Nature Geoscience

Published by Springer Nature
Online ISSN: 1752-0908
Learn more about this page
Recent publications
Sea level rise causes barrier islands to migrate landward. Coastal evolution modelling reveals a centennial-scale lag in island response time and suggests migration rates will increase by 50% within the next century, even if sea level were to stabilize.
Thin-section map and mineral chemistry of EC 002
a, Back-scattered electron image of the studied EC 002 thin section. Large crystals of augite are shown set in a fine-grained ground mass of oligoclase and an SiO2-rich phase. A xenocrystic low-Ca pyroxene rimmed by augite is visible in the section. b, Ternary diagram of EC 002 pyroxenes compared with brachinites and evolved achondrites3,13,32–34. c, Ternary diagram of EC 002 feldspars compared with brachinites and evolved achondrites. Data are from the same sources as b.
Bulk rock lithophile element chemistry of EC 002 and similar achondrite
a, Total alkali (Na2O + K2O) silica (SiO2) diagram for EC 002, evolved achondrites, the terrestrial continental crust and bulk silicate Earth (BSE). Domain boundaries from ref. ³⁵. b, CI-chondrite-normalized REE patterns for the same samples. Data are from GRA 06128/9³, NWA 11119⁴, a dacitic clast in a howardite¹⁰, the high-Si Almahata Sitta clast ALM-A⁹, high-Si achondrites NWA 11575¹¹ and NWA 2191³⁶, terrestrial continental crust¹ and BSE³⁷.
Bulk rock CI-chondrite-normalized HSE abundances of achondrite meteorites and the terrestrial UCC
Note the low and fractionated HSE pattern of EC 002 relative to Brachina and GRA 06128/9. Angrite and eucrite HSE patterns show highly depleted and fractionated patterns similar to EC 002. Sources of data: GRA 06128/9 and brachinites3,13,32,33, diogenites¹⁹, angrites²⁰ and UCC⁶. HSE are ordered from left to right by 50% condensation temperature, with Re being the most refractory.
Unlike the other terrestrial planets, Earth has a substantial silica-rich continental crust with a bulk andesitic composition. A small number of meteorites with andesitic bulk compositions have been identified that are thought to be the products of partial melting of chondritic protoliths, a mode of petrogenesis distinct from that of Earth’s continental crust. Here we show, using geochemical analyses, that unlike other known andesitic meteorites, Erg Chech 002 has strongly fractionated and low abundances of the highly siderophile elements and mineralogy consistent with origin from a melt. The meteorite’s bulk composition, which is similar to terrestrial andesites, cannot be explained by partial melting of basaltic lithologies and instead requires a metal-free chondritic source. We argue that Erg Chech 002 probably formed by ~15–25% melting of the mantle of an alkali-undepleted differentiated asteroid. Our findings suggest that extensive silicate differentiation after metal–silicate equilibration of chondritic parent bodies was already occurring within the first 2.25 million years of Solar System history and that andesitic crust formation does not necessarily require plate tectonics.
Rocket emissions and debris from spacecraft falling out of orbit are having increasingly detrimental effects on global atmospheric chemistry. Improved monitoring and regulation are urgently needed to create an environmentally sustainable space industry.
Analyses of the 2014 Iceland–Holuhraun volcanic eruption revealed the emitted aerosols induced a 10% increase in cloud coverage above the region, suggesting anthropogenic aerosols might strongly cool the Earth’s climate by increasing the cloud coverage.
Comparison between ML-MODIS predictions and MODIS observations
a–d, Validation against non-perturbed observations (excluding 2014) of cloud properties: Nd (cm–3) (a), reff (μm) (b), LWP (g m–2) (c) and CF (unitless) (d). e–h, Volcanic perturbation signals in October 2014, indicated by the difference between the machine-learning predictions and the observations: Nd (cm–3) (e), reff (μm) (f), LWP (g m–2) (g) and CF (unitless) (h). October MODIS observations from Aqua (2002–2020) and Terra (2001–2020) are analysed. Colour indicates the normalized data density function with a maximum value of 1.0, with 80% of the data being contained within the black dashed area.
Source data
Changes in cloud properties caused by the volcanic perturbation estimated using machine-learning predictions and MODIS observations for October 2014
a–c, The spatial distribution and zonal means of the changes (Δ) in Nd (a), reff (b) and CF (c). d–f, Probability density functions (so that the areas under the curves are equivalent) for MODIS and ML-MODIS for Nd (d), reff (e) and CF (f).
Source data
Responses of cloud properties to the volcanic aerosol perturbation in October 2014
The ACI signals of responses (fingerprints) are indicated as the ratios between MODIS (Aqua and Terra) observations and machine-learning predictions: ratio = MODIS divided by ML-MODIS. Uncertainties of non-perturbed baseline references are estimated using a Monte Carlo method and are shown in black (Methods; based on non-volcanic October datasets spanning 2001–2020). The variability of the cloud responses to the Holuhraun volcanic aerosol perturbation is shown in pink. The box plots show 10th, 25th, 75th and 90th percentiles and median (Med.); the mean value is indicated by a dot. The susceptibilities of reff, LWP and CF to changes in Nd are given in a green colour, median [90% confidence interval]. Area (in units of km²) weighted averaging is used to calculate average cloud properties over the geographical region (Fig. 2), in order to estimate an unbiased large-scale response signal. Therefore, the ratios shown here are slightly different from the slopes shown in Fig. 1, in which area-weighted averaging is not applied.
