Cortinariaceae supply most class II peroxidase transcripts in boreal humus

A question about who switches the genes on

Class II peroxidases are the heavy oxidative machinery of the Agaricomycetes, the fungal class within which they evolved: enzymes that attack lignin by way of highly oxidising Mn3+ ions and expose the cellulose and hemicellulose locked behind it. Textbooks hand them to white-rot saprotrophs. A paper published online on 13 August 2026 in New Phytologist, and registered with Crossref as volume 252, issue 1, pages 400–413 of the October 2026 print issue, asked which fungi in a boreal forest soil are actually transcribing those genes. The answer, measured in the field rather than in a culture flask, is that most of the transcripts come from ectomycorrhizal fungi in the family Cortinariaceae — the webcap family.

Florian Barbi, Uwe Menzel, Domenico Simone, Tuula Niskanen and Björn D. Lindahl, working between the Institute of Microbiology of the Czech Academy of Sciences, the Swedish University of Agricultural Sciences in Uppsala and the Finnish Museum of Natural History at the University of Helsinki, sampled Fiby Urskog: a 64-hectare old-growth reserve 16 km west of Uppsala, left without major human intervention for several centuries. Sixteen plots of 8 by 8 metres, eight of them fertile, nitrogen-rich and dominated by Norway spruce, eight poor and dominated by Scots pine. Cores 3 cm across were taken through the entire depth of the organic horizon on four days — 4, 7, 17 and 27 October 2016 — four cores per plot each time, sixteen per plot pooled into a single composite sample. Live mosses, freshly fallen litter and mineral soil were stripped off and each core frozen on dry ice within seconds. The fieldwork is a decade older than the paper. The full text, open under CC BY 4.0, carries the whole protocol.

Around 125 million sequences per sample, some 2 billion reads in total. Rather than assemble a whole metatranscriptome, the team built a pipeline of its own — RING, for Reads Isolator from Nucleotidic Giga-dataset — to pull out the reads belonging to three gene families: AA2 (class II peroxidases), GH6 (cellobiohydrolases, which degrade cellulose) and GT48 (glucan synthases, a proxy for living fungal biomass), plus beta-tubulin as the normalising denominator. Identification was restricted to the most-expressed contigs: 146 for AA2, 63 for GH6 and 267 for GT48, each set accounting for 80% of that family's mapped reads. Their first hypothesis was that peroxidase transcription would be lower in the poor plots, where saprotrophic Agaricomycetes are scarcer.

What the 78% is a percentage of

The sentence the paper's title rests on reads that on average 78% of the AA2 transcripts originated from ectomycorrhizal fungal genera, mostly from the family Cortinariaceae, and then adds a bracket: 70% in the nitrogen-rich plots, 73% in the poor ones. That bracket belongs to the Cortinariaceae, not to the ectomycorrhizal total, which is how the paper's parallel sentence about GT48 is built, where the family's share is given as 28% and 25%. Saprotrophic genera supplied on average 25% of the identified AA2 transcripts, led by Mycena at 10% and 9.5% respectively. Below 5% each, AA2 transcripts also turned up in the ectomycorrhizal genera Russula, Hebeloma and the underground-fruiting Hysterangium.

Those figures are quoted exactly as published, and no arithmetic is performed on them here. They do not close: 78% plus 25% is 103%, and 70 and 73 average to 71.5 rather than 78. They are also shares of reads mapping to identified, most-expressed contigs, not shares of every peroxidase transcript in the soil.

A second number makes a different claim. Overall AA2 expression, normalised against beta-tubulin, was more than twice as high in the poor soil (0.67 ± 0.12) as in the fertile soil (0.30 ± 0.06), at P = 0.007 — the opposite of what the authors had predicted. The Results text leaves that spread unlabelled; the caption to Figure 2 says the error bars are standard deviations. Overall GH6 and GT48 expression did not differ between the two forest types. What changed for GH6 was its ancestry. In the fertile spruce plots, 63% of the sequences mapping to identified GH6 contigs came from saprotrophic Agaricomycetes, with Mycena at 29% and Rhodocollybia at 15%; in the poor pine plots basidiomycetes accounted for only 25%, ascomycetes for 73%, and the class Leotiomycetes alone for 45%. It is also the thinnest dataset behind any figure in the paper: GH6 rests on 50,150 mapped reads, kept at the lowest of the four mapping-quality thresholds used (samtools Q 1), against 237,605 reads for AA2 at Q 19 and 430,598 for GT48 at Q 22.

