Skip to main content

Analysis data for "The effect of agricultural land retirement on pesticide use"

Cite this dataset

Larsen, Ashley et al. (2024). Analysis data for "The effect of agricultural land retirement on pesticide use" [Dataset]. Dryad.


Agricultural land retirement generates risks and opportunities for ecological communities and ecosystem services. Of particular interest is the influence of retired cropland on agricultural pests and pesticides, as these uncultivated lands may directly shift the distribution of pesticide use and may serve as a source of pests and/or natural enemies for remaining active croplands. Few studies have investigated how agricultural pesticide use is impacted by land retirement. Here we couple field-level crop and pesticide data from over 200,000 field-year observations and 15 years of production in Kern County, CA, USA to investigate: 1) how much pesticide use and applied toxicity are avoided annually due to the direct effects of retirement, 2) whether surrounding retirement drives pesticide use on active cropland and what types of pesticides are most influenced, and 3) whether the effect of surrounding retirement on pesticide use is dependent on the age or revegetation cover on retired parcels. Our results suggest about 100 kha are idle in any given year, which equates to about 1.3-3 M kg of pesticide active ingredients foregone. We also find retired lands lead to a small increase in total pesticide use on nearby active lands even after controlling for a combination of crop-, farmer-, region- and year-specific heterogeneity. More specifically, the results suggest a 10 % increase in retired lands nearby results in about a 0.6 % increase in pesticides, with the effect sizes increasing as a function of the duration of continuous fallowing, but decreasing or even reversing sign at high levels of revegetation cover. Our results suggest increasingly prevalent agricultural land retirement can shift the distribution of pesticides based on what crops are retired and what active crops remain nearby.

README: Analysis data for "The effect of agricultural land retirement on pesticide use"

These data are used to understand the impact of land retirement on pesticides in Kern County, California. For details on the analysis, please see

Description of the data and file structure

Below is a basic dictionary to understand variable naming convention.

Data File 1 RetirementStats_postRR_Dryad.csv

45 columns/variables and 211,820 rows/observations

  • IHS indicates inverse-hyperbolic sine transformed
  • IHSKgHa[pesticide]Fill100 indicate different pesticide use variables that have been IHS transformed.
  • Fill indicates the variable includes 2021 data from Kern County Agricultural Commissioner's Office since CDPR data are not available (see text)
  • 100 indicates the variable was pre-multiplied by 100 prior to IHS transformation (see text)
  • KgHa[]AI indicates kg/ha of pesticide active ingredients ("AI") for different types of pesticides.
  • Hati_[] indicate toxicity to different taxa
  • IHSHaUncultYrFtT indicates IHS transformed annually uncultivated agriculture nearby (see text)
  • IHSHaUncultRS[] indicate IHS transformed variables of different remote sensing metrics of uncultivated land (quantiles of revegetation, RS[]FracQ1-Q4, where buff indicates the interior buffer and full indicated all fields; see text)
  • IHSHaUncultYrFtT[] indicates IHS transformed metrics of uncultivated agriculture of different age bins. Variables lacking "IHS" prefix indicate the variable prior to IHS transformation.
  • IHSHaPrmtOrig indicates IHS transformed hectares permitted (field size, based on permit). Variables lacking "IHS" prefix indicate the variable prior to IHS transformation.
  • Year indicates year
  • MTRfill indicates Public Land Survey Township
  • commcode indicates commodity code
  • permitNumEncode indicates the farmer ID (encoded)
  • permitsiteEncode is the unit of observation (encoded) for the statistical analyses

Data File 2 ForegonePU_post.csv

10 columns/variables and 17 rows/observations

  • ForegPestAIKgYrComPUR1k Foregone pesticide use (active ingredients) based on commodity-year specific averages, scaled by 1000.
  • ForegToxNetKgYrComPUR1T Foregone net toxicity based on commodity-year specific averages, scaled by 1Trillion. Mammal (Mam), Bird (Brd), Worm (Worm) and Bee (Bee) toxicity take the same form
  • HectareFallowedYr1k Annually uncultivated agriculture, scaled by 1000
  • HectareActive1x1k Actively cultivated cropland cultivated (counted once, see text), scaled by 1000
  • PUR indicates California Department of Pesticide Regulation Pesticide Use Reports (see text).

Sharing/Access information

Data was derived from the following sources:

  • Please text for original data sources and how they were used and processed.


Identifying and characterizing retired lands

Vector files representing crop field boundaries from 1997 to 2021 were downloaded from the Kern County Agricultural Commissioner's Office ( These geospatial data include information such as farmer (“permit”), area in production, and crop type (“commodity”, “commodity code”), from which we derived crop family. Using the set of permitted fields between 1997 and 2021, we identified fields that changed from production to lack of production or vice versa (see SI methods). Retirement was defined as either a parcel permitted as “uncultivated agriculture” with no other crop produced on that physical location during the year, or as a parcel that did not receive a permit in the focal year, but was cultivated at some point in the time series.

Pesticide use data

California mandates the collection of pesticide use data on production agriculture, which includes information such as date of application, product number, and amount of product used, among other data. The field-level, daily pesticide use data were sourced from the California Department of Pesticide Regulation (CDPR) Pesticide Use Reports when available (pre-2021) and the Kern County Agricultural Commissioner's Office for 2021. 

Pesticide applied toxicity

To quantify applied toxicity, we prioritized ecotoxicological observations available through the Pesticide Properties Database (PPDB). The database derives ecotoxicological endpoints from European and United States regulatory agencies where able, and where peer-reviewed literature, ecotoxicity prediction tools, and other resources are consulted, verified datasets receive the highest priority.

Foregone pesticide use and toxicity

From daily, field-level pesticide use data, we calculated mean annual pesticide use (kg ha−1 of active ingredients) by crop-year. We calculated foregone pesticide use based on the crop-year specific pesticide use rates using kg ha−1 average for the crop last produced on the retired fragment.


Data coverage and cloud screening: Revegetation of retired lands was evaluated using multispectral satellite imagery. All available Landsat 5, 7, and 8 images from WRS-2 Path 42, Rows 35 and 36 were downloaded as Collection 2, Level 2 surface reflectance from the USGS EarthExplorer web portal ( Photosynthetic vegetative cover on retired lands was then quantified using spectral mixture analysis (SMA) of Landsat imagery

See the journal article for more information on the methodology

Usage notes






National Science Foundation, Award: 2042526

National Institute of Food and Agriculture, Award: 2022-67019-36397


National Science Foundation, Award: 2226649

NASA, Award: 80NS SC22K0907