README: Analysis data for: "Spillover effects of organic agriculture on pesticide use on surrounding fields"

https://doi.org/10.5061/dryad.vhhmgqp1f

These datasets include information on cropland and pesticide use from Kern County, CA and US wide. For Kern County, information on cropland includes characteristics of the focal crop field (e.g. crop type, field size, organic status) as well as information about the landscape surrounding each focal crop field. For the national-scale analysis, information on pesticide use and crop characteristics are provided at the county level.

Description of the data and file structure

There are 6 data files:

1.     SpatialOrganic_MainText_R2.csv: This file includes the data required to reproduce figures 2-3 in the main text. "" indicates missing data.

2.     SpatialOrganic_SI_R2.csv: This file includes the data required to reproduce analyses in the supplementary information, except SI figure 6 (removing overlapping fields), SI figure 9 (event study), and SI table 11-12 (national scale analysis). "" indicates missing data.

3.     SpatialOrganic_SI_NoOverlap_R2.csv: This file includes the data required to reproduce SI figure 6 (removing overlapping fields). "" indicates missing data.

4.     event_study_replication_data.csv: This file includes all data to replicate the event study. NA indicates missing data.

5.     national_census.csv: This file include all the data to replicate the national level analysis using pesticide use from the USDA Census of Agriculture. NA indicates missing data.

6.     national_usgs.csv: This file include all the data to replicate the national level analysis using pesticide use from the USGS National Pesticide Synthesis Project. NA indicates missing data.

Kern County Files (#1-3 above)

The outcome and righthand side variables inverse hyperbolic sine transformed, and include the prefix "IHS". Variables with small average values are pre-multiplied by 100 (see text), and are given the suffix "100". All outcome variables take the form IHSKgHaAI[pesticide type]100 where KgHa indicates kg/ha, AI indicates "active ingredients" and "pesticide type" is one of: "Pest" (all pesticides), "Insecticide" (all insecticides"), "InsectOnly" (pesticides functioning only as insecticides), "HerbOnly" (pesticides functioning only as herbicides), "FungOnly" (pesticides functioning only as fungicides), and "InsectFung" (pesticides functioning as dual action insecticides / fungicides).

The main specification for the Kern County analysis takes the form,

reghdfe IHSKgHa[pesticide type]100 IHSTotOrgHaNearTemp_3yO IHSAgHaNearTemp IHSHaPrmtMulti IHSAgShareOwn2500 i.cdfa_3ypur_org, absorb (year MTRfill commcode permitNum) vce(cluster permitNum).

reghdfe is the STATA command, which can be installed using "ssc install reghdfe".

IHSKgHa[pesticide type]100: see above.

IHSTotOrgHaNearTemp_3yO: Covariate of interest, the IHS transformed amount of organic agriculture within 2.5km of the focal field (see text).

IHSAgHaNearTemp: The IHS transformed amount of agriculture within 2.5km of the focal field (see text).

IHSHaPrmtMulti: The IHS transformed field size, adjusted for multi-cropping (see text).

IHSAgShareOwn2500: The fraction of agriculture within the 2.5km buffer that is owned by the focal farmer (see text).

cdfa_3ypur_org: Binary variable for whether or not the focal field is organic. For models that are run separately for organic and conventional focal fields, this binary variable is dropped. cdfa_3ypur_org = 1 is an organic field, cdfa_3ypur_org = 0 is a conventional field.

absorb(year MTRfill commcode permitNum): These are dummy variables or "fixed effects" for year, region (Public Land Survey Township, "MTRfill"), commodity ("commcode") and farmer ID code ("permitNum"). "absorb" is associated with the reghdfe command.

vce(cluster permitNum): cluster robust standard errors were used, clustered at the farmer ID code (permitNum) level.

Other specifications used the amount of organic fields in distance annuli from the focal field. These variables are coded as IHSHaOrgNearD[distance]_3yO, where distance is 0-500m ("500"), 500-1000m ("1000"), 1000-1500 ("1500"), 1500-2000 ("2000") and 2000-2500 ("2500"). In the SI, these analyses are extended out to 5000m, following the same naming conventional.

