Dead plants are probably bad for earthworms

Written by Andrew Kniss. Editor’s note: This article was originally posted at the Weed Control Freaks blog.

An earthworm. Source: Wikimedia Commons.

I’ve gotten a number of questions recently about a new study titled “Glyphosate-based herbicides reduce the activity and reproduction of earthworms and lead to increased soil nutrient concentrations” that was recently published in the online journal Scientific Reports. Although the title seems pretty straightforward, there are some flaws in the study design that preclude any broad conclusions from this research. I would especially be cautious of making statements like Roundup is “imperiling streams, lakes, [and] aquifers” or that there are “far-reaching consequences of [glyphosate] use in ecosystems” based on this work. In the abstract the authors state:

In a greenhouse experiment, we assessed the impact of the most widely used glyphosate-based herbicide Roundup on two earthworm species with different feeding strategies.

Except, they didn’t. At least not directly. It is certainly plausible that herbicides (glyphosate or others) might have some direct effect on earthworms, so it seems like a reasonable study. However, because of the study design and methods used in this research, none of the effects they observed can be attributed directly to glyphosate. There are two main issues:

  1. One of the herbicides applied in the study was “Roundup Speed” which contains the herbicide pelargonic acid in addition to glyphosate, so it is impossible to conclude anything about the direct effects of glyphosate.
  2. More importantly, the researchers didn’t include a control treatment where they killed the plants without herbicides. All of the effects on earthworms and nutrients observed in this study could simply be due to killing the plants in the pots. It is perfectly plausible the exact same effects would be observed if the plants were clipped or pulled out of the pots.

First, some comments on the herbicide applications in this study. In the methods section, the authors state (I’ve edited out some parenthetical information about the herbicide concentrations to make it easier to read):

“Eight weeks after planting, mature plants (D. glomerata was about 40 cm high, T. repens 19 cm, T. officinale 31 cm) of half of the mesocosms were treated with the herbicide ‘Roundup®’ (treatment +H), whereas the other half of the mesocosms remained untreated (treatment –H). Each +H mesocosm was sprayed with 7.2 ml of ‘Roundup® Alphée’ on two consecutive days (in sum 14.4 ml), and 10 ml of ‘Roundup® Speed’ two days afterwards.”

There are a couple items of note here. First, the researchers made three separate herbicide applications over a 4 day span. This seems very odd. I’m speculating here, but my guess is that when they were conducting the study, the authors didn’t realize how slowly glyphosate works. It is not uncommon for glyphosate to take 10 to 20 days to kill plants under greenhouse conditions, especially plants of the large size in this study (7.5 to 15 inches tall at application). When they didn’t see immediate results from the first application, they reapplied the herbicide on day 2. Then, after seeing no immediate results from the second application, they went to the store and found a product named “Roundup Speed” which then caused nearly instant results. Again, this scenario is speculative on my part, but would explain the very unrealistic herbicide application pattern they used.

Seed heads of Dactylis glomerata, the plants used in the experiment. Source: Wikimedia Commons

The unfortunate choice to use “Roundup Speed” for the third application brought in a major confounding factor. In addition to glyphosate, Roundup Speed contains a second herbicide active ingredient (pelargonic acid). Pelargonic acid is a very rapidly acting herbicide, and is presumably the active ingredient that puts the “speed” in Roundup Speed. Another formulation of pelargonic acid advertises that “Results are usually visible within minutes after spray application.” So the rapid desiccation of the plants in this study (and possibly the other effects observed) is quite possibly attributable to this compound, and may have nothing at all to do with glyphosate.

But that’s not the only problem with extrapolating these greenhouse results to field applications. I calculated the amount of glyphosate they applied to the pots (adding up the three applications they made) and converted it into the amount of glyphosate per unit area. The amount of glyphosate applied to each pot was equivalent to a field rate of 11.3 pounds of glyphosate per acre (or 12,680 grams per hectare). A typical application rate in a field of Roundup Ready crops would be somewhere between 0.75 to 1.3 pounds per acre. So the amount of glyphosate they applied is about an order of magnitude too high to be relevant to a field situation.

The more important issue with this study was a lack of an appropriate control treatment to determine the effect of plant removal. The authors state in the abstract:

“We demonstrate, that the surface casting activity of vertically burrowing earthworms (Lumbricus terrestris) almost ceased three weeks after herbicide application, while the activity of soil dwelling earthworms (Aporrectodea caliginosa) was not affected. Reproduction of the soil dwellers was reduced by 56% within three months after herbicide application.”

Pelargonic (nonanoic) acid, added to Roundup Speed herbicide, further confounded the results of this experiment.
Pelargonic (nonanoic) acid, added to Roundup Speed herbicide, further confounded the results of this experiment.

I wasn’t able to find any studies that specifically evaluated the impact of plant removal on earthworms, but earthworms can certainly be negatively affected by a reduction in plant species diversity. In this study, the plant species diversity was effectively reduced to zero; they killed all the plants with the herbicide. I’m not an earthworm expert, but I would guess that killing all the plants in an area will have a negative effect on earthworms. Especially if you keep them in the pots for three months. If the authors would have included a treatment where they killed the plants by some other method (like mowing or burning, or perhaps even an application of only pelargonic acid), then they could have isolated the effect of the herbicide. But as it is, there is no way to separate the effects of the herbicide from the confounding effects of dying plants.

In addition to the effects on earthworms, the abstract makes some pretty remarkable claims with respect to nutrients:

“Herbicide application led to increased soil concentrations of nitrate by 1592% and phosphate by 127%, pointing to potential risks for nutrient leaching into streams, lakes, or groundwater aquifers.”

This is a bit of a technical point, but nitrate and phosphate “increases” of this magnitude wouldn’t be possible in this study, since the herbicides they applied wouldn’t contain that much of those compounds. Laws of physics would have been violated to increase these nutrients by this magnitude. What the authors actually mean is that the availability of these nutrients was different between the sprayed and unsprayed treatments. But again, this wasn’t a direct effect of the herbicide, but simply an impact of the plants dying. Using the same logic, the authors could have concluded that herbicide application significantly increased soil moisture, since there was a clear difference in moisture content between the sprayed and unsprayed pots. But this is obviously not the case. Differences in soil moisture were due to having live plants in the unsprayed pots taking up water from the soil, and dead plants not taking up water. It is the same for nitrate and phosphate. Plant matter (roots, leaves, etc.) is made up of soil nutrients (among other things), and therefore when the plant is growing these nutrients are taken up and no longer freely available in the soil. In this study, the plants continued to grow in the unsprayed pots, and so those plants continued to take up nitrogen and phosphorus from the soil. In the herbicide-treated pots, the plants stopped taking up nutrients, leaving more nutrients available in the soil. In addition, as the plants died, it is possible some of the nutrients in the plants may have been mineralized, or returned back to the soil as the plants decomposed.

The herbicide treatment didn’t actually “increase” these nutrients. What the the authors observed is simply a difference in the available nitrogen and phosphate in the soil due to differences in plant growth between the treatments. This is certainly not a novel finding, and would be expected for any study where living plants are compared with dead plants. Which, again, is why a plant removal control treatment (clipping, pulling, eating, whatever) would be necessary to draw any conclusions from this study.

Written by Guest Expert

Andrew Kniss is an Associate Professor of Weed Ecology & Management at the University of Wyoming. He has a PhD in Agronomy with a minor in Statistics. He teaches undergraduate & graduate courses including Ecology of Plant Protection and Weed Science & Technology. His research focuses on developing sustainable weed management programs in agronomic crops.


  1. As Andrew points out, the study has some problems. Nonetheless, it appears that, albeit at very high application rates, there was an effect of the the herbicides on the nightcrawlers, but what about the alternative? Agriculture is not perfect, so we must look at the tradeoffs. If glyphosate is not used, and weeds need to be controlled, what would be the effect of tillage on the activity of these worms? I would guess that it would be much more drastic, and probably affect both worm species. No-till agriculture, even with glyphosate, is known to increase nightcrawler numbers compared to tilled systems.
    Then there is the fact that here in the US, the nightcrawler is an non-native invasive species, which does damage in some forest systems, so couldn’t glyphosate could be looked at as a selective wormicide?

  2. So the herbicide accomplished exactly and no less than the purpose for which it was intended, killing the target plants, reducing the demand on nutrients and water that would be needed were you to have a resistant or subsequently-planted crop? And worms found the elimination of all plants less than salubrious? This seems just to confirm the usefulness of this tool when used for a valid purpose. Raising worms in barren soil doesn’t sound valid.
    That the incredibly excessive application rate didn’t produce more grotesque destruction attests to the safety of the product(s). Salt is probably more dangerous.

