A Carnegie Mellon lab is coring a 178-acre Pittsburgh steel site for pollutant-eating bacteria

The selection argument is old and simple. A gram of soil can hold billions of bacterial cells drawn from thousands of species ^[7]. A compound like benzene is toxic to most of them, while a few already carry metabolic pathways that let them tolerate it or consume it as food, which gives those few an advantage over neighbours at the same site ^[8]. The survivors grow more common, and in bacteria that advantage can move sideways, because the genes pass directly to other bacteria and not only to offspring ^[9]. More than a century of steelmaking on that stretch of the Monongahela, including the Jones & Laughlin Steel Co., ran a long unplanned version of that experiment ^[3].

To read that experiment, the sampling never leaves the site. Soil comes from areas of Hazelwood Green with different historical levels of contamination ^[10], in cores that preserve the layers at each depth ^[11]. Decades of contamination records have been mapped onto the modern site, so microbial abundance can be compared across locations and depths ^[14]. Comparing a polluted site against a clean one elsewhere would change the soil, the setting and the management history at the same time. Staying inside one footprint holds those roughly constant.

It also inherits a confound. The contaminated soil at Hazelwood Green was remediated, and the site is now a centre for research, technology and community engagement ^[6]. The communities in those cores have lived through the pollution and through the cleanup, and a contrast drawn inside one site cannot say which of the two shaped them.

Then there is the distance between a gene and a cleanup. Metagenomic sequencing identifies the microorganisms in an environmental sample directly from its DNA rather than growing and studying one organism at a time ^[12], so what comes back is a catalogue of what is present. The author writes that DNA sequencing helps establish what microbes might be capable of, and that the lab isolates bacteria to test those predictions directly ^[15]. The second step is where a rate would come from: how fast an isolate degrades BTEX at the concentrations actually in the ground.

All of this rests on one site. 178 acres ^[1] works out to about 0.28 square miles ^[18], one industry on one river, studied by a lab two miles away ^[2]. Across Appalachia and the Rust Belt, former industrial sites are being turned into neighbourhoods, research centres and technology hubs ^[20], and a characterisation protocol that asks about resident biology would need more than a single Pittsburgh footprint behind it.

The prediction is worth testing. Hazelwood Green's soil held petroleum hydrocarbons, heavy metals and other pollutants ^[4], among them BTEX, some of it carcinogenic ^[5], which is the kind of sustained pressure that should leave a mark on a microbial community. I would expect a clearer signal in the genes than in the degradation rates, since a gene can persist in a community after the compound that favoured it is gone. For now the account covers the collecting and the comparing ^[16].