Six Stages. It Reaches One.

A western flower thrips passes through six stages on its way to being a problem: egg, two larval stages, a prepupa, a pupa, and an adult that flies.

Cucumeris kills one of them.

Not "prefers" one of them. Not "is most effective against" one of them. In a study that watched individual attacks under a microscope, second-instar larvae of three different thrips species were never successfully attacked at all across the observation windows. The eggs are inserted into plant tissue. The prepupa and pupa are in the soil. The adult is too large and it flies.

Almost everything people find disappointing about this animal falls out of that one sentence — the weeks of lag, the counts that rise before they fall, the reservoir that keeps producing adults, and the reason it is sold as prevention rather than as a cure. It also explains why the usual upgrade advice is wrong, which is the part that surprised us.

It isn't refused. It's attacked, and it fails

The distinction matters because it changes what you expect to see.

A predator that declines to attack large prey is making a choice, and choices can be pushed — starve it, remove the alternatives, and it takes what's there. That is not what happens here.

The foundational work is from 1989. Predatory mites were offered first- and second-stage Thrips tabaci larvae, and capture success was measured against both the predator's hunger and the larva's size. Starved predators attacked more readily — and on contact with a starved predator, second-stage larvae still incurred a lower death risk than first-stage ones. The mechanism was visible: the larvae jerked their abdomens and produced a drop of rectal fluid.

The clever part of that study is the control. When the researchers suppressed the defensive behaviour with anaesthesia, attack success went up — but it did not level out the difference between the two stages. Behaviour explains part of the refuge. Something else, size or handling, explains the rest, and the authors said so rather than papering over it.

Thirty-five years later a different group repeated the logic with a different predator and three thrips species, watching individual attacks in fifteen-minute windows. The result was starker: no second-instar larva of any species was successfully attacked. First instars of two species were sometimes taken — specifically when the predator got to them before they could mount a defence.

Read that carefully. Even the stage cucumeris is sold against sometimes wins. What makes first instars catchable is not that they're defenceless; it's that they're slower to react.

The droplet is a signal, not just a shield

The defensive droplet is more interesting than it first appears. It contains decyl acetate and dodecyl acetate, and it works as an alarm pheromone — larvae exposed to it, natural or synthetic, produce their own droplet faster when subsequently attacked.

Which means a patch of plant with an active predatory mite in it is a patch of primed thrips. The refuge isn't a fixed property of the prey; it strengthens where the hunting is happening.

It is a small piece of chemistry with an unwelcome implication: the harder the predator works, the better the larvae get at surviving it.

The size argument nobody has actually measured

The standard explanation is that second instars are simply too big. That is almost certainly part of it, and it is worth being straight about how well established it is.

An adult cucumeris is 0.5–1.0 mm long. The first-instar larvae used in the standard predation assay were 0.5–0.6 mm long — already about the predator's own body length. Second instars are larger again, and the first larval stage lasts roughly half as long as the second, so the animal spends most of its larval life at the bigger size.

But: no published study measures thrips larval width or mass by instar alongside phytoseiid body size and computes a predator-to-prey size ratio for this pair. The size argument is asserted universally and, as far as we can find, has never been quantified.

The handling-time literature has the same hole from the other side. Handling times for phytoseiids on first-instar thrips are published — between about 1.8 and 5.2 hours depending on species. There is no handling time for any phytoseiid on a second instar, which is unsurprising when successful attacks are too rare to time. The gap is itself the finding.

How much of a thrips population is even reachable

This is the question the whole subject turns on, and nobody has published the answer. So here is the arithmetic, with the workings visible, because a number you can check beats a number you have to trust.

Measured stage durations for western flower thrips at 77 °F (25 °C): egg 4.1 days, the two larval stages together 7.8 days, prepupa 1.3, pupa 2.8 — about sixteen days egg to adult. Adult females live around 26 days. The first larval stage is roughly a third of the combined larval period.

Run those durations two ways and the answer forks — which is the actual finding:

Stage	In a stable population	In a population that's exploding
Egg	9.8%	49.4%
First instar	6.2%	17.8%
Second instar	12.4%	19.0%
Prepupa	3.1%	2.7%
Pupa	6.7%	4.2%
Adult	61.9%	6.9%

These are our calculation from published parameters, not a published result. They assume the measured stage durations, the one-third to two-thirds larval split, an adult lifespan taken from a different temperature to the development times, and no stage-specific mortality — which is a real simplification. Treat them as an order of magnitude.