One genus refused to play its assigned part. Piloderma, long grouped with the nitrogen miners, was absent from the most-expressed AA2 contigs altogether. A check of the full AA2 output against the protein databases at NCBI and the Joint Genome Institute did turn up Piloderma-matching contigs at above 93% similarity, but they amounted to 0.084% of the AA2 transcripts — against a large share of the GT48 expression, which the authors read as an indicator of active fungal biomass. Their results, they write, challenge that characterisation, "at least within the scope of our study". The genes exist. In this humus, in this month, they were barely switched on.

The limits the authors wrote down themselves

Transcripts are not decomposition. No enzyme activity, no mass loss, no nitrogen release and no carbon flux was measured here. The authors' own verbs stay conditional: symbiosis "may enable" ectomycorrhizal fungi to use tree photoassimilates to drive energetically costly oxidation belowground, and that oxidation "may, thereby, enable" trees to regulate decomposition and nutrient cycling indirectly. The distance between switching a gene on and turning humus into carbon dioxide is where all the uncertainty in this study lives.

Nor is ectomycorrhizal dominance of peroxidase expression a phenomenon of poor soil. Sequences mapping to identified AA2 contigs were distributed similarly among fungal genera under both conditions, and the discussion states plainly that ectomycorrhizal fungi dominated class II peroxidase expression in the nitrogen-rich plots as well, albeit at a lower rate of expression. The level shifted with fertility; the cast did not.

The fertility contrast itself cannot be credited to nitrogen alone. Every nitrogen-rich plot is spruce and every nitrogen-poor plot is pine, so soil fertility and host tree are completely confounded — a limitation the paper sets out at length, noting that tree species differ in litter quality, in root-derived carbon inputs, in which ectomycorrhizal partners they favour, and in manganese availability. Extractable manganese was six times higher in the fertile plots; pH there was 4.29 ± 0.06 against 3.71 ± 0.03 in the poor ones, these values reported as standard errors, and the ratios of carbon to total nitrogen and of carbon to ammonium nitrogen were higher in the poor soil by factors of 1.2 and 1.8. Manganese is also one of the authors' own candidate explanations for the headline result: where the metal is six times scarcer, more peroxidase might be needed simply to keep a working pool of oxidised Mn3+. That is a suggestion offered in the discussion, not a measurement.

One reserve, sixteen composite samples, a single autumn, and only the organic horizon, with the mineral soil and the fresh litter deliberately removed: this is not a portrait of boreal forest in general. Neither is it a story about webcaps rotting wood. Cellulase expression in the poor soil did not fall at all — it changed hands.

What it leaves for a reader in the pines

Rank matters here, and it is easy to get wrong. The unit throughout is the family Cortinariaceae, not the genus Cortinarius; the family long held a single genus and was divided into ten in 2022. Within that family exactly one match reaches species level: contig AA2_27, an almost perfect match to a reference sequence of Cortinarius ominosus. Two more sit outside it, AA2_54 matching Mycena sanguinolenta perfectly and AA2_104 matching Rhodocollybia butyracea almost perfectly. All three are identifications of sequences, not of mushrooms collected and named in the reserve.

The layer this study is about is the one under a walker's boots — the dark humus between the loose litter and the mineral sand, a few centimetres thick, from which every core came. Its thickness and colour differ between a fertile spruce stand and a poor pine stand, in Uppland and in Poland alike, and that difference is legible before anyone writes down a species list. The fruit bodies of the Cortinariaceae that appear and vanish each autumn are the smallest and briefest part of a mycelium which, if Fiby Urskog generalises at all, spends the rest of the year reading genes long filed under wood decay. Whether the same holds in a Polish pine forest is not something this paper can answer; it would take the same measurement, in the same layer, made again.