The drought that species-rich grasslands keep at bay

On 15 July 2026, a multi-site synthesis reported the cleanest signal yet that biodiversity does its heaviest lifting precisely when ecosystems are under the most strain. Across thousands of plots from as far apart as the Swiss plateau and the Inner Mongolian steppe, grasslands packed with plant species held their productivity far better during extreme drought than species-poor grasslands did. Forests, the same study found, showed no comparable insurance effect. The pattern reframes a long-running debate: biodiversity is not just a long-run yield question, it is a climate hedge, and the asset most worth hedging turns out to be the one we have spent the last century ploughing up.

The result matters because grasslands cover more of the planet's land surface than forests do, store roughly a third of terrestrial carbon, and supply the bulk of the world's meat and dairy. They are also among the first ecosystems that warming will push past their historic range. Treating species richness as an optional input, the way some agricultural economics still implicitly does, misprices the most drought-sensitive land use on Earth.

What the data actually show

Researchers pulled together standardised drought-experiment plots from long-running European biodiversity experiments and comparable sites in the Inner Mongolian grasslands of northern China, then compared two extremes: plots engineered with very low species counts (one to four species) versus plots built up to the local diversity maximum. In a non-drought year, species-rich plots were already more productive. Under experimentally imposed extreme drought, the productivity gap widened, with the diverse grasslands losing a much smaller share of their biomass than the species-poor ones.

That is not a small lift. The synthesis points to two compounding effects that show up repeatedly in the underlying trials. First, diverse plant communities tend to include species with different rooting depths, leaf phenologies, and water-use strategies, so as the topsoil dries out the system still has access to moisture and active growth at depth. Second, drought years are exactly when canopy gaps open in species-poor swards; a richer seed bank and a wider portfolio of growth forms close those gaps faster, suppressing the bare-soil feedback that locks in yield losses. Forests, with their closed canopies and deeper rooting profile, lost less absolute productivity under drought, so the relative insurance value of diversity was smaller to begin with.

The counter-narrative worth taking seriously

The natural reading of the headline finding would be that farmers should sow ever more seeds and watch the rows turn green even in a bad year. That is not what the paper says, and the authors are careful about it. Variability also rises with diversity in some contexts; not every mix is more stable in every season. Several of the underlying trials record years in which species-rich plots actually under-perform a well-tuned low-diversity sward, particularly on fertile soils where a single high-yield grass can out-compete everything else. The pandemic-era work that initially suggested "insurance" effects from biodiversity later had to be qualified once third-party replications caught up.

A second objection comes from the gap between experimental plots and working farms. Real grasslands are grazed, fertilised, reseeded, sometimes irrigated, and above all managed for a particular product, milk, beef, lamb, hay, that the experimenters do not sell. The insurance value of diversity inside a fenced research plot is a poor predictor of the insurance value a farmer experiences, because farmers are constantly trading stability for peak yield. The honest version of the paper's claim is therefore narrower than the press release: under extreme drought, the productivity advantage of species-rich grassland widens; whether that advantage survives the rest of the management package is a separate question, one the synthesis explicitly flags for follow-up.

Why the pattern sharpens as the climate warms

The synthesis lands in a region that is on course for a step-change in growing-season dryness. Southern and central Europe has just registered a run of summers in which soil-moisture deficits have pushed pasture into dormancy weeks earlier than the 1981-2010 baseline; northern China's steppe belt is modelling a comparable shift. Under those conditions, the absolute penalty for being species-poor climbs. A productivity loss that is immaterial in a mild year can be the difference between a viable farm and a forced sale in a bad one.

That is the structural frame. As average conditions drift toward what used to be the bad end of the distribution, the ecosystems we count on to absorb the shock are increasingly the ones we have industrialised the hardest. Europe has lost roughly half of its species-rich permanent grassland since the 1970s, much of it converted to simplified ryegrass or to cropland. China has lost Inner Mongolian grassland at comparable scale, with policy-driven grazing restrictions now attempting the unusual task of bringing it back. If biodiversity is the cheapest insurance available for a hotter climate, both regions are systematically under-insured.

What to watch next

The synthesis is built on experimental drought, which is a contrived worst case and not a forecast. The next test is whether the same insurance effect shows up under the slower, messier droughts that real weather delivers, in pastures that have been simplified for decades, and in systems where grazing pressure interacts with diversity loss. The European and Chinese groups behind the synthesis have signalled a follow-on design that keeps the multi-site plot architecture but extends monitoring into commercially managed grassland rather than fenced experiments.

For now, the practical read is conservative. Where grasslands are being reseeded or rotated, the evidence supports moving from monoculture ryegrass toward a small portfolio of locally adapted species, including legumes and deep-rooted forbs; the drought pay-off, where it exists, comes from functional diversity, not from maximising species counts. Where policy is choosing between restoring species-rich permanent grassland and converting to other uses, the drought-insurance argument is now strong enough to count as a material factor in the cost-benefit ledger. And where climate adaptation plans still treat biodiversity as an amenity rather than as infrastructure, this synthesis is the cleanest available reason to update the spreadsheet.