Sixty-six years of burning are still visible in a peatland's fungi

A hillside laid out in 1954

Blanket peat on a moor in the northern Pennines has been burned on fixed rotations since 1954, and a paper published online on 7 September 2026 in FEMS Microbiology Ecology reports that the soil fungal community still differs between burned and unburned ground at every depth sampled, down to 60 centimetres. The site is the Hard Hill long-term burning experiment at Moor House-Upper Teesdale National Nature Reserve. Shaun M. Allingham, of the University of Nottingham, the University of Derby and the NERC Environmental Omics Facility in Sheffield, is the first author, with five colleagues at Derby, Manchester Metropolitan University and Mount Royal University in Calgary. Sampling took place in July 2020, sixty-six years after the plots were laid out. The full text is open access.

Three regimes were compared: no burning at all since 1954; twenty-year intervals, which has meant four burns; and ten-year intervals, which has meant seven. Both burned treatments were last burned in 2017. Only the fenced plots were sampled, so grazing drops out of the comparison. Across four blocks the team placed three one-metre quadrats per treatment and took five cores per quadrat at each of 0–20, 20–40 and 40–60 centimetres, pooling every set of five into a single composite sample. Of the 108 samples collected, those returning fewer than a thousand reads were discarded as insufficient; the per-group sample sizes printed with each figure add up to 104, which is why the analyses report 95 residual degrees of freedom. Fungi were identified by sequencing the ITS region on an Illumina MiSeq and matching the reads against the UNITE database.

Fewer taxa, and a different cast

Observed richness differed by burn regime (F(2, 33.35) = 12.04, P < 0.001) and by depth, and the authors' conclusion is explicit: short-rotation burning reduced richness across all sampled depths, not merely in the layer the flames reach. Shannon and Simpson diversity were lower on burned ground than on the control, but the two burned regimes did not differ significantly from each other, so ten-year rotations were not measurably worse than twenty-year ones on those indices. Composition differed by regime and by depth alike (PERMANOVA on Bray–Curtis dissimilarity, R² of 0.135 and 0.083, P = 0.001).

What shifted was not simply a matter of less. Ascomycota averaged 81 per cent of reads and Basidiomycota 16, but Basidiomycota reached 38 per cent in short-rotation topsoil against 4 per cent under long rotations. Leotiomycetes, the dominant class at 61.2 per cent of all reads, was at its lowest in the unburned control at all three depths, and Agaricomycetes peaked where burning was most frequent. Other genera went the other way: in burned topsoil under twenty-year rotations Serendipita fell from 18.7 to 2.6 per cent of reads and Archaeorhizomyces from 10.9 to 6.1, while Venturia was absent from both burned regimes. No genus at all was tied uniquely to the short-rotation plots, which the authors suggest may point to reduced niche differentiation: burning here favoured broadly distributed, disturbance-tolerant fungi rather than a distinctive fire-adapted flora.

The result with the longest reach concerns not who is present but what decides it. The modified stochasticity ratio depended on regime, on depth and on the interaction of the two (F(4, 503.38) = 5.45, P < 0.001), so the authors read it one layer at a time. At 0–20 and 40–60 centimetres the unburned plots sat below the fifty per cent threshold, the mark of assembly governed mainly by deterministic environmental filtering, while the burned plots sat higher, shifted towards chance; at 20–40 centimetres the three regimes looked much alike. Levins' niche widths were broader in unburned plots than in either burned treatment.

What the percentages are not

Every figure above is a share of rarefied ITS amplicon reads: not biomass, not a cell count, not a measure of how much fungus is in the peat. A genus can fall as a proportion simply because others rose, so Serendipita dropping from 18.7 to 2.6 per cent describes a change in the community's composition rather than a measured loss of that fungus from the ground.

Nor is this a study of wildfire. Rotational burning on managed moorland is deliberate, low in intensity and spaced at fixed intervals, and nothing here should be carried across to peat fires that escape control. The design has a limit the authors state plainly: the treatments differ in the interval between burns as well as in their number, and because both burned regimes last burned in 2017, time since fire is identical between them.

Two further cautions belong in the reading rather than in a footnote. Enzyme activity and decomposition rates were not measured; the authors say only that shifts in saprotroph abundance "could potentially translate to changes in decomposition dynamics". And the trophic labels come from FUNGuild, which classified 40 per cent of the sequence variants and whose symbiotroph category lumps ectomycorrhizal genera such as Cortinarius together with ericoid partners and with plant-associated fungi that form no ectomycorrhiza at all, among them Serendipita and Phialocephala, so, in the authors' own words, the assignments "may not fully capture the ecological complexity of all taxa". Chance did not take over everywhere either: with the regime-by-depth interaction significant, the stochasticity ratio has to be read layer by layer, and at 20–40 centimetres the regimes were alike.

Where the record can be checked

Hard Hill is a British moorland management experiment. It carries no advice about Polish peat bogs and none whatsoever about collecting anything. What it offers is a rare chance to see how far a surface practice reaches: the difference the experiment has tracked aboveground for decades, in the cover of heather, graminoids, Sphagnum and other mosses, was also there at 40–60 centimetres, as deep as anyone dug. The variables that best explained community structure changed with depth as well — calcium, manganese, the cover of mosses other than Sphagnum, total carbon and pH at the surface; ammonium and moisture at 20–40 centimetres; iron and lead at 40–60. The raw reads are deposited in the NCBI Sequence Read Archive under BioProject PRJNA1002628.

The publication record is worth checking too, because it is not consistent. The publisher gives 7 September 2026 as the online publication date, in the September 2026 issue, volume 102, issue 9. The PubMed record carries 20 August 2026 in its journal-issue date field, which cannot be right, since the paper was accepted on 28 August 2026. Where catalogues disagree, the publisher's date is the one to use, and it is the one used here.