Biology

Biological threshold policies: how thresholds differ from detections

Definition

A biological threshold policy is the versioned, reviewable rule set that turns biological observations into a management decision: which life stages count, how density or defoliation is assessed, which crop stage applies, and how much uncertainty is tolerated. A detection reports what a model saw; a threshold policy determines whether what was seen justifies action.

This page describes the structure of a threshold policy and where its authorship sits. It states no threshold value, because that is an agronomic judgment and not ours to publish.

Published: August 6, 2026
Last updated: August 6, 2026

What is a biological threshold?

A level of pest presence or crop damage at which a management program considers intervention justified. It is a biological and economic judgment about a crop in a context, not a property of any imaging system.

Thresholds are expressed differently depending on the pest and the program: counts of a particular life stage per plant or per unit of row, a proportion of plants infested, or a percentage of leaf area lost. What they share is that the number means something only alongside the crop stage, the sampling method, and the region it was set for.

How is a threshold different from a detection?

A detection is an observation about an image. A threshold is a rule about a field. Conflating them is the most common way a detection system gets mistaken for a decision system.

Detection and threshold compared
	Detection	Threshold policy
What it describes	Content of an image.	When action is warranted in a field.
Who authors it	A trained model, from data.	Agronomic authorities and the grower's program.
What changes it	Retraining on new data.	Biological review and a versioned revision.
Unit of reference	Pixels, frames, and confidence.	Crop, stage, area, and time.
Failure mode	A wrong label.	A rule applied outside the context it was set for.

What is an economic threshold?

The pest level at which intervention is considered justified because expected crop loss outweighs the cost of acting. It depends on crop value, treatment cost, and growth stage, so the same pest count can sit on either side of it.

This is why a decision platform cannot carry a single global threshold. An economic threshold that is sound in one region, season, and market is not automatically sound in another, and a system that applies one everywhere is asserting an agronomic judgment it has no basis for.

Who sets the thresholds?

Agronomic authorities and the grower's own program — extension services, research stations, agronomists, and crop advisers. Swathmark does not author agronomic thresholds and does not recommend threshold values.

The platform's role is to represent a threshold policy faithfully, apply it only within the context it was written for, version it, and record which version produced a decision. Where a policy does not cover the crop stage, region, or life stage in front of it, the intended behavior is to abstain rather than to extrapolate.

Swathmark does not provide agronomic, pesticide-label, or treatment advice. Threshold selection, product choice, and label compliance remain the responsibility of the grower and their advisers.

Why must a threshold policy be versioned separately from the model?

So a change in biology can be told apart from a change in perception. If the rule is embedded in detector code, nobody can determine whether an outcome shifted because the model was retrained or because someone edited a threshold.

Reviewability: an agronomist can read and challenge a policy without reading model code.
Attribution: a change in decisions can be traced to a policy revision or to a model version, not to an indistinguishable mixture of the two.
Replay: an earlier decision can be re-evaluated against a revised policy while the original remains on record.
Scope control: a policy states the crop, stage, region, and pest it applies to, so applying it elsewhere is a detectable error rather than a silent one.

What does a threshold policy contain?

Its scope, the biological measures it uses, how uncertainty is handled, and its authorship and review history. The values themselves are only meaningful in the presence of the rest.

Intended structure of a threshold policy
Element	Purpose
Scope	Crop, pest, growth stages, region, and season the policy applies to.
Life stages counted	Which stages contribute to the measure and which are ignored.
Measure	Whether the rule is expressed in counts, proportion infested, or defoliation.
Sampling basis	The coverage and assessment method the values assume.
Uncertainty handling	How much ambiguity is tolerated before the outcome is abstain.
Authorship and review	Who wrote and reviewed the policy, and when.
Version	The identifier pinned into every decision the policy produced.

What does a threshold policy not do?

It does not select a product, calculate a rate, interpret a label, or authorize an application. It answers one question — whether observed pest pressure meets the level its authors set for action.

No product selection, rate, timing, or mixing guidance.
No pesticide-label interpretation or regulatory determination.
No resistance-management prescription, though thresholds are one input to such a program.
No authority over the grower's decision, which remains theirs to make.

Inputs, outputs, limitations, and evidence status

Inputs

Biological observations: life stage, counts or density within the imaged area, visible damage.
Crop identity and growth stage at the time of capture.
Recorded uncertainty and asset-validity state from verification.
A threshold policy version whose scope covers the crop, pest, and stage.
The sampling coverage the observation actually represents.

Outputs

An evaluation of the observation against the policy, for a defined area and time.
Treat, no-treat, or abstain, with the policy version pinned to the result.
The specific policy clause that determined the outcome.
An explicit out-of-scope result where no policy version applies.

Limitations

A policy is valid only inside the crop, stage, region, and season it states.
Thresholds are agronomic judgments, and this site publishes no threshold values.
A policy cannot compensate for observations that misrepresent field-scale pest pressure.
No product, rate, timing, label, or regulatory guidance is produced.

Evidence status

The policy structure described here is the intended platform design.
No threshold values, policy libraries, or agronomic recommendations are published.
No endorsement by any extension service, university, or agency is claimed or implied.
The first validation domain is Colorado potato beetle in potatoes.

Terms used on this page

Definitions are shared site-wide. The full list is in the glossary.

Threshold policy: The reviewable, versioned rule set that turns biological observations into a management decision: which life stages count, how density or defoliation is assessed, which crop stage applies, and what uncertainty is tolerated. Held separately from model code.
Economic threshold: The pest level at which intervention is considered justified because expected crop loss outweighs the cost of acting. Economic thresholds are set by agronomic authorities and local programs, not by a software vendor.
Integrated pest management: A management approach that combines monitoring, biological and cultural controls, and threshold-based intervention so that treatment is applied in response to observed pest pressure rather than on a fixed calendar.
Pest pressure: The intensity of pest activity affecting a crop over an area and period, expressed through counts, density, or damage rather than a single sighting. Pressure is a field-scale property; a detection is an image-scale observation.
Defoliation: Loss of leaf area caused by feeding damage, commonly assessed as a percentage of the canopy. Several pest programs express thresholds in defoliation rather than pest counts.
Life stage: The developmental phase of an insect — for example egg mass, larval instar, pupa, or adult. Life stage changes both how a pest appears in imagery and how a management program treats it, so detecting the stage matters as much as detecting the species.