The 82-species story: why ERMI tests 36 of them: ExplainmyERMI

The 82-species MSQPCR panel

In the early 2000s, EPA researchers led by Stephen Vesper and Richard Haugland developed a quantitative PCR methodology (mold-specific qPCR, or MSQPCR) for measuring mold species in settled household dust at the DNA level, an alternative to culture-based methods, which miss non-culturable species and require living spores. The original MSQPCR panel, published in Meklin et al. 2004 (J Environ Monit 6:615–620), included 82 fungal species or related species groups. Each species had a validated assay with its own primer set, probe sequence, and GenBank-deposited reference DNA.

The same 82-assay panel was then applied to the EPA's 2006 American Healthy Homes Survey (AHHS), a nationally-representative dust-sampling program of approximately 1,100 U.S. homes (Vesper et al. 2007 reports 1,083 homes used in the ERMI development analysis). Every species on the original panel was measured in every home in that sample.

How 82 became 36, two papers, two selection criteria

The 82-species panel was reduced to a smaller working set via two distinct papers using related but different selection criteria.

Meklin et al. 2004 applied an exclusion threshold to identify species detected too rarely across moldy and reference homes to support statistical comparison. From the paper: “Species (or assay groups) measured by QPCR at average concentrations of less than 2 spores per 5 mg of dust in either moldy homes or reference homes were eliminated from further analysis.” That works out to ≈ 0.4 spores per mg of dust. Of the 82 species/groups, 57 were excluded, leaving 25 species/groups for the moldy-vs-reference ratio analysis.

Vesper et al. 2007 separately developed the Environmental Relative Moldiness Index using 36 species from the same 82-assay panel. Selection criteria for those 36 included detection frequency in the AHHS sample plus group classification, 26 water-damage indicator species (Group 1) and 10 environmental reference species (Group 2). The 36-species ERMI cohort is not identical to Meklin 2004's 25 retained species; some species Meklin's strict threshold excluded were retained for ERMI calculation, and some species/groups Meklin retained do not appear in the ERMI 36.

The roughly ~46 species not used in the ERMI calculation aren't “untested” or “unmeasured.” They are species the EPA did test using validated MSQPCR assays. They simply weren't selected for the index.

What kinds of species were excluded

The ~46 excluded species fall into three broad categories. Several of them have shown up since on commercial extended panels: see extended species for the eighteen we cover in depth.

Toxigenic species too rare in AHHS

Several mycotoxin producers were on the 82-species panel but detected too rarely to anchor an index. Examples: Aspergillus parasiticus (aflatoxins, full B + G set), Aspergillus terreus (citrinin and some-strain gliotoxin), Memnoniella echinata (simple trichothecenes; the dry-spore Stachybotrys relative), and the Penicillium ochratoxin A producers that were tested within the PenGrp1 group assay (P. aurantiogriseum, P. viridicatum, P. verrucosum).

Plant pathogens with sporadic indoor presence

Fusarium as a genus was not part of the original 82-species panel at all. It is the most consequential omission: Fusarium species are the principal producers of three major mycotoxin families (trichothecenes T-2 / HT-2 / DON / NIV, zearalenone, and fumonisins). Commercial labs have added Fusarium assays (at genus level or for F. solani specifically) precisely because the standard ERMI has this gap.

Less common Aspergillus and Penicillium species

Other species tested in the original panel but excluded by the inclusion threshold include various Aspergillus species outside the most common indoor members (A. carbonarius, A. clavatus, A. tamarii, A. wentii, and others) and additional Penicillium species (P. expansum, P. citrinum, P. roqueforti, P. citreonigrum) that show up on commercial extended panels but didn't make the 36-species cut.

Using the threshold as a reference bound

The exclusion threshold is the most useful fact in the 82 → 36 story. It works as a defensible ceiling on the true AHHS geometric mean (GM) for every excluded species:

The EPA tested the species across the ~1,100-home AHHS reference sample.
The geometric mean came in below the threshold of 2 spores per 5 mg dust (≈ 0.4 SE/mg).
The exact GM is unknown to us, but we know with certainty that it's less than 0.4 SE/mg.

That gives us a conservative elevation factor for any current dust detection of an excluded species. If a lab reports X SE/mg of (say) P. expansum, then:

True elevation = X ÷ GM, where GM < 0.4 SE/mg.
A conservative lower bound on the elevation factor is X ÷ 0.4.
Example: a detection of 4 SE/mg is at least a 10× elevation over the EPA reference ceiling, and likely substantially more, since the actual GM is somewhere below 0.4.
Example: a detection of 10 SE/mg is at least a 25× elevation.

This isn't a percentile band (we don't have the distribution), and it isn't a modified ERMI score (the calculation is per-species, not rolled into the index formula). It's the same detected-over-reference elevation logic the 36 ERMI species use, applied with a defensible upper bound on the reference for excluded species.

Caveat on units and approximation. Meklin 2004 states the threshold in spores per 5 mg of dust (≥ 2 spores per 5 mg, on average, retained; below that, excluded). We rescale to per-mg as 0.4 spores/mg, and treat 1 spore ≈ 1 spore equivalent (SE) for typical ERMI species. The actual cell-to-SE conversion can vary by species (multinucleate spores complicate the ratio), and the threshold itself is an averaging boundary across the AHHS sample, so 0.4 SE/mg should be read as “the order-of-magnitude conservative ceiling,” not a precise constant. The real GM for any excluded species may be substantially smaller, which means the real elevation factor for any given detection is likely larger than this rule of thumb produces.

What this means for a homeowner

If your dust test includes an extended species not on the 36, it's because the lab added a probe for it. The species was already validated by EPA and is real and measurable. It just wasn't common enough across the ~1,100-home AHHS reference sample to anchor an index calculation.

Any detectable level reported for an extended species is unusual relative to the EPA's national reference baseline. That's the qualitative anchor: the EPA did test for it, they characterized it across a national sample, and the answer was “present rarely enough that it doesn't belong in the index.” A lab that reports a hit is telling you the species is detectable in your dust right now.

What this does not mean

Extended-species results don't roll into your ERMI score. The 36-species formula is fixed; adding species changes the math and produces a number that isn't comparable to the EPA's validated index. Any lab-published “extended ERMI” or modified score is not an EPA-validated calculation.
Extended-species results aren't comparable to ERMI percentile bands. The percentile distributions the EPA published (quartiles, AHHS reference distribution) are for the 36 ERMI species. Extended-species hits don't map to those bands.
Extended-species results aren't directly comparable across labs. Each lab calibrates extended assays against its own internal standards, and no public reference distribution exists for any of them. A “★★★” on one lab's report and a “detected” on another's might not mean the same thing.

The 82-species story: why ERMI tests 36.