Why Fern's spider mite course is 37 days

Monty has spider mites. The undersides of his leaves are speckled pale, there's a little webbing where the leaf meets the stem, and Dr. Fern's photo check agrees. Fern suggests a course: spray with neem oil every 11 days, 3 times, then two checks a week apart. About 37 days in all.

That's longer than most of what you'll read online, where "spray for a couple of weeks" is the usual advice. It's also a number people are entitled to question. Why 37, and not 14 or 21 or a round month?

The length is the most important number in a treatment. Stop too early and the eggs you never reached hatch into a fresh infestation two weeks after you put the spray bottle away. Run too long and people spray a plant for nothing, which costs them time and costs the plant, since oils and soaps aren't free for leaves either. A length somebody picked because it sounded about right would be easy to ship and impossible to explain. So Fern works it out, and it shows how.

What actually decides how long it takes

A spray mostly kills what it touches. For a pest, that means the mites, nymphs and larvae out in the open, and often not the eggs. Spider mite eggs sit under webbing on the undersides of leaves. Thrips lay theirs inside the leaf tissue, and their pupae drop into the potting mix. Mealybugs lay into a waxy sac that water rolls straight off.

So a treatment has two jobs. It has to keep going until every egg that was already there has hatched. And the gap between rounds has to be short enough that each new hatch gets hit before it grows up and lays eggs of its own. Get both right and the population can't restart. Miss either and you're back where you began.

Both of those depend on how fast the pest develops, and that depends on temperature. Mites in a warm room grow up much faster than mites in a cool one. Then there's the product's label, which sets how close together you're allowed to spray and sometimes how many times.

None of that is new. Extension services have been writing it in pest notes for decades. What an app needs on top is the numbers themselves, written down with where they came from, in a form it can calculate with.

Collecting the numbers

We gathered them in four parallel research passes: spider mites and thrips, mealybugs and scale, fungus gnats and root rot, and leaf diseases with the sprays people actually have at home. Each pass had the same rules. Open every page you cite, and drop any page that doesn't load. Find at least two independent sources for each life-cycle number. Write down the range the sources give, the value Fern should use, and one line on why. When sources disagree, keep the slower number, because a course a few days too long is the safer mistake.

The result is 105 sources: 59 extension pages (UC IPM, UF/IFAS, NC State, UConn, Purdue, Clemson, UW–Madison and others), 28 peer-reviewed studies, 8 product labels, 5 regulators such as the US EPA and NPIC, 3 developmental databases and 2 pages from biocontrol companies, which we used only to cross-check. Everything lives in one file in our backend, facts.json, which ships with every deploy and is checked when the server starts. If a number cites a source that isn't listed, the tests fail.

The disagreements were the useful part. Ask how long a spider mite egg takes to hatch and extension sheets say about 3 days. Purdue says 3 days at 24 °C. A 1969 life-table study by Laing measured 6.7 days. Both are right. Laing's mites lived on a daily cycle from 15 to 28 °C, which averages close to a living room, and the 3-day figures come from warm greenhouses. At 21 °C, the answer is about 7 days.

The same thing happened with the temperature below which mites stop developing. A 2017 study put it at 11.6 °C under constant temperatures and 8.6 °C when days and nights swing. Other studies gave 10.6 and 12.5. We kept 11.6, because a higher threshold predicts slower mites in cool rooms, and slower means longer, which is the safe side.

Turning facts into a plan

With the facts in place, the plan follows a handful of rules. For spider mites at 21 °C, eggs hatch in about 7 days and newly hatched mites start laying about 12 days later.

The gap between sprays is one day short of that laying time, so 11 days, and never closer than the product's label allows. Neem oil's homeowner label says at least 7 days apart, so 11 is fine.

How many sprays depends on how bad it is. A few mites get 2, lots get 3, webbing everywhere gets 4. A spray never gets every mite, and the extra rounds catch the young of the ones that survived.

After the last spray come two checks, spaced an egg's hatching time apart, so 7 and 7 days. That's long enough for any egg the sprays missed to hatch and show itself.

Put together, Monty's course is 3 sprays 11 days apart, then 14 days of watching: about 37 days, 23 of them spraying and 14 watching. With a few mites it's about 26 days, and with webbing everywhere about 48.

Some of those rules come from the literature and some are ours. "Two sprays for a few, three for lots" is a judgement call. So is "one day before they can lay" and "two clean checks to finish". Fern keeps those in a separate section of the facts file, and the app never presents them as science.