Source data
Aerosol–cloud interactions have a potentially large impact on climate but are poorly quantified and thus contribute a substantial and long-standing uncertainty in climate projections. The impacts derived from climate models are poorly constrained by observations because retrieving robust large-scale signals of aerosol–cloud interactions is frequently hampered by the considerable noise associated with meteorological co-variability. The 2014 Holuhraun effusive eruption in Iceland resulted in a massive aerosol plume in an otherwise near-pristine environment and thus provided an ideal natural experiment to quantify cloud responses to aerosol perturbations. Here we disentangle significant signals from the noise of meteorological co-variability using a satellite-based machine-learning approach. Our analysis shows that aerosols from the eruption increased cloud cover by approximately 10%, and this appears to be the leading cause of climate forcing, rather than cloud brightening as previously thought. We find that volcanic aerosols do brighten clouds by reducing droplet size, but this has a notably smaller radiative impact than changes in cloud fraction. These results add substantial observational constraints on the cooling impact of aerosols. Such constraints are critical for improving climate models, which still inadequately represent the complex macro-physical and microphysical impacts of aerosol–cloud interactions.
Antarctica preserves Earth’s largest ice-sheet, which in response to climate warming, may lose ice mass and raise sea level by several metres. The ice-sheet bed exerts critical controls on dynamic mass loss through feedbacks between water and heat fluxes, topographic forcing, till deformation and basal sliding. Here we show that sedimentary basins may amplify critical feedbacks that are known to impact ice-sheet retreat dynamics. We create a high-resolution subglacial geology classification for Antarctica by applying a supervised machine-learning method to geophysical data, revealing the distribution of sedimentary basins. Hydro-mechanical numerical modelling demonstrates that during glacial retreat, where sedimentary basins exist, the groundwater discharge rate scales with the rate of ice unloading. Antarctica’s most dynamic ice streams, including Thwaites and Pine Island glaciers, possess sedimentary basins in their upper catchments. Enhanced groundwater discharge and its associated feedbacks are likely to amplify basal sliding and increase the vulnerability of these catchments to rapid ice retreat and enhanced dynamic mass loss.
WTL effects on global wetland NEE and total GHG emissions
a, Relationship between NEE and sum of three GHG (CO2, CH4 and N2O) net fluxes in different WTLs, drawn from 174 site-year records that reported three GHGs. b, NEE, CH4, N2O and sum of three net fluxes for different WTL conditions. c–f, Total (c) and individual (CO2 (d), CH4 (e), N2O (f)) GHG fluxes for the six different WTLs considered. Points in each box are sampled from the original dataset (3,672 site-year records total) with 1,000 bootstraps. Different letters in the boxes indicate significant differences (P < 0.01) between various WTLs based on nonparametric Wilcoxon signed-rank tests. Bold vertical lines show the median; boxes indicate the middle two quartiles; horizontal lines indicate the non-outlier range. Note that x axes have been truncated for enhanced readability.
Source data
Nonlinear hydrothermal influence on GHG exchange
a–c, Dependency of GHG emissions in boreal (long-term average air temperature <4 °C) (a), temperate (long-term average air temperature 4–17 °C) (b) and tropical (long-term average air temperature >17 °C) (c) regions on WTL and climate. d, The ‘mean’ groups are calculated from equilateral weighted averages in each climate regime. Dots and shadows represent mean ± 1.96 SEs. e–h, Contribution ratios of NEE, CH4 and N2O to the sum of three GHG net fluxes in the three climatic regions (boreal (e), temperate (f) and tropical (g)) and the mean (h).
Source data
GHG emissions from degraded wetlands under different scenarios
a, Time series of GHG emissions from degraded wetlands under three scenarios since 1950. The emissions are constrained by natural WET index and soil organic carbon pool. ‘History trend’ is history-derived scenario. ‘Rewet (ALL)’ and ‘Rewet (high-OCS)’ are based on the rewetting restoration of all and only high organic carbon stock wetlands, respectively. b, GHG net flux from degraded wetlands in main countries and continents over different periods. ‘Others’ refers to the sum of GHGs from countries that were not in the top ten of GHG emitters. EU, Europe; NA, North America; SA, South America; AS&OA, Asia and Oceania; AF, Africa.
Source data
Spatial pattern of the GHG emissions owing to wetland degradation and reduction potential via rewetting wetlands
a,b, The GHG emissions owing to wetland degradation under history-driven scenario in 1950–2020 (a) and under history-driven, business-as-usual scenario in 2021–2100 (b). (c) The reduction potential under Rewet (ALL) scenario in 2021–2100.
Source data
Carbon and nitrogen losses from degraded wetlands and methane emissions from flooded wetlands are both important sources of greenhouse gas emissions. However, the net-exchange dependence on hydrothermal conditions and wetland integrity remains unclear. Using a global-scale in situ database on net greenhouse gas exchanges, we show diverse hydrology-influenced emission patterns in CO2, CH4 and N2O. We find that total CO2-equivalent emissions from wetlands are kept to a minimum when the water table is near the surface. By contrast, greenhouse gas exchange rates peak in flooded and drained conditions. By extrapolating the current trajectory of degradation, we estimate that between 2021 and 2100, wetlands could result in greenhouse gas emissions equivalent to around 408 gigatons of CO2. However, rewetting wetlands could reduce these emissions such that the radiative forcing caused by CH4 and N2O is fully compensated by CO2 uptake. As wetland greenhouse gas budgets are highly sensitive to changes in wetland area, the resulting impact on climate from wetlands will depend on the balance between future degradation and restoration.
