One State Counts: Pesticide Use Reporting and the Sector Nobody Measures

1. Introduction: the number that exists once

This journal has repeatedly found that a figure everybody cites has one origin. This article is about a case where that is not a failure but the actual situation.

1.1 Made in a published book chapter

And echoed by the regulator, which describes its programme as recognised as the most comprehensive in the world.³⁵

1.1b And the contrast with our efficacy article is direct

That article found a regulator permitted to register a product without requiring evidence it works. This one finds the same regulatory world collecting extraordinarily detailed records of where that product then goes. What is measured and what is tested do not line up.

1.2 What this article argues

That one jurisdiction's returns carry the empirical weight for a global question, that the sector most likely to involve untrained application is exempt from them, and that the system's own analyst puts its coverage at eighty to ninety per cent. Sections 15, 18 and 24 are the case.

2. What it is

A mandatory reporting system.

All pesticide applications in production agriculture must be reported to the appropriate county agricultural commissioner's office, and all pesticide applications made by commercial pest control businesses must be reported.¹

2.0b Two layers of checking rather than one

Validation happens at the county before the data leaves and again at the state after it arrives, which is what produces the error rate in §21.⁴

2.1 County first, then the state

Reports are entered into a county database and checked for errors, after which the commissioner sends required data to the state department, where additional validation takes place.⁴

2.2 And the system is now filed online

Fully web-based, with reports submitted through online forms, which the analyst presents alongside better validation before data arrives, faster entry and improved location accuracy.¹

2.3 Which is why the error checking can happen early

Validation at the point of entry catches what would otherwise be corrected months later or not at all.¹

3. How big

The volume is part of the claim.

The department collects around three million pesticide use records a year, and the database contains more than eighty million records going back to 1990.⁴

3.0b And the two figures in our sources differ

Over 2.5 million records a year in one account and around three million in another, which is not a contradiction if they describe different periods and which we note because this journal counts such things.¹⁴

3.1 With earlier records held apart

Records from 1974 to 1989 are kept in a separate database because those early records vary in the type of information they contain.⁴

3.2 And the reporting is monthly

Which sets the resolution of the whole series: a month is the finest period any use figure drawn from it can describe, whatever the individual records contain.⁵

4. And it is a census

Which is the property that matters.

The programme's analyst lists among its strengths that data are obtained from a census not just samples, and that the data are geographically referenced.¹

4.0b And the alternative everywhere else is sales

Quantity sold tells you what entered a market, not what was applied, where, when or to what. The two questions are related and not the same, and only one of them is what an exposure assessment needs.

That distinction is ours and none of our sources draws it.

4.1 A census answers questions a survey cannot

Our articles on detection probability and on trap catch both turned on what a sample can and cannot support. A complete enumeration removes that whole class of problem.

4.1b And the geographic reference is what multiplies its value

A record of what was applied is useful. A record of what was applied at a located place can be overlaid on water bodies, habitat, schools and residential areas, which is what the uses in §23 require.¹

4.2 Subject to §18

Where the completeness is estimated.

5. Where it came from

Limited reporting was first required by 1950, with the database containing records from 1974, and the current full use reporting system starting in 1990.¹

5.1 Which makes the series older than it looks

Seventy-five years of some kind of reporting, with a usable continuous record running back half a century.¹

5.2 And the length is what makes trends visible

A use dataset that began last year would say what is applied now. One that began in 1990 says what changed, which is a different and more useful thing.

6. Three systems under one name

Set out in a published history.

The system has existed in three distinct forms: aerial pesticide applications from 1934 to 1956, primarily restricted pesticides from 1970 to 1989, and full agricultural use reporting from 1990 to current.²

6.0b And the 1956 to 1970 gap is unexplained

Between the end of aerial reporting and the start of restricted product reporting, the published account we read names no scheme, and we do not know whether none existed or whether the summary simply omits it.²

6.1 Each covering what was then the concern

Aerial application when that was the visible risk, restricted products when hazard classification was the frame, and everything once the question became exposure.

That reading is ours, though the same chapter says the history shows how changes related to changing attitudes towards pesticides.²

7. The statute

Named.

