Methane (CH4) is one of the major greenhouse gases being targeted for reduction by the Kyoto protocol. The focus of recent research in animal science has thus been to develop or improve existing CH4 prediction models to evaluate mitigation strategies to reduce overall CH4 emissions. Eighty-three beef and 89 dairy data sets were collected and used to develop statistical models of CH4 production using dietary variables. Dry matter intake (DMI), metabolizable energy intake, neutral detergent fiber, acid detergent fiber, ether extract, lignin, and forage proportion were considered in the development of models to predict CH4 emissions. Extant models relevant to the study were also evaluated. For the beef database, the equation CH4 (MJ/d) = 2.94 (+/- 1.16) + 0.059 (+/- 0.0201) x metabolizable energy intake (MJ/d) + 1.44 (+/- 0.331) x acid detergent fiber (kg/d) - 4.16 (+/- 1.93) x lignin (kg/d) resulted in the lowest root mean square prediction error (RMSPE) value (14.4%), 88% of which was random error. For the dairy database, the equation CH4 (MJ/d) = 8.56 (+/- 2.63) + 0.14 (+/- 0.056) x forage (%) resulted in the lowest RMSPE value (20.6%) and 57% of error from random sources. An equation based on DMI also performed well for the dairy database: CH4 (MJ/d) = 3.23 (+/- 1.12) + 0.81 (+/- 0.086) x DMI (kg/d), with a RMSPE of 25.6% and 91% of error from random sources. When the dairy and beef databases were combined, the equation CH4 (MJ/d) = 3.27 (+/- 0.79) + 0.74 (+/- 0.074) x DMI (kg/d) resulted in the lowest RMSPE value (28.2%) and 83% of error from random sources. Two of the 9 extant equations evaluated predicted CH4 production adequately. However, the new models based on more commonly determined values showed an improvement in predictions over extant equations.

SUMMARY: Methane, in addition to being a significant source of energy loss to the animal that can range from 0·02 to 0·12 of gross energy intake, is one of the major greenhouse gases being targeted for reduction by the Kyoto protocol. Thus, one of the focuses of recent research in animal science has been to develop or improve existing methane prediction models in order to increase overall understanding of the system and to evaluate mitigation strategies for methane reduction. Several dynamic mechanistic models of rumen function have been developed which contain hydrogen gas balance sub-models from which methane production can be predicted. These models predict methane production with varying levels of success and in many cases could benefit from further development. Central to methane prediction is accurate volatile fatty acid prediction, representation of the competition for substrate usage within the rumen, as well as descriptions of protozoal dynamics and pH. Most methane models could also largely benefit from an expanded description of lipid metabolism and hindgut fermentation. The purpose of the current review is to identify key aspects of rumen microbiology that could be incorporated into, or have improved representation within, a model of ruminant digestion and environmental emissions.

The objective of this study was to investigate the effects of starch varying in rate of fermentation and level of inclusion in the diet in exchange for fiber on methane (CH4) production of dairy cows. Forty Holstein-Friesian lactating dairy cows of which 16 were rumen cannulated were grouped in 10 blocks of 4 cows each. Cows received diets consisting of 60% grass silage and 40% concentrate (dry matter basis). Cows within block were randomly assigned to 1 of 4 different diets composed of concentrates that varied in rate of starch fermentation [slowly (S) vs. rapidly (R) rumen fermentable; native vs. gelatinized corn grain] and level of starch (low vs. high; 270 vs. 530g/kg of concentrate dry matter). Results of rumen in situ incubations confirmed that the fractional rate of degradation of starch was higher for R than S starch. Effective rumen degradability of organic matter was higher for high than low starch and also higher for R than S starch. Increased level of starch, but not starch fermentability, decreased dry matter intake and daily CH4 production. Milk yield (mean 24.0±1.02kg/d), milk fat content (mean 5.05±0.16%), and milk protein content (mean 3.64±0.05%) did not differ between diets. Methane expressed per kilogram of fat- and protein-corrected milk, per kilogram of dry matter intake, or as a fraction of gross energy intake did not differ between diets. Methane expressed per kilogram of estimated rumen-fermentable organic matter (eRFOM) was higher for S than R starch-based diets (47.4 vs. 42.6g/kg of eRFOM) and for low than high starch-based diets (46.9 vs. 43.1g/kg of eRFOM). Apparent total-tract digestibility of neutral detergent fiber and crude protein were not affected by diets, but starch digestibility was higher for diets based on R starch (97.2%) compared with S starch (95.5%). Both total volatile fatty acid concentration (109.2 vs. 97.5mM) and propionate proportion (16.5 vs. 15.8mol/100mol) were higher for R starch- compared with S starch-based diets but unaffected by the level of starch. Total N excretion in feces plus urine and N retained were unaffected by dietary treatments, and similarly energy intake and output of energy in milk expressed per unit of metabolic body weight were not affected by treatments. In conclusion, an increased rate of starch fermentation and increased level of starch in the diet of dairy cattle reduced CH4 produced per unit of eRFOM but did not affect CH4 production per unit of feed dry matter intake or per unit of milk produced.

Mechanistic models (MMs) have served as causal pathway analysis and ‘decision-support’ tools within animal production systems for decades. Such models quantitatively define how a biological system works based on causal relationships and use that cumulative biological knowledge to generate predictions and recommendations (in practice) and generate/evaluate hypotheses (in research). Their limitations revolve around obtaining sufficiently accurate inputs, user training and accuracy/precision of predictions on-farm. The new wave in digitalization technologies may negate some of these challenges. New data-driven (DD) modelling methods such as machine learning (ML) and deep learning (DL) examine patterns in data to produce accurate predictions (forecasting, classification of animals, etc.). The deluge of sensor data and new self-learning modelling techniques may address some of the limitations of traditional MM approaches – access to input data (e.g. sensors) and on-farm calibration. However, most of these new methods lack transparency in the reasoning behind predictions, in contrast to MM that have historically been used to translate knowledge into wisdom. The objective of this paper is to propose means to hybridize these two seemingly divergent methodologies to advance the models we use in animal production systems and support movement towards truly knowledge-based precision agriculture. In order to identify potential niches for models in animal production of the future, a cross-species (dairy, swine and poultry) examination of the current state of the art in MM and new DD methodologies (ML, DL analytics) is undertaken. We hypothesize that there are several ways via which synergy may be achieved to advance both our predictive capabilities and system understanding, being: (1) building and utilizing data streams (e.g. intake, rumination behaviour, rumen sensors, activity sensors, environmental sensors, cameras and near IR) to apply MM in real-time and/or with new resolution and capabilities; (2) hybridization of MM and DD approaches where, for example, a ML framework is augmented by MM-generated parameters or predicted outcomes and (3) hybridization of the MM and DD approaches, where biological bounds are placed on parameters within a MM framework, and the DD system parameterizes the MM for individual animals, farms or other such clusters of data. As animal systems modellers, we should expand our toolbox to explore new DD approaches and big data to find opportunities to increase understanding of biological systems, find new patterns in data and move the field towards intelligent, knowledge-based precision agriculture systems.