Other specifications (see text):

IHSTotOrgHaNearTemp_1y: 1 year organic definition

IHSCropDiv100: Crop diversity covariate

IHSHaFam[]: Area in different crop taxanomic families

IHSAgShareHighAI_2500: Area in crops that are high pesticide use crops

IHSHaOrgFrag, IHSHaAgFrag, IHSAgShareOwnFrag: metrics for IHSTotOrgHaNearTemp_3yO, IHSAgHaNearTemp, and IHSAgShareOwn2500, respectively, but calculated based on rasterized landscapes. The "Poly" suffix, further indicates that the raster calculation was based on field polygon edge rather than field polygon centroid.

Multi-cropped fields are indicated by the binary variable "Multi"

IHSKgHaPrdPest100, IHSTreatedHaPrmt100, IHSHati_net: other measures of net pesticide use based on pesticide products, area treated, and net applied toxicity (see text).

See code and figure legends for additional, minor specifications used exclusively in the SI.

Definition of all variables, as requested by Dryad.

*Note: *SpatialOrganic_MainText_R2 is a subset of variables in SpatialOrganic_SI_R2 to facilitate replication of tables and figures, as presented in the main text and SI of the published paper. SpatialOrganic_SI_NoOverlap_R2 is a subset of the sample from SpatialOrganic_SI_R2, specifically a subset that keeps only one observation of overlapping fields within the same calendar year.

AgHaNearTemp: Hectares of agriculture within 2.5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field. For summary statistics.

AgShareOwn2500: Share of the surrounding area (2.5km from focal field centroid), minus the permitted field area, that was managed by the focal farmer. HaOwnTot/(1963.49-HaPrmtMulti) * 100; see definitions below. Can exceed 100 due to crop rotations. For summary statistics.

CropDiv: Diversity of crop types within the 2.5km buffer of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field. Calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti). For summary statistics.  

DecReg: Binary variable for whether or not the field had a date active in December. Robustness test.

HaOwnTot: Hectares of agriculture within 2.5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field, and had the same farmer ID as the focal field. Calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti). For summary statistics.

HaPrmtMulti: Permitted field size (ha) divided by the number of contemporaneously grown crops on the same field geometry (ie shared farmer, year, field geometry, date active and production status). For summary statistics.

HaPrmtOrig: Permitted field size (ha). For summary statistics.

HaSameComTot: Hectares of the same commodity within 2.5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field. Calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti). Robustness test.

IHSAgHaNearTemp: IHS transformed hectares of agriculture within 2.5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field. Calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti). Main specification.

IHSAgHaNearTemp25_5k: IHS transformed hectares of agriculture between 2.5-5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field. Calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti). Robustness test.  

IHSAgHaNearTempYr: IHS transformed hectares of agriculture within 2.5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field. Calculation accounts for multi-cropped surrounding fields, as in HaPrmtMulti, but without the requirement for shared date active. Robustness test.

IHSAgShareHighAI_2500: IHS transformed share of surrounding cropland within 2.5km of the focal field centroid, that overlapped production based on date active (and inactive, as relevant) of the focal field, with pesticide use rates greater than or equal or that of cotton. Calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti). Robustness test.

IHSAgShareOwn25_5k: IHS transformed share of cropland between 2.5-5km from the focal centroid that was managed by the focal farmer. Area based on HaPrmtMulti. Robustness test.

IHSAgShareOwnFrag: IHS transformed share of cropland 0-2.5km from the focal field (and excluding the focal field) that was managed by the focal farmer. Based on rasterized fields (30m pixels) within a 2500m buffer of the focal field centroid. Robustness test.

IHSAgShareOwnFragPoly: IHS transformed share of cropland 0-2.5km from the focal field (and excluding the focal field) that was managed by the focal farmer. Based on rasterized fields (30m pixels) within a 2000m buffer of the focal field boundary. Robustness test.

