nutrition models ............................................................... 11 Description of mechanistic nutrition models ........................................................... 16 Intercomparison of Model Predictions for Milk Production........................................ 22 Description of the study database ............................................................................ 22 Description of the feedstuff information.................................................................. 23 Description of the animal information ..................................................................... 24 Assessment of model adequacy and regressions ..................................................... 25 Results of the Model Intercomparison ..................................................................... 25 Whole-Farm Modelling ............................................................................................... 30 Conclusion/Recommendations .................................................................................... 32 References .................................................................................................................... 33 Appendix 1. Detailed description of selected nutrition models currently in use ......... 53 6 Acronyms AFRC Agricultural and Food Research Council ANSJE Amino Acid and Nitrogen Supply Jolly Estimator APIM Agricultural Production Systems Simulator ARC Agricultural Research Council Cb Model accuracy CCC Concordance correlation coefficient CH4 Methane CHO Carbohydrate CNCPS Cornell Net Carbohydrate and Protein System CP Crude protein DairyMod Australian Dairy Grazing Systems DCP Digestible crude protein DIESE Discrete Event Simulation Environment DM Dry matter DMI Dry matter intake EE Ether extract FASSET Farm Assessment Tool FC Fibre carbohydrate FiM Feed into Milk FU Feed unit GE Gross energy GHG Greenhouse gas GPFARM Great Plains Framework for Agricultural Resource HPM Hurley Pasture Model IFSM Integrated Farm System Model iNDF Indigestible neutral detergent fibre INRA Institut National de La Recherche Agronomique IPCC Intergovernmental Panel on Climate Change kd Fractional degradation rate kp Fractional passage rate LRNS Large Ruminant Nutrition System MB Mean bias MCP Microbial crude protein ME Metabolisable energy 7 MES Model Evaluation System MP Metabolisable protein MSEP Mean square error of prediction MY Milk yield NDF Neutral detergent fibre NE Net energy NFC Nonfibre carbohydrate NPN Nonprotein nitrogen NRC National Research Council OM Organic matter PAMELA Protozoa and Acid Metabolism Estimator; a Lift to ANSJE PaSim Pasture Simulation pdNDF Potentially digestible neutral detergent fibre RMSEP Root of mean square error of prediction SGS Sustainable Grazing Systems TDN Total digestible nutrients VFA Volatile fatty acids WFM Whole Farm Model 8 Introduction The Agricultural Model Intercomparison and Improvement Project (AgMIP1) is an international endeavour whose purpose is to bring together agricultural modelling communities with cutting-edge information technology to enhance models’ predictions and to foster the development of the next generation of models for the agricultural sector. Rosenzweig et al. (2013) indicated the goals of AgMIP are to improve world food security (e.g., meat and milk) and to enhance adaptation capacity across different regions of the globe in light of climate change. Thus, the intercomparison of models’ adequacy in predicting meat and milk production is an important step. Recent estimates from the Intergovernmental Panel on Climate Change (IPCC) indicate that agriculture was responsible for 13.5% of the global anthropogenic greenhouse gas (GHG) emissions in 2004 (IPCC, 2007). Agricultural methane (CH4) emissions, of which 33 to 39% is ruminant enteric CH4, accounts for about 60% of the global anthropogenic CH4 emissions (Moss et al., 2000). Consequently, about 2.7 to 3.2% of global anthropogenic GHG is due to CH4 emissions from ruminant enteric fermentation. To identify potential strategies for mitigating CH4 emissions to the environment, valid predictions of CH4 emissions by ruminants (e.g., cattle, sheep, and goats) must be available in order to accurately represent their share of GHG emissions (Gerber et al., 2013; Hristov et al., 2013; Tedeschi et al., 2011; Tedeschi et al., 2003). Several attempts have been made to predict CH4 emissions by ruminants, from simple empirical relationship regressions (Axelsson, 1949; Blaxter and Clapperton, 1965; Kriss, 1930) to more robust empirical regressions using dietary nutrient intake and composition (Ellis et al., 2009; Ellis et al., 2007; Jentsch et al., 2007; Moe and Tyrrell, 1979; Moraes et al., 2014; Ramin and Huhtanen, 2012, 2013; Wilkerson et al., 1995; Yan et al., 2009) to more complex systems using biochemical pathways and anaerobic fermentation stoichiometry (Baldwin, 1995; Danfær et al., 2006a; Dijkstra et al., 1992; Gill et al., 1989; Mills et al., 2001; Pitt et al., 1996) and thermodynamics (Janssen, 2010; Kohn and Boston, 2000). Though empirical regressions may generally yield more accurate and precise estimates of animal responses than mechanistic models for practical applications (France et al., 2000), they may 1 http://www.agmip.org 9 not be as useful as mechanistic/dynamic systems in understanding the mechanisms underlying CH4 production and in providing opportunities to discover strategies for mitigating CH4 emissions under different scenarios of production. Benchaar et al. (1998), however, found that mechanistic models provided greater precision and accuracy than empirical regressions and that mechanistic models could be calibrated through adjustment factors to yield even better predictions. Several attributes of commercial ruminant production poses additional complications to developing comprehensive mathematical models. The first issue is methodological: the discrepancy between methods can bias model predictions. For example, the respiration chamber method measures total CH4 (rumen and hindgut), whereas sulphur hexafluoride (SF6) method determines rumen CH4 only, which does not represent complete recovery of the ruminant animal’s total CH4 emission, though the difference can be as low as 4% or less for SF6 (McGinn et al., 2006a). A revised methodology for the SF6 tracer technique has been proposed (Deighton et al., 2014). The second issue is statistical: in addition to predictive errors of mathematical models, the observed data also contains sources of random, unknown errors. The scientific community does not entirely accept the use of meta-analytical techniques to overcome this issue because, when predicting CH4 emission, one cannot remove the random errors associated with a random factor (e.g., studies) if its share of the total variance is considerable. The third issue is related to grazing ruminants: besides the inherent difficulties in modelling grazing ruminants (Teague et al., 2013), the techniques for measuring CH4 emission in pasture conditions are challenging, even the use of open-path laser scanners to determine concentrations of CH4 has produced contradictory or incomplete results (McGinn and Beauchemin, 2012; McGinn et al., 2006b; McGinn et al., 2011; Tomkins et al., 2011). Though mathematical nutrition models for ruminants can predict animal performance under different environmental conditions, it is not entirely clear which nutrition model can adequately predict animal production in different parts of the world. Most intercomparisons of the adequacy of livestock mathematical models’ predictions of animal response have been performed only as needed, and few have used guided and experimentally designed comparisons. Often, model evaluations are perceived as ways to promote the use of one system over another rather than highlight important gaps and limitations of scientific knowledge and how to address them. It is very common to find publications about models that have been “validated”; when in reality models cannot be validated in the sense of proving their correctness and utility for future predictions; they can simply be evaluated on an ad hoc basis (Tedeschi, 2006). For beef cattle, intercomparisons of livestock ruminant models for the performance of growing animals have been conducted (Arnold and Bennett, 1991a, b), but recent models for beef cattle have not been compared. 10 Others have made partial comparisons or critiques of specific elements of nutrition models (Alderman, 2001; Alderman et al., 2001a, b; Bannink et al., 1997; Dijkstra et al., 2008; Sauvant, 1996; Sundstøl, 1993; Tedeschi et al., 2013a), but there have been only limited comparisons about the models’ ability to predict growth or milk yield (MY) in different parts of the world under distinct production scenarios (e.g., feedstuffs, breeds, management, and climatic factors). Furthermore, model comparisons for nutrient excretion and GHG emissions of modern livestock operations are lacking (Tedeschi et al., 2005b), mainly for regions of the world that produce large quantities of meat and milk to support the livelihood of humans. The goal of this manuscript was to update and expand the discussion on the evolution and evaluation of models for milk production by Tedeschi et al. (2014). The first objective is to provide the evolution and a brief synopsis of important mathematical nutrition models that can be used to assess animal performance (i.e., milk production) or CH4 emission of dairy cattle production systems in different parts of the world. The second objective is to a show preliminary comparison of a subset of these mathematical nutrition models on the adequacy of their predictions of MY. The third objective is to discuss the inclusion of ruminant nutrition models into modern whole farm models. Evolution of Nutrition Models Figure 1 shows the chronological evolution of relevant nutrition models and feeding standard systems that were developed to facilitate fundamental research as well as on-farm applications for evaluation and formulation of diets. The models described below are usually classified as empirical systems and they formed the conceptual basis for the development of more complex, modern mathematical nutrition models. Appendix 1 has detailed information about selected nutrition models provided by their developers. Description of empirical nutrition models The American model—National Research Council (NRC) The NRC system is based on the work of the group within the United States Department of Agriculture (Beltsville, Maryland) studying dairy cattle by using respiration chambers (Moe et al., 1972; Moe and Tyrrell, 1974; Moe and Tyrrell, 1975; Moe et al., 1970; Moe et al., 1971; Tyrrell and Moe, 1975a; Tyrrell and Moe, 1975b; Tyrrell et al., 1970) and studies conducted at the University of California-Davis for growing and finishing cattle by using the comparative slaughter technique (Garrett et al., 1959; Lofgreen, 1965; Lofgreen and Garrett, 1968; Lofgreen et al., 1962). The first Recommended Nutrient Allowances for Dairy and Recommended Nutrient Allowances for Beef were published in 1945 (NRC, 1945a; 1945b). 11 12 Figure 1.Chronological evolution of mathematical nutrition models (red boxes) and key references (blue boxes). Year of publication or release is shown on the left. The green boxes represent models not yet released to the public. The solid line represents a direct relationship of influence, and the dashed line represents that at least one other version or edition was released in between the marks. References are: (A1) (NRC, 1945a; NRC, 1945b) (A2) Leroy (1954), (B1) (Blaxter, 1962), (B2) Van Soest (1963a) and Van Soest (1963b), (C1) Nehring et al. (1966), (C2) Lofgreen and Garrett (1968), (C3) Moe et al. (1970), (D1) Schiemann et al. (1971), (D2) Waldo et al. (1972), (D3) Hoffmann et al. (1974), (D4) Ministry of Agriculture, Fisheries and Food (1975), (D5) Van Es (1975), (E1) Baldwin et al. (1977), (E2) Baldwin et al. (1980), (F1) France et al. (1982), (F2) Gill et al. (1984), (F3) Fox and Black (1984), (F4) Conrad et al. (1984), (G1) Danfær (1990), (H1) Illius and Gordon (1991), (H2) France et al. (1992), (H3) Russell et al. (1992), Sniffen et al. (1992), and Fox et al. (1992), (H4) Dijkstra et al. (1992), Neal et al. (1992), and Dijkstra (1993), (H5) Tamminga et al. (1994), (J1) Nagorcka et al. (2000), (J2) Mills et al. (2001), (J3) Fox et al. (2004), (J4) Cannas et al. (2004), (K1) Bannink et al. (2006), (K2) Bannink et al. (2008), and (L1) Gregorini et al. (2013). RNS is the Ruminant Nutrition System. Adapted from Tedeschi et al. (2014). 