The temperature changes everything

Development speeds up roughly in proportion to how far the temperature is above that lower threshold. It's the same degree-day idea growers use to time spraying in orchards. At 11.6 °C mites barely develop. At 21 °C they're 9.4 degrees above it. In a warm room at 28 °C they're 16.4 degrees above it, so everything runs a little under twice as fast.

In a warm room the eggs hatch in about 4 days and new mites lay after about 7. The gap would be 6 days, but neem oil's label says at least 7, so Fern uses 7. Since that's now as long as the laying time, some mites will lay before the next spray, so Fern adds a fourth spray to catch their young. The watching period shrinks to 8 days because eggs hatch faster. The course comes out at about 30 days.

In a cool room at 16 °C the opposite happens. Mites crawl along so slowly that the gap stretches to 25 days and the whole course to about 81. That looks strange, and we'll come back to it.

Fern doesn't know the temperature of your living room. It assumes about 21 °C for plants indoors, and it says so: on the screen that explains the plan, the room temperature carries a dashed "Assumed" tag and a Change button. Pick "Warm room, about 28°C" and the plan redraws before you commit: "With a warm room: about 30 days, was 37". For plants outdoors, Fern can use the coming week's forecast instead.

Where the first rule broke

Spider mites fit the rules neatly, because every stage after the egg feeds in the open where a spray can reach it. Mealybugs don't.

A citrus mealybug takes about 59 days at 21 °C to go from hatching to laying its first eggs. "One day before they can lay" would have meant spraying every 58 days, which is plainly wrong. The reason is that a young mealybug, a crawler, is only easy to reach for its first few days. Once it settles, it covers itself in wax, and soap, oil and alcohol struggle to wet it. The window that matters is how long the young stay reachable, not how long they take to grow up.

So the rule grew two parts. The gap is the shorter of "before they can lay" and "while the young are reachable". And the sprays have to cover the whole time new young keep appearing. A female mealybug keeps laying into her waxy sac for 10 to 20 days according to UC IPM, and about 22 days in one lab study, and the sac protects the eggs the whole time. With rubbing alcohol on a cotton swab every 7 days, that comes to 5 rounds and about 45 days.

No source we found times the crawler window itself at room temperature. For mealybugs and scale, Fern uses the repeat intervals extension services recommend, 7 and 10 days, and the facts file says that's where the number came from.

Scale is the hardest case. Brown soft scale gives birth to live young and keeps releasing crawlers for weeks, 22 to 32 days at 25 °C in the lab and longer in other reports. Covering all of that with a contact treatment needs 8 rounds. Fern caps a course at 90 days, so it suggests 5 and says plainly on the plan that the cap cut it short. That makes a scale course about 81 days. A systemic product, which the plant takes up through its roots so feeding scale insects get it wherever they hide, means 2 applications 8 weeks apart. The plan mentions that as an option, without pushing it.

Diseases don't have eggs

Pests run on a life cycle. Leaf diseases run on spores and wet leaves, and treating them is closer to protecting new growth than to breaking a cycle. So Fern plans them differently.

For powdery mildew, the treatment is a protectant spray every 7 days, the short end of what UC IPM and the product labels give when the disease is active, kept up until no new powder appears. After that comes a clean stretch. The Connecticut Agricultural Experiment Station recommends scouting weekly until plants have stayed mildew-free for at least three weeks, so Fern waits 21 days without new powder. The course is at least 36 days, and "at least" is honest: any new spot restarts the 21 days.

Leaf spot was harder, because no source we found gives a rule for when it counts as cleared. So we worked one out from how long infections take to show. On ficus, bacterial leaf spot formed lesions 7 to 10 days after infection in a UF/IFAS study, and one fungal spot took 10 to 14 days. Fern waits 28 days without new spots, at least two incubation periods, and the plan says that number is derived. Copper helps against the bacterial kind, but it won't heal spots that are already there, and the plan says that too.

Root rot doesn't get a spray schedule at all. UW–Madison's plant clinic doesn't recommend fungicides for root rot on houseplants, and the RHS says much the same. The fix is physical: take the plant out of its pot, cut away the soft, dark roots, repot into fresh mix in a clean pot with a drainage hole, and water only when the top of the soil is dry. After that Fern checks in weekly for 42 days, looking for new wilting and new growth. That window is our judgement. The only timeline we found was a houseplant grower's: several weeks, sometimes a few months.