IHSAgShareOwnYr2500: Same as IHSAgShareOwn2500, except calculation for “multi-cropped” surrounding fields drops the requirement for shared date active (ie rotations ignored). Robustness test.

IHSAgShareOwn2500: IHS transformed share of cropland between 0-2.5km from the focal that was managed by the focal farmer. Area based on HaPrmtMulti and a buffer of 2.5km from the centroid of the focal field. Main specification.

IHSCropDiv100: IHS transformed crop diversity, pre-multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Robustness test.

IHSEdgeLengthOrg: IHS transformed length (m) of shared edge with organic cropland, based on rasterized crop fields. Robustness test.  

IHSHaAgFrag: IHS transformed hectares of cropland, based on rasterized crop fields and a 2.5km buffer of the focal field centroid. Robustness test.

IHSHaAgFragPoly: IHS transformed hectares of cropland, based on rasterized crop fields and a 2km buffer of the focal field polygon boundary. Robustness test.  

IHSHaFam1… IHSHaFam31: IHS transformed hectares of agriculture in different crop taxanomic families. Robustness test.

IHSHaOrgFrag: IHS transformed hectares of organic cropland, based on rasterized crop fields and a 2.5km buffer of the focal field centroid. Robustness test.  

IHSHaOrgFragPoly: IHS transformed hectares of organic cropland, based on rasterized crop fields and a 2km buffer of the focal field polygon boundary. Robustness test. 

IHSHaOrgNearD500_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 0-500m annuli from the focal field. For all IHSHaOrgNearDxxxx_3YO variables, calculation accounts for multi-cropped surrounding fields (see HaPrmtMulti) and is based upon overlapping production dates. Main specification.

IHSHaOrgNearD1000_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 500-1000m annuli from the focal field. Main specification.

IHSHaOrgNearD1500_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 1000-1500m annuli from the focal field. Main specification.

IHSHaOrgNearD2000_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 1500-2000m annuli from the focal field. Main specification.

IHSHaOrgNearD2500_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 2000-2500m annuli from the focal field. Main specification.

IHSHaOrgNearD3000_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 2500-3000m annuli from the focal field. Robustness test.

IHSHaOrgNearD3500_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 3000-3500m annuli from the focal field. Robustness test.

IHSHaOrgNearD4000_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 3500-4000m annuli from the focal field. Robustness test.

IHSHaOrgNearD4500_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 4000-4500m annuli from the focal field. Robustness test.  

IHSHaOrgNearD5000_3yO: IHS transformed hectares of organic cropland, using 3 year definition (see below, see text) in the 4500-5000m annuli from the focal field. Robustness test.

IHSHaOrgNearDYr500_3yO: Equivalent calculation to IHSHaOrgNearDYr500_3yO, except the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.   

IHSHaOrgNearDYr1000_3yO: Equivalent calculation to IHSHaOrgNearDYr1000_3yO, except the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.     

IHSHaOrgNearDYr1500_3yO: Equivalent calculation to IHSHaOrgNearDYr1500_3yO, except the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.    

IHSHaOrgNearDYr2000_3yO: : Equivalent calculation to IHSHaOrgNearDYr2000_3yO, except the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.

IHSHaOrgNearDYr2500_3yO: : Equivalent calculation to IHSHaOrgNearDYr2500_3yO, except the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.

IHSHaOwnTotYr: IHS transformed ha of cropland between 0-2.5km from the focal that was managed by the focal farmer. Area based on definition of “multi-cropped” drops the requirement for shared date active and is based upon overlapping production dates. Robustness test.

IHSHaPrmtMulti: IHS transformed HaPrmtMulti. Main specification.

IHSHaPrmtMultiYr: Equivalent calculation to IHSHaPrmtMulti, except definition of “multi-cropped” drops the requirement for shared date active (see HaPrmtMulti). Robustness test.

IHSHaPrmtOrig: IHS transformed HaPrmtOrig. Robustness test.

IHSHati_net: IHS transformed net applied toxicity based on Parker et al. 2023 (see text). Robustness test.

IHSKgHaAIFungOnly100: IHS transformed kg/ha of active ingredients only functioning as fungicide. All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Main specification.

IHSKgHaAIHerbOnly100: IHS transformed kg/ha of active ingredients only functioning as herbicide. All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Main specification.

IHSKgHaAIInsectFung100: IHS transformed kg/ha of active ingredients with dual action insecticide/fungicide properties. All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Main specification.

IHSKgHaAIInsectOnly100: IHS transformed kg/ha of active ingredients only functioning as insecticide. All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Main specification.

IHSKgHaAIInsecticide100: IHS transformed kg/ha of active ingredients functioning as insecticide. All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Main specification.

IHSKgHaAIInsecticide100Yr: Equivalent calculation to IHSKgHaAIInsecticide100, except field area is based on the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.

IHSKgHaAIPest100: IHS transformed kg/ha of all pesticide active ingredients. All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Main specification.

IHSKgHaAIPest100Yr: Equivalent calculation to IHSKgHaAIInsecticide100, except field area is based on the definition of “multi-cropped” drops the requirement for shared date active. Robustness test.

IHSKgHaPrdPest100: IHS transformed kg/ha of all pesticide products (active ingredients + adjuvants). All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text). Robustness test.

IHSTotOrgHaNearTemp_3yO: IHS transformed amount of organic agriculture (ha), defined as organic based on CDFA data and 3 years of no conventional pesticide use, within 2.5km of the focal field. Main specification.

IHSTotOrgHaNearTemp_1yO: IHS transformed amount of organic agriculture, defined as organic based on CDFA data and 1 year of no conventional pesticide use, within 2.5km of the focal field. Robustness test.

IHSTotOrgHaNearTempYr_3yO: IHS transformed amount of organic agriculture, defined as organic based on CDFA data and 3 year of no conventional pesticide use, within 2.5km of the focal field. Fields were only counted once in this specification (ie no crop rotations). Robustness test.

IHSTreatedHaPrmt100: IHS transformed area treated rate (total treated ha/HaPrmtMulti * 100) . All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text).

IHSorgarea_2ylag: IHS transformed 2-year lag of ha of organic agriculture, based on rasterized fields (same calculation as IHSHaOrgFrag, excepted lagged 2 years). Robustness test.

KgHaAIFungOnly: kg/ha of active ingredients only functioning as fungicide. Robustness test.

KgHaAIHerbOnly: kg/ha of active ingredients only functioning as herbicide. Robustness test.

KgHaAIInsectFung: kg/ha of active ingredients dual functioning as insecticides and fungicides. Robustness test.

KgHaAIInsectOnly: kg/ha of active ingredients only functioning as insecticide. Robustness test.

KgHaAIInsecticide: kg/ha of active ingredients functioning as insecticide. Robustness test.

KgHaAIPest: kg/ha of all pesticide active ingredients. Robustness test.

KgHaPrdPest: kg/ha of all pesticide products (active ingredients + adjuvants). Robustness test.

MTRfill: ID for public land survey Township. Township based on pesticide use reports for fields with non-zero pesticides, and based on the field permits otherwise. Main specification.

Multi: Binary variable for multi-cropped fields, based on field permit, irrespective of whether other crops were reported contemporaneously. Robustness test.  

TotOrgHaNearTemp3yO:  Amount of organic agriculture (ha), defined as organic based on CDFA data and 3 years of no conventional pesticide use, within 2.5km of the focal field and that overlapped production based on date active (and inactive, as relevant) of the focal field.

TreatedHaPrmt100: Area treated rate (total treated ha/HaPrmtMulti * 100). All pesticide use rates were multiplied by 100 prior to IHS transformation to avoid small numbers (see text).

annual: Binary variable indicating annual crop or not.

cdfa_1ypur_org: Binary variable for whether or not the focal field is organic, based on definition of 1 year of no conventional pesticide use and within a polygon (Public Land Survey Section, APN or permitsite) with reported organic fields. Robustness test.

cdfa_3ypur_org: Binary variable for whether or not the focal field is organic, based on definition of 3 years of no conventional pesticide use and within a polygon (Public Land Survey Section, APN or permitsite) with reported organic fields. Main specification.

North: Binary variable for whether or not the focal field is greater than or equal to the median latitude of organic fields.

commcode: commodity ID code

constant: Arbirtrary constant required for specific code to run in a loop.

permitNum: farmer ID code.

Year: year of focal field permit.

Event Study Files (#4 above)

The event study files are a rasterized version of the “Kern County Files (#1-3 above)”. The definitions are the same as above, but the name may vary. Here, we mainly provide the link to the definitions in the previous section. The data files contain missing data. Missing data are indicated by “NA”.

pixel_id: ID of raster/pixel

year: Calendar year

farmer: Same as permitNum (the definition is provided above)

crop: Same as commcode (the definition is provided above)

township: Same as MTRfill (the definition is provided above)

pesticides_ha_ai: Active ingredients of pesticides per hectare. This is the same as KgHaAIPest (the definition is provided above).

insecticide_ha_ai: Active ingredients of insecticides per hectare. This is the same as KgHaAIInsecticide (the definition is provided above).

field_size: Same as HaPrmtOrig (the definition is provided above).

share_own: Same as AgShareOwn2500 (the definition is provided above).

ag_ha: Same as AgHaNearTemp (the definition is provided above).

org_ha: Same as TotOrgHaNearTemp_3yO (the definition is provided above)

or_year: Calendar year the pixel becomes organic.

or_time: Year relative to “or_year” i.e or_time := year - or_year

or_treated: Binary time-constant variable that indicates whether pixels become organic at one point in time (1) or stay conventional agriculture throughout (0)

or_absorbing: Binary time-constant variable that indicates whether pixels stay organic after becoming organic (1) or switch back to conventional (0)

organic: Binary time-varying variable that indicates whether a field is organic (1) or not (0)       

 bo_treated: Binary time-constant variable that indicates whether neighboring fields become organic at one point (1) or all neighboring fields remain conventional (0).

bo_year: Year in which neighboring field becomes organic.

bo_time: Year relative to “bo_year” i.e. bo_time := year -** **bo_year.

National Files (#5, #6 above)

The data files contain missing data. Missing data are indicated by “NA”.

County: County

Year: Year

Insecticides: Acres treated with insecticides

Herbicide: Acres treated with herbicides

Oganic: Acres of organic agriculture

Cropland: Acres of cropland

Corn: Acres of corn

Cotton: Acres of cotton

Soy: Acres of soy

Potatoes: Acres of potatoes

Hay: Acres of hay

GLYPHOSATE: Quantity of glyphosate from the USGS National Pesticide Synthesis Project

ATRAZINE: Quantity of atrazine from the USGS National Pesticide Synthesis Project

CHLORPYRIFOS: Quantity of chlorpyrifos from the USGS National Pesticide Synthesis Project

ALDICARB: Quantity of aldicarb from the USGS National Pesticide Synthesis Project

Sharing/Access information

Data was derived from the following sources:

·       California Pesticide Use Reports

·       California Product Database

·       Kern County Agricultural Commissioner's Spatial Data

·       USDA Census of Agriculture

·       USGS National Pesticide Synthesis Project

Code/Software

Data analyses were conducted in STATA 16 MP and R.

R2MainTextPost.do is the STATA script to conduct the Kern County analyses (Table 1, Figures 2-3) in the main text. It relies on SpatialOrganic_MainText_R2.csv. For csv files associated with STATA scripts, "" indicates missing data.

R2SupplementaryInfoPost.do is the STATA script to conduct the SI analyses besides the event study and national scale analyses. It relies on SpatialOrganic_SI_R2.csv and SpatialOrganic_SI_NoOverlap_R2.csv. For csv files associated with STATA scripts, "" indicates missing data.

“Event” is the R script to run the event study analyses. For csv files associated with R scripts, NA indicates missing data.

“National” is the R script to run the national scale analyses (Including figure 5). For csv files associated with R scripts, NA indicates missing data.

“Simulation” is the R script to run the simulation (Including Figure 4). For csv files associated with R scripts, NA indicates missing data.