Slide 1-3
^ This is a model of glyphosate, the active ingredient in Round up and similar weed killers. ^

Slide 4-6
^ What do you think glyphosate does to your soil? How about your food?
Howdy folks, my name is Walt Sumner. I am a retired family physician and I’m trying to learn how modern food production affects our health. Today I’d like to show you the results of a glyphosate experiment on my farm, funded by the USDA North Central Sustainable Agriculture Research and Education program. ^

Slide 7-8
Glyphosate has two interesting effects. ^ Glyphosate’s main effect is to block this step that plants and microbes use to make amino acids like tyrosine, tryptophan, and phenylalanine. That makes glyphosate an effective ^ antibiotic against some bacteria and fungi, ^ and an effective post-emergent herbicide against many plants. ^ In fact, one use for glyphosate is to kill an entire seed crop all at once in the fall, so that the seeds dry uniformly before harvesting.

People do not make tyrosine this way. Instead, we get tyrosine from our food, or we make it from a similar amino acid, so glyphosate has no direct effect on our tyrosine supply. We also rapidly excrete unchanged glyphosate in urine, so it does not seem likely to be doing much to our bodies, and it is much safer than the herbicides that it replaced. 

Slide 9
^ Although glyphosate is an antibiotic, some microbes can tolerate glyphosate. ^ There are reports that glyphosate kills several beneficial gut bacteria ^ but not some disease-causing bacteria. Healthy plants also cooperate with soil fungi to get a lot of nutrients, ^ so indiscriminate fungicidal effects could force farmers to use more fertilizer, which is getting expensive. However, the practical importance of any changes that glyphosate causes in gut or soil microbiomes is not settled yet. ^

Slide 10-12
Glyphosate’s second interesting effect is chelation. ^ All of those oxygen atoms in glyphosate tend to be negatively charged, so they cling to any positively charged atoms, molecules, or particles in the soil, especially mid-size metal ions with a plus two charge. Glyphosate binds relatively scarce copper and zinc ions more tightly than smaller and more plentiful calcium or magnesium ions. Plants use a lot of divalent cations that they absorb from the soil. ^ You might wonder if a plant can absorb and use a chelated cation. 

Slide 13
^ People sometimes suggest that glyphosate applied to farmers’ fields chelates so many divalent cations that crops will be deficient in these metals. I doubt that. The top six inches of an acre of Southern farmland should weigh about 2,000,000 pounds and could easily contain three pounds each of ^ copper and ^ zinc, ^ one-hundred pounds of magnesium, ^ and one-thousand pounds of calcium. ^ Farmers might spray as much as one and a half pounds of glyphosate on it at once. That’s a ratio of 3,000 calcium and magnesium ions and 5 ions each of copper and zinc for every molecule of glyphosate. ^ Even if it only binds copper and zinc, 90% of those ions still should not be chelated. Also, farmers are targeting weed leaves, not the soil, so a lot of it is initially tied up in weeds. ^ On top of that, in warm healthy soil glyphosate should break down in a few days or weeks, freeing chelated ions. 

If farmers spray glyphosate on a crop to dry it for harvest, it could contaminate the harvested seeds. People or animals that eat contaminated seeds will expose their gut bacteria to glyphosate, probably shifting the gut bacteria in an unfavorable direction, and presumably could impair their absorption of metals. 

Slide 14
^ This experiment looked for effects of early season glyphosate application, not crop desiccation. There were six strips, all about 20’ x 60’, within a few feet of each other. The strips were fertilized as indicated by soil tests, as they have been for several years.  

Slide 15
^ Three were fertilized with NPK fertilizers as recommended by the state extension service online calculator, represented by gray fill colors. 

Slide 16 ^ The other strips were treated with about ¼ inch of compost made from leaves and spent coffee grounds, and mineral salts recommended by an online organic fertilizer calculator called Organicalc. The rich brown fill color indicates regenerative management. 

Slide 17
^ The two plots to the north were tilled to disrupt weed growth, represented by green outlines.

Slide 18
^ The two plots in the middle had a thick winter wheat cover crop, which smothered the hen it and purple deadnettle that can overwhelm the plot. The wheat was cut, mostly dried in place, and then tilled to about four inches, represented by the blue outlines.  

Slide 19 
^ The plots to the south were treated with generic glyphosate solution to kill established weeds in late March, then tilled before spring planting, and treated with glyphosate again in early June, before planting summer crops.  The vivid pink outlines represent glyphosate treatment. No growing crop was treated with glyphosate.  

Slide 20
^ Several spring and summer crops were planted in the same pattern in each strip. ^ Representatives of three staple crops were a hybrid sweet corn, cowpeas, and a hybrid tomato. 

Slide 21
^ The experiment tracked three outcomes: ^ The first was the levels of glyphosate and its breakdown product, AMPA, in summer crops. ^ The second outcome was crop yields for sweet corn, cowpeas, amaranth, and okra. ^ The third outcome was crop elemental chemistry, with particular attention to nutrient metals that glyphosate might bind, and essential elements that microbes might deliver to plants, like phosphate. 