13 The latest revision of the Nutrient Requirements for Dairy Cattle was released in 2001 (NRC, 2001). Nutrient Requirements for Beef Cattle was published in 1996 (NRC, 1996) and updated in 2000 (NRC, 2000). The British model—Agricultural Research Council (ARC) Van Es (1975) compared the feeding standard systems of dairy cows developed by the American (Beltsville), German (Rostock), Dutch (Wageningen), and British groups, and concluded that the starch equivalent system incorrectly evaluated feeds for dairy cows. Van Es (1975) proposed that the newly obtained information from energy balance trials with dairy cows should be used instead. The starch equivalent system assumed that the feeding values of feedstuffs for growing and finishing steers would rank the same when fed to lactating dairy cows. The first livestock system by the ARC was released in 1965 (ARC, 1965), and it was substantially revised in 1980 (ARC, 1980). The 1965 publication relied largely on data from non-lactating ruminants and used the factorial approach (Van Es, 1975). The ARC (1965) adopted the metabolisable energy (ME) feeding system developed by Blaxter (1962). Later, the metabolisable protein (MP) was introduced in the ARC (1980). Technical reports published in 1990 and 1991 by the Agricultural and Food Research Council (AFRC) further modified the ARC (1980). A revised publication in 1993 incorporated these modifications (AFRC, 1993). Agnew and Yan (2000) indicated that the main limitations of the ME system as adopted by the ARC and AFRC publications were the lack of calorimetric data obtained in the UK and the ancient data used to develop these systems. The MP system of the AFRC (1993) was based on basal endogenous nitrogen (N) losses at a maintenance level of intake, and there was no provision to adjust for cows consuming at production level. The Feed into Milk (FiM) project was developed to overcome these limitations, but it still uses the original concepts proposed in the 1960s. Agnew and Newbold (2002) provide a more detailed discussion about the evolution of the feeding systems in the UK. The German model—Rostock Feed Evaluation System Jentsch et al. (2003) and Chudy (2006) describe the evolution of the German feeding standards. Oskar Kellner’s Starch Value System was the main ruminant feeding system until the twentieth century. It was based on Gustav Kühn’s methods and the energy metabolism of adult oxen. Kurt Nehring validated the use of the starch value system to assign energy values to different feeds, using open-circuit respiration chambers at the Oskar Kellner Institut fur Tierernahrung of the Academy of Agricultural Science. This research site used four airconditioned respiration chambers to study the fundamentals of energy metabolism and nutrient utilization in farm animals, to apply the results in feed evaluation, and to develop a complete feed evaluation system (Chudy, 2006). The publications by Nehring et al. (1966), Schiemann et al. (1971), and Hoffmann et al. (1974) formed the basis of the Rostock Feed Evaluation System. In 1971, the system was referred to as the “GDR Feed Evaluation 14 System,” and seven editions were published until 1989. The most current edition (Beyer et al., 2003) was revised by Jentsch et al. (2003). The French model—Institut National de la Recherche Agronomique (INRA) In France, for a long time, the energy value of feeds and the energy requirements of ruminant animals were based on the feed unit (FU) system developed in 1954 by André M. Leroy (Leroy, 1954). In the 1970s, the INRA proposed a new system based on the same principles adopted by the Netherlands (Van Es, 1978) and Switzerland (Bickel and Landis, 1978), systems in which the net energy (NE) values of feeds were estimated from ME and from the partial efficiency of use of ME for maintenance, growth, and lactation. However, it differed in three aspects: the ME content of feeds was computed from gross energy (GE), the NE content of feeds was still expressed in FU equivalent, and the energy allowances for growing and finishing cattle of different breeds were determined using experiments conducted in France (Vermorel, 1978). These modifications were published by the INRA in its 1978 publication Alimentation des Ruminants. The INRA updated the French system in 1988 (INRA, 1988) and 1989 (INRA, 1989). The latest revised publication of the French system was released in 2007 (INRA, 2007). The Australian model—Commonwealth Scientific and Industrial Research Organization (CSIRO) The first publication of feeding standards developed by the CSIRO was released in 1990 (CSIRO, 1990) and revised in 2007 (CSIRO, 2007). These standards are based on the UK feeding standards (AFRC, 1993; ARC, 1965; ARC, 1980). The CSIRO system (CSIRO, 1990) was the foundation of decision support systems for Australian conditions, named GrazPlan2, which included models such as GrazFeed, GrassGro, and AusFarm (Donnelly et al., 2002; Donnelly et al., 1997; Freer et al., 1997; Salmon et al., 2004). Nagorcka et al. (2000) proposed a more mechanistic and dynamic rumen model that would have included variables other than substrate types and ruminal pH to improve the description and accountability of ruminal production of volatile fatty acids (VFA). This model was named AusBeef, and it is to be included in the GrazPlan suite of models. The Dutch Feed Evaluation System As described by Van Es (1978), the energy evaluation system used in the Netherlands was based on the work of Van Es (1975) with modifications to the calculation of the ME content of feeds, the maintenance requirement, and the use of FU equivalency, in which one lactation 2 Available at http://www.grazplan.csiro.au 15 FU contained 1.65 Mcal of NE for lactation. Until 1991, the protein system was based on the digestible crude protein (DCP), and in 1994 the DVE/OEB, which is based on the MP concepts adopted by the INRA (1989), was proposed for inclusion in the Dutch protein system (Tamminga et al., 1994). Description of mechanistic nutrition models The models in the next group are usually classified as mechanistic because they contain conceptual and mechanistic elements in their logical structure. Some are intrinsically dynamic while others are static (i.e., time is not a continuous variable), but all of them are deterministic. Molly Molly3 is a dynamic, mechanistic model based on biochemical reactions in animal metabolism (Baldwin, 1995). However, Molly was not the first mechanistic model; it came after Myrtle and Daisy (France, 2013), which were conceptualized based on the combination of extant models of rumen functions (Baldwin et al., 1977; France et al., 1982) and metabolism (Baldwin et al., 1980; Gill et al., 1984). The Myrtle’s rumen model was developed to address the nutrient supply of North American diets and the metabolism model was designed to describe nutrient partitioning and energy balance of lactating cows (France, 2013). Myrtle and Daisy were described by Baldwin et al. (1987b), Baldwin et al. (1987c), and Baldwin et al. (1987a). Finally, after six years of improvements and modifications to the code (France, 2013), Baldwin (1995) released Molly. The present research programs in Australia (e.g., AusBeef; Nagorcka et al. (2000)) and in New Zealand are in many ways based on Baldwin's work (he spent two sabbatical leaves there with John Black and Bruce Robson), and Molly's successors are in active use there (John P. McNamara, personal communication). More recently, Hanigan et al. (2009) have updated Molly with sophisticated parameters fitting based on new datasets assembled over the last 20 years. Others other have challenged and improved the energy and adipose functions of Molly and integrated Molly with a model of reproductive processes (Boer et al., 2011) to create the first integrated model of nutritional and reproductive processes, known as Jenny (McNamara and Shields, 2013). Cornell Net Carbohydrate and Protein System (CNCPS) The most recent complete version of the CNCPS was published by Fox et al. (2003) and Fox et al. (2004). It includes both beef and dairy cattle with two levels of solution (L1 and L2). 3 Available at http://www.vmtrc.ucdavis.edu/metabolic 16 Modifications have been made to L2 for the Cornell-Penn-Miner Institute (CPM) Dairy as described by Tedeschi et al. (2008), to CNCPS version 6.0 as described by Tylutki et al. (2008), and to subsequent CNCPS versions (Van Amburgh et al., 2010; Van Amburgh et al., 2013; Van Amburgh et al., 2009). The original description of the mechanistic ruminal fermentation submodel of the CNCPS was published in early 1990s (Fox et al., 1992; O'Connor et al., 1993; Russell et al., 1992; Sniffen et al., 1992), and additional modifications and new submodels have been developed since then (Lanzas et al., 2008; Lanzas et al., 2007a; Lanzas et al., 2007b; Seo et al., 2006; Tedeschi et al., 2002a; Tedeschi et al., 2008; Tedeschi et al., 2000a; Tedeschi et al., 2005a; Tedeschi et al., 2013b; Tedeschi et al., 2002b; Tedeschi et al., 2000b; Tedeschi et al., 2001; Tedeschi et al., 2006; Tylutki et al., 1994). Derivative models have been developed and deployed. CPM Dairy was developed for dairy cattle based on the computational engine of the CNCPS version 5.0 with additional features (Boston et al., 2000; Tedeschi et al., 2008) and Chalupa and Boston (2003) provide a historical perspective on the development of CPM Dairy. Similarly, the Large Ruminant Nutrition System4 (LRNS) is based on the calculation logic of the CNCPS version 5.0. The AMTS.Cattle.Pro5 is an implementation of the CNCPS version 6.1. Like the beef NRC (2000), the LRNS has two levels of solution: the L1 uses empirical equations to compute total digestible nutrients (TDN), ME, NE, and MP whereas L2 uses the fractionation of protein, fractional rates of ruminal degradation and ruminal passage, microbial crude protein (MCP) using the microbial growth submodel (Russell et al., 1992; Tedeschi et al., 2000b), and intestinal digestibility to compute MP. The MCP yield is predicted by two groups: those that grow slowly on fibre carbohydrates (FC) and those that grow more rapidly on nonfibre carbohydrates (NFC). Each feed carbohydrate (CHO) fraction (A is sugars, B1 is starch and pectins, B2 is available neutral detergent fibre (NDF), and C is unavailable fibre) and protein fraction (A is nonprotein N (NPN), B1 is soluble true, B2 is non-cell-wall, B3 is available cell wall, and C is unavailable cell wall) has its own fractional degradation rate (kd). Undegraded fractions flow out of the rumen with either the solid or the liquid passage rate (kp). CNCPS version 6.0 (Tylutki et al., 2008) expanded the CHO fractions were expanded to provide separate pools for organic and volatile fatty acids and soluble fibre, as documented by Lanzas et al. (2007a), and to provide new kp empirical equations developed by Seo et al. (2006). In CNCPS version 6.1 (Van Amburgh et al., 2010), peptides were shifted from the NPN to the soluble protein fraction that degrades with a reduced kd, and the liquid kp is used to predict the proportion of this fraction that passes undegraded from the rumen, as documented by Lanzas et al. (2008). 4 Available at http://nutritionmodels.tamu.edu/lrns.html or http://nutritionmodels.com/lrns.html 5 Available at https://www.agmodelsystems.com/AMTS/cattlepro.php 17 Ruminant This model was first described by Herrero (1997) and the latest complete description presented in Herrero et al. (2013); it is largely based on the work of Illius and Gordon (1991), Sniffen et al. (1992) and AFRC (1993). It consists of a dynamic section that estimates intake and the supply of nutrients to the animal from the fermentation kinetics and passage of feed constituents (carbohydrate and protein) through the gastrointestinal tract and their subsequent excretion, whereas another section determines their nutrient requirements using well recognised principles. Feeds are described by four main constituents: ash, fat, carbohydrate, and protein. These are divided into soluble, insoluble but potentially degradable, and indigestible fractions. Carbohydrate fractions represent non-structural carbohydrates (solCHO), potentially digestible cell wall, and the indigestible residue. For concentrate feeds, the proportion of starch in the solCHO is also used. Starch and fat in forages are almost negligible, but they may be important fractions in grains. The protein fractions are the same as those estimated in the MP system (AFRC, 1993), with the difference that their representation in this model is dynamic. The pools of digested nutrients obtained from the model are used to calculate the supply of nutrients to the animals. The model takes as inputs the quantities of fermentable nutrients available in a particular time step and returns as outputs the products of fermentation. The inputs are fermentable carbohydrate separated into simple sugars, starch, and cell wall material; fermentable N separated into ammonia and protein; and lipid, each summed across the various feed constituents, together with the microbial pool size. The outputs are the quantities of new microbial matter, the individual VFA, CH4, ammonia, and unfermented carbohydrates. It is assumed that there is only a single pool of microorganisms of fixed composition. The microbial maintenance requirement was set at 1.63 mM of ATP per gram of microbial dry matter (DM) per hour. The quantities of individual VFA and CH4 produced are calculated according to the quantities of different substrates fermented. There is no fixed upper limit to the quantity of microbial matter produced; the lower limit is zero growth. If fermentable N supply limits the amount of fermentable carbohydrate that can be used, unfermented carbohydrate is returned to the appropriate rumen pool, thus reducing the effective rate of carbohydrate fermentation. The model is generic and can simulate animals of different