Article ID: | iaor20072350 |
Country: | Netherlands |
Volume: | 93 |
Issue: | 1/3 |
Start Page Number: | 115 |
End Page Number: | 142 |
Publication Date: | Mar 2007 |
Journal: | Agricultural Systems |
Authors: | Morlon Pierre, Mac Karen, Munier-Jolain Nicolas, Qur Lionel |
Keywords: | artificial intelligence: decision support, decision: studies |
The aim of this research was to improve the advice given by extension institutions to French farmers and to develop a Decision Support System (DSS) for weed control that would match the practical approach adopted by farmers. Farmers running 15 farms with different farming systems in different regions completed comprehensive interviews which allowed them to explain how they deal with weeds. We built temporal diagrams for crop management sequences and decision making. This paper describes the basic framework common to all the farmers interviewed. Each farmer employed pre-established weed control programmes. When designing these programmes, farmers integrated different time scales: the current year, the rotation, and the long term. In the short term, they considered the risks of yield losses and/or lower harvest quality plus harvesting difficulties. In the medium term, they anticipated the risk of finding a weed species in another crop of the rotation where control would be difficult or costly, weighing the risks of yield loss against the cost and effectiveness of solutions, not only in the current crop but also in subsequent crops, so that once again, the rotation was the central focus of weed control. In the long term, their main aim was to limit the soil seed bank to an acceptable level. The farmers interviewed stated that they would continue to implement a weed control programme that they deemed satisfactory as long as no new problem appeared, and until they could learn about more effective technical solutions. When designing a DSS that will ensure successful, more sustainable weed management practices, it is crucial to take account of both the complexity of the decision-making process and the multicriteria nature of decision making.