You put predatory mites on your plants. Two weeks later there are more pests than there were on the day you started, on a plant now covered in the things that are supposed to be eating them. It reads like the clearest evidence anybody could ask for that you wasted your money.
It is usually the opposite. Most of what was on your plant on release day was not adult pests — it was eggs, already laid and already hidden before the predators ever arrived. Nothing you put out could have prevented them, and they hatch on their own schedule over the following two weeks whatever you bought and however well you used it.
Which is most of the explanation for the rise you are looking at around day twelve to eighteen. The rest of it is arithmetic — how many eggs, hatching across how long, against how fast a predator population can grow from what you put on the plant. That arithmetic is the useful part, because it gives the whole thing a shape: a date by which the line should turn, and a way it should look on the way there.
It matters more than it sounds like, because this is the exact week most people switch products. Doing that resets the clock to zero and buys them the same two weeks over again, usually while blaming the first product for the second delay.
The short version
- What you are seeing
- More visible pests than there were on the day you put the mites out, usually week two or three
- What is happening
- Eggs already on the plant before the mites arrived are now hatching
- Why the predators did not stop it
- They eat eggs and the youngest stages. Those eggs hatched before they were reached
- What it means
- The backlog is draining. This is the shape of it working, not of it failing
- The expensive mistake
- Switching products now, which restarts the whole establishment period
- When to actually worry
- No downward move by about week four, in a room warm enough to breed
Why the numbers go up
On the day you released, your plants held four things: adult pests, larvae, pupae and eggs. You could see the adults. You could not see most of the rest, and the eggs were the largest group by a wide margin.
Predatory mites eat eggs and the youngest larvae. They physically cannot get into anything older — a settled whitefly nymph has thickened up, and a second-stage thrips larva is big enough to fight back and win. Under direct observation, predators failed to kill a single second-stage thrips larva.
They do eat eggs — but eggs are the slow half of the job, because most of a thrips egg sits down inside the leaf. Watched under a lens, a single female took about four and a half thrips eggs in a day against thirteen first-stage larvae. That is the whole problem in one pair of numbers: a backlog laid over three weeks by every adult you could see is simply larger than that rate can clear before it hatches. And it hatches on its own schedule — a few days for some pests, a couple of weeks for others — whatever you released.
Meanwhile the adults you could see on day one are still there. They were never in reach. They are also still laying, for as long as they live.
Put those together and the visible count does something that feels wrong and is completely normal: it dips for a few days, because the predators cleared what was exposed on the leaves, then rises as the pre-existing eggs hatch. That rise is the backlog draining. It is not new ground being lost.
What it looks like when it is working, week by week
Roughly, in a room warm enough for the predators to breed. Everything here stretches in a cool room and compresses in a warm one.
| When | What you see | What is actually happening |
|---|---|---|
| Days 1–5 | A small improvement, or nothing | Predators clear the exposed juveniles. Adults unaffected — they are too large to be eaten |
| Week 2 | Numbers rising | The pre-release eggs are hatching. Predators are also laying, but their generation is behind |
| Week 3 | The peak, usually | Backlog mostly emptied. The predator population is now mostly animals born on your plant |
| Week 4 | A clear fall | Adults ageing out and not being replaced, because the generation behind them was eaten |
| Weeks 5–6 | Low and staying low | Standing population holding. Nothing escalating, which is the point |
The important part of that table is the third column, because the first column looks identical whether things are working or not. That is exactly why the rebound is dangerous — not because it is bad, but because it is indistinguishable from bad without knowing the mechanism.
Switching products now restarts the clock
Here is the sequence, and every step in it is what a reasonable person does.
You release. Week one looks encouraging. Week two is worse than the day you started. You conclude the product did not work — reasonably, because that is what the evidence in front of you looks like. So you either spray something, or you buy a different predator, or both.
Spraying kills the predators, including the generation that hatched on your plant and was about to do the work. Switching products throws away two weeks of establishment and starts the same two-week clock again, with the same rebound waiting at the end of it.
Which is how people end up doing this three times
Each attempt gets abandoned at the same point in the cycle, for the same reason, and each one is judged on the same misleading week.
The pattern is not bad luck and it is not bad product. It is a timing mismatch — a predator that works on the next generation, judged on this one.
What to do for the next two weeks
- Nothing, for another two weeks. Genuinely. This is the hardest instruction in biological control and it is the one that works.
- Do not spray, including the gentle products. Insecticidal soap, horticultural oil and neem kill on contact and a predatory mite is a small soft-bodied animal. There is no spray that distinguishes them.
- Do not wipe or hose the worst leaves. It feels productive and it sets you back — wiping removes the predators along with the visible pests and leaves the eggs, which are glued on, exactly where they were.