The current database was established in 1990 by a food safety act passed the previous year.²

7.0b The vehicle tells you who had to be persuaded

A reporting system arriving inside a food safety statute is a system that had to be justified to people worried about residues on produce, whatever anybody later wanted the data for.²

7.1 Which locates the motivation

Food safety rather than applicator safety or environmental protection, whatever the data has since been used for.²

7.2 And the regulator gives a different framing

That the state became the first to require full reporting in response to demands for more realistic and comprehensive pesticide use data, which is a statement about data rather than about food.⁵

8. What agricultural means there

A definitional point with large consequences.

The state has a broad legal definition of agricultural use, so the reporting requirements include pesticide applications to parks, golf courses, cemeteries, rangeland, pastures, and along roadside and railroad rights-of-way.⁵

8.0b The definition is doing regulatory work

Calling a golf course an agricultural site is not a biological statement. It is a way of pulling a large category of application into a reporting requirement that was written for farms.

That reading is ours.

8.1 Plus several things further from a field

All postharvest treatments of agricultural commodities, all treatments in poultry and fish production, and some livestock applications.⁵

9. Which is broader than it sounds

And the database uses its own categories.

In the database, agricultural uses are applications for producing an agricultural commodity, and non-agricultural uses are all other legally reportable uses, such as rights-of-way, landscape maintenance, golf courses, parks, recreation areas, cemeteries, structural pest control, and vector control.³

9.0b Which means the headline category names mislead

A figure reported as agricultural use includes things done on a golf course, and a figure reported as non-agricultural includes our own trade. Anyone reading a summary table without the definitions will get both wrong.³⁵

9.0c And vector control sits in the same list

Which puts mosquito and tick programmes, structural work and golf course maintenance in one reporting category defined by not being crop production.³

9.1 With the regulator noting the two definitions differ

The database defines agricultural and non-agricultural uses slightly differently from how the governing code defines them, which is the kind of detail our articles on statistic provenance keep finding underneath a clean figure.³

10. Our own trade reports into it

Which is why this article sits in this journal.

Structural pest control appears explicitly in the list of reportable non-agricultural uses.³

10.1 Alongside vector control

Which places public health mosquito and tick work in the same reportable category, and which means the programmes our black fly and West Nile articles described leave a record of exactly what went out.³

10.2 And landscape maintenance too

So the lawn treatments at the centre of our pollinator exposure article are counted when done commercially and not when done by the householder.³⁵

11. What a structural operator files

One account: structural pest control companies must submit a monthly summary report to the County Agricultural Commissioner detailing every application made in the previous month, including the exact product name, the federal registration number, the total quantity used down to the decimal point, and the number of applications performed.⁶

11.1 And the filing is routine rather than exceptional

Monthly, covering the previous month, as an ordinary condition of operating rather than something triggered by an incident or a complaint.⁶

12. Which is a remarkable level of detail

Our assessment, as a business that would have to file it.

Product rather than active ingredient, registration number rather than trade description, quantity to the decimal rather than an estimate. Nothing in the returns we are familiar with approaches it.

12.0b The contrast with how products are usually discussed

Our chemistry articles report on active ingredients, because that is how the literature is organised. A registration number is a narrower object: one formulation, one concentration, one set of approved uses.⁶

12.1 And it is what makes the data usable

A registration number identifies one formulation, which means the record can be linked to a label, a rate and a set of restrictions. An active ingredient name cannot.

12.1b And location is recorded to a survey grid

County, section, township, range, base and meridian, which places an application within a parcel rather than within a region and is what allows the analyses in §23 to be done spatially.²

12.2 The agricultural fields are equally detailed

Including date, hour the treatment was completed, operator and site identification numbers, location by county, section, township, range, base and meridian, acreage planted and treated, and the names of the persons who made and supervised the application.²

13. One source disagrees about scope

Which we flag rather than resolve.

The same account that describes the monthly return says the state requires full reporting of all agricultural use as well as the use of restricted materials by structural pest control operators, which is narrower than every application by a commercial business.⁶

13.0b The difference is not small

Restricted materials are a subset defined by hazard. Every application by a commercial business is everything, including products a consumer could buy. Which of those two describes our trade's obligation changes what the database contains about us.