bodyweights because of the incorporation of allometric rules for scaling passage rates. The model also includes explicit protein–energy interactions, feeding level effects on passage rates, and pH effects on cell wall degradation rates. These aspects are essential for predicting stoichiometry changes, the effect of different supplementation regimes, and the substitution effects of forages and concentrates. This model has been used in a number of systems analysis studies of feeding strategies for ruminants (Castelán-Ortega et al., 2003; Herrero et al., 1999), herd replacement decisions (Vargas et al., 2001), trade-offs in smallholder systems (Waithaka et al., 2006), greenhouse gas emissions an mitigation 18 strategies in livestock systems (Bryan et al., 2013; Havlík et al., 2014; Herrero et al., 2013; Herrero et al., 2008; Thornton and Herrero, 2010). Dutch Tier 3 The rumen fermentation models developed by Baldwin et al. (1977) and Black et al. (1981) had limitations. The model by Beever et al. (1981) was unable to predict duodenal flow of protein diets when low protein content were simulated, and the model by Baldwin et al. (1987c) could not describe fibre fermentation of high-concentrate diets. An attempt to modify these models and improve the predictions of VFA production in the rumen culminated in the development of another model that was described by Dijkstra et al. (1992) and evaluated by Neal et al. (1992), which led to the development of the Amino Acid and Nitrogen Supply Jolly Estimator (ANSJE) model (J. Dijkstra, personal communication). Subsequently, Mills et al. (2001) added an empirical representation of digestion occurring in the small intestine and a mechanistic representation of fermentation occurring in the hindgut. They also included the prediction of CH4 production, including new coefficients for VFA formation that Bannink et al. (2006) obtained from data on lactating cows alone. After these modifications to the original ANSJE model, a computer interface was added, creating COWPOLL, a decision support tool for evaluating dairy cow diets for their pollution impact. Later, Bannink et al. (2008) developed a more mechanistic approach that made the formation of VFA in the rumen dependent on pH. Simultaneously, they developed a model describing the absorption of VFA across the rumen epithelium and metabolism of VFA therein. Concurrently to ANSJE, Dijkstra (1994) developed a rumen model with specific focus on the representation of the presence and activity of protozoa: Protozoa and Acid Metabolism Estimator; a Lift to ANSJE (PAMELA; J. Dijkstra, personal communication), but this version has not been incorporated into the COWPOLL fermentation model. PAMELA was evaluated by Dijkstra and Tamminga (1995). Since 2005, the model published by Mills et al. (2001), which included the VFA formation of Bannink et al. (2008) (which itself replaced that of Bannink et al. (2006)), has been used as a Tier 3 approach to estimating CH4 emission in dairy cattle for the national inventory report in the Netherlands (Bannink et al., 2011). It is commonly called “Dutch Tier 3”. In recent years, further modifications have been made to the rumen and large intestine models, a mechanistic version has been developed for the small intestine submodel, and calculations on manure production and composition have been added. These modifications have not yet been made public (A. Bannink, personal communication). Karoline The dairy cow model Karoline, a dynamic and mechanistic model component of the Nordic feed evaluation system NorFor (Volden, 2011), allocates the feed CHO into eight fractions: forage indigestible NDF (iNDF), forage potentially digestible NDF (pdNDF), concentrate 19 iNDF, concentrate pdNDF, starch, lactic acid, VFA (acetic, propionic, and butyric acids), and a heterogeneous remainder pool that is calculated by subtracting CHO, crude protein (CP), and ether extract (EE) from organic matter (OM), and that most likely contains water-soluble CHO, pectic substances, plant organic acids, and alcohols produced during the silage fermentation process (Danfær et al., 2006a). The feed CP is separated into six fractions: ammonia N, free amino acids, peptides, soluble true protein, insoluble protein, and potentially indigestible protein (Danfær et al., 2006a). The feed fat is converted to fatty acids from feed’s EE by using different equations for forage and concentrate feedstuffs (Danfær et al., 2006a). Karoline allows the user to modify the ruminal kd for forage and concentrate pdNDF, starch, and insoluble true protein, but the other fractions are fixed. Danfær et al. (2006b) indicated that Karoline’s prediction errors for some digestion variables were smaller than those obtained with the Molly model (Baldwin, 1995) as evaluated by Hanigan et al. (2013). Karoline adopted a two-pool ruminal kinetics with selective fibre retention for CHO and a three-pool ruminal kinetics with selective insoluble protein retention for protein as described by Danfær et al. (2006a) and schematized in Figure 2. A selective retention model is used to mathematically account for escapable and non-escapable pools in the rumen (Allen and Mertens, 1988; Mertens, 1989, 2005). The non-escapable pool can only be degraded (i.e., digested) or transferred to another pool, but it cannot escape the rumen, so it represents an intermediate step before escaping ruminal fermentation. There is some evidence that selective retention models more adequately mimic the ruminal kinetics of forage (Huhtanen et al., 2006; Mertens, 1993), concentrate (Mambrini, 1997; Wylie et al., 2000), and starch (Tothi et al., 2003), but the need to obtain an additional fractional rate of release from one pool to another has be taken into account. Karoline also contains a hindgut model that behaves similarly to the rumen model but is simpler. Karoline’s prediction of CH4 is based on pool size of fermentable substrates and anaerobic fermentation kinetics (Sveinbjörnsson et al., 2006). 20 kr CHO Escapable rumen CHO Rumen CHO CHO intake rate kp CHO CHO release rate CHO degradation rate 1 kd CHO 1 CHO escape rate CHO degradation rate 2 kd CHO 2 kr PROT Rumen PROT PROT intake rate PROT release rate kp PROT Escapable rumen PROT kd PROT 2 kd PROT 1 PROT degradation rate 1 Soluble PROT intake rate PROT escape rate Rumen soluble PROT PROT degradation rate 2 Soluble PROT escape rate Soluble PROT degradation rate kp liquid kd soluble PROT Figure 2. Illustration of the ruminal kinetics of carbohydrate and protein fractions based on the Karoline model. Boxes represent state, stock, or level variables (units); double-line arrows represent flows or rates (units/time); and single-line arrows represent variable causation and interrelationships. CHO is carbohydrate, kr is fractional release rate, kd is fractional degradation rate, kp is fractional passage rate, and PROT is protein. Adapted from Danfær et al. (2006a). 21 Intercomparison of Model Predictions for Milk Production The models selected for this preliminary comparison were the LRNS version 1.0.30 (solutions L1 and L2), NRC6 (2001) version 1.1.9, and Molly. Whenever available, the model-predicted MY was the least between the energy-allowable MY or the protein-allowable MY. Description of the study database A database was developed to compare the adequacy of selected nutrition models in predicting MY of dairy cows from six distinct regions around the world—Africa, Asia, Europe, Latin America, North America, and Oceania—based on their animal production characteristics: types of feeds (e.g., silage-based, pasture), feeding system (intensive versus extensive), type of cattle (e.g., Holstein-based, crossbreds), level of intensification, and animal management, among other factors. The database comprised of 50 scientific papers published in peerreviewed journals from 1992 to 2014 (Abdullah et al., 2000; Alvarez et al., 2001; Assis et al., 2004; Auldist et al., 1999; Bargo et al., 2001; Chantaprasarn and Wanapat, 2008; Chen et al., 2008; Colmenero and Broderick, 2006; Danes et al., 2013; Dey and De, 2014; Erasmus et al., 2013; Erasmus et al., 1992; Erasmus et al., 1994; Erasmus et al., 1999; Erasmus et al., 2004; Fatahnia et al., 2008; Grainger et al., 2010; Greenwood et al., 2013; Guo et al., 2013; Heard et al., 2007; Heard et al., 2004; Irvine et al., 2011; Jesus et al., 2012; Kalscheur et al., 1999; Khezri et al., 2009; Kokkonen et al., 2000; Lehmann et al., 2007; Liu et al., 2008; Lunsin et al., 2012; McCormick et al., 2001a; McCormick et al., 2001b; Meeske et al., 2009; Moallem, 2009; Moharrery, 2010; Mosavi et al., 2012; Murphy, 1999; O'Mara et al., 1998; O’Mara et al., 2000; Oguz et al., 2006; Petit and Gagnon, 2011; Piamphon et al., 2009; Sanh et al., 2002; Suksombat and Chullanandana, 2008; Sun et al., 2009; Vafa et al., 2012; Valizadeh et al., 2010; Walker et al., 2010; Yalçın et al., 2011; Yan et al., 2011; Yarahmadi and Nirumand, 2012). The database contained 173 observations (19 for Africa, 45 for Asia, 16 for Europe, 12 for Latin America, 44 for North America, and 37 for Oceania) with the minimum information needed for simulation, such as animal and feedstuff characteristics, dry matter intake (DMI), and milk composition and production. Common feedstuff and animal databases were developed, and functions were created to import and export the data from one model to another using R (R Core Team, 2014). 6 Available at https://nanp-nrsp-9.org/nrc-dairy-model 22 Description of the feedstuff information The feedstuff database contained 173 records obtained from the studies. Missing information on needed dietary composition was obtained from the LRNS feed library, NRC (2000, 2001) feed libraries, American and Canadian tables of feed composition (NRC, 1982), and the Brazilian feedstuff composition repository7. The 10 most common feeds were finely ground dry corn, finely ground soybean meal, corn silage, barley grain, urea, wheat bran, beet pulp shreds, fishmeal, blood meal, and corn gluten. Table 1 summarizes the statistics of the main feed nutrients. Some specific mixes had to be created in order to maintain all ingredients used in the dataset. Table 1. Descriptive statistics of the feedstuff and dietary chemical compositions and animal characteristics Items 1 Median Mean SD 1 Range Min Max Quartiles 25% 75% Diets DM, % as-fed 52.0 57.2 16.6 34.4 90.1 45.7 69.7 Fat, % DM 4.3 4.5 1.7 1.8 11.3 3.2 5.2 Ash, % DM 8.4 8.1 1.6 4.3 11.4 7.0 9.2 CP, % DM 18.0 18.2 2.8 10.1 25.5 16.3 20.2 Soluble CP, % CP 35.0 34.4 6.7 21.5 56.0 28.9 38.4 NPN, % CP 56.5 54.1 21.9 17.4 97.8 29.6 69.5 NDFIP, % CP 14.9 16.0 5.1 6.0 30.9 12.5 18.5 ADFIP, % CP 6.0 5.8 3.0 2.0 19.4 3.1 7.2 Starch, % NFC 70 67.5 14.8 27.2 93.7 59.0 78.9 35.8 34.9 5.8 22.2 46.8 30.3 38.7 8.5 8.3 2.5 3.6 13.4 6.3 11.0 DM, % as-fed 90 79.7 26.9 11.3 100 86 97 Fat, % DM 2.6 6.8 18.3 0 100 0.2 3.9 Ash, % DM 7.5 24.8 37.0 0 100 4.0 13.3 CP, % DM 13.5 25.4 44.0 0 281 5.8 26.8 NDF, % DM Lignin, % NDF Feedstuffs 7 Soluble CP, % CP 21.0 26.4 24.9 0 100 4.0 40.0 NPN, % CP 55.0 46.4 40.1 0 100 0 89.0 NDFIP, % CP 8.0 11.9 13.5 0 75 0 18.0 ADFIP, % CP 2.0 3.9 5.5 0 65 0 6.4 Starch, % NFC 64.0 54.0 39.9 0 100 0 90.0 Available at http://cqbal.agropecuaria.ws/webcqbal/index.php 23 Items 1 NDF, % DM Median Mean SD 1 Range Min Max Quartiles 25% 75% 15.1 21.4 20.3 0 78.9 0 37.8 4.3 6.0 6.4 0 30.7 0 10.4 SBW, kg 567 555 66.4 345 660 522 598 DMI, kg/d 19.1 19.1 3.5 9.1 27.5 17.3 22.1 Lignin, % NDF Animals 1 DIM, days 100 114 64.7 30 265 60 150 MY, kg/d 26.3 26.4 8.3 7.6 45 19.7 32.7 Milk fat, % 3.7 3.7 0.5 2.3 5.0 3.4 4.1 Milk protein, % 3.0 3.1 0.3 2.4 3.9 2.9 3.3 DM = dry matter, CP = crude protein, NPN = nonprotein nitrogen, NDF = neutral detergent fibre, NDFIP = NDF insoluble protein, ADFIP = acid detergent fibre insoluble protein, NFC = non-fibre carbohydrate, SBW = shrunk body weight, DMI = DM intake, DIM = days in milk (i.e., days after calving), MY = milk yield, and SD = standard deviation. Most of the values for kd of the protein and carbohydrate fractions, as well as for mineral and vitamin compositions, were from the LRNS feed library. The approach adopted by Hanigan et al. (2006) was used to obtain the nutrients and fractions needed by Molly, and the ME inputted was the ME predicted by the L1 solution of the LRNS. The feedstuff DMI was calculated as the dry matter (DM) percentage of each feedstuff multiplied by the observed DMI. The data was only used if it was possible to compute DMI for animals with ad libitum access to feeds. Description of the animal information All animal information provided in the studies was used as inputs. However, when relevant information such as mature body weight was not available from the studies, the LRNS default values for each breed were used. When no information was available for pregnancy days and days since calving, values such as less than 100 days and more than 60 days, respectively, were inputted to avoid conflict. Thus, significant pregnancy requirements and negative energy balance were not accounted for. For Molly, the udder cell parameter was estimated as suggested by Palliser et al. (2001) as 179.1 × � 0.053×�� , in which MY is mature daily peak milk yield (L/d). Not enough information was provided to account for the effect of body weight and body condition score changes on predicted MY (Tedeschi et al., 2006). Table 1 summarizes the statistics of the animal characteristics. 