The drench Fern won't suggest

Search for fungus gnats and you'll be told, again and again, to water your pots with diluted hydrogen peroxide. We expected to put it in the gnat course. When the research pass looked for evidence, it found none for gnats at all. The closest was a volunteer Master Gardener answer.

For root rot it found something worse than nothing. In a 2021 trial at Michigan State, a drench of hydrogen peroxide with peroxyacetic acid didn't reduce Pythium root rot on geraniums, and the treated plants rated worse than untreated ones. A hydroponic study saw root injury at concentrations up to 400 mg per litre. The usual home mix, one part 3% peroxide to four parts water, is about 6,000.

So the gnat course uses Bti, a bacterium sold as Gnatrol or Mosquito Bits, that kills the larvae feeding in the top few centimetres of soil. It only stays toxic for about 2 days according to UConn, and the Gnatrol label says weekly, with at least three drenches. Eggs, pupae and adults escape it, so the drenches have to cover a whole generation, about 4 weeks at 21 °C. Along with letting the top of the soil dry between waterings, which is still the most effective thing anyone can do about gnats, that's 5 drenches and about 41 days. Hydrogen peroxide is still on the list of products people can pick for themselves, because it's their plant. Fern just never suggests it.

Showing the working

None of this would matter much if the app just printed "About 37 days" and asked to be trusted. So next to the length there's a "Why" link, and it opens a sheet that shows the whole argument.

It starts with what Fern used. Each input has a small tag saying where it came from: a source, with the study's name on it, or "Assumed", "You said", "From your photo" or "Label". For Monty, "Eggs hatch: In about 7 days at 21°C" carries the tag "Laing 1969". Tap it and you get the fact in plain words, what Fern used it for, and a link to the paper: "Measured on a daily 15 to 28 °C cycle: eggs hatched in 6.7 days, the young took about 10 days to grow up, then 2 days before laying."

Then comes how it adds up, three steps with one reason each: "Spray every 11 days. Before newly hatched mites can lay eggs." No formulas on screen. Then the total, "About 37 days", with a bar split between spraying and watching. Then when it ends, and at the bottom one line: "Guidance, not a rule. Follow your product's label."

You can change the assumptions and Fern redoes the sums in front of you. If you set your own spacing closer than the label allows, Fern keeps your choice and says, without any red, that the label says 7 days.

Checks that change the plan

A plan worked out in advance is still a guess, so the checks ask a question. Before each spray, the course page asks "Mites on Monty?" with three answers: Pests, Fewer, None.

Pests means as many as before, or more, and Fern offers one more round: "Mites on Monty: one more round. +1 spray, +11 days." There are two buttons, Add and Keep as is, and either one is fine. Fewer means the plan is working, so nothing changes. Two clean checks in a row after the last spray brings up "Looks clear — finish?". Fern won't offer to finish before the last spray, even after two clean checks, because finishing early skips the watching period. That's the part that catches the eggs.

Every change goes into a short history on the same sheet, such as "you saw pests → one more round", along with the new length. A course started on a Saturday and stretched twice still explains, weeks later, exactly how it got to its length.

What we left out

Fern uses Gemini to look at a photo and name the problem. It doesn't let it set a schedule. Treatment intervals for something you spray on your plants should come from sources a person can check, the same answer every time, and the facts file is that.

We don't read product labels yet. "What are you using?" offers product types such as neem oil, insecticidal soap, horticultural oil and systemic granules, each with typical label limits. A photo of your own bottle would be better, and it's on the list.

Fern also doesn't measure your room. It asks, and it marks the guess as a guess.

What we still don't know

Most of the life-cycle numbers come from greenhouse and lab studies on crops, not from a monstera on a bookshelf. Mites on eggplant leaves at a constant 21 °C are a reasonable stand-in for mites on a houseplant in a heated room, but they're a stand-in.

The slower-number rule makes some courses longer than the guides most people read. Thirty-seven days for spider mites is longer than the "spray for a couple of weeks" you'll find on many sites. We think the longer course is right, but we haven't proven it. The 81-day plan for a cool room is the one we're least sure of. It's correct by the numbers, because mites barely develop at 16 °C, but it may be too cautious in practice.

The good news is that we'll find out. Every finished course records how it ended: all clear, not better, or didn't make it. Each check records what was seen on each plant. Once enough courses have run, we can compare the lengths Fern suggested with how long it actually took to clear, by problem, product, season and temperature. Where the data disagrees with the facts file, the file will change, and the Why sheet will say so.