A meta-analysis of 281 sites found biological recovery after river restoration and almost no measurable improvement in what those ecosystems actually do.

River restoration in Europe is a large and legally driven enterprise. Fewer than 10 per cent of Germany’s rivers and streams have at least good ecological status by the Federal Environment Agency’s reckoning, and a great deal of money goes into remedying that — reshaping channels, removing bank reinforcement, reinstating meanders and gravel beds.

Whether it works depends on what is being measured. A meta-analysis in Global Change Biology, led from the Helmholtz Centre for Environmental Research with collaborators at 13 research institutions, finds a sharp divergence between two answers.

Species came back. The functions did not measurably follow.

Structure versus function

The distinction is the point of the paper. Biological structure means what is living in a stream — which species, how many, how much biomass. Ecosystem function means what the stream does: how much carbon its algae fix, how it processes nutrients, how fast leaves decompose in it.

The two are related but not equivalent, and restoration assessment has concentrated almost entirely on the first because it is far easier to measure. Counting invertebrates in a kick sample is routine; measuring denitrification rates is not.

“Hydromorphological restoration can improve biological structure, but ecosystem functions do not automatically recover as a result,” said Ulrike Haase, a doctoral researcher at TU Dresden and the study’s lead author. “Our findings show that improvements in stream structure or biological recovery do not necessarily indicate that ecosystem functions have recovered.”

What the numbers were

The team analysed 66 scientific studies covering 281 stream or stream sites across Europe, North America, Asia and Oceania.

They identified 27 functional parameters in that literature, and only nine had enough comparable data to pool — among them gross primary production, nutrient uptake, denitrification and organic matter decomposition.

That attrition from 27 to nine is itself a finding. Two-thirds of the functions anyone has thought to measure in a restored stream have been measured too rarely or too inconsistently to be compared across studies.

Of the nine, exactly one showed a significant difference: denitrification was lower in restored streams than in degraded ones. For the other eight, no clear differences appeared.

The one result, read carefully

A single significant finding pointing the wrong way invites misreading, and the authors head it off.

Denitrification is the microbial conversion of nitrate to nitrogen gas, which removes nitrogen from the water. It sounds like an unambiguously good thing for a stream to be doing, and lower rates sound like a loss.

“Lower denitrification rates may also indicate lower nutrient enrichment,” said Mario Brauns, a river ecologist at UFZ who coordinated the study. Denitrification is limited by how much nitrate is available, so a stream with less nitrogen pollution running through it will denitrify less — not because the microbes have stopped working but because they have less to work on.

Which is to say the one measurable difference may be a sign of improvement recorded as a decline, and that is a problem of interpretation rather than of data.

Brauns names the underlying obstacle: “However, we can assess whether an ecosystem function is actually recovering only if we know its status under appropriate reference conditions.” Without a figure for what denitrification, or primary production, or decomposition ought to look like in a healthy stream of that type, a measurement is a number without a verdict attached.

Whether restoration was asked the right question

There is a reading of these results that is less damning of restoration than the headline suggests, and it concerns what the intervention does.

Hydromorphological restoration changes the physical shape of a stream. It does not change what is coming down the catchment — the fertiliser, the sewage effluent, the sediment from fields. Several of the nine functions examined are governed primarily by water chemistry and nutrient loading rather than by channel form, and expecting them to recover because the banks were reshaped may be asking the wrong thing of the work.

The null results are also nulls. Eight functions showing no clear difference between restored and degraded streams could mean function did not recover; it could equally mean the pooled data lacked the power to detect a change, which is a live possibility when only nine of 27 parameters had enough data to analyse at all.

What is not in doubt is that nobody has been looking. Ecosystem functions have received little attention in restoration assessments, and standardised assessment methods and reference values are still lacking — which is the gap the paper is really about.

Why this matters legally

The timing is not accidental. The EU Nature Restoration Regulation, adopted in 2024, names ecosystem functions as a restoration objective — which means member states are now committed to restoring something nobody has an agreed way to measure.

The authors’ recommendations follow from that: add functional indicators to monitoring, target measures at specific functions rather than at structure alone, and link indicators to reference conditions so that a measurement can be judged.

“The main priority is to further develop and standardise functional indicators for practical use,” Brauns said.

That is a modest-sounding conclusion with real consequences. A regulation that requires function to be restored, assessed with tools that measure structure, will produce projects certified as successful on the available evidence and unexamined on the stated objective.

The release does not state the study’s funding sources.

---

Notes

1. Haase, U., et al. “Hydromorphological restoration of streams does not ensure the recovery of ecosystem functioning.” Global Change Biology, 2026. DOI: 10.1111/gcb.71126.
2. Helmholtz Centre For Environmental Research – UFZ. “Stream restoration improves biodiversity but not necessarily ecosystem functioning.” EurekAlert!, 22 October 2026. https://www.eurekalert.org/news-releases/1148824.

Photo by M. Hasson and M. Lapôtre