Precision farming in extensive livestock systems

Abstract

Despite the strong constraints which apply to extensive livestock farming systems (large herds, large surface areas, pasture-based feeding and eventually low animal productivity), original solutions of precision farming can be developed. Precision farming can especially rely on the individual electronic identification systems (RFID) to recognize or locate the animals. Depending on the utilization of RFID-readers (either stationary or portable devices, held by an operator or carried by a male), it is possible to produce inventories, sorting operations or estrus detection. Animal behavior is most difficult to control at pasture, since the strategic placement of attraction points (water sources, salt blocks, and supplements) can prove insufficient to control the spatial distribution of the grazing pressure in a context of strong pastoral and environmental objectives. In situations where either traditional fences or shepherding cannot be implemented, virtual fences, which are based on a specific behavioral education of the animals, could contribute to a sustainable utilization of rangeland areas. In extensive livestock systems, precision farming does not apply at the individual scale, but implies adjusting management practices to virtual groups of animals sharing the same characteristics, such groups being made and unmade by efficient sorting devices. The utilization of automated herd-monitoring systems based on individual information produces a great amount of data. Whatever the type of sensor used, the data collected must be stored in an information system, then analyzed with appropriate methods (algorithms, statistics…) in order to be translated into decision-support indicators or into actions implemented by machines (sorting gate, weighing, feed distribution). As an alternative to individual monitoring, a parsimonious utilization of sensors can suffice to give alerts (intrusion, predator attack, exit from a given geographical area); the amount of data produced is thus minimized. Further research is needed to propose appropriate methods for data analysis and decision rules, based upon animal behavior modeling. In extensive farming systems, the farmer plays an essential role; when large herds and surface areas are involved, electronic devices and decision-support systems are to be considered as specific aids to be integrated in a larger management strategy. Especially, the farmer has to decide which animals to monitor, where to locate the sensors/readers or virtual fences, and when to implement precision farming, i.e. which are the critical periods. At the moment, commercial solutions for devices and software lack the versatility needed to apply to extensive systems.