Start with a number that runs against what most people assume. On US cropland, average water erosion fell from 3.89 tons per acre per year in 1982 to 2.67 tons in 2017. Wind erosion fell from 3.24 tons to 1.96 over the same 35 years. Taken together, the 2017 National Resources Inventory Summary Report puts the drop in cropland erosion rates at 35 percent.
That report comes from USDA’s Natural Resources Conservation Service, working with Iowa State University’s Center for Survey Statistics and Methodology, and it was published in September 2020. It is a statistical survey of thousands of sample points, not a census of every field, so treat the figures as national averages over a very uneven country.
Soil health is a real subject with real measurements behind it. It has four management principles, a federal agency that publishes literature reviews on it, and a body of peer-reviewed work on what the practices do to water, carbon and soil structure.
What it does for your personal health is a much thinner story, and the honest version of this article says exactly where the thick part ends and the thin part begins.
What soil health means and the four principles behind it
The Natural Resources Conservation Service defines soil health as “the continued capacity of soil to function as a vital living ecosystem.” That phrasing is doing deliberate work. It treats soil as a system with a job rather than as a growing medium, and the job list includes moving water, filtering pollutants, cycling nutrients, housing organisms and supporting plants.
Four management principles sit under that. NRCS presents them as reducing disturbance, maximizing cover, promoting biodiversity and supporting living roots. They make more sense explained in the order a field experiences them.
Living roots are the engine. Plants push a share of what they photosynthesize out through their roots as sugars and other compounds, which feed the bacteria and fungi living in the few millimetres of soil around the root. Bare ground between cash crops means months with the engine switched off. Keeping something alive and rooted for more of the year keeps the biology fed.
Cover is the shield. Residue or living plants on the surface absorb the impact of raindrops, which is what breaks surface aggregates apart and forms the crust that sheds water instead of taking it in. Cover also moderates soil temperature and slows evaporation.
Disturbance undoes both. Tillage, the mechanical breaking and turning of soil before planting, controls weeds and warms a seedbed, and in the same pass it shatters structure, buries residue and exposes organic matter to rapid decomposition. The principle is not that tillage is forbidden. It is that every pass has a cost on the other side of the ledger.
Diversity keeps the system from narrowing. A two-crop rotation selects for the pests, diseases and soil organisms that thrive under two crops. More species above ground tends to mean more variety below it.
The principles work as a set rather than a menu, which is exactly what the research keeps showing. NRCS’s own literature reviews repeatedly find that no single practice performs as well alone as it does in combination, and also note that almost no study has tested the full combination, so the benefit of the combined system is partly inferred.
Organic matter and water: the clearest measured effect
Ask what soil organic matter does and the best-supported answer is about water.
Two terms matter. Available water capacity is the amount of water a soil can hold in a form plants can pull on, the band between the point where gravity stops draining it and the point where roots can no longer extract it. Infiltration is how fast water enters the soil instead of running off.
A figure NRCS cites from a 1994 study, in its 2015 review of conservation practice effects, is the one most worth knowing: as surface soil organic matter rose from 0.5 percent to 3.0 percent, available water capacity in that surface layer more than doubled. A separate NRCS review of the water-holding literature estimates the gain at roughly 8.4 millimetres of additional available water for each 1 percent increase in organic carbon in the top 20 centimetres, and reports manure applications raising available water capacity by 14 to 29 percent.
Infiltration responds too, sometimes dramatically. In one Ohio comparison, water entered plow-tilled soil at 1.2 centimetres per hour and no-tilled soil at 4.6 centimetres per hour. A review NRCS cites reports runoff reductions of 4 to 50 percent under no-till with residue or cover crops. One study found infiltration 43 times greater under no-till, which the NRCS authors themselves flag as atypical rather than representative.
Now the honest part, because this is where popular writing stops. The same NRCS reviews find the results inconsistent and heavily context-dependent. A 32-year study found available water capacity 23 percent higher under no-till with residue retained. A 14-year study found it 19 percent higher with yields 28 percent higher, but conventional tillage out-yielded no-till for the first four years. A nine-year Alberta barley study found 8 percent less available water in the top inch under no-till.
Three cautions run through all of it. Gains usually take years to appear, with the infiltration transition under no-till estimated at roughly two to eight years. Soils above about 40 percent clay may show no gain at all. And more water going in is not automatically good: greater infiltration can carry more nitrate down past the root zone.
Erosion, and what the numbers describe
Two kinds of erosion get measured separately because they work differently.
Sheet and rill erosion is water erosion. Rain detaches soil particles, a thin film of water carries them downslope, and the flow concentrates into small channels. Wind erosion lifts and rolls dry, bare, loose particles, which is why it dominates in drier regions and after a field has been worked.
The mechanism behind reducing both is unglamorous: keep something on the surface. Residue takes the raindrop impact and slows the sheet flow. Crops that leave fragile residue, soybeans and many vegetables among them, give less protection, which is one reason erosion varies as much by rotation as by tillage. Soil structure matters too, since stable aggregates resist being detached in the first place, and one study NRCS cites found manure and organic amendments raising water-stable aggregation by 23 to 27 percent, though in a tropical soil rather than a US field.
Hold two things at once about the national figures. A 35 percent reduction in cropland erosion rates between 1982 and 2017 is a genuine, measured improvement driven partly by conservation programs. And an average of 2.67 tons per acre per year is still a mass of soil leaving a field every year, concentrated on the places most vulnerable to losing it. Averages flatten exactly the variation that matters.
Cover crops and no-till, including the trade-offs
A cover crop is a crop grown to protect and feed soil rather than to sell. NRCS runs it as conservation practice 340, and the standard is more specific than the idea: it requires timely planting, enough biomass to do the job, and termination at a late vegetative stage. Termination timing is not a detail. Kill the cover too late in a dry spring and it has already drunk the moisture the cash crop needed.
No-till means planting directly into undisturbed ground. It is the practice with the best-documented benefits and the one most often described without its costs. The costs are real and they are in the federal literature:
- The transition takes years. Infiltration and aggregate stability generally improve, but studies put the timeline at roughly two to eight years, and several report an early yield penalty.
- Bulk density, a measure of soil compaction, can rise in the first years rather than fall.
- The continuous pores that make no-till soil drain well also move nutrients. One study found no-till increased nitrogen and phosphorus delivery into tile drains for exactly that reason.
- On soils with high shrink-swell clays, runoff can increase rather than decrease. NRCS reviewers judged no-till unsuitable in interior Alaska, where minimum tillage performed better.
- Aggregate stability improvements are not universal. One Georgia sandy soil improved after three years and a southern Piedmont soil improved by about 50 percent under long-term no-till, while other studies found limited or no effect.
There is one more argument worth reporting as an argument rather than settling. Critics have questioned whether no-till systems increase reliance on herbicides and synthetic fertilizer, since weeds that were once buried now have to be controlled another way. The Congressional Research Service records both the criticism and the industry response disputing it. Anyone who tells you that question is closed is ahead of the evidence in one direction or the other. Article 736 covers how claims about these practices are regulated, or rather are not.
Nutrient cycling, phosphorus, and the limits of measurement
Nutrient cycling is the part of soil biology with the most direct link to a farm’s bottom line, and it moves faster than people expect. During the growing season, mineralization converts organic nitrogen to the nitrate form plants take up within one to two weeks.
That speed cuts both ways. Nitrate not taken up is nitrate available to leach. Cover crops reduce that loss by holding residual nitrogen in living tissue, and legume covers add nitrogen of their own.
Phosphorus behaves differently and accumulates. Repeated manure application builds soil phosphorus to levels that raise phosphorus in runoff, which is a water quality problem rather than a soil fertility one.
One study NRCS cites captures the trade-off with unusual precision. On dairy farms, restricting nitrogen cut potential nitrogen losses by 18 to 50 percent but phosphorus losses by only 3 to 15 percent. Restricting phosphorus cut phosphorus losses by 28 to 43 percent and cut farm income by 11 to 23 percent. Both are true. Neither is free.
As for measurement, the structural honesty problem is time. NRCS’s reviews note that most physical soil changes happen gradually, that many published studies run under five years, and that a short study can miss an effect or catch noise. USDA’s own Regenerative Pilot Program, launched in December 2025, requires soil health testing in the first and last year of a contract specifically to set a baseline and record what changed, which is an admission that single snapshots do not tell you much.
From soil health to human health: the gap
One sentence carries the whole argument. The leap from soil health to human health is largely unestablished.
The most common route people take to make that leap is the claim that fruits and vegetables have lost nutrients since the mid-twentieth century. There is a real study behind it, and it is more careful than the headlines built on it.
In 2004, Donald Davis and colleagues compared USDA nutrient data published in 1950 against data published in 1999 for 43 garden crops and 13 nutrients, adjusting for differences in water content. Treated as a group, six nutrients showed statistically reliable declines: protein, calcium, phosphorus, iron, riboflavin and vitamin C. Median declines ran from 6 percent for protein to 38 percent for riboflavin. The other seven nutrients showed no reliable change.
The headline stops there. The paper does not. When the authors looked at individual foods and individual nutrients rather than the group, the authors wrote in the Journal of the American College of Nutrition that “R-values are usually not distinguishable from 1 with current data,” R being their ratio of the 1999 value to the 1950 value. Depending on which estimate of the older data’s error you use, only 20 to 33 percent of those individual ratios differed reliably from no change at all. And about 28 percent of the reliable differences went up, not down.
The authors’ own explanation is the one most often attributed to soil. It is not. They suggested that real declines are most easily explained by changes in cultivated varieties between 1950 and 1999, with possible trade-offs between yield and nutrient content. That is the dilution hypothesis, and it is a hypothesis offered by the authors, not a finding they demonstrated.
Be clear about what this design can and cannot do. It compares two database snapshots 49 years apart, built from different varieties grown in different places and measured with different laboratory methods. Nobody sampled the soil. The study cannot isolate soil management as a cause, and it does not claim to.
For contrast, here is what a controlled experiment looks like. A 2014 meta-analysis in Nature pooled field experiments growing crops at the atmospheric carbon dioxide concentration projected for the middle of this century. C3 grains and legumes, the category that includes wheat, rice and soybeans, had lower zinc and iron concentrations at the elevated level. C3 crops other than legumes also had lower protein, while C4 crops such as maize appeared less affected. Differences between cultivars of the same crop suggested breeding could reduce the sensitivity.
That is a measured manipulation with a control, which sits a tier above an archive comparison. Note what it is about: atmospheric carbon dioxide, not soil management.
So the tiering, stated plainly. Established: soil management changes water movement, erosion rates, aggregate stability and topsoil carbon. Preliminary at best: that those changes alter the nutrient content of harvested food enough to matter. Not established: that any of it measurably changes the health of someone eating the food.
Soil microbiome and gut microbiome are different subjects
The popular version of this story skips all of the above and jumps straight to microbes. Healthy soil has a rich microbiome, your gut has a microbiome, so one must feed the other.
They are both microbial communities, and that is most of what they have in common. The paper most often cited for a connection makes the position clear in its own words. Writing in the journal Microorganisms in 2019, the authors note that soil and the human gut contain roughly comparable numbers of active microorganisms while gut diversity is around a tenth of soil diversity, and then propose an environmental microbiome hypothesis linking the two. Proposing a hypothesis is how a research program starts, not how it concludes. The paper is a review and a proposal, not evidence of a causal chain from a farm’s soil biology to your intestines.
The gut microbiome is a serious subject and other articles in this library handle it. The point here is narrower: do not let a claim about soil biology borrow credibility from gut research, or the reverse.
One soil-to-human-health link is genuinely well established, and it is not microbial. It is direct contact with contaminated soil, lead in older urban soils above all. Article 730 covers that, along with the rest of the garden.
How to judge a soil claim
When someone tells you a farming practice improved soil, five questions separate a measurement from a slogan.
- What exactly was measured: organic matter, infiltration rate, aggregate stability, carbon stock, or just yield?
- To what depth? Surface-only sampling and full-profile sampling routinely disagree.
- Over how many years? Under five years is short for soil, and most physical change is slow.
- Against what baseline, taken when, on the same field?
- Is the comparison the same soil type and climate, or a different region entirely?
Keep two things apart and the subject gets much clearer. Soil health is an agronomy subject, and within that subject the evidence is strongest where it is least glamorous: water goes into healthy soil faster, stays available longer, and carries less soil away with it when it leaves. Those effects are measured, repeatedly, by a federal agency that also publishes the studies that disagree.
The nutrition question is a separate question with weaker evidence, and the microbiome question is a third question that is barely begun. If you are worried about a specific nutrient in your own diet, that is a clinician and a blood test, not a theory about farming. Persistent fatigue with heavy periods, breathlessness on stairs or an odd craving for ice can point toward iron deficiency, which a clinician can check with a blood count and a ferritin level. Guessing at supplements instead is a bad trade, because too much iron causes its own harm and those symptoms overlap with conditions that need a different answer entirely.
Sources
- U.S. Department of Agriculture, Natural Resources Conservation Service. “Soil Health.” nrcs.usda.gov
- U.S. Department of Agriculture, Natural Resources Conservation Service, with the Center for Survey Statistics and Methodology, Iowa State University. “2017 National Resources Inventory Summary Report, September 2020.” nrcs.usda.gov
- U.S. Department of Agriculture, Natural Resources Conservation Service, National Soil Survey Center. “Soil Health Literature Summary: Effects of Conservation Practices on Soil Properties in Areas of Cropland, January 2015.” nrcs.usda.gov
- U.S. Department of Agriculture, Natural Resources Conservation Service. “Effects on Soil Water Holding Capacity and Soil Water Retention Resulting from Soil Health Management Practices Implementation, March 2018.” nrcs.usda.gov
- U.S. Department of Agriculture, Natural Resources Conservation Service. “Effects of Implementation of Soil Health Management Practices on Infiltration, Saturated Hydraulic Conductivity, and Runoff.” nrcs.usda.gov
- U.S. Department of Agriculture, Natural Resources Conservation Service. “Cover Crops for Soil Health.” nrcs.usda.gov
- U.S. Department of Agriculture, Natural Resources Conservation Service. “NRCS Regenerative Pilot Program.” nrcs.usda.gov
- Congressional Research Service. “Regenerative Agriculture and Related Food Product Labeling and Marketing Claims (R48610), 29 July 2025.” everycrsreport.com
- PubMed. “Davis DR, Epp MD, Riordan HD. "Changes in USDA food composition data for 43 garden crops, 1950 to 1999." Journal of the American College of Nutrition, 2004. PubMed. DOI: 10.1080/07315724.2004.10719409.” doi.org
- PubMed. “Myers SS et al. "Increasing CO2 threatens human nutrition." Nature, 2014. PubMed. DOI: 10.1038/nature13179.” doi.org
- PubMed. “Blum WEH, Zechmeister-Boltenstern S, Keiblinger KM. "Does Soil Contribute to the Human Gut Microbiome?" Microorganisms, 2019. PubMed. DOI: 10.3390/microorganisms7090287.” doi.org