22 July 2026 ยท Experimental design
Why I am fencing off four square metres, three ways, for three years
After several posts about carbon fractions, nematode trophic groups, and nitrogen pools, here is the less glamorous physical reality behind all of it: small fenced squares of grass, doing nothing, dotted around working farms, for three years. One farmer has already asked, entirely fairly, why he is paying grass to sit there ungrazed while the field around it keeps working, a good question, and the honest answer is that those quiet little plots are doing most of the heavy lifting in making this research trustworthy.
The design, briefly
Between four and six farms are being selected across a latitudinal climate gradient, roughly the same spread that is already taking me from Dumfries to Devon, chosen specifically to span a genuine range of grazing intensities and weather patterns rather than farms that already agree with each other. Within each farm, four representative fields are selected, and within each field, three grazing exclusion plots of four square metres are permanently fenced off, with four grazing plots adjacent to each exclusion plot digitally mapped and recorded with geotagged photos, but left unfenced and grazed as normal. Altogether that gives a sample size of ninety six to a hundred and forty four across the project, comparing true exclusion against a real spread of grazing intensities rather than two tidy labels.
Why exclusion, rather than simply before and after
A simpler design would merely compare soil samples before and after a change in grazing management on the same field, but a single field, sampled over time, cannot separate the effect of grazing from the effect of that year's rainfall, temperature, or a dozen other factors changing alongside it. Pairing each exclusion plot with grazed ground that has been checked in advance for matching soil physical and chemical properties removes that confounding, so that any difference emerging over the following three years can be attributed to grazing itself, rather than to the plots having started out different, or to a wet summer as against a dry one.
Why three years, not one
Soil food web structure does not shift on a seasonal timescale, and a single sampling round would mostly capture weather noise, a wet spring here, a dry autumn there, rather than a genuine management effect. Three years of seasonal sampling gives the nematode community, the microbial biomass, and the carbon and nitrogen pools time to actually respond to sustained grazing management, rather than to whatever the weather happened to do the week the soil corer came out.
What it all feeds into
Every measurement across every plot, soil physical and chemical properties, vegetation productivity, microbial activity, nematode trophic structure, carbon and nitrogen cycling, gets analysed together using structural equation modelling, testing the full pathway from grazing through to biology as one connected system rather than as isolated pairwise comparisons. None of that works without the unglamorous part first: real farms, real fences, and three years of patience.