The spread is the story. In a population growing fast, first instars are a large minority of everything alive and cucumeris is well placed. In a stable or declining population, first instars are around six percent of individuals, three-fifths of which are adults, and the predator is reaching almost nothing.

Which is a strange and slightly beautiful result: cucumeris is at its best against a population that is getting worse. Its window is the exact moment when a grower would say the problem is escalating, and its worst case is a stable population that has been sitting there for a month.

Why nothing happens for three weeks

Ontario's agriculture ministry puts control with cucumeris or swirskii at five to nine weeks, against three to five for a predatory bug that eats every stage. That gap is not a difference in quality. It's the queue.

Kill a first instar today and you have removed an adult that would have emerged in about ten days. You have not touched the second instars, which will pupate regardless, or the pupae in the substrate, or the adults currently laying. Every one of those completes on schedule while the predator works. The count you are watching — adults on a sticky card — is the last stage in the chain, and it responds last.

There is a real gap in the literature here and it is worth flagging plainly. We could not find a published time series of adult thrips counts following a cucumeris-only release against an untreated control. The best available trial reports larval counts only, at thirty days. So the lag is inferred — from stage durations, from the soil reservoir, and from trials showing that adding a soil agent shortens it — rather than measured in the form everyone implies.

The practical version: judging a cucumeris release at week one is judging a queue by its last position.

What it does instead of killing them

The single most useful thing in this literature is buried in a 2006 nematode paper, and it reframes the whole relationship.

Researchers combining predatory mites on the foliage with entomopathogenic nematodes in the soil observed that the presence of mites on the plants caused large numbers of second-instar thrips larvae to drop off the plants and pupate in the soil — thereby increasing the number of hosts available for the nematodes to attack. Combined, the two agents gave up to 83% control, significantly better than either alone.

So cucumeris does not fail to deal with second instars. It moves them. It cannot kill them, and it makes the plant unpleasant enough that they leave it early and go somewhere else — which happens to be the one place where a different agent is waiting.

That is a much better argument for pairing a foliar mite with a soil agent than the usual one, because it isn't about covering a gap in a table. It's a mechanism: the predator on the leaves is actively delivering prey to the predator in the pot.

How big the soil reservoir actually is

Most of a thrips population leaves the plant to pupate, and "most" is doing some work:

Measured	Crop and conditions
98% went to soil	French bean, microcosms with emergence traps
92–93% in soil	Non-flowering potted chrysanthemum and mini rose
87% in soil	Flowering mini rose
60% in soil	Flowering chrysanthemum — 40% stayed on the plant

Two things follow. First, the flowering state of the crop changes the answer by more than thirty points, so a soil agent is worth much more on some crops than others. Second, there is an unresolved disagreement in the literature — one study reports 60% for flowering chrysanthemum where a review reports 92% for the same crop at the same growth stage. We are not going to pretend that is settled.

And the fraction moves with humidity rather than being a property of the species. Above 80% humidity was necessary for late second instars to survive to pupation, with 90% optimal — and the humidity at which larvae switch to dropping off the plant rather than staying on it was measured at around 81%. Drier air pushes them down. Which means humidity management changes the size of your reservoir, and therefore how much a soil agent is worth to you.

What reaches the stages it can't

It would rather eat pollen, and that's the point

Cucumeris is an omnivore that will complete its entire life cycle on pollen. That is normally presented as a convenience — it survives between pest events — and it is also the thing that makes the whole preventive strategy possible.

It goes further than survival. In a study comparing several supplemental foods against thrips larvae as the reference diet, the supplements produced shorter development times and higher survival than thrips did. Apple and cattail pollen were the best of them; maize was the worst.

A predator that develops better on pollen than on its target pest is not a defective predator. It's a predator you can establish before the pest arrives, which is the only strategy that works given everything above.

The catch is in the other article. Pollen in the crop cut cucumeris predation on thrips by 55% in one study — the mites stay, the mites breed, and they kill fewer thrips. Supplemental food raises predator numbers and lowers per-mite predation at the same time, and whether the net is positive has been demonstrated for swirskii and not for cucumeris.

