Expert response from Community Manager
Friday, 02/10/2015 10:30
You asked about glyphosate effects on gut microbes and specifically call out three studies as providing evidence of disruption. Folta’s answer has some good points but was from a broad perspective, so this answer will be a longer and a more technical answer than you might see typically on a site like this. First, I’ll make some generic comments and then some comments about the specific studies you cited.
Currently, gut microbes are one of the hottest topics in biology. There is no doubt that they are important, but we need to also understand that microbes are dynamic and a change does not necessarily mean that there is a health impact. We’ve known for a long time that there are differences in microbiome due to an individual’s personal physiology and differences due to diets. Many animal models exist, including rodents, but ruminants (like cows and sheep) can be a good model for studying gut microbes, because much early information came from them and most of the gut microbes in ruminants and the human gut are anaerobes, meaning they do not survive in an oxygen rich environment. It takes special skills or equipment to culture these. Also, ruminant microbes are more accessible allowing researchers to take frequent samples and determine temporal changes in great detail.
But scientists also need to determine how things work or how they are affected by the environment by eliminating noise, or all of the confounding factors that make conclusions about results difficult. This is often done in a test tube (in vitro) while recognizing that it only is part of the complexity that exists in the whole animal (in vivo) in a real world exposure. For instance, if I want to look at the impact of drinking coffee on gut microbes, I can add coffee to a microbial culture in a test tube. If it contains a single species, this is known as a mono-culture. I might find out that certain species don’t tolerate this well and with further experiments I could determine if the effects might be due to pH changes (acidity of the coffee) or due to fatty acids from cream, or due to a chemical in the coffee such as caffeine or that dreaded pumpkin spice. I could also do the same experiment with multiple species of microbes, known as a mixed-culture, and I may get a different result. That’s because each bacteria in the culture has different requirements to grow and fills a separate niche making the total metabolic activity in that tube much more complex. To paraphrase a professor of mine, “one microbe’s waste is another microbe’s food”.
There have been numerous long-term rodent studies with glyphosate fed at large amounts. If it had an impact on health, as a result of effects on gut microbes or through another mechanism, these studies are a good way to demonstrate if there is actually an impact on health. We’ve talked a lot about numbers of animals and strains of rodents on this site, but a very important thing to understand from these studies is the number of endpoints that are measured and the detailed observations. Rodents are important model animals in gut microbe research, which is well illustrated by the studies where lean mice are made obese by fecal transplants from obese humans. One unfounded claim is that glyphosate changes microbes in the gut and the pathological sequela would be gut inflammation; however, gut histology from rat toxicology has not indicated this to be an issue.
There are two studies that have examined effects on gut microbiology and microbial function that use mixed-population approaches. One is in vivo, and the other is a more sophisticated in vitro study. The first study  was an in vitro study that used a specialized apparatus for maintaining a normal ruminal microbial population under controlled conditions for a long period . The German Federal Institute for Risk Assessment (BfR) commissioned a study with public money using this technique to investigate two objectives. First, to determine if quantitative composition of ruminal microflora or ruminal metabolism might be altered with glyphosate. Second, to determine if there is evidence of C. botulinum overgrowth (i.e., did treatments favor growth of C. botulinum over other microbial species). To meet the first objective the effects of a formulation of glyphosate containing a tallowamine surfactant on rumen fermentative parameters were studied. No major changes in rumen parameters were detected except slight decreases in ammonia nitrogen (due to microbial breakdown of amino acids) concentrations and increases in isovalerate (due to fermentation of organic compounds) production in response to the higher dosage. There was an increase in Bifidobacterium spp. (generally considered beneficial) but the Clostridia were not affected. In the second trial, growth of C. sporogenes (inoculated to the rumen fluid as a surrogate for C. botulinum) did not affect the resulting Clostridia community.
The second study was an in vivo study that used rumen cannulated sheep fed diets with formulated glyphosate added at the most conservative scenario based on the highest glyphosate residues determined in grass three to eight days after application . An additional treatment had supplemental aromatic amino acids to see if they would reverse potential effects of glyphosate, assuming they would be limiting for bacterial growth. There was no indication that the rumen microbes were affected with or without aromatic amino acids based on rumen pH, NH3-N, and VFAs. Additionally, in situ digestibility of NDF and DM were measured using Dacron bags suspended in the rumens and there was no effect of treatments. These latter endpoints indicate that rumen function was unaffected by Roundup®.
These two studies share some features that more accurately predict real world effects. As stated previously, both use mixed populations of microbes. They also used a continuous (the sheep study) or semi-continuous (the Rusitec study) turnovers. In human and animal intestines, there is a fairly continuous influx of nutrients and there is a similarly steady outflow of gut contents, including microbial cells. This results in the population of microbes being in a steady state condition. This is important not only because it affects microbial responses to treatments, but it eliminates concerns about when you take samples. You could run a system like this for indefinite periods and take samples days or weeks after initiating the experiment. Most important, these studies, which are good models of real world conditions, do not indicate that there are meaningful effects of glyphosate on gut microbes.
What about the papers you cite as “evidence of disruption”?
Poultry - Shehata et al., 2013. Simply put, there are no samples from a chicken given Roundup in this study ; therefore, there is no evidence of disruption. They grew in vitro monocultures of selected bacterial species and measure their “growth” when formulated Roundup was added to the culture media. At best, these results can demonstrate a potential but they don’t do a good job of that. The study is difficult to interpret because they leave out a lot of detail about how the study was conducted. Growth is determined from a single point in time at 48 hours. That time is probably too long, because the bacteria eventually run out of nutrients and die in their own waste products, one cannot ascertain if the value measured is from growth increasing or decreasing. This is very important since growth rate of bacteria has a major influence on the outcome of all such tests. Brown et al.  stated that “an under recognized but major determinant of such physiology is the rate of cell replication”. Shehata et al. suggest that Roundup differentially affects good bacteria vs. bad bacteria and this assumes that good and bad are defined by a susceptible EPSPS enzyme for which there is no basis. There are several examples among their small selection of organisms of a “bad” bacteria being more susceptible than a “good” bacteria and that also assumes that a bad bacteria is always a bad bacteria (not such a great assumption).
Cattle - Krger [sic.] et al., 2013. According to their introduction, the reason the study was done was due to farms in Germany that were suspected of having a rare form of visceral botulism. Normally, cattle get clinical botulism from ingesting a toxin (BoNT) produced by Clostridium botulinum, not due to infection by the organism. Their theory was that due to dysbiosis there is an overgrowth of C. botulinum in the affected cows resulting in BoNT being produced and absorbed from the gastrointestinal tract. No reason was identified and they were hypothesizing, without direct data that this clinical outbreak may have been due to glyphosate. Seyboldt with a team of veterinarians investigated this situation . They sampled 1388 animals from 139 farms in these region and found no evidence that any animal had NoBT in its feces and they considered fecal detection of NoBT to be a requirement of the hypothesis. Kruger et al.  did not study any animals or obtain any samples from animals suspected of clostridia or even dysbiosis. Instead, they did testing of mono-cultures to which glyphosate or formulated Roundup was added to the culture media. They then compared the minimum inhibitory concentrations (MIC) of Roundup or glyphosate added to culture media for mono-cultures of a small number of bacterial species. They found a small difference between the MIC for C.botulinum vs. E. faecalis but these MICs are greater than glyphosate intake and are not informative of similar effects in vivo or on resulting health impacts. These conclusions are a big leap of faith. Additionally, they don’t rule out if these effects are due to surfactants (formulation) or pH (glyphosate), either is a possibility. It is noteworthy that if you read the abstract, they don’t present any data whatsoever from the study and carefully limit their conclusion to glyphosate possibly contributing to a health malady that had not even been shown to occur.
Swine - Carman et al., 2013. This study  was addressed previously on GMO Answers. There are no measurements related to microbiology in this paper. The authors discuss “red” stomachs that are not statistically different when analyzed using all of the data. There were no cultures to show infections and no tissue biopsies to show inflammation. Moreover, they did not report if the pigs in this study ever consumed glyphosate. But even assuming for the sake of argument that they did, the authors state that there was no difference in body weight gains between test and control animals, suggesting there were no health effects. Other signs of clinical illness were not reported.
These studies share that fact that there was no indication among any data of an abnormal health finding related to glyphosate. The information that led to the question answered by Dr. Folta is taken from a series of correlations between glyphosate use and various diseases. The disease data are then correlated with various purported causes in the scientific literature. For an example of how misleading such correlations can be, see http://rounduproulette.com. Seneff’s work is based on supposition derived from correlations identified using data obtained from disparate sources and there is no actual demonstration of real world exposures of glyphosate to back her claims.
1. Riede S, Schafft H, Lahrssen-Wiederholt M, Breves G (2014) Effects of a glyphosate-based herbicide on in vitro ruminal fermentation and microbial community with special attention to clostridia. Proc Soc Nutr Physiol 23: 34.
5. Brown MRW, Collier PJ, Gilbert P (1990) Influence of growth rate on susceptibility to antimicrobial agents: modification of the cell envelope and batch and continuous culture studies. Antimicrobial Agents and Chemotherapy 34: 1623-1628.
8. Carman J, Vlieger H, Ver Steeg L, Sneller V, Robinson G, et al. (2013) A long-term toxicology study on pigs fed a combined genetically modified (GM) soy and GM maize diet. Journal of Organic Systems 8: 38-54.
How GMOs are Researched, Developed and Tested
Learn how plant biotechnology works through the scientists who research, develop and test GMO crops at Dupont.