Neogene and late Miocene climate changes
a, Benthic foraminifera δ¹⁸O stack for the Neogene⁶¹. b, Collection of published CO2 estimates for the past 20 Myr using phytoplankton (δ¹³C of phytoplankton compounds, blue)13,25–27,37,38,62–66, boron isotopes (orange)12,32,50,62,64,67–72, C3 plants (green)²⁹, palaeosols (teal)73–77 and stomata (purple)78–83. Error bars denote reported 2 s.d. Data compilation built on previous CO2 compilation by ref. ³⁵. c,d, Twenty-point running mean δ¹⁸O (c) and δ¹³C (d) of benthic foraminifera records from ODP (Ocean Drilling Project) 999 (green)⁸⁴, IODP (International Ocean Drilling Project) U1338 (light blue)²², ODP 982 (red)⁴⁰ and ODP 1147 (dark blue)²¹. An equilibrium correction (+0.64‰) applied to all δ¹⁸O records85,86. e, SST records used in late Miocene temperature stack recalibrated using BAYSPLINE⁴³. Records are coloured according to latitude: >50° N, pink¹⁶; 30–50° N, yellow¹⁶; the tropics, green16,87–89; 30–50° S, purple¹⁶. Site information and citations for all datasets are available in Supplementary Table 1.
Records of late Miocene climate change
a, δ¹¹Β of T. trilobatus (this study, blue circles). Dashed lines represent mean δ¹¹B for 7–6 Ma and 6–5 Ma. Previous planktic foraminifera δ¹¹B data from ODP 926 (red circles) and ODP 1000 (red triangles)¹². Error bars denote 2 s.d. b, Mg/Ca SST estimates from ODP 926 (filled blue circles, this study), ODP 1146 (yellow²¹; orange⁹⁰) and IODP U1338 (empty blue circles)²³. c, CO2 estimates (this study, blue filled circles) using δ¹¹B (data in a). Previous δ¹¹B CO2 estimates from ODP 926 (red circles) and ODP 1000 (red triangles)¹². Error bars denote 2 s.d. Previous estimates using phytoplankton as shown in Fig. 1b (Previously collected phytoplankton CO2 estimates shown as red crosses⁶⁴, filled blue diamonds⁶⁶, empty blue diamonds²⁶, blue plus³⁷, blue triangles²⁷; errors bars omitted for clarity but displayed in Fig. 1b). Previous estimates using C3 plants (green circles)²⁹. d, ΔFCO2 using δ¹¹B (data in a; blue filled circles) with smoothing spline (bold blue line; Methods). e, ΔSST stack. f, ΔGMST stack. Error bands encompass 68% (dark red/blue) and 95% (light red/blue) of 10,000 Monte Carlo simulations for δ¹¹B-derived CO2 and ΔFCO2 estimates and ΔSST/ΔGMST stacks (see Methods for full details). See Supplementary Table 1 for full ΔSST/ΔGMST output and Supplementary Table 2 for raw data and CO2/Mg/Ca SST/ΔFCO2 estimates.
Latest Miocene climate sensitivity regression and key climates sensitivity studies throughout the past 70 myr
a, Cross-plot of smoothed ΔFCO2 against ΔGMST for the latest Miocene with 95% confidence interval (error bars). Regression lines fitted by SIMEX regression (solid grey line) with 95% confidence interval (dashed grey lines; Methods)⁹¹. b, Probability density function of latest Miocene ECS calculated when scaling ESS to ECS. Bold, regular and dashed vertical lines denote the median and 66% and 95% confidence. Red lines indicate ECS from the IPCC for the twenty-first century (1.5–4.5 °C warming per doubling of CO2 at 66% confidence). c, Most likely ECS and ESS and their distributions for the present day (ECS, red5,92), Last Glacial Maximum (ECS, orange10,58,93–98; ESS, pale orange¹¹), Pleistocene (ECS, yellow8,9,49,50; ESS, pale yellow48,50,51,99), Pliocene (ECS, green46,50,100; ESS, pale green50,101,102), Miocene (ECS, blue54,103; ESS, pale blue⁵⁴; arrows denote estimates from this study) and the rest of the Cenozoic (>20 Ma; ECS, purple104–106; ESS, pale purple3,9,47,107). Boxes and whiskers represent reported 66% and 95% confidence intervals, respectively. Black lines represent reported most likely ECS/ESS. Black squares denote estimates incorporating modelling data. ECS from the IPCC for the twenty-first century (1.5–4.5 °C warming per doubling of CO2 at 66% confidence, red vertical lines). See Supplementary Table 3 for data and references.
Earth’s climate cooled markedly during the late Miocene from 12 to 5 million years ago, with far-reaching consequences for global ecosystems. However, the driving forces of these changes remain controversial. A major obstacle to progress is the uncertainty over the role played by greenhouse gas radiative forcing. Here we present boron isotope compositions for planktic foraminifera, which record carbon dioxide change for the interval of most rapid cooling, the late Miocene cooling event between 7 and 5 Ma. Our record suggests that CO2 declined by some 100 ppm over this two-million-year-long interval to a minimum at approximately 5.9 Ma. Having accounted for non-CO2 greenhouse gasses and slow climate feedbacks, we estimate global mean surface temperature change for a doubling of CO2—equilibrium climate sensitivity—to be 3.9 °C (1.8–6.7 °C at 95% confidence) on the basis of comparison of our record of radiative forcing from CO2 with a record of global mean surface temperature change. We conclude that changes in CO2 and climate were closely coupled during the latest Miocene and that equilibrium climate sensitivity was within range of estimates for the late Pleistocene, other intervals of the Cenozoic and the twenty-first century as presented by the Intergovernmental Panel on Climate Change. Climate sensitivity in the late Miocene was comparable to the late Pleistocene and twenty-first century, with cooling at the time coupled to declining carbon dioxide, according to a CO2 record determined from boron isotopes in planktic foraminifera
African hydroclimate compared with global change over the past 11 Myr
a, Atmospheric pCO2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$p_{{\mathrm{CO}}_2}$$\end{document} reconstructions (references in Supplementary Information): purple, estimates from planktonic foraminiferal boron isotopic signatures; pink, alkenone-derived estimates. b, Cenozoic global reference benthic foraminfera oxygen isotope dataset, with 20 kyr smoothing²⁹. c, Regional stacks of sea surface temperature (SST) difference from modern annual means²³: blue, >50° N; red, tropics. d, δ¹³C signature of n-alkanes from Site 659 (dark purple, this study; light purple, ref. ³⁶) and offshore East Africa⁴⁸ (grey). Error bars indicate 1σ. e, Major events in hominid evolution. Solid bars are taxon ranges; dashed lines represent confidence interval on taxon origin⁴⁷. f, [Al + Fe]/[Si+K + Ti] of calibrated elemental abundances. Modern endmember values updated from ref. ¹⁷ (Supplementary Information). Brown dashed line marks mean over past 1 Myr. g, ln[Zr/Rb] XRF core scan ratios. Dashed line marks mean over past 1 Myr. h, Site 659 estimated dust flux (3-point smoothed). Median value in red; 1st, 5th, 25th, 75th, 95th and 99th percentiles also shown in shades of orange/yellow. i, Major global climate events. MPT, mid-Pleistocene transition; iNHG, intensification of Northern Hemisphere glaciation; mPWP, mid-Pliocene warm period; MSC, Messinian salinity crisis. Vertical grey shading indicates transitions between climate stages.
