Imagine you're running a restaurant kitchen with 20 cooks, each with a slightly different specialty. The old assumption was that after about 5-6 cooks, you've got all the skills covered — the rest are just backup. This paper says no: cook number 18 is still making unique dishes the first 17 couldn't. Remove them, and the menu shrinks. That's functional redundancy, and the world's largest test of it just came back negative. The committed claim: across 23 categories of ecosystem services and functions, biodiversity-benefit curves do not saturate. They keep climbing. The 'functional redundancy' assumption — that species substitute for one another, so a few losses barely register — is overestimated at global scale. This matters because conservation triage and economic cost-benefit models have long leaned on redundancy to justify acceptable losses. If the curve doesn't flatten, every extinction carries more marginal cost than the models priced in. The study combined 423 published studies and 222,829 data points spanning land, freshwater, marine, and estuarine ecosystems — a database more than twice the size of the previous largest effort. The team, spanning King's College London, Imperial College London, the Natural History Museum, and The Alan Turing Institute, built predictive models across all 23 service categories. The architecture is meta-analytic synthesis: pooling effect sizes from hundreds of primary studies, then fitting biodiversity-service response curves. This is classical ecological statistics at massive scale, not a novel algorithmic family — its power comes from sheer data breadth and geographic coverage. The most striking single result: ocean carbon sequestration showed the strongest positive response to biodiversity of any service measured. Blue carbon sinks — coral reefs, salt marshes, mangroves, phytoplankton-driven carbon pumps — depend on marine diversity far more steeply than other services depend on their respective species pools. The authors flag this result honestly: it rests on a small number of underlying datasets and they identify it as an urgent knowledge gap. If you're counting on blue carbon for climate mitigation, this is a load-bearing finding built on thin evidence. Not everything scales with diversity. Coastal hazard protection — flood and erosion control — proved relatively insensitive to biodiversity because it often hinges on one or two foundational species (the shrubs that stabilize dunes, the mangroves that break waves). This exception actually strengthens the paper's credibility: the authors aren't claiming a monotonic relationship everywhere. They're saying the default assumption of saturation is wrong for most services, while correctly noting that some services depend on keystone rather than portfolio effects. The team linked their database to IPCC socioeconomic scenarios and produced a forward projection: biological pest control on farmland declines under fossil-fuel-heavy development paths compared with lower-emission trajectories, with the sharpest losses hitting countries with rapid population growth and lower development levels. This is where the paper crosses from ecology into policy: the people who can least afford to replace free ecosystem pest control with synthetic alternatives are projected to lose it fastest. The full database and model forecasts are publicly available, which is the right move for a study making claims this broad. The question now is whether independent teams can replicate the response-curve finding with different analytical choices — meta-analyses are sensitive to inclusion criteria and effect-size extraction. The ocean carbon result, in particular, needs primary data before anyone should build policy on it.
