Drought redraws the underground: six Kansas years push prairie soils toward a narrower microbiome

For nine consecutive growing seasons at the Konza Prairie Biological Station in the Flint Hills of northeastern Kansas, sheets of plastic troughs have intercepted roughly two-thirds of every rainfall, simulating a drought that the surrounding tallgrass has rarely, if ever, experienced in the instrumental record. The troughs came down in year seven. The rain came back. The soil, a multi-author team reported this week, did not.

The finding, published in a peer-reviewed ecology journal and summarised by Phys.org on 14 July 2026, is one of the clearest empirical signals yet that the underground half of grassland ecosystems is slower to recover than the green canopy above it. Over six years of experimental drought, bacterial diversity in the treated plots declined measurably and community composition shifted toward taxa adapted to dry conditions. When the precipitation exclusion ended, those shifts did not simply reverse.

That is the result that matters for two very different audiences: ecologists trying to predict what the North American prairie will look like under a warmer, drier climate regime, and the much larger cohort of land managers, agronomists and policymakers who treat soil biodiversity as a quietly banked asset. Both groups have spent more than a decade assuming that microbial communities are, on climate-relevant timescales, elastic. Six years of Kansas data suggest the elasticity has limits, and that the limits may show up first in the soil.

The Konza experiment

The work draws on a long-running rainfall-exclusion experiment at Konza, a site the Nature Conservancy and Kansas State University have stewarded since the early 1970s as part of the US Long-Term Ecological Research network. Starting in the 2017 growing season, the team installed fixed rainout shelters over a set of native tallgrass plots, capturing about 66 percent of incoming precipitation during the growing season while leaving temperature, sunlight and ambient humidity largely untouched. The design, in other words, isolates drought as a variable from the confounding factors that bedevil observational climate studies.

"Six years of consecutive growing-season precipitation exclusion produced a reduction in microbial diversity and a shift in bacterial community composition towards more drought-tolerant taxa," the authors write, summarising the headline result. They sampled soils to roughly 15 centimetres across treatment and control plots, sequenced bacterial DNA, and tracked changes year over year. The control plots, rainfed throughout, gave the experiment its baseline.

The pattern is consistent with a body of shorter-term work on grassland and forest soils, but the multi-year frame matters. Single-season drought experiments often produce reversible effects; the longer the manipulation runs, the harder it becomes to separate a transient stress response from a genuine community reorganisation. The Konza plots have now passed the threshold where that distinction is unavoidable.

Counter-narrative: not yet collapse

A collapsing-soil narrative would be simpler to write, and a few pickup outlets have gestured in that direction. The data do not support it. Bacterial abundance, measured as total biomass, did not crater in the treated plots; diversity fell but the community remained functional, and the more drought-tolerant taxa that rose in the rankings include organisms associated with nitrogen cycling and organic-matter turnover. For ranchers and conservationists in the region, that distinction is not academic: a less diverse but still operative microbial workforce can keep the prairie's nutrient economy turning over, at least within the timescale of the experiment.

There is also a counter-narrative worth flagging on the question of recovery. The authors sampled the post-drought years, and they report that several diversity and composition metrics did not snap back to control-plot levels once the troughs were removed. But "did not return to baseline" is not the same as "continued to decline." The experiment does not yet answer whether the trajectories stabilise, partially rebound, or drift further with another decade of observation. Konza's long-term infrastructure makes that question answerable, which is part of why the site exists.

Structural frame: the slow carbon economy underneath

Soils hold somewhere between two and three times as much carbon as the atmosphere does, and grasslands like the Konza system are carbon sinks in part because their microbial communities are slow and inefficient at finishing the job of decomposition. Drought changes the terms of that bargain. As bacterial communities retool toward taxa that thrive on less water, the rates at which soil organic matter breaks down shift with them, with consequences for the carbon flux the biosphere sends back to the atmosphere.

That is the structural point that lifts the new paper out of its local setting. The tallgrass prairie is not a marginal biome; it stretches from the Texas Gulf coast to the Canadian prairies, and similar precipitation-exclusion experiments at other LTER sites have been finding convergent signals. Put plainly, what happens beneath the Flint Hills matters to anyone modelling the next century of the North American carbon budget. The microbial community is, in the language climate researchers increasingly prefer, a slow variable with a long memory.

A secondary structural point concerns inference. Most public discussion of climate-driven biodiversity loss runs on charismatic megafauna, range maps and tree-line shifts. Soil microbial communities are described as though they will absorb whatever the climate throws at them, on the grounds that bacteria reproduce fast and disperse everywhere. Multi-year experiments like Konza's erode that assumption. The slower the variable, the more it can cross a threshold invisibly, and the harder it is to reverse once crossed.

What to watch next

Three follow-up questions are worth tracking. First, whether the diversity decline continues, plateaus, or partially reverses in the post-2017 recovery years, which the Konza team will presumably sample as long as funding holds. Second, whether similar experiments at the more arid end of the prairie biome, where baseline microbial communities are already adapted to chronic water stress, show the same pattern or a muted one. Third, and most consequential, how the bacterial shifts map onto measured carbon and nitrogen fluxes, which the new paper does not yet report.

For policymakers, the take-away is unglamorous and important. Drought-resilience strategies aimed only at the visible prairie, the canopy, the forage value, miss the half of the system that takes a decade to break and possibly longer to fix. The Konza data argue for treating soil microbial communities as infrastructure, with the monitoring budget and the institutional patience that designation implies.

The Flint Hills will get its next growing season whether or not the troughs come back. The soil underneath, the new paper suggests, will carry the memory of a six-year drought that the rain has already stopped blaming itself for.

This publication approached the Konza study through Phys.org's coverage; the underlying paper supplies the headline figures and experimental design. Where the wire summary did not specify, we have used qualifying language rather than asserting numbers.