Relationship between lithophile element ratios and hydrogen and carbon isotopic compositions of plant waxes at Site 659
a, Last glacial cycle. b, Late Pliocene. c, Early Pliocene. [Al + Fe]/[Si + K + Ti] of bulk sediment in orange (this study). Leaf wax C31n-alkane δ¹³C data in green39,40, and δD in blue39,40; error bars mark 1 s.d. High [Al + Fe]/[Si + K + Ti] and low n-alkane δD values both indicate more humid conditions. d–f, Violin plots indicating the distribution of leaf wax C31n-alkane δ¹³C (green) and δD (blue) data in high-resolution snapshots in the late Quaternary (d) and Pliocene (e) (data from refs. 39,40), and low-resolution records covering the past 11 Myr (f, data from this study and ref. ³⁶). Dots indicate median values; black rectangles indicate interquartile range.
Strong response of African hydroclimate to astronomical forcing recorded at Site 659
Example from the late Miocene. a, [Al + Fe]/[Si + K + Ti] of calibrated elemental abundances. b,c, ln[Zr/Rb] (b) and ln[Ca/Fe] (c) XRF core scan ratios. d, Summer insolation at 65° N as estimated by the La2004 orbital solution⁴⁹. Background colour shows the composited sediment core image. Darker sediments signal dry/dusty conditions; lighter sediments signal more humid intervals (Extended Data Fig. 4 and Supplementary Information).
Radiogenic isotope signature of Site 659 lithic fraction compared with values of PSAs reveals consistent source
a, Site 659 bulk sediment data (black crosses) compared with source region measurements (circles) coloured by PSA³¹. Data point size indicates dust-source activation frequency (DSAF)²⁸. Crosses indicate mean signatures of each PSA weighted by annual DSAF; bars denote ± one weighted standard deviation³¹. Samples for which only εNd or ⁸⁷Sr/⁸⁶Sr data exist are included in the calculation of mean values but are not plotted. b, Map of North African PSAs. Dust sources with activation frequencies >5% are shown in bold colours, <5% in pale colours. Panel b adapted with permission from ref. ³¹, Elsevier.
The Sahara is the largest hot desert on Earth. Yet the timing of its inception and its response to climatic forcing is debated, leading to uncertainty over the causes and consequences of regional aridity. Here we present detailed records of terrestrial inputs from Africa to North Atlantic deep-sea sediments, documenting a long and sustained history of astronomically paced oscillations between a humid and arid Sahara from over 11 million years ago. We show that intervals of strong dust emissions from the heart of the continent predate both the intensification of Northern Hemisphere glaciation and the oldest land-based evidence for a Saharan desert by millions of years. We find no simple long-term gradational transition towards an increasingly arid climate state in northern Africa, suggesting that aridity was not the primary driver of gradual Neogene expansion of African savannah C4 grasslands. Instead, insolation-driven wet–dry shifts in Saharan climate were common over the past 11 Myr, and we identify three distinct stages in the sensitivity of this relationship. Our data provide context for evolutionary outcomes on Africa; for example, we find that astronomically paced arid intervals predate the oldest fossil evidence of hominid bipedalism by at least 4 Myr. Pulses of Saharan dust have been entering the North Atlantic since at least 11 Ma, a result of astronomically paced cycles between arid and humid conditions in northern Africa, according to a terrigenous input record from an ocean core off west Africa.
Distribution of DOP concentrations and the relationship between the upper 50 m DOP stocks and P* over the global ocean
a, Observations of mean upper 50 m DOP concentrations (µM) (coloured circles) overlain on the 2° × 2° mean DOP concentrations from the average of three machine-learning algorithms, with predictors of P*, the chlorophyll a concentration and NPQ-corrected φsat, see Methods. b, Correlation between observed upper 50 m DOP stocks and surface ocean P* (µM) computed from World Ocean Atlas 2013 (ref. ²⁴) with a Type II linear regression model best-fit line (black) and a 95% confidence level (blue lines). The annual mean upper 50 m DOP stock from the BATS site (open blue circle) and station ALOHA (open red circle) are shown for reference. n, number of samples.
