Long-term trends in the intensity and relative toxicity of herbicide use

Abstract

Herbicide use is among the most criticized aspects of modern farming, especially as it relates to genetically engineered (GE) crops. Many previous analyses have used flawed metrics to evaluate herbicide intensity and toxicity trends. Here, I show that herbicide use intensity increased over the last 25 years in maize, cotton, rice and wheat. Although GE crops have been previously implicated in increasing herbicide use, herbicide increases were more rapid in non-GE crops. Even as herbicide use increased, chronic toxicity associated with herbicide use decreased in two out of six crops, while acute toxicity decreased in four out of six crops. In the final year for which data were available (2014 or 2015), glyphosate accounted for 26% of maize, 43% of soybean and 45% of cotton herbicide applications. However, due to relatively low chronic toxicity, glyphosate contributed only 0.1, 0.3 and 3.5% of the chronic toxicity hazard in those crops, respectively.

Figure 1. Herbicide area-treatments for six crops in the United States 1990 to 2015.

Area-treatments are an estimate of the number of herbicide treatments applied to each field. Linear regression P value for soybean=0.271; all other crops P<0.001.

Figure 2. Toxicity of herbicides used between 1990 and 2015 in the United States.

Chronic toxicity (a) represented by 24-month oral rat no observable effect level (NOEL); acute toxicity (b) represented by oral rat LD₅₀. Toxicity is presented on a log₁₀ scale. For boxplots, bold vertical line represents median toxicity value, box encloses the inter-quartile range, and lines extend from minimum to maximum values. Each point represents a unique herbicide active ingredient, colour coded by WSSA site of action group number. Acute toxicity categories are those defined by US EPA.

Figure 3. Herbicide chronic hazard quotients for six crops 1990 to 2015.

Hazard quotient is the number of chronic rat (oral, 24 month) NOEL values applied per hectare. Mann-Kendall tests for monotonic trend: maize (tau=0.50, P=0.006); soybean (tau=−0.69, P<0.001); cotton (tau=0.60, P=0.001); rice (tau=−0.87, P=0.024); spring wheat (tau=0.02, P=0.964); winter wheat (tau=0.10, P=0.620).

Figure 4. Herbicide acute hazard quotients for six crops 1990 to 2015.

Hazard quotient is the number of acute rat (oral) LD₅₀ values applied per hectare. Mann-Kendall tests for monotonic trend: maize (tau=−0.93, P<0.001); soybean (tau=−0.71, P<0.001); cotton (tau=−0.52, P=0.006); rice (tau=−0.73, P=0.060); spring wheat (tau=0.13, P=0.500); winter wheat (tau=0.53, P=0.005).