  3. Hi there,
    Your blog was recently recommended to me by a colleague who reads it. I am an organic chemist gone chemical biologist and my colleague is a bioengineer gone chemist. So this is not exactly my field of expertise, but I am interested in agriculture and its future, so I will follow your blog from now on.
    Anyway; very interesting criticism of this paper. It is indeed surprising to see this published in -Nature-(!) Scientific Reports, when the control configurations of their study are poorly thought through (as you point out). Another low-quality peer reviewing process?
    Also, would you (everyone) care to speculate why they state that they used “a lower-than-recommended dose of glyphosate-based herbicide.”, when your (Andrew K.) calculations show that they don’t?
    Next, if the integrity of results of the above study are severely impacted by the poor design of their study – what would you say about this paper?
    Effect of repeated low doses of biocides on the earthworm Aporrectodea caliginosa in laboratory culture.
    J.A. Springett, R.A.J. Gray, Soil Biology and Biochemistry, 1992, 24, 12, 1739 – 1744
    Abstract: The growth rates of Aporrectodea caliginosa (Savigny) were measured over a 100-day period in soil in culture chambers which were treated with common biocides singly and in combination. The biocides used were: the fungicide Captan, the herbicide, Glyphosate and the insecticide, Azinphos-methyl. The biocides were applied at intervals of 14 days and each treatment was replicated six times. The results are variable, all biocides depressed growth when applied alone but some combinations reduced the effect of other biocides. Azinphos-methyl and Glyphosate applied alone, reduced growth the most over the 100 days and at all rates of application. Azinphos-methyl applied at the highest rate killed worms. Captan applied alone had the least effect on growth and mortality. In combination, Glyphosate and Captan had a lesser effect than Glyphosate alone. Azinphos-methyl and Captan had less effect than Azinphos-methyl alone. After 100 days the combination of all three biocides reduced growth to the same degree as Glyphosate alone.
    Thanks for your insights and contributions!
    best regards

  4. But this is an artificial system in multiple ways. One of the strangest things, though, is that if we were talking about situations with GMOs, obviously we wouldn’t have plant death, right? So a good control for GMO systems would have been to include those types of plants.
    Maybe this has some insight into non-GMO use of glyphosate, but that’s not what seems to aerate everybody.

  5. Åsmund, I had a look at the paper and there are quite a few things that worry me about it. There is a bit of lack of clarity about how the experiments were done, which makes it hard to interpret the results.
    Firstly, the authors claim that the recommended field rate of glyphosate is 14.4 g/ha. I know of no use pattern for glyphosate that uses so little herbicide. Indeed, such a rate would be unlikely to control any plants. I think the authors must have made a mistake, but then that means I don’t know what their actual applications mean.
    The authors don’t state how they applied the pesticides, what nozzles were used, what water rates were used. This is important as it reflects coverage and if high water rates are used, the could actually wash the pesticides into the soil. The authors also don’t state what brands of products were used. One of the things about glyphosate is that the formulations contain detergents and different formulations contain different detergents and different amounts of detergents.
    The pesticides were sprayed every 2 weeks, which means that 7 applications were made across the course of the experiment. This is completely unrealistic. Glyphosate takes about 3 weeks to work.
    There are also some oddities about the results. For some of the treatments, there was no dose response. The lowest dose was more effective than other doses. Some of the mixtures gave odd results. We know azinphos-methyl is toxic to earthworms, but adding this to glyphosate increased growth compared to glyphosate alone or azinphos-methyl alone. There appears to be an element of randomness about the results obtained.
    As a result of the lack of information about the methods, the incorrect claim about use rates of glyphosate, the unrealistic application timing and the randomness of the results, I don’t know what, if anything, the research tells us.

  6. Hi Chris,
    Thanks a lot for your reply and pardon the slow response. I agree with several of the points you highlight regarding this article, especially the lack of stringence in the chemical characterization of what they are treating the plants with (formulations of glyfosate).
    What is really needed is worm metabolomic data from untreated and treated worms, in paralell with a study that can establish steady state concentrations of glyfosate in soil beneath plants treated with commonly used amounts of herbicide.
    Have a nice weekend.

Comments are closed.