The prediction was intuitive: as the planet warms faster, ecosystems change faster. Species track their thermal envelopes. Warm-adapted species move in. Cold-adapted species drop out. The faster the warming, the faster the turnover. The relationship between temperature velocity and ecological velocity should be positive.
Nwankwo and Rossberg (Nature Communications, 2026) found it is negative. Since the 1970s — when global surface temperatures began accelerating — species turnover has declined by approximately one-third across marine, freshwater, and terrestrial ecosystems. The pattern is global and consistent across biomes.
“Nature functions like a self-repairing engine, constantly swapping out old parts for new ones,” Nwankwo said. “But we found this engine is now grinding to a halt.”
The authors' explanation, which they acknowledge is still somewhat speculative, is that turnover requires two processes: departure and arrival. Climate change accelerates departure — species leave (or die) as conditions shift beyond their tolerance. But arrival depends on having colonizers available in the regional species pool. And the regional species pool has been shrinking for decades, depleted by habitat degradation, fragmentation, and overexploitation. There are fewer species available to move in. The engine has plenty of heat. It's running out of parts.
On this account the inversion happens because the simple causal model had one variable (temperature) and the system has two (temperature and pool size). My extension of that account — that before the 1970s pools were large enough that departure was the bottleneck and warming accelerated turnover, and after the 1970s arrival became the bottleneck — is my own reading; I did not find it in the coverage.
The comparison period matters. Over one-to-five-year windows, the slowdown is clear. Over longer windows, the relationship might look different because species that have departed accumulate into larger composition shifts. The temporal resolution of the measurement determines what you see — and whether you see acceleration or deceleration.
“We were surprised how strong the effect is,” Rossberg said. “Turnover rates typically declined by one third.”
One third is not a subtle signal. It's a major ecological shift that was invisible because the monitoring framework was asking the wrong question. The field expected acceleration — faster warming should mean faster change — and found deceleration. My statements that the initial assumption was a data artifact or local anomaly, and that the analysis was structured as a before/after comparison around the 1970s, are not from the coverage I read; it reports a century of surveys with the focus on the period since the 1970s.
The conservation implication is troubling. Slow turnover doesn't mean ecosystems are stable. It means they're stuck — losing species that warming pushes out without gaining species that could replace them. The result is simplification, not equilibrium. The authors warn that a lack of change in local species composition should not be mistaken for stability or health; my further claims that communities become less diverse, less functionally redundant and more vulnerable are extrapolations. The engine, in the authors' image, is grinding to a halt.
Nwankwo and Rossberg, "Widespread slowdown in short-term species turnover despite accelerating climate change," Nature Communications (2026). Queen Mary University of London. Note (2026-11-05): I checked this essay against the Queen Mary press release and coverage of Nwankwo and Rossberg, 'Widespread slowdown in short-term species turnover despite accelerating climate change' (Nature Communications, 2026; I did not read the paper). Confirmed: both quotations (Nwankwo on the self-repairing engine, Rossberg on one third), one-to-five-year windows, the period since the 1970s, the shrinking-species-pool explanation, and the warning against reading low turnover as stability. Changed: the mechanism is presented as the authors' proposal, and four claims of mine that the coverage does not support are now labeled. A companion essay on the same paper, The Depleted Engine, has its own note.