The USGS had some great data on pesticide usage in the United States that causes great envy among Australian pest researchers. In this post we have a look at some of this data and present some spatial and temporal trends in an important pesticide group that has been the subject of much controversy in recent history.
Such data allows researchers to better estimate selection pressure for resistance evolution, as well as other off-target effects of pesticide usage.
The USGS provides a low and high estimate of pesticide usage as per the methodology described here. For simplicity we use the lower estimate in the following.
First, we load the county level pesticide usage data and a shape file that allows matching of estimated neonicotinoid usage in space.
#Import libraries library(tidyverse) library(sf) library(tmap) # Common neonicotinoid compounds neonics = toupper( c("acetamiprid", "clothianidin", "dinotefuran", "imidacloprid", "thiamethoxam")) # US pesticide usage us = "data/USA land usage/US_pesticide_usage/EPest.county.estimates.2016.txt" %>% read_delim(delim = "\t") %>% mutate(atlas_stco = paste0(STATE_FIPS_CODE, COUNTY_FIPS_CODE)) %>% mutate(usage_kg = EPEST_LOW_KG) %>% filter(COMPOUND %in% neonics) %>% group_by(atlas_stco) %>% summarise(usage_kg = sum(usage_kg, na.rm=T)) # US shape file shp = st_read("data/USA land usage/US_counties/CoUS17_WGS84WMAS.shp")
## Reading layer `CoUS17_WGS84WMAS' from data source `C:\Users\james\Documents\git\jmaino_website\content\post\data\USA land usage\US_counties\CoUS17_WGS84WMAS.shp' using driver `ESRI Shapefile' ## Simple feature collection with 3070 features and 8 fields ## geometry type: MULTIPOLYGON ## dimension: XY ## bbox: xmin: -13884420 ymin: 2820030 xmax: -7452828 ymax: 6340332 ## projected CRS: WGS 84 / Pseudo-Mercator
# join usage to shape file for chem and scale for usage intensity uspest = shp %>% mutate(atlas_ha = atlas_acre * 0.404686) %>% # metric FTW left_join(us) %>% mutate(usage_g_ha = 1e3*usage_kg/atlas_ha)
Then after scaling neonicotinoid usage by county area to estimate usage intensity, we can plot this data on a linear scale, and then a log scale to better see variation
# range of values logrange = uspest %>% filter(! (is.na(usage_g_ha)|usage_g_ha==0)) %>% pull(usage_g_ha) %>% log10 %>% range %>% round # plot usage intensity pal = rev(viridis::magma(10)) mytmtheme = tm_layout(legend.position = c("right", "bottom"), bg.color = "#37314a", scale = 0.5,outer.bg.color = "#37314a", frame = F) # linear tm_shape(uspest) + tm_fill(col="usage_g_ha", title = "County level 2016\nneonicotinoid usage (g/ha)", style = "cont", palette = pal) + mytmtheme
#log tm_shape(uspest) + tm_fill(col="usage_g_ha", title = "County level 2016\nneonicotinoid usage (g/ha)", style = "cont", breaks = 10^(logrange:logrange), palette = pal) + mytmtheme
The USGS provides historical data through time at a crop level that allows us to look at trends in usage the crop level.
Here we can see that neonicotinoid usage increased rapidly but dropped off precipitously after 2014. Soybean and corn production account for most neonicotinoid usage in the United States.
# get usage by crop type through time crop_usage = "data/USA land usage/US_pesticide_usage/LowEstimate_AgPestUsebyCropGroup92to16.txt" %>% read_delim(delim = "\t") %>% filter(Compound %in% neonics) %>% gather(crop, usage, -State_FIPS_code, -State, -Compound, -Year, -Units) %>% group_by(crop, Year) %>% summarise(usage_t = sum(usage, na.rm=T)/1e3, .groups="drop") # plot usage by crop type through time crop_usage %>% mutate(crop = gsub("_", " ", crop)) %>% ggplot(aes(Year, usage_t, color=reorder(crop, desc(usage_t)))) + geom_line(stat="identity", size=1) + # theme_bw() + mydarktheme + theme(panel.background = element_blank()) + scale_color_manual(values=rainbow(10, s = 0.5)) + ylab("Neonicotinoid usage (tonnes)") + guides(color = guide_legend(""))
Total neonicotinoid usage by year is summarised in the following table.
# total usage crop_usage %>% group_by(Year) %>% summarise(`Neonicotinoid Usage (tonnes)` = sum(usage_t, na.rm=T), .groups="drop") %>% arrange(desc(Year)) %>% knitr::kable()
|Year||Neonicotinoid Usage (tonnes)|