Aerosols from experimentally infected pigs were sampled over 24 days using an Andersen cascade impactor separating particles from 0.4 to 10 micron. Influenza A was detected in every size range, from 5.5 times ten squared RNA copies per cubic metre in the 1.1 to 2.1 micron range up to 4.3 times ten to the fifth in the 9.0 to 10.0 micron range. PRRSV was detected in every range except 0.7 to 2.1 micron, from 6 times ten squared copies per cubic metre at 0.4 to 0.7 micron up to 5.1 times ten to the fourth at 9.0 to 10.0 micron. PEDV was detected across all particle sizes and at higher quantities than either of the others. For all three, the highest concentrations were in the largest particles.
The paper's own conclusion is that airborne PEDV, influenza A and PRRSV are found across a wide range of particle sizes but that virus viability is particle size dependent. PRRSV concentration was significantly higher in large particles in the 4.7 to 9 micron range. That is the finding the carrier particle argument rests on: what a filter is asked to intercept is not a bare virion travelling alone but a virion on a host matrix of droplet nuclei, dust and respiratory secretions, and viability in transit depends on that matrix.
One number requires care. The figure 2.1 micron appears in this paper as the upper edge of a PRRSV RNA detection gap and as a stage cut point of the impactor used for sampling, which is an instrument bin edge rather than a biological or filtration threshold. It is not an infectivity floor, and it should not be restated as one. Infectivity was assessed by virus isolation on all positive and suspect samples, and the per size range isolation results sit in the paper's Table 2. Any statement about which particle sizes yielded viable virus must come from that table.