13.1 And the other sources say commercial applications generally

All applications made by commercial pest control businesses must be reported.¹

13.2 The source that narrows it is a law firm page

Which we would ordinarily trust on a legal question and which contradicts a regulator's own analyst, so we record the conflict and rely on the analyst for §2.¹⁶

14. The exemptions

The primary exceptions to the reporting requirements are home-and-garden use and most industrial and institutional uses.⁵

14.1 Primary suggests there are others

The wording names the main exceptions rather than a complete list, and we did not find the complete list.⁵

15. Home and garden first

Named first among the exceptions.⁵

So the most comprehensive record of pesticide use in the world does not include what people put on their own lawns, in their own kitchens, or under their own sinks.

15.1 Which is a defensible rule to write

Requiring a household to file a monthly return for a can of wasp spray would be unenforceable and absurd, so the exemption is not an oversight. The consequence is nonetheless that the sector has no data.

Both halves of that are ours.

16. Why that matters most

Our argument, and the centre of the article.

Every other sector in the system involves a licensed person applying a product they were trained on, under a label they can be enforced against, with a record they must file. The exempt sector involves untrained people applying products bought in a supermarket in the buildings they sleep in.

16.0b And the asymmetry runs the wrong way throughout

The more training a person has, the more thoroughly their applications are recorded. The relationship should run the other way if the purpose is to find out where the risk is.

16.1 Which is where the exposure is closest

Our applicator exposure article concerned people with equipment and training. This concerns residential interiors with neither.

16.1b And the products are not trivial ones

The consumer shelf holds pyrethroids, neonicotinoids, borates and the fumigant our last consumer article examined, which are the same chemistries our other articles discuss in a professional context.

16.2 And the number of applications is presumably enormous

We say presumably because the exemption means nobody knows.

17. And it compounds a problem we found earlier

In this journal's most recent consumer article.

That article found a registered pesticide whose label restricts it to an airtight container being used across attics, crawlspaces and gardens against a dozen vertebrates, illegally, with no prevalence figure available for how often.

17.0b And the same gap sits under several of our articles

Our work on consumer products, on minimum risk exemption products and on retail insecticides each reached a point where the question was how much of this is actually being used, and each stopped there.

17.1 The exemption in §14 is why there is no figure

A sector exempt from reporting produces no denominator, so misuse within it cannot be expressed as a rate.

17.1b And the same reasoning applies to resistance

Our resistance articles describe selection pressure building from repeated exposure. Exposure occurring in the exempt sector counts toward that pressure and appears in no record of it.

17.2 Which is a structural rather than an accidental gap

Both are ours.

18. The coverage estimate

Among the data quality statements in the programme analyst's own presentation: probably 80 to 90% of actual use is reported.¹

18.1 Presented as a data quality statement

Sitting in the same section of the presentation as the error rate and the census property, which is to say offered as a characteristic of the system rather than as a caveat buried elsewhere.¹

19. Eighty to ninety per cent

Of the best system in the world.

Which means the most complete pesticide use data anybody has is short by somewhere between a tenth and a fifth, by the estimate of the person presenting it.¹

19.0b And the missing portion is not random

Whatever causes an application to go unreported, it is unlikely to be evenly distributed across products, operators and regions, which means the gap is a bias rather than noise.

That is our inference and the source offers no view on it.¹

19.1 Probably is doing work in that sentence

The estimate is itself an estimate, with no method given in what we read.¹

20. Which is the honest number and a troubling one

Both at once, and we want to give the first its due.

A programme that publishes an estimate of its own incompleteness is behaving well. Most of the figures this journal has traced came with no such statement attached.

20.0b The comparison this journal usually makes runs the other way

We have criticised figures presented with more confidence than their basis supports. This is a figure presented with less confidence than its collection would justify, in a field where that is rare.

20.1 And the trouble is what sits underneath it

If the world's most complete census is missing a tenth to a fifth, then every jurisdiction without such a system is missing an amount nobody has characterised.

20.1b And the comparison cannot be made

Nobody can say whether a place without reporting captures nothing or captures a great deal informally, because the question of coverage only arises where there is something to have coverage of.

20.2 Which is not a criticism of the programme

It is a statement about everywhere else.

21. The error rate is separate

And much better.