24 Assessment of model adequacy and regressions The adequacy of the models was assessed by using the Model Evaluation System8. The models were compared on the following statistics were used: mean bias (MB); concordance correlation coefficient (CCC); model accuracy (Cb); model precision (r2); mean square error of prediction (MSEP) and its decomposition into mean bias, systematic bias, and random errors; and MSEP square root (RMSEP) (Tedeschi, 2006). Statistical analyses were conducted with R version 3.1 (R Core Team, 2014) and graphics were generated with the ggplot2 package (Wickham, 2009). Linear regressions used observed values as the dependent variable (Y-axis) and the predicted values as the independent variable (X-axis), and the ordinary least square linear regressions were obtained with the lm function (R Core Team, 2014). For random coefficient models, studies were assumed to be a random effect, and the parameter estimates were obtained with the generalized linear mixed-effects regressions using the lme function of the nlme package (Pinheiro et al., 2014). The variance components of the random coefficient models (i.e., random errors and study errors) were estimated using a diagonal positive-definite matrix constructor and only a random intercept parameter was fitted. Results of the Model Intercomparison Figure 3 has the boxplots of the residue of observed minus predicted MY for each model and region. Models had different MY residue distribution, and mean and median values across regions, but the MY residue for North America were the most consistent with the least variation. The MY residue for Latin American had the largest variations. Figure 4 depicts scatter plots between observed and predicted MY using the selected nutrition models. There was a disproportionate number of studies that had adequate information to execute the selected nutrition models. Quantitative information of feed nutrition models has long been lacking (Arnold and Bennett, 1991b; Sauvant, 1996), and the problem persists today. Most mechanistic models are detail-oriented systems that attempts to account for as many biological concepts and relational structures as possible. This poses a problem when evaluating these models: the needed information may not be available at all times for all production conditions. In fact, the levels of aggregation differs significantly among nutrition models for lactating ruminants (Sauvant, 1996). Model reduction techniques, such as the replacement of model variables with constants, might be an alternative for situations in which complex models have to be used with limited information (Crout et al., 2009). In fact, the simplest of the models selected for this study, LRNS L1, seemed to have the best graphical representation (i.e., less scatter around the Y = X line; Figure 4). 8 Available at http://nutritionmodels.tamu.edu/mes.html or http://nutritionmodels.com/mes.html 25 26 Figure 3. Boxplots of observed minus predicted milk yield (kg/d) for five models (Large Ruminant Nutrition System (LRNS) using solution levels 1 and 2, NRC (2001), and Molly) for six regions around the world (Africa, Asia, Europe, Latin America, North America, and Oceania). The box represents the first and third quartile, the whiskers (vertical lines) represent the minimum and maximum values, the horizontal line within the box represents the median, the asterisk represents the mean, and the solid dots represent outliers. 27 Each model has a distinct predictive behaviour, their direct comparison difficult and incomplete. This finding agrees with previous comparisons between CNCPS-based and Molly-based nutrition models (Kohn et al., 1994). These two models differ substantially in their modelling scope (i.e., applied versus biochemical) and logical structure (i.e., dynamic versus empirical) Figure 4. Relationship between observed milk yield (Y axis) and study-unadjusted predicted milk yield (X axis) by the Large Ruminant Nutrition System (LRNS) using solution levels 1 (A) and 2 (B), NRC (2001) (C), Molly (D), and Ruminant (E) for data collected from six regions around the world (Africa, ; Asia, ; North America, ; Europe, , Latin America, ; and Oceania, ). Studies are represented by different colours and linear trendlines. The dashed line represents the linear regression of all data points and the shaded area represents the 95% confidence interval of the linear regression. The solid diagonal line represents the Y = X line. 28 Table 2 lists the statistics of the models’ adequacy in predicting MY. When study effect was not considered (i.e., data points within studies were assumed to be uncorrelated), MB varied from -4.06 (Molly) to 0.87 (LRNS L1) kg/d, and the RMSEP ranged from 5.6 (LRNS L2) to 8.07 (Molly) kg/d. These results suggest that, depending on the nutrition model used, a single-point prediction of MY might be between ±5.6 and ±8.07 kg/d different from the observed MY, but on average it can vary from -4.06 to 0.87 kg/d. Model precision (i.e., r2) was low to moderate and varied from 0.55 (Molly) to 0.69 (NRC, 2001). Although model accuracy (i.e., Cb) was high (> 0.88), the CCC was high for NRC (2001) (0.81), moderate for LRNS L2 (0.77) and LRNS L1 (0.79), and low for Molly (0.66) due to low model precision. The inadequacy of these models’ predictions (i.e., MSEP) was mostly due to random errors for LRNS L1, LRNS L2, and NRC (2001), whereas MB was 25.4% of MSEP for Molly. These diagnostics are not that different from those reported by Tedeschi et al. (2008), who evaluated the CPM Dairy model with data on high-producing dairy cows. They found an r2 of 0.798, CCC of 0.89, Cb of 0.997, and RMSEP of 5.14 kg/d. Tylutki et al. (2008) evaluated CNCPS version 6.0 and reported improved statistics compared to ours (e.g., r2 > 0.847, CCC > 0.918, and RMSEP < 4.5 kg/d). Their evaluations indicated greater model accuracy and precision most likely because their dataset was more homogenous and their feedstuffs were standard and came with detailed physicochemical descriptions. For grass-based diets, Dijkstra et al. (2008) compared the dietary energy value predicted by the AFRC, FiM, the Dutch NE system, and a version of the Dutch Tier 3 model. They reported that the Dutch Tier 3 model (a mechanistic model) was more precise and accurate than the other three models (which, essentially, were empirical, static models). Others have found that empirical, static models can also predicted dietary energy values accurately (Kaustell et al., 1997). For each graph in Figure 4, the inconsistency in the direction of the linear trendlines within studies suggests that the models were not accurate (and maybe not precise) within studies in predicting MY. Some study linear trendlines even have directions opposite to the Y = X line mark, indicating that as observed MY increased, models predicted less MY, or vice-versa. For other studies, a model predicted a change in MY, but observed MY was constant. In fact, most of the random variation (> 66%) was due to study effects (Table 2). This suggests that the lack of adequate inputs caused incomplete representation of the production scenarios, and that the models were unable to simulate the data because important variables were not part of the model or simply because the considerable variation within each study limited the predictability of MY. Other factors were likely not accounted for, such as inability to account for the impact of changes in body condition score on MY (Tedeschi et al., 2006). More complex models exist to predict such changes (Tedeschi et al., 2013b), but they would require even more specific inputs. 29 Table 2. Model adequacy statistics and variance component analysis of five models’ predictions of milk yield 1 Statistics N Mean (kg/d) Predicted (X) Observed (Y) MB (kg/d) r2 CCC Cb MSEP Root (kg/d) MB (%) Slope (%) Random (%) Variances (kg2/d2) σ2 (OLS) σ2 + σ2Study (GLS) σ2Study (GLS) % of σ2 + σ2Study 1 Milk yield not adjusted for study effect LRNS NRC Molly Level 1 Level 2 (2001) 173 173 173 164 Milk yield adjusted for study effect LRNS NRC Molly Level 1 Level 2 (2001) 173 173 173 164 25.5 26.4 0.87 0.68 0.79 0.96 28.1 26.4 -1.72 0.63 0.77 0.98 27.9 26.4 -1.49 0.69 0.81 0.98 30.1 26.1 -4.06 0.55 0.66 0.88 25.5 26.2 0.63 0.85 0.85 0.92 28.1 26.2 -1.88 0.84 0.86 0.94 27.9 26.1 -1.78 0.84 0.86 0.94 30.1 25.4 -4.75 0.93 0.77 0.80 6.21 1.95 40.8 57.3 5.6 9.39 9.36 81.3 5.38 7.72 18.1 74.2 8.07 25.4 29.9 44.7 5.04 1.56 69.1 29.3 3.95 22.8 32.9 44.3 4.29 17.2 42.8 39.9 6.35 55.9 36.2 7.87 22.3 27.9 18.7 67.0 25.9 27.8 19.3 69.2 21.7 27.2 18.1 66.5 29.5 37.9 33.9 89.4 7.53 — — — 6.98 — — — 7.44 — — — 3.21 — — — MB = mean bias, CCC = concordance correlation coefficient, Cb = model adequacy, MSEP = mean square error of prediction, OLS = ordinary least squares (using linear models, LM), GLS = generalized least squares (using linear mixed-effects model, LME), LRNS = Large Ruminant Nutrition System, NRC = National Research Council. 30 When the observed MY was adjusted for the random effects of studies (Table 2) using the random coefficient models with variance components for intercept and slope, as expected, the model precision increased (> 0.84), CCC increased (> 0.77), and RMSEP decreased (< ±6.35 kg/d). These statistics are more similar to those reported by Tedeschi et al. (2008), who also had previously adjusted MY for the random effects of studies. Whole-Farm Modelling The animal module of process-based whole-farm models that is used to estimate the flow of elements (e.g., C, N, and P) has posed an enduring model development challenge mainly for those systems that deal with grazing conditions. This challenge is caused by the intrinsic problem of estimating the amount and quality of the forage consumed by the animal. Several whole-farm simulation models have been developed and their animal modules differ considerably. Examples of whole-farm models include Agricultural Production Systems Simulator (APIM9) (Moore et al., 2007), Australian Dairy Grazing Systems (DairyMod) (Johnson et al., 2008) and Sustainable Grazing Systems (SGS) (Johnson et al., 2003) (both collectively referred to as AgMod10), DairyNZ Whole Farm Model, Discrete Event Simulation Environment (DIESE) (Martin-Clouaire and Clouaire, 2009), EcoMod (Johnson et al., 2008), Farm Assessment Tool (FASSET11) (Berntsen et al., 2003), Great Plains Framework for Agricultural Resource Management (GPFARM) (Andales et al., 2003), GRAZPLAN (Donnelly et al., 1997; Moore et al., 1997), Hurley Pasture Model (HPM), Integrated Farm System Model (IFSM12) (Rotz et al., 2005; Rotz et al., 1999), LINCFARM, Pasture Simulation (PaSim13) (Graux et al., 2011), PROGRASS, and Whole Farm Model (WFM) among many other systems. Bryant and Snow (2008) reviewed nine pastoral simulation models (APSIM, EcoMod, FASSET, GRAZPLAN, GPFARM, HPM, IFSM, LINCFARM, and WFM) and concluded that there was a need to include pests and diseases on pasture production as well as improved animal performance predictions, including a more mechanistic model for voluntary feed intake and ruminal fermentation processes. More 9 Available at http://www.apsim.info/ 10 Available at http://www.imj.com.au/consultancy/index.html 11 Available at http://www.fasset.dk/ 12 Available at http://www.ars.usda.gov/SP2UserFiles/Place/19020000/ifsmreference.pdf 13 Available at https://www1.clermont.inra.fr/urep/modeles/pasim.htm 31 recently, Snow et al. (2014) provided a brief summary of six of these models (APSIM, AgMod, DIESE, FASSET, GRAZPLAN, and IFSM) and compared their different approaches to model forage mixtures in the paddocks, animal-forage interactions, N transfers by the animal in the paddocks, management of the whole farm, and future prospects. They also provided ideas and solutions for the imminent limitations of these six models. Del Prado et al. (2013) indicated that whole-farm models are the appropriate scale for mitigating GHG emissions because the farm represents the unit at which management decisions are made. They analysed different approaches for modelling GHG. Most of these reviews discussed the strengths and drawbacks of whole-farm models, but there is a lack of model intercomparison under different production systems. Based on our intercomparison of ruminant nutrition models and the complexity of whole-farm models, our recommendation is that simple models such as the level 1 solution of the LRNS are used with whole-farm models to predict GHG emissions. More complex nutrition models can be implemented into whole-farm models if additional needed information is available and the complexity of the model does not impede or bias the interpretation of the simulations. Conclusion/recommendations In the first part of this manuscript, we highlighted that though mathematical nutrition models share similar assumptions and calculations, they have different conceptual and structural foundations inherent to their intended purposes. A direct comparison among these models was further complicated by the different models requiring unique inputs that are very often not available, and the low reliability of the inputs prevents an unbiased assessment of the models’ predictions. Very few studies have collected the necessary information to run more mechanistic systems, and users have to rely on standard information to simulate MY using many models. Study effect was a critical source of variation that limited our ability to conclusively evaluate the models’ applicability under different scenarios of production around the world. Only after study variation was removed from the database did the adequacy of the models’ predictions of milk production improve, but deficiencies still existed. Based on these analyses, we conclude that not all models are suitable for predicting milk production and that simpler systems might be more resilient to variations in studies and production conditions around the world. Improving the predictability of milk production by mathematical nutrition models is a prerequisite to further development of systems that can effectively and correctly estimate the contribution of ruminants to GHG emissions and their true share of the global warming event. 