Relationships between upper 50 m DOP stocks, surface chlorophyll a concentration and NPQ-corrected φsat
a–c, Upper 50 m DOP stocks (a), surface chlorophyll a concentration (b) and NPQ-corrected φsat values (c) for the GO-SHIP P18-2016 cruise in the eastern Pacific. d–f, Corresponding plots for the BIOSOPE cruise in the eastern South Pacific²³, where one sample with [PO4³⁻] > 1.5 µM was excluded from this analysis. g–i, Corresponding plots for the GOM2019 cruise in the Gulf of Mexico. j–l, Corresponding plots for the GO-SHIP P06-2017 cruise in the South Pacific. m–o, Corresponding plots for the AMT17, AMT14 and 36N cruises in the Atlantic Ocean², where open circles/triangles represent samples from the North Atlantic and solid circles/triangles represent samples from the South Atlantic. p–r, Corresponding plots for the KH12-3 cruise in the western North Pacific³¹. Panels a, d, g, j, m and p created in Ocean Data View⁵¹. The surface convergence zones in the eastern South Pacific and South Atlantic are denoted by the red boxes (j and m) and samples from these regions are represented by red circles/triangles in plots k, l, n and o. Samples with climatological nFLH < 0.003 mW cm⁻² µm⁻¹ sr⁻¹ are represented by triangles. All black lines are lines of best fit determined using a Type II linear regression model. Details of correlations and sample sizes from each cruise are listed in Table 1. [Chlorophyll a], chlorophyll a concentration.
Conceptual model of factors influencing surface ocean DOP distributions with representative ocean regions
PO4³⁻ stress increases along the x axis, while iron stress increases along the y axis. Regions explored in this study are classified into four quadrants according to their conditions of PO4³⁻ stress and iron stress to show whether DOP accumulation or loss occurs. See texts for details. ‘>’ and ‘>>’ in the quadrants mean ‘greater than’ and ‘much greater than’, respectively.
Dissolved organic phosphorus (DOP) has a dual role in the surface ocean as both a product of primary production and as an organic nutrient that fuels primary production and nitrogen fixation, especially in oligotrophic gyres. Although poorly constrained, the geographic distribution and environmental controls of surface ocean DOP concentrations influence the distributions and rates of primary production and nitrogen fixation in the global ocean. Here we pair DOP concentration measurements with a metric of phosphate stress, satellite-based chlorophyll a concentrations and a satellite-based iron stress proxy to explore their relationship with upper 50 m DOP stocks. Our results suggest that phosphate and iron stress work together to control surface ocean DOP concentrations at basin scales. Specifically, upper 50 m DOP stocks decrease with increasing phosphate stress, while alleviated iron stress leads to either surface DOP accumulation or loss depending on phosphate availability. Our work extends the relationship between DOP concentrations and phosphate availability to the global ocean, suggests a linkage between marine phosphorus cycling and iron availability and establishes a predictive framework for DOP distributions and their use as an organic nutrient source that supports global ocean fertility. Production and consumption of dissolved organic phosphorus in the surface ocean is controlled by the interplay between phosphate and iron stress, according to global analyses of the distribution of marine nutrients.
Maps of field-sampling locations and peat swamp forest predictions in the central Congo Basin
a, Map of field-sampling locations. Points indicate transects, coloured by region. The Congo and Ruki River regional groups appear to be in largely river-influenced peatlands, predominating in DRC, sampled for this study. The Likouala-aux-Herbes River and Ubangi River regional groups are in largely rain-fed interfluvial basins, predominating in ROC, from ref. ⁹. The base map, in green, shows the first-generation peat swamp forest map⁹. Inset: location of the central Congo Basin peatlands. b, Predicted land-cover classes across the central Congo Basin from this study as the most likely class per pixel (>50%), using a legend identical to ref. ⁹ to facilitate comparison. In both panels, national boundaries are black lines; sub-national boundaries are grey lines; non-peat-forming forest includes both terra firme and non-peat-forming seasonally inundated forests. Panel a adapted from ref. ⁹, Springer Nature Limited.
Comparison of peat swamp predictions from this study with a previous map
Peat swamp forest predictions from this study and ref. ⁹ using the most likely class per pixel. White indicates peat in both studies; red indicates peat in this study only; blue indicates peat only in ref. ⁹. Open water is dark grey. National boundaries are black lines; sub-national boundaries are grey lines. Adapted from ref. ⁹, Springer Nature Limited.
Maps of peat thickness and uncertainty across the central Congo Basin
a, Median prediction of peat thickness (m) from 100 RF regression models with four predictors: distance from the peatland margin, precipitation seasonality, climatic water balance and distance from the nearest drainage point. b, Relative uncertainty (%) of the peat-thickness estimate, expressed as ± half the width of the 95% CI as percentage of the median. Black lines represent national boundaries; grey lines represent sub-national administrative boundaries.
Maps of below-ground peat carbon density and uncertainty across the central Congo Basin
a, Median prediction of below-ground peat carbon density (MgC ha⁻¹), obtained from applying 20 normally distributed thickness–carbon-density regressions (Extended Data Fig. 7) to 100 peat-thickness estimates (Fig. 3a), generating 2,000 carbon-density estimates. b, Relative uncertainty (%) of the carbon-density estimate, expressed as ± half the width of the 95% CI as percentage of the median. Black lines represent national boundaries; grey lines represent sub-national administrative boundaries.
The world’s largest tropical peatland complex is found in the central Congo Basin. However, there is a lack of in situ measurements to understand the peatland’s distribution and the amount of carbon stored in it. So far, peat in this region has been sampled only in largely rain-fed interfluvial basins in the north of the Republic of the Congo. Here we present the first extensive field surveys of peat in the Democratic Republic of the Congo, which covers two-thirds of the estimated peatland area, including from previously undocumented river-influenced settings. We use field data from both countries to compute the first spatial models of peat thickness (mean 1.7 ± 0.9 m; maximum 5.6 m) and peat carbon density (mean 1,712 ± 634 MgC ha⁻¹; maximum 3,970 MgC ha⁻¹) for the central Congo Basin. We show that the peatland complex covers 167,600 km², 36% of the world’s tropical peatland area, and that 29.0 PgC is stored below ground in peat across the region (95% confidence interval, 26.3–32.2 PgC). Our measurement-based constraints give high confidence of globally significant peat carbon stocks in the central Congo Basin, totalling approximately 28% of the world’s tropical peat carbon. Only 8% of this peat carbon lies within nationally protected areas, suggesting its vulnerability to future land-use change.