Slide 22
^ These are the western strips at the end of September. You can see two of the summer staple crops. ^ Sweet corn had spontaneously dried. ^ Cowpeas are still producing.  Two more summer crops were ^ amaranth with the red tops, ^ and okra with the elegant white flowers. The tomatoes are hard to see, blending into the okra. ^

Slide 23
First result. ^ None of the three crops checked had detectable levels of glyphosate or its main metabolite. Again, the crops were not terminated with glyphosate, but glyphosate applied in the spring definitely did not persist to contaminate the fall harvest. ^

Slide 24
However, with a few exceptions, crops that grew elsewhere ^ struggled to grow in the NPK plot treated with glyphosate. That plot developed large cracks in the soil and usually yielded the least. There is some preliminary evidence that the microbiome suffered from the combination of NPK fertilizer and glyphosate, but that was not part of the original experiment, and we need more robust data.

Slide 25
^ This graph will show yields as a percentage of the strip with the highest yield.

Slide 26
^ The tilled regenerative strip grew vigorous plants and had good yields of all crops 

Slide 27
^ The regenerative strip with the cover crop had the best corm yield and good yields for other crops.

Slide 28
^ The regenerative strip treated with glyphosate also grew many vigorous plants and had the highest cowpea yield. However, sweet corn yield was substantially lower. 

Slide 29
^ The tilled NPK strip had a very low okra yield, while other yields were modestly impaired.

Slide 30
^ The NPK strip with the cover crop had an excellent okra yield, while other yields were modestly impaired. 

Slide 31
^ The NPK strip treated with glyphosate had very good cowpea and okra yields, but terrible corn and amaranth yields.  

Slide 32
^ The chemistry assays return values for a lot of elements from each of the 124 samples. We want to know if glyphosate treatment changes concentrations of 16 different elements, but we will get misled if we look for every difference with a 5% chance of happening by chance. It would be like rolling a fair 20-sided die 16 times and claiming that the die is unfair if you see any 20’s. You will often get one or two 20’s by chance. To avoid that, these data are batched into bigger, meaningful groups. The whole group is checked with a test called MANOVA. If the MANOVA result indicates a difference, each element in the group is checked with an ANOVA test. MANOVA and ANOVA let us specify a predictive model. 
^ Our model is that element levels are influenced by the product, like radish root or radish leaf, the kind of soil management, and the kind of weed control. When these tests find a difference with less than a 5% chance of happening by chance, we mark it with an asterisk. Three asterisks mark differences with less than a 1 in 1,000 chance of happening by chance. Most of these are much smaller than 1 in a million, so something triggered the test result - hopefully a real influence on chemical composition.  I’ll use red to show higher concentrations, and blue to show lower. 
^ The second column will show which elements soil management affected. Higher with NPK is in red, lower with NPK in blue. 
^ The third column will show which elements weed control affected. 
^ The second row is the group of elements that were supposed to be adequate in both soil management conditions - the elements were either plentiful in the soil or added in both conditions.
^ The MANOVA test suggests a difference due to soil management, and manganese and nickel levels are higher with NPK, while P was lower.  
^ Weed control made no difference. 
^ The third row is the group of elements that were supposed to be deficient in NPK soil management  because the elements were deficient in the soil and only added in the regenerative condition.
^ The MANOVA test suggests a difference due to soil management, and boron levels are higher with NPK, while molybdenum, sulfur, and zinc were lower. Of these, sulfur and zinc are suspected to be clinically meaningful. 
^ Weed control wanted to make a difference, but really didn’t. 
^ The fourth row is heavy metals, which have been added to the regenerative strips in very tiny amounts in AZOMITE for a few years, but have never been deliberately added to the NPK plots.
^ The MANOVA test suggests a difference due to soil management, and cadmium levels are higher with NPK, but mainly in leaves. 
^ And guess what? Weed control made no difference.

Slide 33
^ Next I want to show you a few box plots of the raw data. For nutrients I will superimpose a blue curve showing my guess at the range of values that should be nutritious for a person eating four pounds of food per day, assuming that all of the food has that concentration of the element, and is about 95% water. Most food is 90-95% water. Age, health, and activity could easily change a curve’s position - these are just educated guesses. A plateau suggests a range of concentrations that would supply all of the element that adults need without being toxic.


Slide 34
^ I’ll also use a red curve to show an informed guess at the intolerable level for cadmium. Same idea.

Slide 35
^ I’ll start with a box plot that shows an obvious difference that has been repeating for years now. ^ The brown boxes show the middle 50% of values for nickel, in parts per million, for the aligned crop grown in the regenerative plots. The blue highlights show the values for the staple crops. ^ The gray boxes show the middle 50% of nickel values for the aligned crop when grown with NPK fertilizer. The staples are outlined in red. ^ All of these crops clearly have higher nickel levels in the NPK condition. The boxes don’t even overlap. The nutrition curve suggests that somewhere over 1 part per million of nickel is too much. This currently seems likely for people with systemic nickel allergy syndromes, although the maximum tolerable dose for everyone else is not clear, as far as I know. 

