This Nature Plants paper by researchers from agroecological and agronomical research institutions in France used a statistical modelling approach to predict the effects of reducing pesticide use on the productivity and profitability of French arable farms.
The paper begins by outlining the ways in which pesticide over-use poses significant health and environmental problems, and describes international and national pesticide reduction targets. Agroecology - not explicitly defined by the paper, but, from the context, referring to integrated, non-chemical-based pest management systems - as a straight alternative to pesticide use may require a massive and logistically difficult overhaul of current production systems, so the researchers sought to identify levels of pesticide reduction which could feasibly be achieved without replacement by such alternative pest management systems, and which would not reduce yields.
The subjects of the study were 946 non-organic arable farms in France, which ranged widely in their pesticide use. The statistical modelling approach taken by the study was as follows:
- A large number of regression models was created, linking pesticide usage (quantified by Treatment Frequency Index, TFI) and 22 different quantitative descriptors of the biophysical and socio-economic situations of the farms, to the farms’ productivity (yields converted into giga-Joules per hectare per year) and profitability (Euros per hectare per year, based on ten crop product and farm input price scenarios).
- A statistical method called Lasso was applied to simplify these many complex regression models into a simpler model which links the most (statistically) important inputs and farm characteristics to the outputs of interest (productivity and profitability).
- This model could then be applied to each farm in turn to assess the importance of pesticide use (TFI) in determining the productivity/profitability of each farm. This was assessed by modelling, for each farm, the effect on productivity/profitability of increasing TFI by one unit, referred to as the TFI effect. (Note that the researchers looked at total TFI as well as TFI for specific pesticide types, e.g. herbicides, insecticides, fungicides)
Figure 1 Adapted from Lechenet et al. (2017). Maps showing model predictions for association between TFI and productivity or profitability. Red = positive marginal TFI effect, yellow = no significant marginal TFI effect, green = negative marginal TFI effect.
The key findings of the paper were (Fig. 1):
- Total pesticide use effects on productivity:
o 6% of farms showed a positive TFI effect – i.e. an increase in pesticide use of one TFI unit is associated (by the model) with an increase in productivity (suggesting that on these farms reducing pesticide use may harm productivity).
o 39% of farms showed a negative TFI effect – i.e. an increase in pesticide use of one TFI unit is associated with a decrease in productivity (suggesting that decreasing pesticide use could actually increase productivity).
o 55% of farms showed no significant TFI effect – i.e. productivity would not be affected by a one unit increase in pesticide use (suggesting that pesticide use could be reduced without affecting productivity).
Total pesticide use on profitability (same explanations as above):
o 22.2% of farms showed a positive TFI effect
o 11.1% of farms showed a negative TFI effect
o 66.6% of farms showed no significant TFI effect
- In terms of productivity, when separated by pesticide type, insecticide TFI showed particular potential for being reduced, with no farms having a positive insecticide TFI effect, and 86% of farms having a negative TFI effect. The percentages for fungicide TFI was close to the effect of total TFI, while the herbicide TFI effect was mostly not significant (65%) or positive (29%), suggesting that decreases in herbicide use would be less feasible than decreases in insecticides or fungicides.
- However, in terms of profitability the specific herbicide, fungicide and insecticide TFI effects were largely non-significant (71, 66 and 76% respectively) or positive (21, 35 and 17% respectively), suggesting a slightly bigger negative impact of pesticide reduction on profitability than productivity.
The researchers then investigated the model further to establish, for the farms with negative or no significant TFI effects (i.e. farms in which reducing pesticide usage was predicted to be positive or neutral for productivity/profitability), how much pesticide use could be reduced by without any adverse effects on crop productivity/profitability. For the farms examined in this manner, the average potential pesticide reduction was calculated to be 42%.
The authors conclude that pesticide reductions with neutral or positive effects on productivity/profitability are theoretically possible, but note that their analysis does not examine the environmental effects of total pesticide reduction or reduction in specific pesticides., and that risk aversion of farmers and policy makers may hamper efforts to reduce pesticide use. They recommend that further and deeper study focusses on how reducing pesticide use may affect the agricultural landscape and markets.
Achieving sustainable crop production while feeding an increasing world population is one of the most ambitious challenges of this century. Meeting this challenge will necessarily imply a drastic reduction of adverse environmental effects arising from agricultural activities. The reduction of pesticide use is one of the critical drivers to preserve the environment and human health. Pesticide use could be reduced through the adoption of new production strategies; however, whether substantial reductions of pesticide use are possible without impacting crop productivity and profitability is debatable. Here, we demonstrated that low pesticide use rarely decreases productivity and profitability in arable farms. We analysed the potential conflicts between pesticide use and productivity or profitability with data from 946 non-organic arable commercial farms showing contrasting levels of pesticide use and covering a wide range of production situations in France. We failed to detect any conflict between low pesticide use and both high productivity and high profitability in 77% of the farms. We estimated that total pesticide use could be reduced by 42% without any negative effects on both productivity and profitability in 59% of farms from our national network. This corresponded to an average reduction of 37, 47 and 60% of herbicide, fungicide and insecticide use, respectively. The potential for reducing pesticide use appeared higher in farms with currently high pesticide use than in farms with low pesticide use. Our results demonstrate that pesticide reduction is already accessible to farmers in most production situations. This would imply profound changes in market organization and trade balance.
Lechenet, M. et al. (2017). Reducing pesticide use while preserving crop productivity and profitability on arable farms. Nature Plants, 17008, 1–6.
Read the full paper here.