32 References Abdullah, M., J. W. Young, H. D. Tyler, and G. Mohiuddin. 2000. Effects of feeding high forage diets and supplemental fat on feed intake and lactation performance in dairy cow. Asian-Australasian Journal of Animal Sciences. 13:457-463. Agnew, R. E., and J. R. Newbold. 2002. Nutritional Standards for Dairy Cattle. Report of the British Society of Animal Science Nutritional Standards Working Group: Dairy Cows. 42 p. Agnew, R. E., and T. Yan. 2000. Impact of recent research on energy feeding systems for dairy cattle. Livest. Prod. Sci. 66:197-215. Agricultural and Food Research Council. 1993. Energy and Protein Requirements of Ruminants. Agricultural and Food Research Council. CAB International, Wallingford, UK. Agricultural Research Council. 1965. The Nutrient Requirements of Farm Livestock. No. 2, Ruminants. H.M. Stationery Office, London, UK. Agricultural Research Council. 1980. The Nutrient Requirements of Ruminant Livestock. Agricultural Research Council. The Gresham Press, London. Alderman, G. 2001. A critique of the Cornell Net Carbohydrate and Protein System with emphasis on dairy cattle. 1. The rumen model. J. Anim. Feed Sci. 10:1-24. Alderman, G., J. France, and E. Kebreab. 2001a. A critique of the Cornell Net Carbohydrate and Protein System with emphasis on dairy cattle. 2. The post-rumen digestion model. J. Anim. Feed Sci. 10:203-221. Alderman, G., J. France, and E. Kebreab. 2001b. A critique of the Cornell Net Carbohydrate and Protein System with emphasis on dairy cattle. 3. The requirements model. J. Anim. Feed Sci. 10:361-383. Allen, M. S., and D. R. Mertens. 1988. Evaluating constraints on fiber digestion by rumen microbes. J. Nutr. 118:261-270. Alvarez, H. J., F. J. Santini, D. H. Rearte, and J. C. Elizalde. 2001. Milk production and ruminal digestion in lactating dairy cows grazing temperate pastures and supplemented with dry cracked corn or high moisture corn. Animal Feed Science and Technology. 91:183-195. Andales, A. A., L. R. Ahuja, and G. A. Peterson. 2003. Evaluation of GPFARM for Dryland Cropping Systems in Eastern Colorado. Agron. J. 95:1510-1524. Arnold, R. N., and G. L. Bennett. 1991a. Evaluation of four simulation models of cattle growth and body composition: Part I - Comparison and characterization of the models. Ag. Syst. 35:401-432. 33 Arnold, R. N., and G. L. Bennett. 1991b. Evaluation of four simulation models of cattle growth and body composition: Part II - Simulation and comparison with experimental growth data. Ag. Syst. 36:17-41. Assis, A. J., J. M. S. Campos, S. C. V. Filho, A. C. Queiroz, R. P. Lana, R. F. Euclydes, J. M. Neto, A. L. R. Magalhaes, and S. S. Mendonca. 2004. Polpa citrica em dietas de vacas em lactação. 1. Consumo de nutrientes, produção e composição do leite. Revista Brasileira de Zootecnia. 33:242-250. Auldist, D. E., K. L. Atkinson, D. W. Dellow, M. J. Silvapulle, and G. H. McDowell. 1999. Utilisation of white clover silage fed alone or with maize silage by lactating dairy cows. Austr. J. Exp. Agric. 39:237-246. Axelsson, J. 1949. The amount of produced methane energy in the European metabolic experiments with adult cattle. Ann. R. Agric. Coll. Sweden. 16:404-419. Baldwin, R. L. 1995. Modeling Ruminant Digestion and Metabolism. Chapman & Hall, New York. Baldwin, R. L., J. France, D. E. Beever, M. Gill, and J. H. M. Thornley. 1987a. Metabolism of the lactating cow. III. Properties of mechanistic models suitable for evaluation of energetic relationships and factors involved in the partition of nutrients. J. Dairy Res. 54:133-145. Baldwin, R. L., J. France, and M. Gill. 1987b. Metabolism of the lactating cow. I. Animal elements of a mechanistic model. J. Dairy Res. 54:77-105. Baldwin, R. L., L. J. Koong, and M. J. Ulyatt. 1977. A dynamic model of ruminant digestion for evaluation of factors affecting nutritive value. Ag. Syst. 2:255-288. Baldwin, R. L., N. E. Smith, J. Taylor, and M. Sharp. 1980. Manipulating metabolic parameters to improve growth rate and milk secretion. J. Anim. Sci. 51:1416-1428. Baldwin, R. L., J. H. M. Thornley, and D. E. Beever. 1987c. Metabolism of the lactating cow. II. Digestive elements of a mechanistic model. J. Dairy Res. 54:107-131. Bannink, A., J. France, S. Lopez, W. J. J. Gerrits, E. Kebreab, S. Tamminga, and J. Dijkstra. 2008. Modelling the implications of feeding strategy on rumen fermentation and functioning of the rumen wall. Anim. Feed Sci. Technol. 143:3-26. Bannink, A., J. Kogut, J. Dijkstra, J. France, E. Kebreab, A. M. Van Vuuren, and S. Tamminga. 2006. Estimation of the stoichiometry of volatile fatty acid production in the rumen of lactating cows. J. Theor. Biol. 238:36-51. Bannink, A., M. W. van Schijndel, and J. Dijkstra. 2011. A model of enteric fermentation in dairy cows to estimate methane emission for the Dutch National Inventory Report using the IPCC Tier 3 approach. Anim. Feed Sci. Technol. 166-167:603-618. Bannink, A., H. Visser, and A. M. Van Vuuren. 1997. Comparison and evaluation of mechanistic rumen models. Br. J. Nutr. 78:563-581. 34 Bargo, F., D. H. Rearte, F. J. Santini, and L. D. Muller. 2001. Ruminal digestion by dairy cows grazing winter oats pasture supplemented with different levels and sources of protein. Journal of Dairy Science. 84:2260-2272. Beever, D. E., J. L. Black, and G. J. Faichney. 1981. Simulation of the effects of rumen function of the flow of nutrients from the stomach of sheep: Part 2—Assessment of computer predictions. Ag. Syst. 6:221-241. Benchaar, C., J. Rivest, C. Pomar, and J. Chiquette. 1998. Prediction of methane production from dairy cows using existing mechanistic models and regression equations. J. Anim. Sci. 76:617-627. Berntsen, J., B. M. Petersen, B. H. Jacobsen, J. E. Olesen, and N. J. Hutchings. 2003. Evaluating nitrogen taxation scenarios using the dynamic whole farm simulation model FASSET. Ag. Syst. 76:817-839. Beyer, M., A. Chudy, H. L. Hoffman, W. Jentsch, W. Laube, K. Nehring, and R. Schiermann. 2003. Rostock Feed Evaluation System; Reference numbers of feed value and requirement of the base of net energy. Gottlob Volkhardtsche Druckerei, Amorbach. Bickel, H., and J. Landis. 1978. Feed evaluation for ruminants. III. Proposed application of the new system of energy evaluation in Switzerland. Livest. Prod. Sci. 5:367-372. Black, J. L., D. E. Beever, G. J. Faichney, B. R. Howarth, and N. M. Graham. 1981. Simulation of the effects of rumen function on the flow of nutrients from the stomach of sheep: Part 1—Description of a computer program. Ag. Syst. 6:195-219. Blaxter, K. L. 1962. The Energy Metabolism of Ruminants. Hutchinson, London, UK. Blaxter, K. L., and J. L. Clapperton. 1965. Prediction of the amount of methane produced by ruminants. Br. J. Nutr. 19:511-522. Boer, H. M. T., C. Stötzel, S. Röblitz, P. Deuflhard, R. F. Veerkamp, and H. Woelders. 2011. A simple mathematical model of the bovine estrous cycle: Follicle development and endocrine interactions. J. Theor. Biol. 278:20-31. Boston, R. C., D. G. Fox, C. J. Sniffen, R. Janczewski, R. Munsen, and W. Chalupa. 2000. The conversion of a scientific model describing dairy cow nutrition and production to an industry tool: The CPM Dairy project. Pages 361-377 in Modelling Nutrient Utilization in Farm Animals. J. P. McNamara, J. France and D. Beever, eds. CABI Publishing, Oxford. Bryan, E., C. Ringler, B. Okoba, J. Koo, M. Herrero, and S. Silvestri. 2013. Can agriculture support climate change adaptation, greenhouse gas mitigation and rural livelihoods? insights from Kenya. Climatic Change. 118:151-165. Bryant, J. R., and V. O. Snow. 2008. Modelling pastoral farm agro‐ecosystems: A review. New Zealand Journal of Agricultural Research. 51:349-363. 35 Cannas, A., L. O. Tedeschi, D. G. Fox, A. N. Pell, and P. J. Van Soest. 2004. A mechanistic model for predicting the nutrient requirements and feed biological values for sheep. J. Anim. Sci. 82:149-169. Castelán-Ortega, O. A., R. H. Fawcett, C. Arriaga-Jordán, and M. Herrero. 2003. A Decision Support System for smallholder campesino maize–cattle production systems of the Toluca Valley in Central Mexico. Part II—Emulating the farming system. Ag. Syst. 75:23-46. Chalupa, W., and R. Boston. 2003. Development of the CNCPS and CPM models: The Sniffen affect. Pages 15-24 in Proceedings of Cornell Nutrition Conference for Feed Manufacturers, Syracuse, NY. New York State College of Agriculture & Life Sciences, Cornell University. Chantaprasarn, N., and M. Wanapat. 2008. Effects of sunflower oil supplementation in cassava hay based-diets for lactating dairy cows. Asian Australasian J. Anim. Sci. 21:42-50. Chen, P., P. Ji, and S. L. Li. 2008. Effects of feeding extruded soybean, ground canola seed and whole cottonseed on ruminal fermentation, performance and milk fatty acid profile in early lactation dairy cows. Asian Australasian J. Anim. Sci. 21:204-213. Chudy, A. 2006. Rostock Feed Evaluation System - An Example of the Transformation of Energy and Nutrient Utilization Models to Practical Application. Pages 366-382 in Nutrient Digestion and Utilization in Farm Animals: Modelling Approaches. E. Kebreab, J. Dijkstra, A. Bannink, W. J. J. Gerrits and J. France, eds. CABI Publishing, Cambridge, MA. Colmenero, J. J. O., and G. A. Broderick. 2006. Effect of dietary crude protein concentration on milk production and nitrogen utilization in lactating dairy cows. Journal of Dairy Science. 89:1704-1712. Commonwealth Scientific and Industrial Research Organization. 1990. Feeding Standards for Australian Livestock. Ruminants. Commonwealth Scientific and Industrial Research Organization, Melbourne, Australia. Commonwealth Scientific and Industrial Research Organization. 2007. Nutrient Requirements of Domesticated Ruminants. Commonwealth Scientific and Industrial Research Organization, Collingwood, VIC. Conrad, H. R., W. P. Weiss, W. O. Odwongo, and W. L. Shockey. 1984. Estimating net energy lactation from components of cell solubles and cell walls. J. Dairy Sci. 67:427-436. Crout, N. M. J., D. Tarsitano, and A. T. Wood. 2009. Is my model too complex? Evaluating model formulation using model reduction. Environmental Modelling & Software. 24:1-7. 36 Danes, M. A., L. J. Chagas, A. M. Pedroso, and F. A. Santos. 2013. Effect of protein supplementation on milk production and metabolism of dairy cows grazing tropical grass. Journal of Dairy Science. 96:407-419. Danfær, A. 1990. A dynamic model of nutrient digestion and metabolism in lactating dairy cows. PhD Dissertation, National Institute of Animal Science, Foulum, Denmark. Danfær, A., P. Huhtanen, P. Udén, J. Sveinbjörnsson, and H. Volden. 2006a. The Nordic Dairy Cow Model, Karoline - Description. Pages 383-406 in Nutrient Digestion and Utilization in Farm Animals: Modelling Approaches. E. Kebreab, J. Dijkstra, A. Bannink, W. J. J. Gerrits and J. France, eds. CABI Publishing, Cambridge, MA. Danfær, A., P. Huhtanen, P. Udén, J. Sveinbjörnsson, and H. Volden. 2006b. The Nordic Dairy Cow Model, Karoline - Evaluation. Pages 407-415 in Nutrient Digestion and Utilization in Farm Animals: Modelling Approaches. E. Kebreab, J. Dijkstra, A. Bannink, W. J. J. Gerrits and J. France, eds. CABI Publishing, Cambridge, MA. Deighton, M. H., S. R. O. Williams, M. C. Hannah, R. J. Eckard, T. M. Boland, W. J. Wales, and P. J. Moate. 2014. A modified sulphur hexafluoride tracer technique enables accurate determination of enteric methane emissions from ruminants. Anim. Feed Sci. Technol. 197:47-63. Del Prado, A., P. Crosson, J. E. Olesen, and C. A. Rotz. 2013. Whole-farm models to quantify greenhouse gas emissions and their potential use for linking climate change mitigation and adaptation in temperate grassland ruminant-based farming systems. Animal. 7:373-385. Dey, A., and P. S. De. 2014. Influence of condensed tannins from Ficus bengalensis leaves on feed utilization, milk production and antioxidant status of crossbred cows. Asian Australasian J. Anim. Sci. 27:342-348. Dijkstra, J. 1993. Mathematical Modelling and Integration of Rumen Fermentation Processes. Dissertation, University of Wageningen, Wageningen. Dijkstra, J. 1994. Simulation of the dynamics of protozoa in the rumen. Br. J. Nutr. 72:679699. Dijkstra, J., E. Kebreab, A. Bannink, L. A. Crompton, S. López, P. A. Abrahamse, P. Chilibroste, J. A. N. Mills, and J. France. 2008. Comparison of energy evaluation systems and a mechanistic model for milk production by dairy cattle offered fresh grass-based diets. Anim. Feed Sci. Technol. 143:203-219. Dijkstra, J., H. S. S. C. Neal, D. E. Beever, and J. France. 1992. Simulation of nutrient digestion, absorption and outflow in the rumen: model description. J. Nutr. 122:22392256. Dijkstra, J., and S. Tamminga. 1995. Simulation of the effects of diet on the contribution of rumen protozoa to degradation of fibre in the rumen. Br. J. Nutr. 74:617-634. 37 Donnelly, J. R., M. Freer, L. Salmon, A. D. Moore, R. J. Simpson, and T. P. Bolger. 2002. Evolution of the GRAZPLAN decision support tools and adoption by the grazing industry in temperate Australia. Ag. Syst. 74:115-139. Donnelly, J. R., A. D. Moore, and M. Freer. 1997. GRAZPLAN: Decision support systems for Australian grazing enterprises-I. Overview of the GRAZPLAN project, and a description of the MetAccess and LambAlive DSS. Ag. Syst. 54:57-76. Ellis, J. L., E. Kebreab, N. E. Odongo, K. Beauchemin, S. McGinn, J. D. Nkrumah, S. S. Moore, R. Christopherson, G. K. Murdoch, B. W. McBride, E. K. Okine, and J. France. 2009. Modeling methane production from beef cattle using linear and nonlinear approaches. J. Anim. Sci. 87:1334-1345. Ellis, J. L., E. Kebreab, N. E. Odongo, B. W. McBride, E. K. Okine, and J. France. 2007. Prediction of methane production from dairy and beef cattle. J. Dairy Sci. 90:34563466. Erasmus, L. J., Z. Bester, and R. J. Coertze. 2013. Milk composition as technique to evaluate the relative bioavailability of a liquid rumen protected methionine source. S. Afr. J. Anim. Sci. 43:S86-S92. Erasmus, L. J., P. M. Botha, and A. Kistner. 1992. Effect of yeast culture supplement on production, rumen fermentation, and duodenal nitrogen flow in dairy cows. J. Dairy Sci. 75:3056-3065. Erasmus, L. J., P. M. Botha, and H. H. Meissner. 1994. Effect of protein source on ruminal fermentation and passage of amino acids to the small intestine of lactating cows. J. Dairy Sci. 77:3655-3665. Erasmus, L. J., I. Smith, A. Muller, and D. O’Hagan. 1999. Effects of lasalocid on performance of lactating dairy cows. J. Dairy Sci. 82:1817-1823. Erasmus, L. J., R. Venter, and R. J. Coertze. 2004. The effect of a liquid rumen protected lysine on the productivity of Holstein cows. S. Afr. J. Anim. Sci. 34:89-91. Fatahnia, F., A. Nikkhah, M. J. Zamiri, and D. Kahrizi. 2008. Effect of dietary fish oil and soybean oil on milk production and composition of holstein cows in early lactation. Asian Australasian J. Anim. Sci. 21:386-391. Fox, D. G., and J. R. Black. 1984. A system for predicting body composition and performance of growing cattle. J. Anim. Sci. 58:725-739. Fox, D. G., C. J. Sniffen, J. D. O'Connor, J. B. Russell, and P. J. Van Soest. 1992. A net carbohydrate and protein system for evaluating cattle diets: III. Cattle requirements and diet adequacy. J. Anim. Sci. 70:3578-3596. Fox, D. G., L. O. Tedeschi, T. P. Tylutki, J. B. Russell, M. E. Van Amburgh, L. E. Chase, A. N. Pell, and T. R. Overton. 2004. The Cornell Net Carbohydrate and Protein System model for evaluating herd nutrition and nutrient excretion. Anim. Feed Sci. Technol. 112:29-78. 38 Fox, D. G., T. P. Tylutki, L. O. Tedeschi, M. E. Van Amburgh, L. E. Chase, A. N. Pell, T. R. Overton, and J. B. Russell. 2003. The Net Carbohydrate and Protein System for evaluating herd nutrition and nutrient excretion: Model documentation. Mimeo. No. 213. Animal Science Dept., Cornell University, Ithaca, NY. 292 p. France, J. 2013. Application of mathematical modelling in animal nutrition, physiology and energy balance. Pages 517-519 in Proceedings of the 4th International Symposium on Energy and Protein Metabolism and Nutrition, Sacramento, CA. Wageningen Academic Publishers. France, J., M. K. Theodorou, R. S. Lowman, and D. E. Beever. 2000. Feed evaluation for animal production. Pages 1-9 in Feeding Systems and Feed Evaluation Models. M. K. Theodorou and J. France, eds. CABI Publishing, New York. France, J., J. H. M. Thornley, R. L. Baldwin, and K. A. Crist. 1992. On solving stiff equations with reference to simulating ruminant metabolism. J. Theor. Biol. 156:525-539. France, J., J. H. M. Thornley, and D. E. Beever. 1982. A mathematical model of the rumen. J. Agric. Sci. 99:343-353. Freer, M., A. D. Moore, and J. R. Donnelly. 1997. GRAZPLAN: Decision support systems for Australian grazing enterprises-II. The animal biology model for feed intake, production and reproduction and the GrazFeed DSS. Ag. Syst. 54:77-126. Garrett, W. N., J. H. Meyer, and G. P. Lofgreen. 1959. The comparative energy requirements of sheep and cattle for maintenance and gain. J. Anim. Sci. 18:528-547. Gerber, P. J., A. N. Hristov, B. Henderson, H. Makkar, J. Oh, C. Lee, R. Meinen, F. Montes, T. Ott, J. Firkins, A. Rotz, C. Dell, A. T. Adesogan, W. Z. Yang, J. M. Tricarico, E. Kebreab, G. Waghorn, J. Dijkstra, and S. Oosting. 2013. Technical options for the mitigation of direct methane and nitrous oxide emissions from livestock: A review. Animal. 7:220-234. Gill, M., D. E. Beever, and J. France. 1989. Biochemical bases needed for the mathematical representation of whole animal metabolism. Nutrition Abstract Review. 2:181-200. Gill, M., J. H. M. Thornley, J. L. Black, J. D. Oldham, and D. E. Beever. 1984. Simulation of the metabolism of absorbed energy-yielding nutrients in young sheep. Br. J. Nutr. 52:621-649. Grainger, C., R. Williams, T. Clarke, A. D. G. Wright, and R. J. Eckard. 2010. Supplementation with whole cottonseed causes long-term reduction of methane emissions from lactating dairy cows offered a forage and cereal grain diet. J. Dairy Sci. 93:2612-2619. Graux, A. I., M. Gaurut, J. Agabriel, R. Baumont, R. Delagarde, L. Delaby, and J. F. Soussana. 2011. Development of the Pasture Simulation Model for assessing livestock production under climate change. Agric. Ecosys. Environ. 144:69-91. 39 Greenwood, J. S., M. J. Auldist, L. C. Marett, M. C. Hannah, J. L. Jacobs, and W. J. Wales. 2013. Ruminal pH and whole-tract digestibility in dairy cows consuming fresh cut herbage plus concentrates and conserved forage fed either separately or as a partial mixed ration. Animal Production Science. 54:(in press). Gregorini, P., P. C. Beukes, M. D. Hanigan, G. Waghorn, S. Muetzel, and J. P. McNamara. 2013. Comparison of updates to the Molly cow model to predict methane production from dairy cows fed pasture. J. Dairy Sci. 96:5046-5052. Guo, Y. Q., Y. Zou, Z. J. Cao, X. F. Xu, Z. S. Yang, and L. Shengli. 2013. Evaluation of coarsely ground wheat as a replacement for ground corn in the diets of lactating dairy cows. Asian Australasian J. Anim. Sci. 26:961-970. Hanigan, M. D., J. A. D. R. N. Appuhamy, and P. Gregorini. 2013. Revised digestive parameter estimates for the Molly cow model. J. Dairy Sci. 96:3867-3885. Hanigan, M. D., H. G. Bateman, J. G. Fadel, J. P. McNamara, and N. E. Smith. 2006. An ingedient-based input scheme for Molly. Pages 328-348 in Nutrient Digestion and Utilization in Farm Animals: Modelling Approaches. E. Kebreab, J. Dijkstra, A. Bannink, W. J. J. Gerrits and J. France, eds. CABI Publishing, Cambridge, MA. Hanigan, M. D., C. C. Palliser, and P. Gregorini. 2009. Altering the representation of hormones and adding consideration of gestational metabolism in a metabolic cow model reduced prediction errors. J. Dairy Sci. 92:5043-5056. Havlík, P., H. Valin, M. Herrero, M. Obersteiner, E. Schmid, M. C. Rufino, A. Mosnier, P. K. Thornton, H. Böttcher, R. T. Conant, S. Frank, S. Fritz, S. Fuss, F. Kraxner, and A. Notenbaert. 2014. Climate change mitigation through livestock system transitions. Proc. Natl. Acad. Sci. 111:3709-3714. Heard, J. W., C. R. Stockdale, G. P. Walker, C. M. Leddin, F. R. Dunshea, G. H. McIntosh, P. M. Shields, A. McKenna, G. P. Young, and P. T. Doyle. 2007. Increasing selenium concentration in milk: Effects of amount of selenium from yeast and cereal grain supplements. J. Dairy Sci. 90:4117-4127. Heard, J. W., G. P. Walker, P. J. Royle, G. H. McIntosh, and P. T. Doyle. 2004. Effects of short-term supplementation with selenised yeast on milk production and composition of lactating cows. Australian Journal of Dairy Technology. 59:199-203. Herrero, M. 1997. Modelling Dairy Grazing Systems: An Integrated Approach. Dissertation, University of Edinburgh, Edinburgh, UK. Herrero, M., R. H. Fawcett, and J. B. Dent. 1999. Bio-economic evaluation of dairy farm management scenarios using integrated simulation and multiple-criteria models. Ag. Syst. 62:169-188. Herrero, M., P. Havlík, H. Valin, A. Notenbaert, M. C. Rufino, P. K. Thornton, M. Blümmel, F. Weiss, D. Grace, and M. Obersteiner. 2013. Biomass use, production, feed 40 efficiencies, and greenhouse gas emissions from global livestock systems. Proc. Natl. Acad. Sci. 110:20888-20893. Herrero, M., P. K. Thorton, R. Kruska, and R. S. Reid. 2008. Systems dynamics and the spatial distribution of methane emissions from African domestic ruminants to 2030. Agric. Ecosys. Environ. 126:122-137. Hoffmann, L., R. Schiemann, W. Jentsch, and G. Henseler. 1974. Die verwertung der futterenergie für die milchproduktion. Archiv für Tierernaehrung. 24:245-261. Hristov, A. N., J. Oh, C. Lee, R. Meinen, F. Montes, T. Ott, J. Firkins, A. Rotz, C. Dell, A. Adesogan, W. Yang, J. Tricarico, E. Kebreab, G. Waghorn, J. Dijkstra, and S. Oosting. 2013. Mitigation of Greenhouse Gas Emissions in Livestock Production; A review of technical options for non-CO2 emissions. FAO Animal Production and Health Paper. No. 177. Food and Agriculture Organization, Rome, Italy. 206 p. Available at: http://www.fao.org/docrep/018/i3288e/i3288e.pdf. Accessed on: December 31, 2014. Huhtanen, P., J. Nousiainen, and M. Rinne. 2006. Recent developments in forage evaluation with special reference to practical applications. Agricultural and Food Science. 15:293-323. Illius, A. W., and I. J. Gordon. 1991. Prediction of intake and digestion in ruminants by a model of rumen kinetics integrating animal size and plant characteristics. J. Agric. Sci. 116:145-157. Institut National de la Recherche Agronomique. 1988. Alimentation Des Bovins, Ovins & Caprins. INRA-Quae, Paris, France. Institut National de la Recherche Agronomique. 1989. Ruminant nutrition. Recommended allowances and feed tables. Institut National de la Recherche Agronomique, John Libbey Eurotext, Montrouge, France. Institut National de la Recherche Agronomique. 2007. Alimentation des bovins, ovins et caprins. Besoins des animaux. Valeurs des aliments. Editions Quae, Versailles, France. Intergovernmental Panel on Climate Change. 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II, and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Core Writing Teams, R. K. Pachauri and A. Reisinger, eds. Cambridge University Press, Geneva, Switzerland. 104 p. Available at: http://www.ipcc.ch/publications_and_data/ar4/syr/en/contents.html. Accessed on: November 16, 2014. Irvine, L. D., M. J. Freeman, D. J. Donaghy, I. Yoon, G. Lee, and J. R. Roche. 2011. Short communication: Responses to supplemental Saccharomyces cerevisiae fermentation product and triticale grain in dairy cows grazing high-quality pasture in early lactation. J. Dairy Sci. 94:3119-3123. 41 Janssen, P. H. 2010. Influence of hydrogen on rumen methane formation and fermentation balances through microbial growth kinetics and fermentation thermodynamics. Anim. Feed Sci. Technol. 160:1-22. Jentsch, W., A. Chudy, and M. Beyer. 2003. Rostock Feed Evaluation System: Reference Numbers of Feed Value and Requirement on the Base of Net Energy. Plexus Verlag, Miltenberg-Frankfurt, Germany. Jentsch, W., M. Schweigel, F. Weissbach, H. Scholze, W. Pitroff, and M. Derno. 2007. Methane production in cattle calculated by the nutrient composition of the diet. Arch. Anim. Nutr. 61:10-19. Jesus, E. F., L. H. A. Conti, T. Tomazi, M. E. Diniz, A. Migliano, L. F. P. e. Silva, F. P. Renno, and M. V. d. Santos. 2012. Intake, nutrient digestibility and milk yield of dairy cows fed urea and two levels of crude protein in diets with sugar cane. Journal of Animal and Veterinary Advances. 11:4135-4142. Johnson, I. R., D. F. Chapman, V. O. Snow, R. J. Eckard, A. J. Parsons, M. G. Lambert, and B. R. Cullen. 2008. DairyMod and EcoMod: biophysical pasture-simulation models for Australia and New Zealand. Austr. J. Exp. Agric. 48:621-631. Johnson, I. R., G. M. Lodge, and R. E. White. 2003. The sustainable grazing systems pasture model: Description, philosophy and application to the sgs national experiment. Austr. J. Exp. Agric. 43:711-728. Kalscheur, K. F., J. H. Vandersall, R. A. Erdman, R. A. Kohn, and E. Russek-Cohen. 1999. Effects of dietary crude protein concentration and degradability on milk production responses of early, mid, and late lactation dairy cows. Journal of Dairy Science. 82:545-554. Kaustell, K., M. Tuori, and P. Huhtanen. 1997. Comparison of energy evaluation systems for dairy cow feeds. Livest. Prod. Sci. 51:255-266. Khezri, A., K. Rezayazdi, M. D. Mesgaran, and M. Moradi-Sharbabk. 2009. Effect of different rumen-degradable carbohydrates on rumen fermentation, nitrogen metabolism and actation performance of holstein dairy cows. Asian Australasian J. Anim. Sci. 22:651-658. Kohn, R. A., R. Boston, J. D. Ferguson, and W. Chalupa. 