Map of water-column Hg profiles in the Arctic Ocean
a, Literature data marked with pink circles1,16,17,20,21,23,25,50. Stations (P1–P7), sampled during the 2019 summer and winter cruises, marked with black diamonds. b, Close-up Map of sampling stations with warmer Atlantic Water (AW) advected into the Barents Sea from the south and west, and Polar Water (PW) advected from the north and east²⁷. Figure created using Ocean Data View⁵¹.
THg concentrations along the shelf–deep-basin gradient
Stations P1–P7 (latitude) are on the x axis for summer 2019 and winter 2019 cruises. Figure created using Ocean Data View⁵¹.
ΔiHg and ΔMeHg at specified depth intervals for each station
iHg 0–100 m is in light orange, iHg 100–200 m is in dark orange, MeHg 0–100 m is in light blue and MeHg 100–200 m is in dark blue. A positive Δ value indicates a temporal gain in the iHg or MeHg pool while a negative Δ value indicates a temporal loss in the iHg or MeHg pool. Error bars represent combined standard uncertainty (±1σ). Values associated with Fig. 3 are compiled in Supplementary Table 1. *Stations P2 and P5 were integrated from 100 m to sample depth less than 200 m. Figure created using Microsoft Excel.
Total MeHg concentrations along the shelf–deep-basin gradient
a,b, Stations P1–P7 (latitude) are on the x axis for summer 2019 (a) and winter 2019 (b) cruises. Figure created using Ocean Data View⁵¹.
High biota mercury levels are persisting in the Arctic, threatening ecosystem and human health. The Arctic Ocean receives large pulsed mercury inputs from rivers and the atmosphere. Yet the fate of those inputs and possible seasonal variability of mercury in the Arctic Ocean remain uncertain. Until now, seawater observations were possible only during summer and fall. Here we report polar night mercury seawater observations on a gradient from the shelf into the Arctic Ocean. We observed lower and less variable total mercury concentrations during the polar night (winter, 0.46 ± 0.07 pmol l−1) compared with summer (0.63 ± 0.19 pmol l−1) and no substantial changes in methylmercury concentrations (summer, 0.11 ± 0.03 pmol l−1 and winter, 0.12 ± 0.04 pmol l−1). Seasonal changes were estimated by calculating the difference in the integrated mercury pools. We estimate losses of inorganic mercury of 208 ± 41 pmol m−2 d−1 on the shelf driven by seasonal particle scavenging. Persistent methylmercury concentrations (−1 ± 16 pmol m−2 d−1) are probably driven by a lower affinity for particles and the presence of gaseous species. Our results update the current understanding of Arctic mercury cycling and require budgets and models to be reevaluated with a seasonal aspect.
Reconstructed benthic δ¹³C and modelled anomalies for the mid-Pliocene Warm Period
The benthic δ¹³C anomaly (‰) is calculated from the OC3 database²⁰ with late Holocene core-top (diamonds) and modern δ¹³C ocean-water (squares) values. Symbols for open ocean sites are large and outlined in black; symbols for marginal sites possibly influenced by boundary effects are small and outlined in grey. a–c, Horizontal basin-wide δ¹³C anomalies for 1,000–1,500 m (a), 2,200–2,800 m (b) and 3,200–4,000 m (c) water depths. The largest anomaly is found in the west due to a deep western boundary current (for example, ref. ⁴) related to western intensification of ocean circulation.
Observed benthic δ¹³C and modelled zonal-mean anomalies for the mid-Pliocene Warm Period
The vertical cross section across the Pacific shows the core and spatial extent of the PMOC.
Benthic isotope records of Prob-stack and Site 882 sediment records CaCO3 MAR and opal MAR
a, Benthic isotope records of a probabilistic Pliocene-Pleistocene stack (Prob-stack)⁴⁵. b,c, Site 882 sediment records⁴¹ CaCO3 mass accumulation rate (MAR) (b) and opal MAR (c).The grey shading indicates the mid-Pliocene Warm Period. Northern Hemisphere glaciation (NHG) and Marine Isotope Stages KM3 and M2 are labelled.
Geologic intervals of sustained warmth such as the mid-Pliocene Warm Period can inform our understanding of future climate change, including the long-term consequences of oceanic uptake of anthropogenic carbon. Here we generate carbon isotope records and synthesize existing records to reconstruct the position of water masses and determine circulation patterns in the deep Pacific Ocean. We show that the mid-depth carbon isotope gradient in the North Pacific was reversed during the mid-Pliocene compared with today, which implies water flowed from north to south and deep water probably formed in the subarctic North Pacific Deep Water. An isotopically enabled climate model that simulates this North Pacific Deep Water reproduces a similar carbon isotope pattern. Modelled levels of dissolved inorganic carbon content in the North Pacific decrease slightly, although the amount of carbon stored in the ocean actually increases by 1.6% relative to modern due to an increase in dissolved inorganic carbon in the surface ocean. Although the modelled Pliocene ocean maintains a carbon budget similar to the present, the change in deep ocean circulation configuration causes pronounced downstream changes in biogeochemistry. Marine carbon isotope patterns point to substantial deep water formation in the North Pacific during the mid-Pliocene Warm Period, according to a synthesis of carbon isotope records and isotope-enabled climate modelling.
Map of the study area
The map includes tectonic plates, arc volcanoes (red triangles), state boundaries and topography as shaded relief with ocean as light blue and continent as light gray. The inset shows the location along the west coast of North America. Forearc geologic terranes are outlined with dashed lines, distinguished by colour. Small filled circles show magnetotelluric site locations, with different colours for each data source, as indicated in legend. The Olympic Peninsula (OP), Puget Sound (PS) and Vancouver Island (VI) are marked in the northern part of Siletzia. Other geographic names referred to in text are: OR, Oregon; WA, Washington; CA, California; BC, British Colombia.
3D view of the forearc portion of the model from NNW
Resistivity cross-sections are shown at one degree latitude increments with superposed isosurfaces for conductive material (ρ < 16 Ω m). The isosurfaces are restricted to the deep forearc up-dip of the forearc mantle corner (FMC) (42.2° N to 47.8° N, 125.2° W to 122.8° W and 7.5 km to 50 km depth). Curved grey surfaces mark the top of the subducting Juan de Fuca plate⁶¹. Coastlines and state borders are shown with white lines, and the outline of Siletzia³⁶ by the green line, both at z = 0. Principal conductive and resistive features discussed in text are labelled C1–C3, and R1−R4, respectively. An animation showing this 3D view from different angles is provided in Supplementary Fig. 2.
2D views derived from the 3D resistivity model
Slab depth contours⁶¹ are plotted as thin black lines, and the outline of Siletzia³⁶ is shown by the green dashed line. a, Depth to the bottom of the 300 Ω m isosurface. The crustal column between this depth and ~5 km depth is almost always at least 300 Ω m (Fig. 2 and Supplementary Fig. 1). Crustal faults¹⁸ are shown by white lines and volcanic vents⁶² are marked by red symbols. The resistive blocks discussed in text are labelled a–g. b, Conductance of the 10 km layer above the plate interface. Magnetotelluric sites are shown by small white circles, and conductive features discussed in the text are labelled A–I. c, The positions of highest (red crosses) and second highest (blue crosses) conductance peaks as a function of latitude are plotted over the ETS density for 2009–2019 (ref. ⁶³). Patches where resistive material extends to the plate interface (from a) are outlined in green and the estimated position of the FMC⁴⁹ is shown by the thick white dashed line.
Source data
Profiles of conductance in a 10-km-thick layer above the upper plate interface
a, Conductance as a function of slab depth at fixed latitudes corresponding to the sections shown in Fig. 2 (coloured lines). The black dashed line shows average along-margin computed depth contours. The three nominal peaks (seaward, FMC and arc) are marked as 1, 2 and 3, respectively. The mean slab depths of the peak ETS density and FMC position are also marked by vertical grey dashed lines. b, Conductance from b averaged over non-overlapping bands of slab depths (17.5–27.5 km and 27.5–37.5 km) as a function of latitude.
Source data
Conceptual model for fluid transport and storage
Yellow and orange features represent zones of fluid storage in the overriding crust, with orange indicating relatively higher concentrations. Light blue arrows represent bound and trapped fluids transported down with the subducting slab, orange arrows mark zones of fluid release and white arrows denote fluid diffusing through the crust, with enhanced flow in a more permeable channel along the megathrust (thin arrows). a, An impermeable block of Siletzia near the deeper fluid release zone forces stronger fluid flow up-dip along the megathrust, leaking out along the western edge of Siletzia where the fluids are sequestered. b, Fluids released near the ETS zone are able to diffuse into the overriding crust, resulting in a deeper and more pronounced zone of fluid accumulation. More fluid accumulates in the shallower conductive feature in a because both shallow and deep sources contribute.
Subduction of hydrated oceanic lithosphere can carry water deep into the Earth, with consequences for a range of tectonic and magmatic processes. Most of the fluid is released in the forearc where it plays a critical role in controlling the mechanical properties and seismic behaviour of the subduction megathrust. Here we present results from three-dimensional inversions of data from nearly 400 long-period magnetotelluric sites, including 64 offshore, to provide insights into the distribution of fluids in the forearc of the Cascadia subduction zone. We constrain the geometry of the electrically resistive Siletz terrane, a thickened section of oceanic crust accreted to North America in the Eocene, and the conductive accretionary complex underthrust along the margin. We find that fluids accumulate over timescales exceeding 1 My above the plate in metasedimentary units, while the mafic rocks of Siletzia remain dry. Fluid concentrations tend to peak at slab depths of 17.5 and 30 km, suggesting control by metamorphic processes, but also concentrate around the edges of Siletzia, suggesting that this mafic block is impermeable, with dehydration fluids escaping up-dip along the megathrust. Our results demonstrate that the lithology of the overriding crust can play a critical role in controlling fluid transport in a subduction zone. The lithology of the overriding plate plays a critical role in determining fluid transport in subduction zones, according to magnetotelluric imaging of the impact of the dry, mafic Siletzia terrane on fluids in the Cascadia subduction zone, North America.
Historical shoreline changes along the VBI between 1851 and 2017
a, Map of the shoreline position at different times. The background digital elevation model is derived from 2016 Lidar data collected by the USGS. Numbers in parentheses following island names are average island-wide shoreline change rates for the 1851–2017 period (in m yr⁻¹). CE, Common Era. b, System-wide mean (± standard error) and 10-year running average shoreline-change rate along the full VBI from northern Wallops to southwestern Fisherman’s islands. Shoreline position data for the VBI system prior to 1880 are incomplete and hence more uncertain. Observed variability through time is largely due to autogenic behaviours at the barrier-chain scale.
Comparison of predicted and measured topobathymetry for the VBI
Comparison of barrier morphology for different baseline SLR rates (Ro) and at different times after SLR increases instantaneously to 4 mm yr⁻¹ (Ri). The measured values are for the VBI in 2010²⁷. The beach berm in the VBI is ~1.5 m above MSL²⁸. The maps above the graphs are a snapshot at t = 80. The simulation with Ro = 1.5 mm yr⁻¹ and t = 80 yr is taken as representative for the comparison to the 2010 VBI.
