Derek Lowe has a sensible article about glyphosphate, the herbicide otherwise known as Roundup. Glyphosate is scary: it’s a chemical, don’t you know, and it kills weeds, so who knows what it’s doing to your children and your cats; even scarier, some crops are being genetically modified to be resistant to glyphosate, and then proteins that protect against Roundup might end up in your cornflakes.
And now some people are raving that glyphosate causes autism, because of course every chemical compound that they don’t understand causes behavioral problems that we don’t like. We must have a scapegoat. It doesn’t matter that it has never been found to have an effect on humans.
An extensive scientific literature indicates that glyphosate is specifically not genotoxic, is not a carcinogen or a teratogen, nor has any specific adverse health effect ever been demonstrated to have been caused by exposure to or low-level consumption of glyphosate. It has little effect on non-target organisms other than plants; a contributing factor to this is that glyphosate inhibits an enzyme found in plants. This enzyme is not found in humans, other mammals, birds, fish, or insects.
The use of glyphosate on herbicide tolerant crops has proven problematic to anti-GMO activists since adoption of the technology promotes the switch to a chemical with a lower environmental impact quotient and lower toxicity.
Lowe explains the statistical nature of risk, and the cautious style of chemical classifications, that allows almost any chemical to be judged as risky to some degree, and feeds sensationalist misreadings. All the data really seems to be saying that it does nothing to animals, but let’s cover our bets and keep an eye on it.
I even have an anecdote about Roundup. We tried to see if it has any early teratogenic effects. The results are sadly unpublishable (for very bad reasons) so it’s safe to summarize them here.
We have a simple assay for developmental errors. Zebrafish pop out a bunch of eggs every morning when the lights come on, and we clean them up and separate them out into beakers, with about 100ml of water for 50 embryos. For our controls, we use fish tank water, the same stuff the adults are swimming around in. It’s got fish pee in it, bacteria, fungal spores, even tiny invertebrates (check your home aquarium water — would you drink it?). We use that because it does have some challenges for growing embryos, and provides a good background for comparisons — we lose 5-10% of the embryos, usually to fungal growth, in these situations.
I had a student who wanted to test local water sources for potential teratogens. So they collected jugs of water from nearby ponds and streams, which are rich with agricultural runoff. We then grew embryos in simple, unfiltered water from Lake Crystal, or the Pomme de Terre river, or nearby ponds, just to see if we had any preliminary effect worth pursuing. This is why we use crude tank water for the controls — those sources would also be complex and biologically rich.
Here’s the boring result: nothing happened. Fish grew happily in water from a shallow pond full of duck poop with an ethanol plant on one side and a dairy farm on the other, with no detectable disorders or effects on the rate of development. In fact, the pond water embryos were healthier in one sense — they had reduced mortality from fungal infections than embryos in tank water. Tentative explanation for that: tank water might specifically be a breeding ground for fungi that thrive on fish, or the fungi might be more sensitive to agricultural chemicals than the fish are. Anyway, it was a negative result.
Then we thought to push it, and see if we could get any deleterious effect from those agricultural chemicals, so I bought a gallon of Roundup at the hardware store, and we did a dilution series. Nope, nothing. We had embryos growing in a concentration of several percent glyphosate, and they didn’t seem to mind at all. We used concentrations that were approximately ten times what Monsanto recommends that you spray directly on your lawn, and the zebrafish didn’t care.
Now of course this was a limited and preliminary experiment. All we were examining was survival and basic morphology, and we were only looking at early developmental events, like gastrulation and neurulation and the earliest twitching behaviors, and we can say with some confidence that those were unaffected. We did not look at older animals, so if it were an endocrine disruptor (it isn’t) for instance, we wouldn’t know it. We also don’t have a test for fish autism.
I can also say that I wouldn’t drink glyphosate, but not because I’m afraid it would give me cancer. It’s because the straight stuff is kind of oily and smells nasty. So those stunts where people give Monsanto executives a glass of Roundup and dare them to drink it are really misleading — they’re not going to drink it because concentrated-just-about-anything is unpleasant.
I think the bottom line is that making a claim about the deleterious effects of a substance requires actual data, and not cherry-picking suggestive and vaguely defined effects.
It also says that the file drawer effect is a problem. I suspect there have been lots of preliminary experiments that see nothing, and are abandoned as unpublishable, like ours. That effect is also complicated. You can’t tell me just to take the data we got and publish that, because it really was just a quick pilot experiment to see if there was something worth pursuing. It’s not just that a negative result is unpublishable, but that we didn’t see enough of an effect to make it worth our while to invest enough time and effort to make the results thorough and robust enough to even consider getting it into publishable shape. And thus science staggers on.