Colorado potato beetle: life stages, detection, and validation scope

What is the Colorado potato beetle?

A leaf-feeding beetle whose larvae and adults consume potato foliage, and which also feeds on other solanaceous crops including tomato and eggplant. Heavy feeding removes leaf area, and severe defoliation reduces the plant's capacity to produce yield.

Adults overwinter in soil and emerge to feed and lay eggs as host plants become available, with the number of generations in a season varying by region and season length. It is among the most studied insect pests in temperate agriculture, which is one reason it is a reasonable domain to be measured against: the biology is not in dispute.

This is general background. Local pest development, thresholds, and management guidance come from the grower's extension service and advisers, not from this page.

Which life stages matter for detection?

Egg masses, the larval instars, and adults are all visually distinguishable to a trained observer, but not equally easy to image. Difficulty rises as the target gets smaller and as it sits further under the canopy.

Why does life stage matter for a treatment decision?

Because stage changes both what a threshold policy counts and how a population will develop. A count that does not distinguish stages is a weaker input than the same count that does.

Stage composition carries timing information that a total count discards: a population dominated by egg masses and early instars is at a different point in its development than one dominated by late instars and adults. Which stages a policy counts, and how, is stated by the policy — not inferred by the detector.

How is Colorado potato beetle monitored in practice?

By scouting: inspecting sampled plants along a route, recording stages and counts or estimating defoliation, and comparing the result to a threshold set by an extension program or adviser.

The reason thresholds dominate this pest's management is resistance. Colorado potato beetle populations have developed documented resistance to multiple insecticide classes, so extension guidance emphasises resistance management — intervening on observed pressure rather than on the calendar, and rotating insecticide modes of action so a population is not repeatedly exposed to the same selection pressure. Programmes also anticipate timing with degree-day accumulation from a spring biofix, rather than waiting to find a generation already established.

Swathmark does not provide agronomic, pesticide-label, or treatment advice, and states no threshold values. Monitoring method, threshold selection, product choice, and label compliance remain with the grower and their advisers.

Why did Swathmark start with Colorado potato beetle?

Because it is a hard, honest test of a full capture-to-decision loop rather than an easy demonstration. The pest is visible, biologically staged, and threshold-managed, so every part of the chain is exercised and can be checked against known biology.

• Visible on foliage, so the detection question is genuinely a vision question.
• Biologically staged, so life-stage detection matters rather than being decorative.
• Threshold-managed already, so a decision layer is answering a question growers actually ask.
• Well studied, so a claim about it can be checked against an established literature.
• Occlusion, small early stages, and look-alike ambiguity are all present — the failure modes are real, not contrived.

The data, policy, evaluation, and equipment interfaces are designed to extend to further visible, threshold-managed pests and specialty crops. That extension is a design direction, not a claim of current performance beyond this domain.

What is the validation scope?

Colorado potato beetle in potatoes. Any statement about another crop, pest, region, or capture system is outside the scope Swathmark is working within, and the platform is designed to abstain rather than extrapolate across that boundary.

Scope is stated at this level of detail because it is the part most often left vague. A model evaluated on one crop, in one region, with one capture geometry has been evaluated on exactly that, and the honest description of anything beyond it is that it has not been tested.

What evidence exists today?

No detection performance figures are published for this or any pest. Accuracy, recall, precision, calibration, and false-treatment rate are conditional claims requiring an approved evidence record, and none has been approved for publication.

When figures are published they will state the dataset, the field and region scope, the crop and pest, the model and policy versions, the operating conditions, and the known limitations. A number without those is not evidence — it is a number, and a grower cannot use it to decide anything.