Coupling a sow herd model with a bioclimatic model of gestation rooms: development and evaluation
Abstract
Global warming and the increased frequency of heat waves intensify the risk for reproductive sows to be exposed to temperatures above their thermoneutrality zone. This affects the overall sow herd performance through feeding behaviour perturbation, reduced prolificacy, unsuccessful insemination, or increased mortality. A model was developed on Python to simulate dynamic interactions between ambient temperature in gestation rooms and individual performances of sows in the herd. Dynamics of ambient temperature result from thermal balance between heat losses (walls, air renewal) and heat sources (animals, heaters) in the room, depending on feeding management, the fan and heating systems management, walls’ thickness and conductivity, and outside temperature. The model runs on an hourly time step and couples an individual-based module of herd management and feeding practices with a model of thermal balance in the room. The herd module applies discrete events associated to sow reproduction and farmer’s practices. It also describes rooms with reproduction stage, number of animals and occupation time. The heat production of each sow is modelled as a function of its gestation stage, live body weight, production objectives (backfat thickness and litter size at farrowing) and diet. Data from three batches of sows were used to evaluate the model’s ability to predict sows’ performance and the effective ambient temperature within a gestation room. The sows’ ingestion and live weight evolution were similar between observed and simulated data. Simulated temperatures were similar to the observed temperatures (mean error=0.3, 0.2 and 0.8 °C; RMSEP=0.8, 0.9 and 1.2 °C, respectively for the three batches). However, the simulated variation amplitude is larger than the observed one, especially during heat peaks. A better description of the hourly distribution of feed intake, individual activity level, and regulation rules of fan and heating systems should improve the accuracy of model predictions. The model will be integrated into a decision support tool to assess the vulnerability of swine systems to climate change.