The design does two jobs, and only the first is about geography. Step one sequenced the Antarctic Arthrobacter strains and compared their genomic traits against strains held in public databases [7], which is a search for absence: 90 percent of the Antarctic strains matched nothing anywhere else [8]. A failed match is only ever as good as the collection you failed to match against, so the comparison set carries more weight here than the percentage does. The team drew its comparators from the places most likely to break the hypothesis rather than the places most likely to flatter it, sampling the Tibetan Plateau, Svalbard and Chile's Atacama Desert [5], and pitting nearly 500 strains from elsewhere against more than 100 from the continent [6]. That is about 600 strains in total, with the rest of the world outnumbering Antarctica roughly five to one [14]. Cold, dry soils were the hardest ground for an endemism claim to survive on, and it survived there.
Run the percentage the other way and roughly ten of the Antarctic strains did have counterparts elsewhere [15]. Those are the ones I would want the raw data on, because under the old assumption they are unremarkable and under the new result they are anomalies worth explaining.
Step two took the strains into Petri dishes under mimicked Antarctic temperature and dryness, where the Antarctic isolates grew more slowly but proved very resilient [10]. That result speaks to how well a strain endures once it has landed, a separate question from how it got there in the first place. Microbes ride updrafts into the atmosphere and ocean currents out to sea [17], so nobody doubted that cells reach the continent; the question was always what happens after they land. Slow, tough, locally suited strains are consistent with dispersal that is rare, and equally consistent with dispersal that is common and almost always fatal. The two readings imply different contamination risks for anyone tracking human introductions, and this experiment does not separate them.
This result carries a clear limit: it speaks to one cultivable genus and says nothing quantitative about the rest of Antarctic soil life. Arthrobacter was chosen because it is a widely distributed soil genus and because it grows readily in a lab [4], and lab tractability is exactly the trait that makes a group unrepresentative of the soil microbiome. Byron Adams of Brigham Young University, a co-author, puts the finding at strain level rather than group level: Arthrobacter is global, individual Antarctic strains are not [11]. Fierer is careful in the same direction, noting that only a few endemic microbes had been found before, in very specific environments, and that whether other microbial groups show the pattern is an open question [9]. Dragone's own framing is that Antarctica has long been treated as the likely exception to "everything is everywhere" and that these results support the idea [3].
My read, according to the account published with the paper: this is the sturdiest counterexample the assumption has faced, and it is a counterexample for one cultivable genus, resting on decades of Antarctic sampling by international teams [12] and on the coverage of databases whose gaps are not in the figure. Treat it as a bound on a general law rather than a replacement for one, and be suspicious of any bioprospecting or contamination claim that leans on "found nowhere else" without arguing how hard anyone looked.