1994. The integration and comparison of dairy cow models. Pages 117-128 in Proceedings of the 4th International Workshop on Modeling Nutrient Utilization in Farm Animals, Foulum, Denmark. Kohn, R. A., and R. C. Boston. 2000. The Role of Thermodynamics in Controlling Rumen Metabolism. Pages 11-24 in Modelling Nutrient Utilization in Farm Animals. J. P. McNamara, J. France and D. E. Beever, eds. CABI Publishing, New York, NY. 42 Kokkonen, T., M. Tuori, V. Leivonen, and L. Syrjälä-Qvist. 2000. Effect of silage dry matter content and rapeseed meal supplementation on dairy cows. 1. Milk production and feed utilisation. Animal Feed Science and Technology. 84:213-228. Kriss, M. 1930. Quantitative relations of the dry matter of the food consumed, the heat production, the gaseous outgo, and the insensible loss in body weight of cattle. J. Agric. Res. 40:283-295. Lanzas, C., G. A. Broderick, and D. G. Fox. 2008. Improved feed protein fractionation schemes for formulating rations with the Cornell Net Carbohydrate and Protein System. J. Dairy Sci. 91:4881-4891. Lanzas, C., C. J. Sniffen, S. Seo, L. O. Tedeschi, and D. G. Fox. 2007a. A revised CNCPS feed carbohydrate fractionation scheme for formulating rations for ruminants. Anim. Feed Sci. Technol. 136:167-190. Lanzas, C., L. O. Tedeschi, S. Seo, and D. G. Fox. 2007b. Evaluation of protein fractionation systems used in formulating rations for dairy cattle. J. Dairy Sci. 90:507-521. Lehmann, M., R. Meeske, and C. W. Cruywagen. 2007. Milk production and in sacco disappearance of pasture NDF in grazing Jersey cows receiving a barley based concentrate. South African Journal of Animal Science. 37:81-89. Leroy, A. M. 1954. Utilization de l'energie des aliments par les animaux. Annales de Zootechnie. 3:337-372. Liu, Z. L., D. P. Yang, P. Chen, W. X. Dong, and D. M. Wang. 2008. Supplementation with selenium and vitamin E improves milk fat depression and fatty acid composition in dairy cows fed fat diet. Asian Australasian J. Anim. Sci. 21:838-844. Lofgreen, G. P. 1965. A comparative slaughter technique for determining net energy values with beef cattle. Pages 309-317 in Proceedings of the 3rd Symposium of Energy Metabolism, Troon, Scotland. Academic Press. Lofgreen, G. P., and W. N. Garrett. 1968. A system for expressing net energy requirements and feed values for growing and finishing beef cattle. J. Anim. Sci. 27:793-806. Lofgreen, G. P., J. L. Hull, and K. K. Otagaki. 1962. Estimation of the empty body weight of beef cattle. J. Anim. Sci. 21:20-24. Lunsin, R., M. Wanapat, and P. Rowlinson. 2012. Effect of cassava hay and rice bran oil supplementation on rumen fermentation, milk yield and milk composition in lactating dairy cows. Asian Australasian J. Anim. Sci. 25:1364-1373. Mambrini, M. 1997. Retention time of feed particles and liquids in the stomachs and intestines of dairy cows. Direct measurement and calculations based on faecal collection. Reproduction Nutrition Development. 37:427-442. Martin-Clouaire, R., and J. Clouaire. 2009. Modelling and simulating work practices in agriculture. International Journal of Metadata, Semantics and Ontologies. 4:42-53. 43 McCormick, M. E., D. D. Redfearn, J. D. Ward, and D. C. Blouin. 2001a. Effect of protein source and soluble carbohydrate addition on rumen fermentation and lactation performance of Holstein cows. J. Dairy Sci. 84:1686-1697. McCormick, M. E., J. D. Ward, D. D. Redfearn, D. D. French, D. C. Blouin, A. M. Chapa, and J. M. Fernandez. 2001b. Supplemental dietary protein for grazing dairy cows: Effect on pasture intake and lactation performance. Journal of Dairy Science. 84:896907. McGinn, S. M., and K. A. Beauchemin. 2012. Dairy farm methane emissions using a dispersion model. J. Environ. Qual. 41:73-79. McGinn, S. M., K. A. Beauchemin, A. D. Iwaasa, and T. A. McAllister. 2006a. Assessment of the sulfur hexafluoride (SF6) tracer technique for measuring enteric methane emissions from cattle. J. Environ. Qual. 35:1686-1691. McGinn, S. M., T. K. Flesch, L. A. Harper, and K. A. Beauchemin. 2006b. An approach for measuring methane emissions from whole farms. J. Environ. Qual. 35:14-20. McGinn, S. M., D. Turner, N. Tomkins, E. Charmley, G. Bishop-Hurley, and D. Chen. 2011. Methane emissions from grazing cattle using point-source dispersion. J. Environ. Qual. 40:22-27. McNamara, J. P., and S. L. Shields. 2013. Reproduction during lactation of dairy cattle: Integrating nutritional aspects of reproductive control in a systems research approach. Animal Frontiers. 3:76-83. Meeske, R., P. R. Botha, G. D. v. d. Merwe, J. F. Greyling, C. Hopkins, and J. P. Marais. 2009. Milk production potential of two ryegrass cultivars with different total nonstructural carbohydrate contents. South African Journal of Animal Sciences. 39:1521. Mertens, D. R. 1989. Evaluating alternative models of passage and digestion kinetics. Pages 79-97 in Proceedings of the 3rd International Modelling Ruminant Digestion and Metabolism, Lincoln College, Canterbury, New Zealand. Lincoln University. Mertens, D. R. 1993. Kinetics of cell wall digestion and passage in ruminants. Pages 535-570 in Forage Cell Wall Structure and Digestibility. H. G. Jung, D. R. Buxton, R. D. Hatfield and J. Ralph, eds. American Society of Agronomy, Madison, WI. Mertens, D. R. 2005. Rate and extent of digestion. Pages 13-47 in Quantitative Aspects of Ruminant Digestion and Metabolism. J. Dijkstra, J. M. Forbes and J. France, eds. CAB International, Wallingford, UK. Mills, J. A. N., J. Dijkstra, A. Bannink, S. B. Cammell, E. Kebreab, and J. France. 2001. A mechanistic model of whole-tract digestion and methanogenesis in the lactating dairy cow: Model development, evaluation, and application. J. Anim. Sci. 79:1584-1597. 44 Ministry of Agriculture, Fisheries and Food,. 1975. Energy allowances and feeding systems for ruminants. Technical Bulletin. No. 33. Her Majesty's Stationery Office, London, UK. 77 p. Moallem, U. 2009. The effects of extruded flaxseed supplementation to high-yielding dairy cows on milk production and milk fatty acid composition. Animal Feed Science and Technology. 152:232-242. Moe, P. W., W. P. Flatt, and H. F. Tyrrell. 1972. Net energy value of feeds for lactation. J. Dairy Sci. 55:945-958. Moe, P. W., and H. F. Tyrrell. 1974. Prediction of the net energy value of feeds for lactation. Pages 201-204 in Proceedings of the 6th Energy Metabolism of Farm Animals, 14, Stuttgart. European Association for Animal Production. Moe, P. W., and H. F. Tyrrell. 1975. Efficiency of conversion of digested energy to milk. J. Dairy Sci. 58:602-610. Moe, P. W., and H. F. Tyrrell. 1979. Methane production in dairy cows. J. Dairy Sci. 62:1583-1586. Moe, P. W., H. F. Tyrrell, and W. P. Flatt. 1970. Partial efficiency of energy use for maintenance, lactation, body gain and gestation in the dairy cow. Pages 65-68 in Proceedings of the 5th Energy Metabolism of Farm Animals, 13, Vitznau, Switzerland. EAAP. Moe, P. W., H. F. Tyrrell, and W. P. Flatt. 1971. Energetics of body tissue mobilization. J. Dairy Sci. 54:548-553. Moharrery, A. 2010. Effect of particle size of forage in the dairy ration on feed intake, production parameters and quantification of manure index. Asian Australasian J. Anim. Sci. 23:483-490. Moore, A. D., J. R. Donnelly, and M. Freer. 1997. GRAZPLAN: decision support systems for Australian grazing enterprises. III. Pasture growth and soil moisture submodels, and the GrassGro DSS. Ag. Syst. 55:535-582. Moore, A. D., D. P. Holzworth, N. I. Herrmann, N. I. Huth, and M. J. Robertson. 2007. The Common Modelling Protocol: A hierarchical framework for simulation of agricultural and environmental systems. Ag. Syst. 95:37-48. Moraes, L. E., A. B. Strathe, J. G. Fadel, D. P. Casper, and E. Kebreab. 2014. Prediction of enteric methane emissions from cattle. Global Change Biology. 20:2140-2148. Mosavi, G. H. R., F. Fatahnia, H. R. Mirzaei Alamouti, A. A. Mehrabi, and H. Darmani Kohi. 2012. Effect of dietary starch source on milk production and composition of lactating Holstein cows. South African Journal of Animal Science. 42:201-209. Moss, A. R., J.-P. Jouany, and J. Newbold. 2000. Methane production by ruminants: Its contribution to global warming. Annales de Zootechnie. 49:231-253. 45 Murphy, J. J. 1999. The effects of increasing the proportion of molasses in the diet of milking dairy cows on milk production and composition. Animal Feed Science and Technology. 78:189-198. Nagorcka, B. N., G. L. R. Gordon, and R. A. Dynes. 2000. Towards a More Accurate Representation of Fermentation in Mathematical Models of the Rumen. Pages 37-48 in Modelling Nutrient Utilization in Farm Animals. J. P. McNamara, J. France and D. E. Beever, eds. CABI Publishing, New York, NY. National Research Council. 1945a. Recommended Nutrient Allowances for Beef Cattle. Recommended nutrient allowances for domestic animals. National Academy Press, Washington, DC. National Research Council. 1945b. Recommended Nutrient Allowances for Dairy Cattle. Recommended nutrient allowances for domestic animals. National Academy Press, Washington, DC. National Research Council. 1982. United States-Canadian Tables of Feed Composition. National Academy Press, Washington, DC. National Research Council. 1996. Nutrient Requirements of Beef Cattle. (7th ed.). Nutrient requirements of domestic animals. National Academy Press, Washington, DC. National Research Council. 2000. Nutrient Requirements of Beef Cattle. (updated 7th ed.). Nutrient requirements of domestic animals. National Academy Press, Washington, DC. National Research Council. 2001. Nutrient Requirements of Dairy Cattle. (7th ed.). Nutrient requirements of domestic animals. National Academy Press, Washington, DC. Neal, H. S. S. C., J. Dijkstra, and G. Margaret. 1992. Simulation of nutrient digestion, absorption and outflow in the rumen: model evaluation. J. Nutr. 122:2257-2272. Nehring, K., R. Schiemann, and L. Hoffmann. 1966. Vorschlag eines neuen systems der energetischen bewertung des futters auf der grundlage der nettoenergie-fett. Pages 1930 in Sitzungsberichte Deutschen Akademie der Landwirtschaftswissenschaften, Berlin, Germany. O'Connor, J. D., C. J. Sniffen, D. G. Fox, and W. Chalupa. 1993. A net carbohydrate and protein system for evaluating cattle diets: IV. Predicting amino acid adequacy. J. Anim. Sci. 71:1298-1311. O'Mara, F. P., J. J. Fitzgerald, J. J. Murphy, and M. Rath. 1998. The effect on milk production of replacing grass silage with maize silage in the diet of dairy cows. Livestock Production Science. 55:79-87. O’Mara, F. P., J. J. Murphy, and M. Rath. 2000. The effect of concentrate supplements differing in ruminal protein degradability on milk production and blood metabolite concentrations of dairy cows grazing perennial ryegrass pasture. Livestock Production Science. 64:183-191. 46 Oguz, F. K., M. N. Oguz, T. Buyukoglu, and S. Sahinduran. 2006. Effects of varying levels of whole cottonseed on blood, milk and rumen parameters of dairy cows. Asian Australasian J. Anim. Sci. 19:852-856. Palliser, C. C., K. P. Bright, K. A. Macdonald, J. W. Penno, and M. E. Wastney. 2001. Adapting the MOLLY cow model to fit production data from New Zealand animals. Proceedings of the New Zealand Society of Animal Production. 61:234-236. Petit, H. V., and N. Gagnon. 2011. Production performance and milk composition of dairy cows fed different concentrations of flax hulls. Animal Feed Science and Technology. 169:46-52. Piamphon, N., C. Wachirapakorn, M. Wanapat, and C. Navanukraw. 2009. Effects of protected conjugated linoleic acid supplementation on milk fatty acid in dairy cows. Asian Australasian J. Anim. Sci. 22:49-56. Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team. 2014. nlme: Linear and Nonlinear Mixed Effects Models. Available: http://CRAN.Rproject.org/package=nlme. Accessed on May 10, 2014. Pitt, R. E., J. S. Van Kessel, D. G. Fox, A. N. Pell, M. C. Barry, and P. J. Van Soest. 1996. Prediction of ruminal volatile fatty acids and pH within the net carbohydrate and protein system. J. Anim. Sci. 74:226-244. R Core Team. 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available at: http://www.R-project.org. Accessed on: May 10, 2014. Ramin, M., and P. Huhtanen. 2012. Development of non-linear models for predicting enteric methane production. Acta Agriculturae Scandinavica, Section A – Animal Science. 62:254-258. Ramin, M., and P. Huhtanen. 2013. Development of equations for predicting methane emissions from ruminants. J. Dairy Sci. 96:2476-2493. Rosenzweig, C., J. W. Jones, J. L. Hatfield, A. C. Ruane, K. J. Boote, P. Thorburn, J. M. Antle, G. C. Nelson, C. Porter, S. Janssen, S. Asseng, B. Basso, F. Ewert, D. Wallach, G. Baigorria, and J. M. Winter. 2013. The Agricultural Model Intercomparison and Improvement Project (AgMIP): Protocols and pilot studies. Agricultural and Forest Meteorology. 170:166-182. Rotz, C. A., D. R. Buckmaster, and J. W. Comerford. 2005. A beef herd model for simulating feed intake, animal performance, and manure excretion in farm systems. J. Anim. Sci. 83:231-242. Rotz, C. A., D. R. Mertens, D. R. Buckmaster, M. S. Allen, and J. H. Harrison. 1999. A dairy herd model for use in whole farm simulations. J. Dairy Sci. 82:2826-2840. 