Search

International audience; The objective of the Feed-a-Gene project (2015-2020; H2020 grant agreement 633531) was to adapt the feed, the animal, and the feeding techniques to improve the efficiency and sustainability of monogastric livestock production systems. For ‘feed’, we concluded that non-GMO European soybeans can be processed in small-scale processing plants to obtain soybean meals with similar or even better nutritional quality as imported soybean meals. Also, there is a potential for the use of green biomass to extract protein for monogastric animals to replace imported soybean meal in organic farming. Concerning the ‘animal’, monogastric animals have to rely more on alternative feed resources that are less or not in competition with other uses. These resources are typically more variable and of poorer nutritional quality. With new molecular and automated records, animals can be selected that do best with feed that is worse and transform these alternative feeds into meat, respecting their welfare and robustness and limiting environmental impacts. Precision livestock feeding is a novel technology that can be using to provide tailored diets to animals according to their (individual) requirements, thereby reducing nutrient excretion and the environmental impact of livestock production. The potential of precision livestock farming is much larger than what we explored in Feed-a-Gene (e.g. precision medication, design of housing systems). It is not a question of ‘if’ but of ‘when’ these systems will be implemented, and ‘how’ society will perceive trade-offs of improved resource use, reduced environmental impact, improved welfare versus a perceived further industrialization of livestock production.

International audience; The objective of Feed-a-Gene (H2020 grant agreement 633531) was to adapt the feed, the animal and the feeding techniques to improve the efficiency and sustainability of monogastric livestock production systems. Efficiency is a complex trait affecting the components of sustainability. Imported soybean meal is a high-quality protein source, but its use can have environmental and social impacts. Technological treatments applied to European-grown protein sources (i.e. soybean meal, rapeseed meal, and green biomass) allow to obtain fractions with different nutritional characteristics that can be fed to different livestock species. Replacing Brazilian soybean meal with locally-produced protein sources reduced energy cost and impact on climate change, but resulted in a transfer of land-use and in more arable land use. Novel traits indicative for feed efficiency were developed and tested (e.g. individual feed intake in broilers, genomic information on nitrogen efficiency, large-scale measurements of digestibility in individual pigs, blood metabolites, and behavioural traits). These traits can be used in livestock management strategies, and digestibility, microbiota, and blood biomarkers appeared promising for genetic selection. Different mathematic models were developed to assess digestive and metabolic efficiency, animal robustness, and variation among animals. A userfriendly tool was developed allowing users to work with these models. Prototypes of precision feeding systems were developed for broilers, growing pigs, and sows. These systems adjust the type of feed to the daily changing nutritional requirements of individual or groups of animals. Precision feeding in pigs reduced key environmental impacts and increased profitability compared to conventional feeding systems.

The aim of the study was to investigate whether a pig growth model is suitable to be modified and adapted for broilers. As monogastric animals, pigs and poultry share many similarities in their digestion and metabolism, many structures (body protein and lipid stores) and the nutrient flows of the underlying metabolic pathways are similar among species. For that purpose, the InraPorc model was used as a basis to predict growth performance and body composition at slaughter in broilers. Firstly, the backbone of InraPorc was translated to Excel and examined whether the equations were suitable for growing birds by evaluating the parameters and model behavior. After determining of differences the model was modified for predicting growth in broilers in relation to the nutrient supply. The model core is very generic in terms of representing the most important nutrient flows and the transformation from absorbed nutrients to protein and fat accretion during growth. The idea of nutrient utilization is mainly based on concepts used in net energy and ideal protein systems. The model is driven by feed intake, the partitioning of energy between protein and lipid deposition, and availability of dietary protein and amino acids. Parameters of the Gompertz function were adjusted using literature data to express maximum feather-free body protein deposition. A separate Gompertz equation was used to estimate feather growth and protein content of feather was assumed to be a nonlinear function of age. The amino acid loss with feathers was considered as a part of the maintenance requirement and the fasting heat production was used as the NE requirement for maintenance (460 kJ/(kg BW)0.70/day). Feed intake for broilers was expressed on a NE bases and estimated by a Gamma-function (which expresses the NE intake as multiples of maintenance), as in InraPorc. Parameters of the Gamma function were adjusted to experimental data from different broiler studies. This study is part of the Feed-a-Gene project and received funding from the European Union’s H2020 program under grant agreement no. 633531.