The reports are screened for about forty kinds of errors, with an error rate of less than half of one per cent.¹

21.0a Forty kinds of error is a specific claim

It implies an enumerated list of failure modes, each with a check written for it, which is a more developed quality regime than a single plausibility screen.¹

21.0b And only one of them can be measured internally

A system can check its own records against each other and find errors. It cannot find an application that was never reported, which is why the coverage figure in §18 is an estimate and the error rate is a count.¹

21.1 Two different quantities

Coverage is what share of applications get reported. Error rate is how accurate the reported ones are. A system can be excellent on the second and incomplete on the first.¹

22. And the analyst names the risk

Under challenges.

Even a few errors can have large effects on an analysis.¹

22.0b And it cuts against the reassurance the error rate offers

Half a per cent of eighty million records is four hundred thousand bad rows, distributed somewhere in the set, and a query returning a few hundred records may contain several of them.¹⁴

That arithmetic is ours.

22.1 Which is a point about use rather than about collection

A database of eighty million records will be queried for a particular product in a particular place, and a subset small enough to answer a question is small enough for a handful of bad records to distort it.

That elaboration is ours.

22.2 Which is a fair warning rather than a weakness

A programme that tells users where its data can mislead them is doing the part of the job that usually gets left to whoever is misled.

23. What depends on this

The regulator lists its data as used by academic and government scientists, growers, registrants, non-profits, policy makers, legal firms, journalists and the general public, to assess risk, food quality, worker exposure, endangered species protection, water and air quality, pest management, and land-use trends.³

23.1 Legal firms appear on that list

Which is worth noticing, since it means the record is used to establish what was applied at a particular place and time for purposes other than research.³

23.2 And the record is public

Available through a public portal and as downloadable archived data files, which is what allows that range of users to exist at all.³

24. Which makes one state the world's evidence base

Our conclusion.

When a study reports how much of a compound is applied, where, and in what pattern over time, the underlying records are very often these ones, because there is no comparable set to use instead.

24.0b And the consequence is a quiet one

Nobody has to decide to rely on a single source. It happens because the alternative is having nothing to cite, and the citation looks the same either way.

24.0c And our own articles have relied on it without knowing

Any figure this journal has quoted about application volumes or patterns may well trace to these records, since they are what exists, and we did not check that when quoting them.

24.1 We did not verify that claim against a bibliography

It follows from the sources describing this system as the most complete in the world rather than from any count of papers, and §29 records it.²³

25. The generalisation problem

Which nobody can solve.

One state with a particular climate, crop mix, regulatory culture and building stock supplies the use data, and conclusions drawn from it are applied to places with none of those properties.

25.0b The properties that differ are the ones that matter

Pest pressure, building construction, heating season, product availability and the mix of licensed against unlicensed application all vary between places, and each of them changes what gets applied and how much.

25.1 Our brown recluse article made the same structural point

That a geographic evidence base gets read as a general one when it is the only base available.

25.1b The honest position is to state the provenance

A figure for how much of a compound is applied is a figure for how much is applied in one place, and saying so is cheap and rarely done.

25.2 And the alternative is worse

Reasoning from no data rather than from data collected elsewhere.

26. What we take from it

Three things.

The comprehensive record exists and it is singular. Section 1.²

It reaches our trade and not the consumer. Sections 10 and 15.³⁵

And its own analyst puts coverage at eighty to ninety per cent. Section 18.¹

26.0b And one habit this suggests

When reading any statement about how much of something is applied, asking which jurisdiction's records it came from, because the answer is usually the same one.

26.1 And one question it leaves open

Whether the exempt sector could be estimated some other way, from retail sales, from survey work or from waste stream sampling, none of which appeared in what we read.

27. Our own position

The disclosure.

An article noting that licensed operators report and householders do not is an article whose conclusion favours licensed operators, and we are one. Section 16's argument would hold if a regulator wrote it.

27.0b And we would still argue for it

Because the data in §23 cannot be produced any other way, and because a trade that objects to being counted has a harder time claiming its practices are defensible.

27.1 And the reporting burden is real

A monthly return itemising every product and quantity is administrative work, and a company describing it as remarkable detail in §12 should say that it would also be a cost to us.⁶

28. The Manitoba position

28.1 We could not find an equivalent

No Canadian or Manitoba system comparable to this appeared in our searching, and we did not establish what reporting, if any, applies to structural use here.

28.1b And our own licensing obligations are the place to start

A structural licence here carries record-keeping requirements, and whether those records are ever aggregated into anything resembling a use dataset is a different question from whether they are kept.

28.2 Which is a gap in this article rather than a finding

Not finding something is not the same as establishing it does not exist, and this is the kind of question a licensing authority could answer directly.

28.2b The practical consequence for a local argument

Any claim here about whether pesticide use has risen or fallen, in any sector, has to be made without the kind of record this article describes.

28.3 And our own provincial article touched the edge of it

Our article on cosmetic pesticide regulation here found the evidence arc turning on what was applied and where, which is exactly the quantity a reporting system would supply.

29. Limitations and open questions

The coverage estimate comes from one slide. The eighty to ninety per cent figure that §§18 to 20 rest on appears in a presentation deck, with no method, no date qualifier and no uncertainty attached, and we found it reproduced identically on three sites.¹

That is the most important limitation because it is the article's central number and it is exactly the kind of figure this journal has criticised elsewhere for travelling without its basis.

We did not read the regulations. Everything about what must be reported comes from summaries, a book chapter abstract and a law firm page rather than from the code sections themselves, which are cited in one source and which we did not open.²⁶

The scope conflict in §13 is unresolved. Whether structural operators report all applications or only restricted materials is a question our sources answer differently, and it matters for §§10 to 12.¹⁶

The exemption statement comes from a project record. Section 14 rests on a science activity description held in a public database rather than on a statutory list of exclusions.⁵

And we made no comparison with other jurisdictions. The claim that this is the only such system is repeated by its own operators and by a book chapter, and we did not test it against what other countries collect.²³

Sections 4.1, 6.1, 12, 16, 17, 20, 21.1, 22.1, 24 and 25 are our reasoning. The census argument, the account of why the exempt sector matters most, the connection to the consumer misuse problem, and the reading of one state as a global evidence base are ours rather than sourced positions.

30. Conclusion

One American state has required full pesticide use reporting since 1990, under a statute passed the year before, after earlier schemes covering aerial applications from the 1930s and restricted products from 1970.¹² It collects around three million records a year, holds more than eighty million, obtains them as a census rather than a sample, and is described in a published book chapter as the largest and most complete database on pesticide use in the world.²⁴¹ Structural pest control reports into it, and one account of the monthly return has operators filing the exact product, the registration number, the quantity to the decimal point and the number of applications, which is a degree of detail nothing in our own filing approaches. The data support risk assessment, worker exposure, endangered species protection, water and air quality and pest management, for scientists, regulators, journalists and the public.³⁶

Two things sit underneath that. The primary exceptions are home and garden use and most industrial and institutional uses.⁵ So the world's most complete record of what pesticide goes where omits the sector in which untrained people apply products bought off a shelf inside the buildings they live in, which is the sector with the closest exposure and the least training, and which is therefore the one nobody can put a number to. Our last consumer article found a registered pesticide being used illegally across attics and gardens with no prevalence figure available. The exemption is why no such figure exists.

And the programme's own analyst estimates that probably eighty to ninety per cent of actual use is reported.¹ Publishing that estimate is creditable, and most of the figures this journal has traced arrived without any such admission. But it means the best number available anywhere is short by a tenth to a fifth, and that every jurisdiction without such a system, including our own as far as we could establish, is short by an amount nobody has characterised at all. The error rate within the reported records is under half a per cent, which is a different and much better quantity, and the analyst's own listed challenge is that even a few errors can have large effects on an analysis.¹ We should add that the eighty to ninety per cent figure reaches us from a single slide with no method attached, which is precisely the kind of provenance this journal spends its time complaining about.

References

1. California's Pesticide Use Reporting System. Presentation deck by a named analyst at the state pesticide regulator, found in identical form hosted by a state agriculture department, a regulatory research organisation and another state's agriculture department. Official technical material in slide form, which we flag because the article's central number comes from it without a stated method. Source for the statement that the state first required limited reporting by 1950, that the database contains records starting in 1974, and that the current full use reporting system started in 1990 with over 2.5 million records each year since; for the statement that all pesticide applications in state production agriculture must be reported to the appropriate county agricultural commissioner's office and that all pesticide applications made by commercial pest control businesses must be reported; for the data quality statements that the reports are screened for about forty kinds of errors with an error rate of less than 0.5 per cent, that probably 80 to 90 per cent of actual use is reported, that data are obtained from a census rather than samples, and that the data are geographically referenced; and for the listed challenge that even a few errors can have large effects on an analysis. https://www.cdfa.ca.gov/oefi/nutrientmanagement/docs/4_Wilhoit.pdf
2. History of Pesticide Use Reporting in California. Chapter in a professional society symposium series, read as its abstract and opening rather than in full. Source for the statement that the state regulator's pesticide use report is the largest and most complete database on pesticide use in the world and contains information on production agricultural use and on some non-agricultural use; for the account that the system has existed in three distinct forms, namely aerial pesticide applications from 1934 to 1956, primarily restricted pesticides from 1970 to 1989, and full agricultural use reporting from 1990 onward; for the statement that the current database was established in 1990 by a food safety act passed the previous year; and for the list of data that must be reported for production agricultural applications, including date, operator of the property treated, location by county, section, township, range, base and meridian, hour the treatment was completed, operator and site identification numbers, total acreage planted, the persons who made and supervised the application, whether it was made by an agricultural pest control business, the crop or site treated, the total acreage or unit treated, and the federal or state registration number. https://pubs.acs.org/doi/10.1021/bk-2018-1283.ch001
3. Pesticide use reporting page published by the state pesticide regulator. Official regulatory material. Source for the statement that the programme is recognised as the most comprehensive in the world and has required detailed reporting since 1990 for nearly all types of agricultural use and many non-agricultural uses; for the note that the database defines agricultural and non-agricultural uses slightly differently from the governing code; for the definition of non-agricultural uses in the database as all other legally reportable uses, listing rights-of-way, landscape maintenance, golf courses, parks, recreation areas, cemeteries, structural pest control and vector control; and for the list of users and purposes, namely academic and government scientists, growers, registrants, non-profits, policy makers, legal firms, journalists and the general public, using the data to assess risk, food quality, worker exposure, endangered species protection, water and air quality, pest management and land-use trends. https://www.cdpr.ca.gov/pesticide-use-in-california/pesticide-use-reporting
4. Summary of pesticide use report data for one year, published by the state pesticide regulator. Official regulatory publication, read as an extract. Source for the description of the data flow, in which reports are entered into the county permit database and checked for errors before the commissioner electronically sends required data to the state department for additional validation; for the statement that the department collects on average around three million pesticide use records a year and that the database currently contains over eighty million records going back to 1990; for the note that earlier records from 1974 to 1989 are kept in a separate database because those early records vary in the type of information they contain; and for the statement that the data are used by state, regional and local agencies, scientists and public interest groups, and support development and adoption of integrated pest management practices in both agricultural and urban settings. https://www.cdpr.ca.gov/wp-content/uploads/2024/12/pur_data_summary_2018.pdf
5. Science activity record describing the pesticide use reporting programme, held in a public project database maintained by a state council. Administrative project material rather than a statutory source, which we flag because the exemption list in this article comes from it. Source for the statement that the programme is recognised as the most comprehensive in the world and that in 1990 the state became the first to require full reporting of agricultural pesticide use in response to demands for more realistic and comprehensive pesticide use data; for the account that all agricultural use must be reported monthly to county agricultural commissioners who report onward to the state department; for the statement that the state has a broad legal definition of agricultural use so that reporting requirements include applications to parks, golf courses, cemeteries, rangeland, pastures and roadside and railroad rights-of-way, together with all postharvest treatments of agricultural commodities, all treatments in poultry and fish production and some livestock applications; and for the statement that the primary exceptions to the reporting requirements are home-and-garden use and most industrial and institutional uses. https://sciencetracker.deltacouncil.ca.gov/node/50082
6. Page on state environmental regulations covering pesticide use and compliance, published by a law firm. Commercial legal marketing material, cited as attributed material and flagged because it conflicts with an official source on scope. Source for the statement that the state operates the most extensive pesticide use reporting system in the world and that since 1990 it has required full reporting of all agricultural pesticide use as well as the use of restricted materials by structural pest control operators; and for the description of the structural return, in which companies must submit a monthly summary report to the county agricultural commissioner detailing every application made in the previous month, including the exact product name, the federal registration number, the total quantity used down to the decimal point, and the number of applications performed. https://www.lawincalifornia.com/business-compliance/environmental-regulations-pesticide-use/