Deliverable D4.6 Prototypes of precision feeding systems gestating and lactating sows

Objectives

Nutritional studies indicate that nutrient requirements for pregnant and lactation differ largely among sows and according tbo the stage of pregnancy or lactation, whereas in practice the same feed is generally fed to all sows in a given herd, for a given physiological stage (i.e., gestation or lactation). The availability of new technologies for high-throughput phenotyping of sows and their environment, and of innovative feeders that allow the distribution of different diets, offers opportunities for a renewed and practical implementation of prediction models of nutrient requirements to improve animal performance and the efficiency of nutrient utilisation, and to reduce feed costs and environmental impacts. The objective of this task was to develop prototypes of precision feeding systems, including the development of the decision support system (DSS) for gestating and lactating sows allowing the distribution of a tailored ration to each sow on each day.

Rationale

The precision feeding system is based on the development of a decision support system (DSS) connected to an automated feeder for feed distribution. This DSS informs the feeder with the amount of each of the different diets to be fed to a given sow over a given day or period, generally two diets differing in their nutrient content. To take that decision, the DSS uses the flow of information relative to the individual sow, her housing conditions and the general feeding strategy in the farm. This information may be provided by sensors and or by the farmer and is stored in a herd database. From the available information, which may vary according to the equipment available on the farm and the data management system, the DSS builds the "best guess" decision to be transmitted to the automated feeder. This involves two steps: (i) the determination of the energy, amino acid, and mineral requirements and (ii) the determination of the amount and composition of the ration to be fed. This ration is prepared by mixing (generally two) different diets available in the automated feeder.

Energy and nutrient requirements are determined according to a factorial approach. The metabolizable energy requirement is calculated as the sum of the requirements for maintenance, physical activity, thermoregulation, growth, and the constitution of body reserves, and (i) development of foetuses and uterine contents during gestation or (ii) synthesis of milk during lactation. These calculations were adapted from the InraPorc® model with some improvements based on recent literature results.

Farm data were used for an in silico evaluation of the precision feeding systems. For gestation, the database was obtained from an experimental farm and contained data from 2511 gestating sows with information about their body condition at mating (i.e., body weight (BW) and backfat thickness (BT) and, at farrowing, their performance (i.e., prolificacy, piglet birth weights). For lactation, the database was obtained from two experimental farms with data from 633 and 817 lactating sows (i.e., parity, body condition at farrowing, and daily feed intake) and their litter (i.e., litter size throughout lactation and piglet weights at birth and at weaning). These data were used to calibrate the parameters of InraPorc® for these phenotypes.

The databases were used to evaluate the interest of precision feeding strategy through simulation approaches. A conventional 1-phase feeding strategy (CF) was compared to a precision feeding (PF) strategy consisting of the mixing of two diets with either a low (L) or a high (H) nutrient content.

For gestation, the CF, L, and H diets contained 4.8, 3.0, and 6.0 g/kg standardized digestible (SID) lysine, and 14, 9, and 16% crude protein, respectively. On average, the level of incorporation of diet L in the PF strategy was 84%, the value being lower in first parity sows (67%). The level or incorporation of diet L decreased during gestation from almost 100% in the first week to less than 30% in the last week, in agreement with the change in amino acid requirements. Compared to the CF strategy, the PF strategy resulted in a 27% decrease in total SID lysine supply and in a 24% decrease in the total crude protein supply. The nitrogen excretion was reduced by 30%, whereas the feed cost decreased by 4.6%. The proportion of sows that were underfed in the last two weeks of lactation decreased from more than 60% with CF to less than 5% with PF. For first parity sows, the difference was even more marked. Conversely, the proportion of sows that were overfed was drastically reduced with PF strategy.

For lactation, the CF, L, and H diets contained 8.5, 11.5 and 6.5 g/kg SID lysine, respectively. On average, the level of incorporation of diet L in the PF strategy was 87%, the value being lower in first parity sows, which have lower feed intake. This resulted in a reduction by about 8% of the average lysine intake, a reduction of 7.1% of N excretion, and a reduction of 1.5% in feed cost. With CF, about 60% of the sows received lysine that was more than 10% in excess of their requirements, and about 20% of the sows received less than 90% of their requirements. With PF, up to 60% sows were adequately fed (i.e., between 90% and 110% of their requirements), and about 10% of the sows were underfed. With PF, protein intake was reduced on average by 5.1%, N excretion was reduced by 8.5%, and feed cost was reduced by 1.5%.

The DSS were validated in silico using a large number of real farm data. For both gestation and lactation, a conventional 1-phase feeding strategy (CF) was compared to a precision feeding (PF) strategy consisting in the mixing of two diets with either a low (L) or a high (H) nutrient content. With precision feeding during gestation, protein intake was reduced by 24% compared to conventional feeding and N excretion was reduced by almost 30%. The average feed cost was decreased by 4.6%. The proportion of sows that were underfed in the last two weeks of gestation was drastically reduced with precision feeding, whereas in the beginning of gestation, the proportion of overfed sows was reduced. With precision feeding during lactation, protein intake was reduced by 5.1%, N excretion was reduced by 8.5%, and feed cost was reduced by 1.5%.

Three different prototypes of precision feeding systems have been implemented on the basis of the adaptation of industrial equipment, one at IFIP for gestation and two at INRA for gestation and lactation. They are now available for demonstration.