47 Russell, J. B., J. D. O'Connor, D. G. Fox, P. J. Van Soest, and C. J. Sniffen. 1992. A net carbohydrate and protein system for evaluating cattle diets: I. Ruminal fermentation. J. Anim. Sci. 70:3551-3561. Salmon, L., J. R. Donnelly, A. D. Moore, M. Freer, and R. J. Simpson. 2004. Evaluation of options for production of large lean lambs in south-eastern Australia. Anim. Feed Sci. Technol. 112:195-209. Sanh, M. V., H. Wiktorsson, and L. V. Ly. 2002. Effect of feeding level on milk production, body weight change, feed conversion and postpartum oestrus of crossbred lactating cows in tropical conditions. Livestock Production Science. 77:331-338. Sauvant, D. 1996. A comparative evaluation of models of lactating ruminant. Annales de Zootechnie. 45:215-235. Schiemann, R., K. Nehring, L. Hoffmann, W. Jentsch, and A. Chudy. 1971. Energetische Futterbewertung und Energienormen: Dokumentation der wissenschaftlichen Grundlagen eines neuen energetischen Futterbewertungssystems. Deutscher Landwirtschaftsverlag, Berlin, Germany. Seo, S., L. O. Tedeschi, C. G. Schwab, and D. G. Fox. 2006. Development and evaluation of empirical equations to predict feed passage rate in cattle. Anim. Feed Sci. Technol. 128:67-83. Sniffen, C. J., J. D. O'Connor, P. J. Van Soest, D. G. Fox, and J. B. Russell. 1992. A net carbohydrate and protein system for evaluating cattle diets: II. Carbohydrate and protein availability. J. Anim. Sci. 70:3562-3577. Snow, V. O., C. A. Rotz, A. D. Moore, R. Martin-Clouaire, I. R. Johnson, N. J. Hutchings, and R. J. Eckard. 2014. The challenges – and some solutions – to process-based modelling of grazed agricultural systems. Environment Modelling & Software. 62:420-436. Suksombat, W., and K. Chullanandana. 2008. Effects of soybean oil or rumen protected conjugated linoleic acid supplementation on accumulation of conjugated linoleic acid in dairy cows’ milk. Asian Australasian J. Anim. Sci. 21:1271-1277. Sun, T., X. Yu, S. L. Li, Y. X. Dong, and H. T. Zhang. 2009. Responses of dairy cows to supplemental highly digestible rumen undegradable protein and rumen-protected forms of methionine. Asian Australasian J. Anim. Sci. 22:659-666. Sundstøl, F. 1993. Energy systems for ruminants. Icel. Agr. Sci. 7:11-19. Sveinbjörnsson, J., P. Huhtanen, and P. Udén. 2006. The Nordic Dairy Cow Model, Karoline - Development of Volatile Fatty Acid Sub-Model. Pages 1-14 in Nutrient Digestion and Utilization in Farm Animals: Modelling Approaches. E. Kebreab, J. Dijkstra, A. Bannink, W. J. J. Gerrits and J. France, eds. CABI Publishing, Cambridge, MA. 48 Tamminga, S., W. M. Van Straalen, A. P. J. Subnel, R. G. M. Meijer, A. Steg, C. J. G. Wever, and M. C. Blok. 1994. The Dutch protein evaluation system: The DVE/OEB-system. Livest. Prod. Sci. 40:139-155. Teague, R., F. Provenza, U. Kreuter, T. Steffens, and M. Barnes. 2013. Multi-paddock grazing on rangelands: Why the perceptual dichotomy between research results and rancher experience? Journal of Environmental Management. 128:699-717. Tedeschi, L. O. 2006. Assessment of the adequacy of mathematical models. Ag. Syst. 89:225247. Tedeschi, L. O., C. Boin, D. G. Fox, P. R. Leme, G. F. Alleoni, and D. P. D. Lanna. 2002a. Energy requirement for maintenance and growth of Nellore bulls and steers fed highforage diets. J. Anim. Sci. 80:1671-1682. Tedeschi, L. O., T. R. Callaway, J. P. Muir, and R. Anderson. 2011. Potential environmental benefits of feed additives and other strategies for ruminant production. Revista Brasileira de Zootecnia. 40:291-309. Tedeschi, L. O., L. F. L. Cavalcanti, M. A. Fonseca, M. Herrero, and P. K. Thornton. 2014. The evolution and evaluation of dairy cattle models for predicting milk production: an agricultural model intercomparison and improvement project (AgMIP) for livestock. Anim. Prod. Sci. 54:2052-2067. Tedeschi, L. O., W. Chalupa, E. Janczewski, D. G. Fox, C. J. Sniffen, R. Munson, P. J. Kononoff, and R. C. Boston. 2008. Evaluation and application of the CPM Dairy nutrition model. J. Agric. Sci. 146:171-182. Tedeschi, L. O., D. G. Fox, L. E. Chase, and S. J. Wang. 2000a. Whole-herd optimization with the Cornell net carbohydrate and protein system. I. Predicting feed biological values for diet optimization with linear programming. J. Dairy Sci. 83:2139-2148. Tedeschi, L. O., D. G. Fox, and P. H. Doane. 2005a. Evaluation of the tabular feed energy and protein undegradability values of the National Research Council nutrient requirements of beef cattle. Prof. Anim. Scient. 21:403-415. Tedeschi, L. O., D. G. Fox, M. A. Fonseca, and L. F. L. Cavalcanti. 2013a. Models of protein and amino acid requirements for cattle. Pages 1-45 (CD format) in Proceedings of the 50th Annual Meeting of the Brazilian Society of Animal Science, Campinas, São Paulo, Brazil. Sociedade Brasileira de Zootecnia (SBZ). Available at: http://www.sbz.org.br/reuniao2013/?lang=en. Accessed on: November 16, 2014. Tedeschi, L. O., D. G. Fox, and P. J. Kononoff. 2013b. A dynamic model to predict fat and protein fluxes associated with body reserve changes in cattle. J. Dairy Sci. 96:24482463. Tedeschi, L. O., D. G. Fox, A. N. Pell, D. P. D. Lanna, and C. Boin. 2002b. Development and evaluation of a tropical feed library for the Cornell Net Carbohydrate and Protein System model. Scientia Agricola. 59:1-18. 49 Tedeschi, L. O., D. G. Fox, and J. B. Russell. 2000b. Accounting for the effects of a ruminal nitrogen deficiency within the structure of the Cornell net carbohydrate and protein system. J. Anim. Sci. 78:1648-1658. Tedeschi, L. O., D. G. Fox, R. D. Sainz, L. G. Barioni, S. R. Medeiros, and C. Boin. 2005b. Using mathematical models in ruminant nutrition. Scientia Agricola. 62:76-91. Tedeschi, L. O., D. G. Fox, and T. P. Tylutki. 2003. Potential environmental benefits of ionophores in ruminant diets. J. Environ. Qual. 32:1591-1602. Tedeschi, L. O., A. N. Pell, D. G. Fox, and C. R. Llames. 2001. The amino acid profiles of the whole plant and of four residues from temperate and tropical forages. J. Anim. Sci. 79:525-532. Tedeschi, L. O., S. Seo, D. G. Fox, and R. Ruiz. 2006. Accounting for energy and protein reserve changes in predicting diet-allowable milk production in cattle. J. Dairy Sci. 89:4795-4807. Thornton, P. K., and M. Herrero. 2010. Potential for reduced methane and carbon dioxide emissions from livestock and pasture management in the tropics. Proc. Natl. Acad. Sci. 107:19667-19672. Tomkins, N. W., S. M. McGinn, D. A. Turner, and E. Charmley. 2011. Comparison of opencircuit respiration chambers with a micrometeorological method for determining methane emissions from beef cattle grazing a tropical pasture. Anim. Feed Sci. Technol. 166-167:240-247. Tothi, R., P. Lund, M. R. Weisbjerg, and T. Hvelplund. 2003. Effect of expander processing on fractional rate of maize and barley starch degradation in the rumen of dairy cows estimated using rumen evacuation and in situ techniques. Anim. Feed Sci. Technol. 104:71-94. Tylutki, T. P., D. G. Fox, and R. G. Anrique. 1994. Predicting net energy and protein requirements for growth of implanted and nonimplanted heifers and steers and nonimplanted bulls varying in body size. J. Anim. Sci. 72:1806-1813. Tylutki, T. P., D. G. Fox, V. M. Durbal, L. O. Tedeschi, J. B. Russell, M. E. Van Amburgh, T. R. Overton, L. E. Chase, and A. N. Pell. 2008. Cornell Net Carbohydrate and Protein System: A model for precision feeding of dairy cattle. Anim. Feed Sci. Technol. 143:174-202. Tyrrell, H. F., and P. W. Moe. 1975a. Effect of intake on digestive efficiency. J. Dairy Sci. 58:1151-1163. Tyrrell, H. F., and P. W. Moe. 1975b. Production efficiency in the high producing cow. J. Dairy Sci. 58:602-610. Tyrrell, H. F., P. W. Moe, and W. P. Flatt. 1970. Influence of excess protein intake on energy metabolism of the dairy cow. Pages 69-72 in Proceedings of the 5th Energy Metabolism of Farm Animals, 13, Vitznau, Switzerland. EAAP. 50 Vafa, T., A. A. Naserian, A. Heravi Moussavi, R. Valizadeh, and M. Danesh Mesgaran. 2012. Effect of supplementation of fish and canola oil in the diet on milk fatty acid composition in early lactating holstein cows. Asian-Australasian Journal of Animal Sciences. 25:311-319. Valizadeh, R., M. Behgar, M. Mirzaee, A. A. Naserian, A. R. Vakili, and S. Ghovvati. 2010. The effect of physically effective fiber and soy hull on the ruminal cellulolytic bacteria population and milk production of dairy cows. Asian Australasian J. Anim. Sci. 23:1325-1332. Van Amburgh, M. E., L. E. Chase, T. R. Overton, D. A. Ross, R. J. Rechtenwald, R. J. Higgs, and T. P. Tylutki. 2010. Updates to the Cornell Net Carbohydrate and Protein System v6.1 and implications for ration formulation. Pages 144-159 in Proceedings of Cornell Nutrition Conference for Feed Manufacturers, Syracuse, NY. New York State College of Agriculture & Life Sciences, Cornell University. Van Amburgh, M. E., A. Foskolos, E. A. Collao-Saenz, R. J. Higgs, and D. A. Ross. 2013. Updating the CNCPS feed library with new feed amino acid profiles and efficiencies of use: Evaluation of model predictions - version 6.5. Pages 59-76 in Proceedings of Cornell Nutrition Conference for Feed Manufacturers, Syracuse, NY. New York State College of Agriculture & Life Sciences, Cornell University. Van Amburgh, M. E., T. R. Overton, L. E. Chase, D. A. Ross, and R. J. Rechtenwald. 2009. The Cornell Net Carbohydrate and Protein System: Current and future approaches for balancing of amino acids. Pages 28-37 in Proceedings of Cornell Nutrition Conference for Feed Manufacturers, Syracuse, NY. New York State College of Agriculture & Life Sciences, Cornell University. Van Es, A. J. H. 1975. Feed evaluation for dairy cows. Livest. Prod. Sci. 2:95-107. Van Es, A. J. H. 1978. Feed evaluation for ruminants. I. The systems in use from May 1977onwards in The Netherlands. Livest. Prod. Sci. 5:331-345. Van Soest, P. J. 1963a. Use of detergents in analysis of fibrous feeds. I. Preparation of fiber residues of low nitrogen content. J. AOAC Intl. 46:825-829. Van Soest, P. J. 1963b. Use of detergents in analysis of fibrous feeds. II. A rapid method for the determination of fiber and lignin. J. AOAC Intl. 46:829-835. Vargas, B., M. Herrero, and J. A. M. van Arendonk. 2001. Interactions between optimal replacement policies and feeding strategies in dairy herds. Livest. Prod. Sci. 69:17-31. Vermorel, M. 1978. Feed evaluation for ruminants. II. The new energy systems proposed in France. Livest. Prod. Sci. 5:347-365. Volden, H. 2011. NorFor - The Nordic Feed Evaluation System. Wageningen Academic Publishers, Wageningen, The Netherlands. 51 Waithaka, M. M., P. K. Thornton, M. Herrero, and K. D. Shepherd. 2006. Bio-economic evaluation of farmers’ perceptions of viable farms in western Kenya. Ag. Syst. 90:243-271. Waldo, D. R., L. W. Smith, and E. L. Cox. 1972. Model of cellulose disappearance from the rumen. J. Dairy Sci. 55:125-129. Walker, G. P., F. R. Dunshea, J. W. Heard, C. R. Stockdale, and P. T. Doyle. 2010. Output of selenium in milk, urine, and feces is proportional to selenium intake in dairy cows fed a total mixed ration supplemented with selenium yeast. J. Dairy Sci. 93:4644-4650. Wickham, H. 2009. ggplot2 Elegant Graphics for Data Analysis. Use R! Springer, New York. Available at: http://had.co.nz/ggplot2/book. Accessed on: December 31, 2014. Wilkerson, V. A., D. P. Casper, and D. R. Mertens. 1995. The prediction of methane production of Holstein cows by several equations. J. Dairy Sci. 78:2402-2414. Wylie, M. J., W. C. Ellis, J. H. Matis, E. M. Bailey, W. D. James, and D. E. Beever. 2000. The flow of forage particles and solubles through segments of the digestive tracts of cattle. Br. J. Nutr. 83:295-306. Yalçın, S., S. Yalçın, P. Can, A. O. Gürdal, C. Bağcı, and Ö. Eltan. 2011. The nutritive value of live yeast culture (Saccharomyces cerevisiae) and its effect on milk yield, milk composition and some blood parameters of dairy cows. Asian Australasian J. Anim. Sci. 24:1377-1385. Yan, R., R. Zhang, X. Zhang, C. Jiang, J.-g. Han, and Y.-j. Zhang. 2011. Changes in milk production and metabolic parameters by feeding lactating cows based on different ratios of corn silage: Alfalfa hay with addition of extruded soybeans. Asian Australasian J. Anim. Sci. 24:800-8009. Yan, T., M. G. Porter, and C. S. Mayne. 2009. Prediction of methane emission from beef cattle using data measured in indirect open-circuit respiration calorimeters. Animal. 3:1455-1462. Yarahmadi, H. M., and H. Nirumand. 2012. The influence of feeding plant oils on milk production and fatty acid content. African Journal of Agricultural Research. 7:426432. 52 The CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) is a strategic initiative of CGIAR and Future Earth, led by the International Center for Tropical Agriculture (CIAT). CCAFS is the world’s most comprehensive global research program to examine and address the critical interactions between climate change, agriculture and food security. For more information, visit www.ccafs.cgiar.org Titles in this Working Paper series aim to disseminate interim climate change, agriculture and food security research and practices and stimulate feedback from the scientific community. CCAFS is led by: Fund Research supported by: Fund Strategic partner: