This will be a long, and long-winded, post, so be warned.

The purpose of this post is to talk about Science and to a lesser extent the, shall we say, stupidity of human nature. As a bit of background: I majored in biology and I’m well versed in the scientific method.

Over the past several decades, Japanese beetles (Popillia japonica) have become more and more of a problem in this area, and many others. The adults feed on a wide variety of ornamental plants and crops, skeletonising leaves and destroying flowers. The larvae feed on grass roots (and the roots of other plants), causing unsightly yellow patches in lawns. In my garden, they are particularly annoying on the roses and (more recently) the witch hazel (Hamamelis virginiana). I don’t have much of a lawn – I dug up the useless waste of space to put in a real garden – so the grubs don’t bother me much. They (the adults) are also fond of feeding on Virginia creeper (Parthenocissus quinquefolia). Since I planted one the beetles are slightly less pernicious on the roses, although that doesn’t solve the problem as such. But I ramble.

In searching for organic ways to control (or better yet, eradicate) Japanese beetles in my garden, I found multiple claims in various Internet forums and websites concerning the plants commonly known as four o’clocks (Mirabilis jalapa). These claims ardently allege that Japanese beetles are highly drawn to feed on four o’clocks, whereupon they die, presumably because the plants are also toxic to them.

I also found many counterclaims saying that four o’clocks had neither effect on Japanese beetles. I was intrigued – of course, any intelligent person knows that the Internet is full of misinformation, half-truths, and outright hoaxes, but I wanted to consider the possibility that there was some truth to the claims. Wishful thinking and all that.

So, I decided to try an experiment.

Let’s talk about experiments. Experiments are a key component of the scientific method, basically consisting of manipulating conditions in a controlled manner to see the resultant effects on the subject. Many people, and I must admit that includes me, use the word ‘experiment’ very loosely, i.e. trying something just to see what happens. However, a procedure must fulfil several criteria to be considered a scientifically valid experiment.

First, there must be a hypothesis. This, in a nutshell, defines the purpose of the experiment: it is the question that the experimenter attempts to answer, the idea that he/she/it (I’ll stick with ‘he’ from now on) seeks to confirm, or the statement that he wishes to prove. Note this it is very difficult to scientifically disprove something; in order to do so, one must completely exhaust all the possibilities and show them to be invalid. Hypotheses can be very simple or very complex, possibly containing multiple corollary or auxiliary hypotheses. They often also include or are based upon a number of assumptions.

For my so-called ‘experiment’, the hypothesis would be, “To determine if specimens of Mirabilis jalapa have an especially attractive effect on Popillia japonica and a lethally toxic effect on such individuals that feed upon them.” In this case, there is an implied assumption that there is at least the possibility of some truth to the Internet claims – otherwise I wouldn’t be bothering. There is also an assumption that, if the claims are true, the plants must be producing some substance(s) that attract and then kill the beetles upon ingestion.

An experiment must also contain a certain sample size. If just one beetle feeds on just one particular plant and dies, that only proves that that beetle fed on that plant and died. Once is once and twice is coincidence, as the saying (mangled) goes. Increasing the sample size (experimental subjects) increases the validity of the results.

An experiment should contain a control group. This is a subset of your test subjects that is not exposed to the specific condition that you are trying to investigate. The purpose of this is to provide an in-house (as it were) comparison of your results. If one is investigating if beans grow better if one plays music to them, the control group would consist of the same type of beans grown under identical conditions except they are not exposed to any music. In this case, the hypothesis might make allowances for various types of music, so the experiment might include different groups of beans being exposed to different types of music – but only and always the same type of music for a given group, plus the control group that is given no music.

The results of an experiment must be objectively obtained, and objectively recorded. What the experimenter thinks or believes is irrelevant; what the experimenter measures and records is what matters. It’s fine to say, for example, “Beans in group A grew a mean average of five centimetres in height during week 1.” Saying, “I think the beans in group A are looking nicer than the ones in group B,” is useless.

An experiment should be replicable. You, or better yet someone else, should be able to exactly copy your method and obtain a similar or identical result. This goes back to the same concept underlying sample sizes: once is once and twice is coincidence. If someone does copy your procedure and consistently gets a different outcome, the validity of what you originally did comes into question. There are instances where an experiment might not be replicable, such as when the conditions are highly dangerous, rare, or unique, but these are exceptions and must be regarded as such.

Once the experiment is complete, the carefully recorded results are organised and analysed. I won’t go into the black art of statistical analysis – it gives me a headache and makes me feel unclean. As the saying goes, there are lies, damned lies, and statistics. But the results of the experiment will determine whether or not the hypothesis was valid; they might alternatively show that the entire experiment was a flop because of poor methodology or mistaken assumptions. Such is life and such is science. Success or failure (proof of the hypothesis or lack thereof), merits and drawbacks of the procedure, and logical deductions are all discussed, then conclusions are drawn. Modifications to the experiment may be required, and the hypothesis might be refined and tested again and again in a whole series of experiments.

Back to my garden. Given the above, what I did would never count as a scientifically valid experiment; at best it might be considered a field test. But here is what I did.

