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.

The expected increase in the global demand for livestock products calls for insight in the scope to increase actual production levels across the world. This insight can be obtained by using theoretical concepts of production ecology. These concepts distinguish three production levels for livestock: potential (i.e. theoretical maximum) production, which is defined by genotype and climate only; feed-limited production, which is limited by feed quantity and quality; and actual production. The difference between the potential or limited production and the actual production is the yield gap. The objective of this paper, the first in a series of three, is to present a mechanistic, dynamic model simulating potential and feed-limited production for beef cattle, which can be used to assess yield gaps. A novelty of this model, named LiGAPS-Beef (Livestock simulator for Generic analysis of Animal Production Systems – Beef cattle), is the identification of the defining factors (genotype and climate) and limiting factors (feed quality and available feed quantity) for cattle growth by integrating sub-models on thermoregulation, feed intake and digestion, and energy and protein utilisation. Growth of beef cattle is simulated at the animal and herd level. The model is designed to be applicable to different beef production systems across the world. Main model inputs are breed-specific parameters, daily weather data, information about housing, and data on feed quality and quantity. Main model outputs are live weight gain, feed intake and feed efficiency (FE) at the animal and herd level. Here, the model is presented, and its use is illustrated for Charolais and Brahman × Shorthorn cattle in France and Australia. Potential and feed-limited production were assessed successfully, and we show that FE of herds is highest for breeds most adapted to the local climate conditions. LiGAPS-Beef also identified the factors that define and limit growth and production of cattle. Hence, we argue the model has scope to be used as a tool for the assessment and analysis of yield gaps in beef production systems.

The model LiGAPS-Beef (Livestock simulator for Generic analysis of Animal Production Systems – Beef cattle) has been developed to assess potential and feed-limited growth and production of beef cattle in different areas of the world and to identify the processes responsible for the yield gap. Sensitivity analysis and evaluation of model results with experimental data are important steps after model development. The first aim of this paper, therefore, is to identify which parameters affect the output of LiGAPS-Beef most by conducting sensitivity analyses. The second aim is to evaluate the accuracy of the thermoregulation sub-model and the feed intake and digestion sub-model with experimental data. Sensitivity analysis was conducted using a one-at-a-time approach. The upper critical temperature (UCT) simulated with the thermoregulation sub-model was most affected by the body core temperature and parameters affecting latent heat release from the skin. The lower critical temperature (LCT) and UCT were considerably affected by weather variables, especially ambient temperature and wind speed. Sensitivity analysis for the feed intake and digestion sub-model showed that the digested protein per kg feed intake was affected to a larger extent than the metabolisable energy (ME) content. Sensitivity analysis for LiGAPS-Beef was conducted for ¾ Brahman×¼ Shorthorn cattle in Australia and Hereford cattle in Uruguay. Body core temperature, conversion of digestible energy to ME, net energy requirements for maintenance, and several parameters associated with heat release affected feed efficiency at the herd level most. Sensitivity analyses have contributed, therefore, to insight which parameters are to be investigated in more detail when applying LiGAPS-Beef. Model evaluation was conducted by comparing model simulations with independent data from experiments. Measured heat production in experiments corresponded fairly well to the heat production simulated with the thermoregulation sub-model. Measured ME contents from two data sets corresponded well to the ME contents simulated with the feed intake and digestion sub-model. The relative mean absolute errors were 9.3% and 6.4% of the measured ME contents for the two data sets. In conclusion, model evaluation indicates the thermoregulation sub-model can deal with a wide range of weather conditions, and the feed intake and digestion sub-model with a variety of feeds, which corresponds to the aim of LiGAPS-Beef to simulate cattle in different beef production systems across the world.

On-farm implementation of IWM includes the biological and economic rationalization of weed management decisions. This, in turn, requires the ability to predict how weed populations and pressures will change with farm management options such as tillage method. We have developed a set of algorithms that simulate seedling emergence of mixed populations of green and redroot pigweed on the basis of the ecophysiological responses of seed germination and shoot elongation to temperature and, for germination, to moisture. The algorithms were calibrated using field data from 1993 and evaluated with data from two locations in 1994 and 1995. Over both sites and years, for four tillage systems, cumulative emergence was predicted with an overall root mean square error of 1.7%. Overprediction in one year was attributed to moisture shortage. Errors were greater for moldboard plowed plots than for those no-tilled. This decreased ability to predict emergence with increased tillage (soil disturbance) suggests that the algorithms should be modified to account for increased heterogeneity in weed seed distribution and soil moisture within disturbed soil. The algorithms, which were explicitly designed for incorporation within a crop growth model (e.g., CROPSIM), could become a useful part of a decision support system to rationalize weed management. Importantly, they could help predict changes in weed populations as farm management is adjusted, thereby reducing economic and environmental risk in agroecosystems.

Growth is a key characteristic of animals and can be defined as any change in body size per time unit, and is influenced by genotype and environment. Mathematical functions called ‘growth models’ have been used to explain the growth patterns of poultry species. These semi-mechanistic growth models have a non-linear structure, sigmoid shape, and certain biologically meaningful parameters. In poultry science, Gompertz, Logistic, Richards and von Bertalanffy functions have been commonly used to model the growth patterns of birds. In this review, the studies concerned have been summarised under the titles ‘determination of the best-fitting growth model’, ‘a comparison of the growth of poultry species or various experimental groups’, and ‘genetic parameter estimates for growth curve parameters’. This review discusses existing and new approaches to growth modelling.

This review covers the egg production models used in poultry. Similarities and discrepancies among the models are illustrated using a real data obtained from a layer breeder flock. Egg laying in poultry begins at sexual maturity and quickly reaches peak production, and then declines with hen age. For many years, egg production studies have been carried out for the purpose of modelling. Some of the functions were developed for this purpose, such as nonlinear regression equations (Gamma, McNally, McMillan, Adams-Bell, Compartmental, Modified Compartmental, logistic-curvilinear, Gloor, Lokhorst, Narushin-Takma) and some Multiphasic (Segmented Polynomial, Persistency, Individual). Almost all of these functions have been developed to allow modelling on the basis of flock averages. Most models having empirical structure and a small number of parameters are considered biologically meaningful. New models are currently required to be useful for both individual egg yields and to contain biologically relevant parameters.