The other noteworthy thing about this box plot is that some of ^ the regenerative values are in or below the sufficient range, while ^ the corresponding NPK values are in the estimated “too much” range. ^ A person on a low nickel diet can eat my regenerative soil okra, ^ but not my NPK soil okra. That suggests that this could be clinically important. Now let’s look at the glyphosate effect. 

Slide 36
^ These box plots illustrate the weed control effect on nickel. Although the statistical test indicates a significant effect, it is much less clear what that effect would be from the graph. This is typical of one-star effects. Tilled is usually above the other conditions, but it is a small effect. 

^ Moreover, similar pictures emerged for calcium, copper, iron, magnesium, manganese, molybdenum, and zinc, as well as heavy metals arsenic, cadmium, antimony, and lead. Glyphosate had little or no effect on these divalent cations. 

Slide 37
^ What about the phosphorus? Generally it is higher in the plots treated with glyphosate. ^ Could it be that the phosphorus in the chemical is getting released and taken up by the crop? Maybe, but if an acre of soil has around 100 pounds of phosphorus and gets one and a half pounds of glyphosate, you would expect that to add maybe 1% to the phosphorus levels. Probably something else is happening here. 


Slide 38
^ You might wondered how copper was affected, since it is relatively scarce in soil, binds tightly to glyphosate, and had a statistically significant change related to weed control. Here it is. The tilled strip frequently is higher than the glyphosate strip, and the cover crop strip is all over the place. Not too compelling. 

Slide 39
^ Here’s zinc. With the exception of beet leaves seemingly picking up a lot in the cover crop strip, there is a lot of overlap for every crop, including the three staple crop representatives. 

Slide 40
^ The soil management affected zinc about as dramatically as it did nickel. All of the regenerative crops have higher zinc levels than the NPK crops.

Slide 41
^ Sulfur box plots tell the same story. ^ Again, many crops from regenerative strips contain higher sulfur levels ^ and the middle 50% of values do not even overlap the corresponding NPK box plots. 

Slide 42
^ And finally, cadmium, a heavy metal that is pretty toxic at low levels, ^ is consistently below detection limit in regenerative strips, ^ but it is elevated in the leaves of vegetables grown in NPK strips. 

Slide 43
^ Limitations. Just in case this idea has been lost recently, let me mention that scientists are supposed to be the first to point out the contradictory realities, uncertainties, and doubts. ^ I am not the embodiment of science, ^ agriculture, or medicine. Wise policy makers would find credible replications or strong corroborating evidence and then run a lot of very public pilot studies before ruining anyone’s life or livelihood with bold new dictates. ^ This experiment is not like farming at scale, although I hope that it contains information of use to large scale farmers. ^ There is some chance that there is a farm somewhere that is unlike a gentle slope in eastern Missouri. ^ I have not yet assessed any effects on concentrations of essential nutrient compounds, because that’s very expensive. ^ Any soil regeneration or degradation from any of my management strategies has not yet been quantified. 

Slide 44
^ Glyphosate, like so many things, involves trade offs. That means that an intelligent, informed and conscientious decision maker could make context sensitive choices about how and when to use it. Those decisions could change over time, especially with new discoveries about glyphosate, rising costs for chemicals with amine and phosphate groups, or new weed control strategies. 

Slide 45
^ Glyphosate can be a defensible weed control option. ^ It is pretty safe, especially in contrast to acutely toxic herbicides, ^ although you don’t want to eat or breathe it. Adjuvants may impose additional health risks. ^ Without glyphosate, some other weed control is necessary to hold weeds at bay. That is likely to be labor, technology, or material intensive, with significant attendant costs. ^ Glyphosate had minor effects on crop’s elemental composition, ^ but may have seriously damaged the soil microbiome in the absence of a layer of living compost. 

Slide 46
There are other weed control options. ^ Cover crops are applicable to every scale operation from raised bed gardens to industrial farms. ^ Shallow tilling is a fast and easy way to disrupt weeds around planting time, but exposes new seeds and may leave viable weed roots. Wheel hoes are nice wherever practical. ^ Thick or solid opaque covers discourage weeds. ^ Animals that eat weeds without compacting the ground are great if you can maintain them. Goats, small cattle, sheep and chickens are candidates. ^ Other remedies have niche applications, and you want to be careful not to directly damage a crop.

Slide 47
^ The best idea might be to change the game whenever you can by planting perennial crops that start above the weeds in their second year. ^ Lots of farms and gardens could stand to invest in rebuilding topsoil depth and biology. ^ Cover crops are probably the most cost-effective way to do this, although you might want a more diverse cover crop than just winter wheat - a lot of regenerative farmers use oats, a radish, and winter peas, and there are mixed cover crop seed packages sold for home gardeners. ^ The compost we use supports a lot of soil life, reversed a lot of negative glyphosate effects, especially for corn and amaranth, and seems to prevent crops from accumulating heavy metals. That compost is made from fallen deciduous leaves layered with coffee grounds. Our local Starbucks discards about 50 pounds of coffee grounds every day. It would be hard to overstate how useful that stuff is for making compost and restoring degraded soil, especially in the absence of large animals. Whether or not you plan to use glyphosate, compost made with a lot of coffee grounds is useful to have.

Thanks for watching.