Prediction of pig performance from data on nutrient intake and animal properties makes it easier to obtain a better productivity. It provides tools to arrive at desired outputs, or to calculate required inputs. Thus it enables production to be flexible, safe and less erratic. It is to be expected that the results will give a more profitable pig production. In practice, different types of models are used, mostly by feed producers, but also in farm management programmes. Each of these existing models was designed to meet a certain objective. The classification of different types of models, and the benefits of using them, are presented in the literature overview of the thesis. After a general overview of modelling, a critical evaluation was provided on existing models. It was concluded from the literature, that a comprehensive model, which predicts the chemical composition in different parts of the body, like in lean or in the meat, does not exist. It was also concluded that mechanistic approach should be used to modelling growth. The conceptual basis of a mechanistic model was developed in accordance with basic properties of protein and lipid metabolism. Since nutrients are almost exclusively absorbed in the hydrolyzed form, simulation of use of nutrients for growth should, at least to some extent, make use of biochemical pathways. Therefore, a biological approach to simulation of anatomical body composition is pretended as it follows nutrients from ingestion through intermediary metabolism to deposition as body fat and protein, preferably in distinct tissues or tissue groups. Prediction of anatomical body composition therefore has to be based on deposition of the chemical entities.Therefore the scope of the present thesis was 1) to develop a mechanistic-dynamic model for growing and fattening pigs which predicts anatomical and chemical body composition at slaughter; 2) to determine which model parameters are sensitive to changes in the model; 3) to determine the model accuracy by quantitative and qualitative prediction of the model tested with independent data; 4) to complete an experiment to define fat production potential of different energy sources at low and high feeding levels, and 5) to study the effect of different energy sources at two feeding levels on the distribution of fat deposition during the fattening period.The thesis presents both the description and the evaluation of the growth model. It was concluded that the developed model predicts growth rate as well as chemical and anatomical body compositions of gilts in the 20-105 kg live weight range, from nutrient intake. The model represents partitioning of nutrients from feed intake through intermediary metabolism to synthesis of body protein and body fat. State variables of the model are lysine, acetyl-CoA equivalents, glucose, VFA, and fatty acids as metabolite pools, as well as protein in muscle, hide, bone and viscera and body fat as body constituent pools. It is assumed that fluxes of metabolites follow saturation kinetics depending on metabolite concentrations. Anatomical body composition is predicted from chemical body composition and accretion. Partitioning of protein, fat, water and ash into muscle, organs, hide and bone fractions are described by allometric equations, driven by rates of muscle protein and body fat deposition. Two experiments were used in the model calibration process, one with 95 growing pigs (20-45 kg) fed different ileal digestible lysine intakes at two feeding levels, and another with 100 growing and fattening pigs (20-105 kg), which received different energy intakes. Differential equations were solved numerically for a given set of initial conditions and parameter values. The integration interval used was 0.01 day, with the fourth-order fixed-step-length Runge-Kutta algorithm. The muscle protein and body fat deposition rates were considered in different weight ranges and for the whole fattening period. Results presented were not sensitive to small changes in initial conditions, or to smaller integration step sizes.In the model evaluation the predicted response of the pigs to changes in model parameters, and to changes in nutrient intakes, are shown. As a result of the sensitivity analysis, the model was responsive to changes in a number of the model parameters examined. Changes in maintenance energy requirements, and the fractional degradation rate of muscle protein, have the largest impact on tissue deposition rates. The model is highly sensitive to changes in the maximal velocity and steepness parameter of lysine utilisation for muscle protein synthesis. Those parameters which directly affect the size of the lysine pool generally have a considerable influence on model predictions. Furthermore, it should be noted that results of this sensitivity analysis depend on nutrient intakes of the reference simulation. The model was relatively insensitive to changes of parameters regarding energy metabolism. It was concluded that the probable reason was that protein and/or lysine was more limiting within the simulated conditions. The model was further tested by independent published data. In general, the model satisfactorily predicted qualitative pig responses to a wide range of variations in nutrient supply. The predicted chemical and anatomical body composition, and also the distribution of protein and fat, were satisfactory in model testing. In most cases, errors in the predicted parameters attributed to the deviation of the regression slope were minor. It was assumed that the major factors contributing to the relatively large bias, observed for most predicted growth characteristics, was variation in pig performance among genotypes, or differences in environmental conditions. Based on the comparison of model simulations with independent data sets, it was recommended to improve the model regarding prediction of protein and fat deposition rates from nutrient intake of different energy sources.It was found that literature data on the effect of different energy sources on fat deposition was limited. In non-protein energy fraction of the diet, dietary lipids, starch and rapidly fermentable non-starch polysaccharides (NSP) are major energy sources. Lipids are absorbed as long-chain fatty acids and starch as glucose. Dietary NSP is fermented and the short-chain fatty acids produced enter intermediary metabolism as an energy source. Equal intakes of energy from glucose, long-chain fatty acids and short-chains fatty acids might result in different fat deposition rates, and quite likely, result in different distributions of body fat over the tissues. There is, however, little quantitative data available on effects of energy source on partitioning of body lipids. Therefore a fattening trial was completed to: 1) study the effect of extra energy intake from fermentable NSP, digestible starch and digestible fat used for fat deposition under protein limiting conditions; 2) determine the location of the fat deposition resulting from extra intake fermentable NSP, digestible starch and digestible fat; 3) determine if the extra fat deposition from different energy sources depends on the level of feed intake, and 4) quantify potential interactions between feed intake level and energy source on the location of extra body fat deposition.A total of 58 hybrid individually housed pigs were used in the trial with an initial body weight of 48±4 kg. The experimental treatments were arranged in a 3x2 factorial design, with three energy sources (i.e. fermentable NSP, digestible starch and digestible fat, all added to a control diet) at each of two energy levels. Within each energy level, daily nutrient intakes were the same with regard to digestible protein, ileal digestible lysine and other amino acids, vitamins and minerals. Treatments had an isocaloric proportion of daily nutrient intake derived from each energy source (0.2 MJ DE/kg 0.75 ), in addition to the nutrients from control diet. It was equal with 11 g/kg 0.75 highly fermentable NSP, 11 g/kg 0.75 starch or 5 g/kg 0.75 digestible fat daily. The DE intakes were 2.0 and 3.0 maintenance requirement in control groups. The additional energy from different sources increased DE intake up to 2.4 and 3.4 times maintenance requirement at low and high feeding levels, respectively. To obtaininitial values, ten pigs were slaughtered at 48±4 kg and the treatment pigs at 106±3 kg body weight. Each body was dissected into four fractions being: 1) lean, 2) organs, 3) hide and subcutaneous fat, and 4) offal. Chemical body composition was determined in each body fraction. The differences between fat deposition of body parts in the control group, and the other treatments, resulted in the additional energy derived from each energy source. As a conclusion from the study, under protein limiting conditions, extra energy intake from fermentable NSP, digestible starch and digestible fat resulted in similar fat deposition. Preferential deposition of extra energy intake in various fat depots did not depend on the energy source. The extra fat deposition from fermentable NSP, digestible starch and digestible fat deposited as body fat was similar at both the low and high levels of feed intake.In the General Discussion, some consequences of the mechanistic approach were discussed and then substantial attention was devoted to the practical aspects of the model. The later part of the General Discussion focuses on representation of different energy sources as an aspect of the model. The energetic efficiency of the different dietary energy sources is discussed, based on data from Chapter 6. The growth model is further evaluated by results of the fattening study. Moreover, data from the fattening study are analysed regarding the distribution of fat deposition. Consequences of the fattening study on the model is discussed regarding the effect of energy sources on energetic efficiency, and on location of fat deposition in the pig. Finally, a new application of the present model is introduced in addition to development of feeding strategies and identifying research priorities.

The calcium and phosphorous retention of layers was evaluated at different production levels (45 - 75 - 95 % of final production peak) with Hy-Line Brown hybrid layers during the first twelve weeks of the laying cycle. 32 layers evenly distributed over 4 treatments (Trts). The Ca content of the diets was identical across all Trts (32,5 g/kg), while P contents and phytase activities differed. In the first Trt the P content of the diet was 2.5 g/kg non-phytate P, without any phytase added. In the second Trt the P content of the diet was reduced by 40 % compared to Trt 1 (1.5 g/kg non-phytate P) without phytase supplementation. In the third and fourth Trts the P content of the diets was the same as in 2 (1.5 g/kg non-phytate P) but the diets were supplemented with phytase (3-phytase produced by trichoderma reesei) at a rate of 250 PPU/kg (Trt 3) and 500 PPU/kg (Trt 4), respectively. According to the results the different P content of the diets affected (P≤0.05) the rate of Ca retention at all trial phases. In the case of reduced P level diets the different phytase inclusion rates did not influence the amount of Ca retained by the birds (P≥0.05). Within the same Trt Ca retention increased (P≤0.05) with increased production intensity. At reduced P content phytase improved (P≤0.05) the P retention of the layers at all trial phases. There was no difference between both phytase application rates. As a result of phytase supplementation P retention reached the P retention of positive control birds (Trt 1) in all trial phases.

The objective of the present study was to investigate the effect of an essential oil product on growth performance and cellular immune response of 28-day-old weaned piglets. A total of 348 piglets (50% gilts, 50% barrows) were assigned to three dietary treatments (6 pens/trt). Th e basal diet was a commercial feed that was supplemented without any growth promoter (NC), with antibiotic growth promoter of 40 ppm avilamycin (PC), or with 0.25 g of an essential oil product (EO) per kg of feed. All pigs were immunized by inactivated Aujeszky’s disease virus vaccine at week one and three of the experiment (28- and 44-days-age, respectively). Blood samples were taken four times (on day one, 16, 24, 32 of the experiment) for lymphocyte stimulation (LST) tests with ConA, PWM, PHA used as non-specific and Aujeszky virus used as specific mitogens from 2 pigs/pen. All piglets were individually weighed on day 0, 8, 16, 24 and 32 of the trial. There was no significant difference among average daily gain, feed intake and feed conversion ratio of piglets fed different dietary treatments. The non-specific LST test at the 4th blood sampling showed higher values in pigs received feeds with essential oil supplementation (EO) than that of the positive (PC) and negative control (NC) groups (P

The aim of the present trial was to study the effect of two different dietary phosphorus (P) levels on the laying performance in long-term (17 months) egg production and egg shell quality in terms of strength and thickness from the 7th to 17th month of laying. In case of eggshell quality parameters the time effect as well as the interaction between dietary P levels and time were also examined. Sixty Tetra-SL layers were fed diets with two different levels of P (4.9 or 4.4 g/kg total phosphorus) at constant 38.5 g/kg calcium level. In the course of the trial egg production %, egg weight, feed intake and body weight were recorded. At every 4th weeks 20 eggs per treatment were broken to deretmine the strength and thickness of eggshell. Results showed that the examined two levels of dietary P did not affect the percentage of egg production and the feed conversion ratio (kg feed/kg egg mass). The egg weight significantly increased and eggshell strength was significantly lower when hens received the lower dietary P feed. During the laying period, both the eggshell strength and thickness gradually decreased as laying period time went on. In conclusion phosphorus content of the layer diet can be lowered, without reducing the egg production, nevertheless approaching the end of the long laying period, higher P-content feed suggested, whereby eggshell strength can be improved.

As researchers search for tools to automate individual animal tracking, particularly in the livestock industry, they come across papers that make strong claims about the ability of such programs to accurately track individual animals within group-housed laboratory settings and imply that this ability will translate to other situations and species. We selected four of these programs (idTracker, ToxTrac, BioTracker, and Ctrax) and tested their ability to track single, pairs, and groups of pigs in an indoor pen typical of commercial farms and a pen modified to improve lighting, contrast, and camera location. We concluded that at present, these four tracking programs do not perform robustly enough to be adapted for use processing videos from commercial-style pig farms, including smaller research farms. It is hoped that identifying limitations of detection programs can help make systems more robust and user-friendly for future use.

ispartof: Practical Precision Livestock Farming: Hands-on experiences with PLF technologies in commercial and R&D settings pages:71-83 ispartof: pages:71-83 status: Published online

Modern livestock production systems can heavily challenge the welfare of animals. A way to manage farm animal welfare in real-time is through the use of Precision Livestock Farming (PLF) systems. The aim of this chapter is to present the authors’ experience in conducting research towards the development of a PLF system that can provide early warnings to pig farmers when the risk of tail damage is high. During our work, we identified challenges related to: (1) choosing the gold standard; (2) choosing the appropriate technology; (3) performing model validation and implementation. Choosing the gold standard for tail damage is challenged by the fact that tail damage will have different stages of development. Also, choosing the optimal time to raise the warning is a compromise between warning early enough to provide farmers’ with sufficient time to prevent the problem and late enough so that the probability of developing into a real problem is high. The gold standard must also be easy to standardise between different observers and conditions. Choice of appropriate technology needs to consider that tail damage is a multifactorial welfare problem with multiple risk factors and thus, several sensor technologies may be needed. Also the nature of the animals to possibly explore and destroy a sensor as well as the harsh environment of livestock buildings is a challenge. Major barriers for performing model validation and implementation are that already in the beginning, data needs to be saved for internal validation, and collaboration contracts with research herds across nations, farmers and information technology companies needs to be in place; to ensure data for external validation and prototype development. Developing a PLF system for welfare monitoring demands that several considerations and decisions are made already in the beginning. These considerations will take time and decisions made will often be a compromise.