Adapting the feed, the animal and the feeding techniques to improve the efficiency and sustainability of monogastric livestock production systems
Adapting the feed, the animal and the feeding techniques to improve the efficiency and sustainability of monogastric livestock production systems

Deliverable D3.6 Detailed specification for the calculation “engine” to be used in the DSS in Task 4.1

Kaposvár University, IRTA, INRA
Publication date: 
25 February 2019
Full title: 
Deliverable D3.6 Detailed specification for the calculation “engine” to be used in the DSS in Task 4.1
Publishing information: 
Feed-a-Gene, February 2019


The objective of the deliverable is to indicate how (elements of) the conceptual models that have been developed in WP3 can be used in the Decision Support Systems (DSS) developed in Task 4.1 for pigs and poultry.


A first version of an animal model to be used for the development of a DSS in Task 4.1 was provided as deliverable D3.1. This model was a modified version of the InraPorc model (for growing pigs) and addressed aspects of phosphorus and calcium metabolism. Elements of this deliverable have been used in Task 4.1 to develop decision support tools for the real-time determination of animal nutritional requirements (see deliverable D4.1). A real-time model was developed to calculate the appropriate feed and nutrient supply for an individual animal for the next day or period. This real-time model is based on the predicted feed intake and body weight gain from which the nutrient requirements (e.g., lysine) are determined.

The modelling approaches of WP4 and WP3 are fundamentally different: WP4 uses real-time modelling using predictions of feed intake and body weight gain, while WP3 worked on modelling “once we have the data” and by “looking back”. Despite these differences, a number of issues and recommendations resulting from WP3 can be used in the real-time modelling of nutritional requirements.

WP3 aimed to develop mathematical models with different approaches including mechanistic dynamic models to simulate nutrient digestion (Task 3.1) and the post-digestive metabolism of nutrients in monogastric animals (Task 3.2). At this stage, the complete digestion model (deliverable D3.2) is probably not suited to be integrated in a decision support system for precision feeding. However, it has been generic in design for both pigs and poultry and the (simulated) effects of body weight and diet composition on digestibility can be used to construct empirical equations that can adjust (constant) table values of nutrient digestibility that are currently used in models and decision support systems.

This concepts proposed in deliverable D3.1 have been developed further to address the effect of different environmental and nutritional aspects on feed intake and predictions of the fatty acid profile in different body components. Also, deliverable D3.1 did not include models for broilers and laying hens, and these aspects are addressed in the current deliverable.

Adaptations to the post-digestive metabolic model to including phosphorus (P) metabolism and adaptations to the feed intake module: The P-module represent phosphorus metabolism and predicts the effect of dietary digestible P supply on P retention and urinary P excretion. The model is integrated in InraPorc and simulates the regulatory mechanism of P supply such as that an insufficient P supply P limits growth and reduces the appetite of the animal. Also, in the original approach, daily feed intake is a phenotypic trait and driven by body weight, but it does not take explicitly into account that dietary and environmental factors can limit the actual feed intake. The updated feed intake module quantifies the effect of specific environmental effects of feed intake (e.g., ambient temperature, stocking density). Feed intake is very variable on a day-to-day basis (see deliverable D4.5), and we suggest that cumulative feed intake could provide a better basis for feed intake predictions. Cumulative feed intake represents a trajectory that the animal may seek to achieve. Rather than attempting to predict what an animal is going to eat tomorrow, it may be better to forecast feed intake for a longer time window, in which cumulative feed intake could be used as a target. The methodologies developed in Task 3.4 and reported in deliverable D3.5 can be used to construct confidence intervals for the predictions of feed intake and body weight gain. This can be used to determine the period for which reasonable predictions for body weight and (cumulative) feed intake can be made, which would also results in reasonable predictions of nutrient requirements in a decision support tool for precision feeding.

A broiler growth and laying hen model based on InraPorc: the InraPorc model predicting post-digestive nutrient utilization for growing pigs was adapted to broilers. Without major changes to the core structure of the model, species-specific parameters and equations were identified. The original pig model uses a generic approach to nutrient partitioning, and the adapted model was able to represent the similar, underlying mechanisms of nutrient partitioning for broilers. The same model core was also used for layers assuming that a laying hen is more mature than a broiler (i.e., with a lower body protein deposition) and has a priority to produce eggs. The model predicts the effect of the digestible nutrient supply, including energy, amino acids, Ca and P, on egg production.

The above-mentioned models are ready to be included in a DSS for precision livestock farming in different ways: i) the model-generated outputs can be compared with real-time collected data, and ii) the real-time daily or, preferable, the cumulative feed intake can be used as input in the simulation model to predict the performance for the next day(s). A difference between the simulation (expected trajectory) and real-time measurements can be used as inputs to an early warning system that can be used to identify technological or health problems. The concept of the robustness module (task 3.3 and deliverable D3.4) can also be considered for the DSS developed in task 4.1.

Media category: