Soil microbiome studies

Overview

This project came to life during my summer internship on a local regenerative farm in Fauquier County, Virginia. During my months there, I learned about the unique principle that drove the whole farm - which emphasized that healthier soil comes from letting the microbiome beneath the surface rebuild itself without any disturbance.

But that raised a question I couldn't shake off. What was actually underneath the ground? Over two days, I took soil samples from a historically disturbed orchard on the farm and had the DNA in each one sequenced.

A dense clump of common ragweed, its finely divided leaves standing above the grass, with young orchard trees in the background.
A cluster of common ragweed in the orchard area.
Looking straight down at a narrow hole cut into red soil with a spade, red clover and grass growing around it.
Cutting a soil sample beneath red clover.
A shrink-wrapped blue rack of sample tubes resting on a printed handling sheet, its text softened, with a hand-written note reading Sample number 1 to 13, labeled on tubes and rack.
The soil samples in individual numbered tubes, on the way to DNA sequencing!

Summarized results from the samples, explained simply

A teaspoon of soil holds about a billion bacteria of thousands of kinds, plus fungi and a crowd of tiny predators and animals. Two of the five plants sampled, red and white clover, were legumes - essentially plants that harbor rhizobia, a type of bacteria that take nitrogen from the air and make the host plant its own fertilizer.
Fescue, chicory and ragweed, the other plant types sampled, are not legumes and don't have this ability. Did the soil life under one type of plant look any different from another?

Across teaspoon-sized samples of soil,

distinct kinds of bacteria and archaea were recorded!

Every soil sample held about the same variety of life. It just wasn't the same cast of characters

If you simply counted how many kinds of bacteria and archaea lived under each plant, the totals would've come out to about even (p = ). But once you looked at which kinds were there, the communities would began to sort themselves by the plant growing above them (R² = , p = ). Two scoops of soil from under the same plant tended to resemble each other more than two scoops from underneath different plants. Basically like two libraries with the same number of books but different titles on the shelves.

Two samples under the same type of plant looked more alike

Each dot compares one pair of soil cores; the higher the dot, the more different the pair. The black bar is the group average and the whiskers are the 95% interval.

Under clover, the nitrogen microbes swap places

The clearest result was about the microbes that handle nitrogen production. A group of archaea (microbes on a separate branch from bacteria) does the first step of turning ammonia into the nitrogen that plants take up.
They were scarce under the two groups of clovers, making up about % of the DNA against % under the other plants, with no overlap between the two groups of cores (q = ).
The rhizobia, the bacteria that make clover its fertilizer, went the other way: % under clover, and % under the rest (q = ). One went down where the other went up, which fits a plant that makes its own nitrogen and leaves less ammonia around.
This counted DNA and not the literal nitrogen activity, so it's a lead worth chasing rather than a proven effect. The full story is on the research page.

The nitrogen microbes, legumes against the rest

Each point is one soil sample shaped by plant. The box is the middle half of the group, the line through it the median, and the whiskers reach the most extreme samples within one and a half times that spread.

It had a food web and it wasn't just random soil soup

A second gene picked up the bigger, more complex organisms: mostly fungi, with single-celled hunters that eat bacteria and tiny animals that eat the hunters. Fungi made up about % of those reads. Something down there was eating something else. Unlike the bacteria, these did not vary by plant (p = ).

Who exactly is down there?

The share of complex-celled life in this soil, across cores. "Other" is everything the reference library could not name, which in soil is never a small pile.

And ragweed, the plant the farm asked about

Common ragweed is the weed anyone would pull. The bacteria under it looked much like the bacteria under everything else (p = ). One fungus, Plectosphaerella, was more common under ragweed than under red clover or fescue (% against % and %), and more common still under the one chicory core (%). One statistical test called that real and another did not, so it is a lead and not a confirmed result (yet). All the limitations of this project are listed on the research page.

On this site

Research

The whole argument: the census, what separates the patches, the nitrogen pattern under clover, ragweed, and every caveat.

Methods

How each number was computed, which checks came back negative, and what is not documented.

Gradient

The community varies along one dominant gradient, and what the archaeal share is really tracking.

Continental drift

A separate study: 20 NEON sites over five years, changing in step, with chemistry rather than distance saying how closely.

Directions

Two follow-ups the sequences already support.

Ground truth

What it would take to turn a read ratio into a measured result.

Sources

Every external work this study stands on, and what each was used for.

About me

Who did this, a semester on the farm, and whose data it is.