- Count something, on the same leaf, weekly. Mark one leaf, or put up a sticky card. You need a number to compare against, because at this stage your impression of the plant is not a reliable instrument.
- If you are running sachets, put the second one in on schedule. The overlapping second sachet is what covers the tail of the backlog — the same sachets you started with. Skipping it because the first one "did not work" is the same error one step later.
And if you did already spray — that happens, and it is better said than hidden. Wait out the residual period for whatever you used before releasing again. Some last days; some sit in the plant for weeks.
When it really has failed: no drop by week four
A rebound that never turns is a real signal, and "just wait" without a stopping condition is not advice; it is a way of never being wrong.
Investigate if the count has not started falling by about week four, measured on whatever you started counting at release, in a room that has stayed warm enough for the predators to breed.
At that point the cause is almost always on a short list, and each one has a fingerprint: the room is too cold for the population to grow, the air is dry enough that the predators' own eggs are not hatching, something was sprayed, the pest is not one this species controls, or the plants stand far enough apart that the predators never reached most of them. The species-specific version is in what goes wrong on a first release.
Tips
Getting through the rebound
Write down a number on release day. Anything countable on one marked leaf, or a sticky card. Two weeks later you will be comparing your memory of the plant against a worse-looking plant, and memory loses.
Put a note in your calendar for week four, not week two. It is the earliest date the count is worth judging, and having the date written down is most of what stops the week-two decision.
Expect the peak to be the worst it looks. Weeks two and three are usually the high point, not a trend. If you can name that in advance it stops being alarming.
Do not tidy the plant. Wiping and hosing take out predators and leave eggs. It is the most understandable counterproductive thing you can do.
If you are going to give up, change one thing, not everything. Switching product and spraying and moving the plant at once means you learn nothing and the next attempt starts from the same guesswork.
A second overlapping release beats a replacement first release. The tail of the backlog is what a second sachet or a follow-up bottle is for. Starting over gets you the same rebound two weeks later.
Common questions
My predatory mites made the problem worse. What happened?
Almost certainly nothing did. Pest numbers commonly rise around week two or three after a release, because the eggs that were already on your plants when you released are hatching on their own schedule. Predatory mites eat eggs and the youngest larvae, so anything laid before they arrived was already past them, and the adults you could see were always too large to be eaten. The rise is the backlog draining, and it usually peaks in week three and falls through week four.
Should I add more predators now?
If you are on a sachet program, put the next one in on schedule — the overlapping release is what covers the tail of the backlog. Starting a whole new release instead usually is not the answer, because you are two weeks into an establishment period and restarting it gets you the same rebound two weeks later.
Can I spray just the worst plant while the mites work on the others?
Not usefully. Spot-treating leaves a patch your predators will not survive crossing, and on a shelf where foliage touches, that is a hole in the middle of the area they are working. If a plant is genuinely beyond the program, isolating it away from the others is the better move than treating it in place.
How long before I should actually worry?
About week four, measured against whatever you counted on release day, in a room warm enough for the predators to breed. Before that there is usually nothing to see even on a release that is working perfectly. After that, the causes are a short and diagnosable list.
Is it too late to save this, or do I start over?
Almost always neither — you are mid-cycle rather than at an endpoint. The exception is if something was sprayed after release, in which case the predators are mostly gone and the honest answer is to wait out the product's residual period before releasing again.
References
- Beretta, G.M., Zandbergen, L., Deere, J.A., Messelink, G.J., Muñoz Cárdenas, K. & Janssen, A. (2024). Predator–prey interactions: how thrips avoid predation. Biological Control 188: 105437. Direct observation of which stages a predatory mite can reach — the reason the adults present at release are never the ones eaten. doi.org
- Nguyen, V.H., Song, Z., Nguyen, D.T., Van Leeuwen, T. & De Clercq, P. (2024). Functional response of four phytoseiid mites to eggs and first-instar larvae of western flower thrips, Frankliniella occidentalis. Insects 15(10): 803. Where the four-and-a-half versus thirteen comes from, and the note that thrips eggs are mostly embedded in leaf tissue. doi.org
- Lee, H.-S. & Gillespie, D.R. (2011). Life tables and development of Amblyseius swirskii (Acari: Phytoseiidae) at different temperatures. Experimental and Applied Acarology 53(1): 17–27. Why the timeline above is stated as depending on the room rather than as fixed weeks. doi.org
- Our own support record. The week two to three rebound is one of the two most frequent situations customers write to us about, and the wording in this article is taken from the reply we have been sending for it rather than written fresh — so the article, the packing insert and the answer you would get by email all say the same thing.
That is the point where it is worth diagnosing rather than waiting, and the causes are a short list.
