Active pharmaceutical ingredients are used worldwide to improve health outcomes but are also ubiquitous environmental contaminants. However, a detailed picture of pharmaceutical risk in global surface waters is largely missing owing to inherent challenge in monitoring pharmaceuticals across all surface waters and assessing mixture effects. Here we generated a 2-km snapshot map of acute mixture toxicity for 53 pharmaceuticals to an invertebrate, alga and bacterium in global rivers, using a machine learning model trained on 1,017 laboratory toxicity assays and 24 geospatial predictors. We predicted pharmaceutical exceedance (mixture toxicity ≥5% regulatory threshold) in 13% of riverine grid cells, with hotspots in Southern China, Northern India, Central and Western Europe and the Southeastern USA. Source proximity and hydrochemistry directly drove pharmaceutical exceedance, while socioeconomic and climatic factors exerted cascading effects by shaping source distribution, hydrochemistry and topography. Encouragingly, counterfactual scenarios indicated that targeted management of six key sources could reduce the river exceedance area by 54–92%, with the greatest benefits from livestock, aquaculture operations and wastewater treatment plants. Moreover, we identified nine priority pharmaceuticals, contributing 54–90% of mixture toxicity. Our work provides a policy atlas for monitoring, intervention and regulation and offers a transferable framework to address other contaminants and endpoints. Pharmaceutical mixture toxicity exceeds ecological thresholds in 13% of global rivers, and targeted control of key pollution sources could substantially reduce these risks, according to an assessment of pharmaceutical pollution in rivers worldwide.
The interannual variability of the atmospheric carbon dioxide growth rate has become increasingly tightly coupled to fluctuations in the land carbon sink, yet the regional basis of this strengthening remains unclear. Here we combine global atmospheric carbon dioxide records with multiple estimates of land carbon fluxes from atmospheric inversions, process-based vegetation models and observation-constrained datasets to quantify changes in regional contributions to the detrended interannual correlation between atmospheric carbon dioxide growth and the land carbon sink over 1959–2022. We find a pronounced spatial reorganization of these contributions. In the atmospheric inversions, the land area reinforcing the correlation increased from 78.1% to 82.7%, whereas the process-based model ensemble showed a slight contraction in area but a stronger contribution from the regions that remained. Both datasets identify concentrated reinforcing contributions in the tropics and frequent sign reversals at higher latitudes, consistent with strengthening trends in the global correlation of −0.026 per decade and −0.031 per decade, respectively. These changes were associated with shifts in land carbon sink variability and its sensitivity to temperature and water availability. Together, the results suggest that under intensifying climate variability, regional land carbon anomalies are becoming increasingly coherent, amplifying their influence on atmospheric carbon dioxide variability. Changes in the regional contribution to the flux of carbon from land have become more coupled to interannual atmospheric carbon dioxide fluctuations from 1959 to 2022, making land carbon sink variability more detectable, according to models and observations.
Efforts to better recognize technical staff within academia are gaining momentum, but further changes — such as clearer career pathways, stable funding, and greater inclusion in research leadership — are needed to future-proof their role in advancing geosciences.
The Early Eocene Climatic Optimum (53 to 49 million years ago) was an interval in Earth’s history when reconstructed temperatures and atmospheric CO2 concentrations were among the highest of the Cenozoic Era (that is, the last 66 million years). The causes of the high CO2 levels remain unclear. CO2 changes over recent ice age cycles appear to be largely controlled by the biologically mediated build-up and release of carbon from the Southern Ocean, a process that might have also played a role earlier in the Cenozoic. Here we test this possibility with foraminifera-bound nitrogen isotope measurements. Our data show higher surface nitrate concentrations in the Southern Ocean, and thus a more inefficient biological sequestration of carbon during peak Early Eocene Climatic Optimum. The associated transfer of CO2 from the deep ocean to the atmosphere would have been amplified by the low seawater buffering capacity of the Eocene, due to the lower pH and high Ca ion concentration. Previously proposed low silicate rock weatherability during the Eocene may have weakened the weathering thermostat, allowing the reconstructed change in biological pump efficiency in the Southern Ocean to explain at least half of the atmospheric CO2 increase observed during the Early Eocene Climatic Optimum. Warming in the early Eocene reduced the efficiency of the Southern Ocean biological carbon pump, limiting carbon sequestration in the ocean interior and enhancing atmospheric CO2 levels, according to foraminifera-bound nitrogen isotope records.
Vincent Thiéry and Migueli Chedrewih discuss how studying ettringite’s natural, geologic occurrences can inform understanding of concrete’s material properties and ageing processes.
