Most ecosystem benefits keep climbing with biodiversity across 423 pooled studies

The paper, in Nature Ecology & Evolution, is a synthesis ^[8]. The team pooled results other researchers had already published: 222,829 data points drawn from 423 studies ^[9]. That averages about 527 points per study ^[15]. Points from one study usually share a site and a protocol, so they are not independent observations. A few large studies can pull hard on a pooled curve.

The test was the shape of the curve. If species substitute for one another, a benefit such as pest control should rise with the first few species and then level off. Across most of the 23 categories it kept rising ^[1], and the authors conclude that this functional redundancy has been overestimated ^[2]. "Our findings challenge the comfortable assumption that ecosystems carry plenty of backup, so losing a species here and there doesn't change much," said Emma Moffett, the study lead and a lecturer in geography at King's College London ^[3].

The thing this account doesn't tell you is how steep those curves are, or how many of the 423 studies changed species numbers in controlled plots instead of comparing sites that already differed ^[9]. The second point decides how far a slope can be read as cause. A species-rich site may also be wetter or less disturbed, and either condition can raise a service by itself.

The ocean figure needs the most care. Carbon sequestration at sea had the strongest response of anything measured ^[4]. The authors say that result rests on a small number of datasets, and they identify it as an urgent gap in current knowledge ^[5]. A slope fitted to few datasets has wide error bars, so it can move a long way in either direction as data accumulate. "Carbon capture at sea responded to biodiversity more strongly than anything else we measured, and yet it's one of the least studied areas we found," Moffett said ^[6].

Flood and erosion protection behaved differently. It often depends on one or two foundational species, such as the shrubs that stabilise dunes ^[11]. For those services, which species survive matters more than how many. The authors argue conservation has to protect overall diversity and those irreplaceable species together ^[12].

The team also ran the database forward against IPCC socioeconomic scenarios ^[14]. "We can now predict the benefits biodiversity provides anywhere," said Will Pearse, an associate professor of evolutionary ecology at Imperial College London ^[7]. That reach is bounded by where the 423 underlying studies were done, and the marine data gap carries into the forecasts too ^[5]. The database and forecasts are public ^[13], so other groups can rerun them.

I think the pooled data justify dropping, for most services, the planning default that a few local extinctions cost little ^[1]. That view holds on one condition: the slopes in the paper have to be steep enough to matter over the handful of species a single land-use decision removes.