Assessment of the diversity of crop rotations based on network analysis indicators

More diversified crop rotations are a key factor in reducing weed, disease and pest pressure while reducing the use of phytosanitary products. The increase in available data calls for the development of new methods and indicators to characterize crop rotations. This paper presents an application of network analysis to assess the diversity of crop rotations from the Land Parcel Identification System, which now provides field boundaries and type of crops grown in a given year for farmers receiving subsidies from the European Common Agricultural Policy. Different indicators are presented to compare the diversity of crop rotations for the 22 regions of mainland France (corresponding to the boundaries of former administrative regions) and the influence of some methodological choices are discussed. Using LPIS data, previous-following crop pairs have been identified for two thirds of the French fields for three crop successions (from 2017 to 2018, from 2018 to 2019 and from 2019 to 2020). These data were used to build crop rotation networks for each region. Crop rotations were simulated from those networks in order to estimate the return time of soft winter wheat, which is the most cultivated crop in the country. Crop rotation networks were similar between the three crop successions compared, but differed among regions. Ignoring the most uncommon previous-following crop pairs, the mean number of precedents per crop ranged from three to nine depending on the region. The estimated return times for winter wheat increase when taking into account grasslands in crop rotation networks, but in any case the use of phytosanitary products was negatively correlated with the return time. The methodology developed in this article provides some initial guidelines for developing relevant agronomic indicators from crop rotation network analysis. It has been shown that the estimated return time could be an indicator of the intensity of use of phytosanitary products, and could therefore be used to guide public policies aimed at reducing the use of these products.