Impact on the spatio temporal distribution of weeds in a decision support system for applying herbicides on soya corn

Project Code PRR07-010

Project Lead

Bernard Panneton - Agriculture and Agri-Food Canada

Objective

To conduct a spatial analysis of weed coverage in corn and soybean fields treated with various herbicides and rates and validate weed management strategies using localized and reduced rate herbicide applications

Summary of Results

Background

Herbicides continue to dominate pesticide sales in Canada, accounting for 78% of all sales amounting to over $1.3M in 2005. Moreover, among all pesticides used, it is the herbicides that are detected most frequently to contaminate ground water. This form of contamination may increase with intensification of corn production in response to increased demand for bio-fuels.

In conventional corn and soybean fields (fall plowing and pre-seeding harrowing), weeds often grow in pockets and the level of infestation can vary greatly from one field to the next. This suggests that it may not be necessary to treat the entire field with herbicide. Previous studies have shown that relative coverage (RC) of a field by weeds may be used as an indicator to determine whether to treat against weeds. In these studies, localized and/or reduced-rate herbicide applications based on RC reduced annual herbicide use by up to 85%. There is then a possibility to economize on herbicide use if a decision support tool can be developed to enable growers determine the areas in the field where application of herbicides could be avoided.

This project aimed at using advanced imaging technology such as photographic sampling and spatial weed coverage data to combine two levels of herbicide application decisions, i.e. areas to be treated and herbicide rate, under a soybean/corn rotation system.

Approaches

Field studies were conducted from 2004 to 2007 on two corn and soybean crop rotations in Beaumont and St-Jean-sur-Richelieu, Quebec. Sections of fields were treated with varying in-crop herbicides, with a full constant dose. Variable doses (no herbicide, half or full doses) were based on the calculated RC value.

The RC of weeds was measured using photographs taken with a standard colour camera, aimed vertically at the soil. Surfaces covered by weeds and crops were measured on the photographs. The ratio of these two surfaces estimated the RC. A value of zero (0) indicated that there were no weeds and a value of one (1) indicated that the surface occupied by weeds was as much as that occupied by the crop. Previous experimental data indicated that when RC was below 0.4, yield losses were negligible, and herbicide applications were therefore based on this index.

Results

In its first year of application, the RC approach helped maintain yields for that year, and it did so by reducing herbicide use by 85%. Unfortunately, for the next three years, herbicide savings only affected 10% of the areas. That was likely due to the fact that residual weed populations renewed the seed bank at levels enabling an increase in densities for the following year. These results clearly show that the RC-based model and the maintenance of yields do not represent a satisfactory long-term solution.

The next step was to determine, an RC threshold under which weed infestation can be tolerated. Analysis helped establish that the threshold is very low (0.077%). A method, based on oblique colour photographs, was developed to rapidly and accurately assess the RC for weeds. Unfortunately, it was not precise enough to measure infestation thresholds corresponding to a 0.077% weed cover. Using oblique photography to assess the RC for weeds is therefore not recommended.

Although results were disappointing, this project established several important facts. Especially, the study revealed that within the crop system studied (conventional corn-soybean rotation) the potential for herbicide savings is low when based on the estimate of weed coverage for a given field (at least over several years).

For further information, please contact Dr. Bernard Panneton.

Date modified: