This paper by John Chase and crew posted recently on PeerJ takes a deep look at the office microbiome composition and concludes that the biggest factors affecting differences in swabbed samples are the geography of the office and where the sample was taken inside it. Their experiment showed minimal impact from other factors that are often intricately tied with these two, including human interaction, seasonal variation, material type, and environmental parameters such as light, temperature, and humidity.
Since carpet is not often found on ceilings, they designed an experiment to pull apart such elements rather than take samples from already existing spaces. They put swatches of different materials on the ceilings, walls, and floors of three office buildings with similar occupancy schedules in each of three geographically distinct cities, for a total of nine offices, and sampled over the following seasons for 16S (bacteria) and ITS-1 (fungal.) They even sampled the samplers!
Their results supported previously shown observations that human skin is important for these environments. Comparing samples from offices within and between cities, they found that the city itself, aka geography, was a clearer predictor of the microbes present than was the office itself. Few fungi found, material matters minimally, and unweighted UniFrac undertaken was not united with weighted Unifrac when weighing where the sample was in the workplace. This last observation is described by the authors in the following passage:
“Interestingly, within a location, the pattern of differences among samples varied between floors and walls/ceilings. Using unweighted UniFrac, floor samples were more similar to other floor samples than wall/ceiling samples were to other wall/ceiling samples (Figure 2). In contrast, using weighted UniFrac, floor samples were less similar to one another. Thus, community differences among floor samples were driven primarily by differences in the relative abundance of the same OTUs, while the differences among wall/ceiling samples were driven more by the presence or absence of particular OTUs. The same pattern was statistically significant in all cities, but strongest in Flagstaff.”
There are a couple important possibilities for discrepancy that the paper addresses. For example, the authors are sure to acknowledge that what they are picking up from the swabs, especially on the horizontal surfaces, might be an accumulation of dead matter. This is important because if there is a high amount of non-living DNA being sequenced in these samples, it could interfere with the signal of a less abundant and more unique living community that the ceiling and wall samples might be picking up, falsely distinguishing location as a major driver for the microbiome composition. Another important sequencing variable that the authors found (by putting theoretical standards into each of the runs) was that the three sequencing runs they used were not entirely consistent, so they were careful to consider that when comparing samples across runs.
There is a lot more in the paper!