One more caution worth carrying: pollen quality is not transferable between studies. Two laboratories tested castor bean pollen and one recorded 100% juvenile mortality while the other ranked it the best of seven. They also disagree about maize, where one found the lowest but positive growth rate and the other found no egg-laying at all. "Cucumeris can live on pollen" is true. "Cucumeris can live on this pollen" is a separate question nobody can answer for your crop.

The upgrade argument is wrong

Here is where the research changed our mind, and it is the reason this article exists.

The standard advice — including in comparison tables we have published ourselves — is that swirskii and limonicus are better thrips predators because they take first and second instars where cucumeris takes only first. It is in nearly every table on the internet.

It does not survive contact with the primary literature.

• The canonical source for the idea is a 1995 paper which states that second instars defend themselves well and were therefore not measured. It reports a pilot experiment with no data table.
• The 2024 attack-observation study found Amblyseius swirskii — the species sold as the upgrade — unsuccessful against every second-instar larva of all three thrips species tested.
• The one quantitative second-instar figure we could find is for limonicus: 0.60 second instars per day against 3.10 first instars, and in a choice test it fed on first instars exclusively.
• A peer-reviewed species review of swirskii lists first-instar thrips only. The manufacturer-side review of why swirskii succeeded commercially does not mention prey-stage range at all — it cites multi-pest range, pollen feeding, and cheap mass rearing.

Ontario's ministry does state that swirskii feeds to a lesser extent on second instars. It gives no citation, and the behavioural study directly contradicts it.

We would rather correct this than keep repeating it. The case for the other species is real; it is just not the case that gets made.

What follows

1. Release before you have a problem. Not as a platitude — as the direct consequence of six stages and one reachable one. A predator present when the first eggs hatch is working on a growing population, which is the situation it is good at.
2. Read the trend at weeks three and four, not week one. The adult count is the last link in a chain the predator is working on at the front.
3. Pair it with something that reaches the soil — and understand the reason. It isn't that a table has a gap. It's that your mites are actively pushing second instars off the plant, and something should be waiting for them.
4. Add sticky cards, always. Nothing you put on the plant or in the pot touches adult thrips. Cards are not monitoring equipment here; they are the only thing acting on that stage at all.
5. Don't switch species expecting second-instar coverage. You will not get it. Switch for diapause, establishment or temperature range — those differences are real.
6. Watch what stage is on your leaves. Small translucent larvae mean the predator has something to work with. Mostly larger yellowish ones mean it doesn't, and no rate increase fixes that.

Common questions

Why doesn't cucumeris eat bigger thrips larvae?

It tries. Second-instar larvae swing their abdomens and produce a droplet of rectal fluid, and the attack fails — in one study watching individual attacks, no second instar of any of three thrips species was successfully attacked. Part of the refuge is that behaviour and part is size, and the study that separated the two showed suppressing the behaviour did not close the gap entirely.

Does swirskii or limonicus eat second instars?

On the published evidence, essentially no. Swirskii was unsuccessful against every second instar tested in the same behavioural study. Limonicus managed 0.60 per day against 3.10 first instars, and fed exclusively on first instars when offered a choice. The claim that these species cover the second instar is repeated everywhere and traces back to a pilot experiment that published no data. Their genuine advantages are elsewhere — no diapause, better establishment, higher populations.

How long before I should expect to see fewer adult thrips?

Ontario's agriculture ministry puts it at five to nine weeks for cucumeris or swirskii, against three to five for Orius, which eats adults directly. The lag is structural: killing a first instar removes an adult that would have appeared about ten days later, while every second instar, pupa and adult already present completes on schedule regardless.

If it only reaches one stage, why does anyone use it?

Because that stage is the throat of the bottle in a growing population, and because cucumeris can be established on pollen before the pest arrives — it develops faster on pollen than on thrips. Used preventively it intercepts each new cohort as it hatches. Used as a rescue on an established population it is reaching about six percent of what is alive, which is why it disappoints in exactly that situation.

Do I really need a soil agent as well?

The mechanism is good: 87–98% of prepupae and pupae are in the substrate depending on crop, and mites on the foliage have been observed driving second instars off the plant and into the soil early, which increases what a soil agent can reach. But the crop-level evidence is genuinely mixed — one review reports foliage plus soil predators did no better than foliage alone, while a cyclamen trial found the combination reduced thrips in flowers faster. Flowering crops keep more thrips on the plant, so the value of a soil agent is not constant.