  1. Grow four o’clocks. I hadn’t grown them before; I mostly try to grow native species for ornamentals (roses being a major exception), and Mirabilis jalapa is native to tropical South America. They are not winter-hardy in Ontario. They are evergreen or perennial in warmer climates, but here they are usually grown as annuals. There is a related native species, M. nyctaginea, but when people talk about four o’clocks they mean M. jalapa. I bought seeds of a variety called ‘Sunshine Yellow’ and sowed them in spring (this all occurred years ago, by the way). I actually planted them in a large pot, partly so I could move it to various spots around the garden, and partly to see if I could overwinter them in a frost-free place. There ended up being half a dozen plants.
  2. Wait for the plants to grow and see what they do, if anything, to Japanese beetles. The beetles show up in mid-late June in most years, so the plants had some time to grow before being put to the test.

And, well, nothing happened. I saw no hordes of Japanese beetles descend in a suicidally voracious horde upon the four o’clocks, and found no masses of dead beetles on the ground around them. As usual there were swarms of them on the roses. There was some very minor feeding damage on the four o’clocks, but again, no evidence that the beetles responsible paid with their lives. Oh well. At least I tried, and at least I wasn’t any worse off than before.

But why didn’t it work? Well, several possibilities come to mind. The following are dependent on the initial assumptions being correct (the claims are valid, and the plants produce substances that attract and kill the beetles).

  1. As I said before, M. jalapa is native to a tropical habitat, and evergreen or perennial in warm climates; they are grown as annuals here. Possibly the active substances are only produced by plants of a certain age or size, and my first-season plants weren’t old or big enough.
  2. Many experienced gardeners can attest that plants grown in containers often perform differently than those grown in the ground. Perhaps my pot-grown four o’clocks produced no toxin, but would have if grown in the ground.
  3. I certainly didn’t keep track of the places where the people who made the claims lived and compare them. Possibly the claimants live in similar environmental zones and the plant toxins are only produced in response to specific environmental triggers in such zones. Possibilities could include a certain temperature for a certain length of time, day/night length, latitude (which affects sunlight intensity), or soil chemistry. The trigger might even be another organism, such as another plant, bacteria, fungus, insect, or anything that lives in such climates but not mine.
  4. Related to the preceding point, humans are fond of making broad general statements based on very limited observations (‘evidence’), and assuming that what applies here and now also applies everywhere and everywhen. However, every garden is different, even ones next to each other. There is no guarantee that what works in one garden will work to the same degree (or at all) in another.
  5. Four o’clocks have been cultivated in gardens for hundreds of years and it is a fact that when cultivars are bred and selected for certain traits, other traits are often lost. Possibly the variety I chose (‘Sunshine Yellow’) has lost the ability to produce the toxin.
  6. Remember that the hypothesis contains two separate components: that four o’clocks are both highly attractive and toxic. What if the plants are individually only mildly attractive and it takes a large biomass (i.e. lots of plants) to really draw the beetles in? Or what if they’re only mildly toxic and the beetles need to ingest quite a bit before they die? Or both? In this case the claims are based on a grain of truth, but exaggerated. People are happy to see what they want to see, and authors are happy to word things in such a way as to support their views. Politicians do it, magazine and newspaper writers do it, marketers do it. ‘Researchers suggest’ is NOT AT ALL the same as confirmed fact, but is often portrayed as such. Possibilities are not necessarily realities. This is (one way) how garden myths get started. Remember also that controlled conditions (as in a lab experiment) are not present in a garden, where any number of variables hold sway .The preceding points should amply illustrate how important it is to control the variables to get satisfactory results in an experiment.
  7. It might be any or all of the above, or it might be none. It could be any number of things that I haven’t thought of and would never occur to me.

There are other possibilities of course: namely that one or both of the basic assumptions are wrong.

  1. Assumption One is wrong, which also invalidates Assumption Two for all intents and purposes. As I said, the Internet is full of misinformation, and the claims might have been made by people who were mistaken about any number of things or just outright out to spread lies. It could even be one person under a number of different screen names.
  2. Assumption One is right, but Assumption Two is not. What if it isn’t the plants that are killing the beetles at all? What if the toxins are produced by some bacteria or fungus or other microorganism that lives on four o’clocks and happens to be ingested when the beetles feed on the leaves? What if the ‘toxin’ is actually a toxic reaction in the beetle’s gut that only happens if the beetle feeds on something else before moving on to four o’clocks? What if it rains chocolate doughnuts?

Thus, based on the outcome of my test (not experiment), it would be misleading to make assertions as to whether or not four o’clocks kill Japanese beetles. Unlike many people, I try not to make general claims without hard evidence, and both my scientific training and honesty disincline me from portraying personal opinion as fact. I cannot say, “OMG it works best thing evar,” or, “Oh noez total scam don’t waste ur tiem.”

So, do four o’clocks kill Japanese beetles or not? I simply don’t have enough confirmed information to make an assertion that applies to all or most gardens. I can say that the four o’clocks I grew, under the conditions I provided, were neither unusually attractive nor toxic to Japanese beetles. Or to put it another way, it didn’t work for me.

But they did smell nice.

As an aside, I have grown four o’clocks every year since then, same variety. I’ve even managed to overwinter them to regrow in successive years. The results regarding Japanese beetles have never changed.

Oh, there’s another aside: I’m personally inclined to doubt the claims have more than a small grain of truth. As a biologist and a gardener I’ve run into too many myths that promise such an easy quick fix.