Predicted barrier island response to different increases in SLR rate
a, Equilibrated barrier-island geometry after 5,000 years of SLR with a baseline Ro of 1.5 mm yr⁻¹. b, Snapshots at different times since the increase in SLR rate from Ro to Ri, for different values of Ri. The red lines in b indicate the shoreline position at t = 0.
Predicted barrier-island retreat over the next 500 years
a, Shoreline position (modelled with CoastMorpho2D) as a function of time after a simulated increase in SLR rate from Ro to Ri at t = 0. b, Simulated system-averaged barrier shoreline retreat rates (smoothed with a 40-yr centred window). In both a and b, the dashed black lines are predictions assuming that the retreat rate instantaneously adapts to the new SLR rate, and the solid black lines are the predictions from the simplified disequilibrium model (equation (1)). For each scenario (that is, each Ro), the relaxation time α is calculated by the best fit between the simplified model (equation (1)) and the CoastMorpho2D predictions for all Ri. The values below the legend are the equilibrated retreat rate Φo (equal to Ro divided by the substrate slope) and the relaxation time α.
The response of coastal barrier islands to relative sea-level rise (SLR) is a long-debated issue. Over centennial and longer periods, regional barrier retreat is generally proportional to the rate of relative SLR. However, over multi-decadal timescales, this simplification does not hold. Field observations along the USA East Coast indicate that barrier retreat rate has at most increased by ~45% in the last ~100 years, despite a concurrent ≥200% increase in SLR rate. Using a coastal evolution model, we explain this observation by considering disequilibrium dynamics—the lag in barrier behaviour with respect to SLR. Here we show that modern barrier retreat rate is not controlled by recent SLR (last decades), but rather by the baseline SLR of the past centuries. The cumulative effect of the baseline SLR is to establish a potential retreat, which is then realized by storms and tidal processes in the following centuries. When SLR accelerates, the potential for retreat is first realized through removal of geomorphic capital. After several centuries, barrier retreat accelerates proportionally to the increase in SLR. As such, we predict a committed coastal response: even if SLR remains at present rates, barrier retreat in response to SLR will accelerate by ~50% within a century. The lag dynamics identified here are probably general, and should be included in predictions of barrier-system response to climate change. Coastal evolution simulations suggest that the modern retreat of coastal barrier islands is controlled by cumulative sea-level rise over the past several centuries and will accelerate by 50% within a century, even if sea-level rise remains at present rates.
Shrubs act as thermal bridges to conduct heat through the tundra snowpack, fostering heat loss from the ground in winter and heat gain in the spring.
Thermal bridging through shrub branches in winter and spring
a, In autumn and winter, in the absence of sunlight, thermal bridging through frozen shrub branches cools the ground. b, In spring, light absorption by shallow buried shrub branches heats up the branches. The resulting heat is transferred through the branches to the ground, whose warming is thus accelerated.
Snow depth at shrub and herb tundra sites during 2018–2019
Observed snow depth at the shrub site (SALIX) and the herb tundra site (TUNDRA) in the Canadian high Arctic.
Thermal variables at SALIX and TUNDRA for 2018–2019
a, One-week running mean air temperature and one-week running mean temperature difference between both sites. b, Snowpack thermal insulance. c, One-week running mean temperature of the ground at 15 cm depth at both sites. These data have not been corrected for air temperature differences between both sites. d, Temperature difference of the ground at 15 cm depth between SALIX and TUNDRA, corrected for air temperature. Negative values indicate that at equal air temperature, the ground temperature is colder at SALIX than at TUNDRA. The dashed horizontal lines are the 0 °C lines, added as visual aids.
Comparison of measured and simulated snow and ground temperatures
Simulations were performed with the MFM at 0 and 5 or 5.8 cm heights and at 15 cm depth. a, At TUNDRA. b, At SALIX. In both cases, an inset illustrates the quality of simulation of the phase at 0 cm. The shift is less than 2 h at TUNDRA and more than 10 h at SALIX. meas, measured; sim, simulated.
Source data
Finite element simulations of temperatures at SALIX
Simulated snow and ground temperatures resulting from heat transfer through a snow cover with and without shrubs are compared. Measured temperatures are also shown. a–c, At 5.8 cm height in the snow (a), at the snow–ground surface (b) and in the ground at 15 cm depth (c). In b, the inset shows the quality of the phase simulation, with the dotted vertical lines marking the maxima of temperature. The green dashed curve is the temperature of a point halfway between two shrubs, while the red curve is the temperature in the middle of a shrub (Extended Data Fig. 9), to which 2 °C was added to facilitate comparison. The right-hand temperature scale (in blue) is for the blue curve only.
Source data
Considerable expansion of shrubs across the Arctic tundra has been observed in recent decades. These shrubs are thought to have a warming effect on permafrost by increasing snowpack thermal insulation, thereby limiting winter cooling and accelerating thaw. Here, we use ground temperature observations and heat transfer simulations to show that low shrubs can actually cool the ground in winter by providing a thermal bridge through the snowpack. Observations from unmanipulated herb tundra and shrub tundra sites on Bylot Island in the Canadian high Arctic reveal a 1.21 °C cooling effect between November and February. This is despite a snowpack that is twice as insulating in shrubs. The thermal bridging effect is reversed in spring when shrub branches absorb solar radiation and transfer heat to the ground. The overall thermal effect is likely to depend on snow and shrub characteristics and terrain aspect. The inclusion of these thermal bridging processes into climate models may have an important impact on projected greenhouse gas emissions by permafrost. Arctic shrubs cool permafrost in winter by acting as a thermal bridge through the snowpack, according to ground temperature observations and heat transfer simulations.
The bulk crustal porosity of the lunar highland may have been generated early in the Moon’s history by basin-forming impacts and then declined exponentially. A new porosity evolution model constrains the timing and sequence of basin formation.
Map of the fractional contribution of each radical loss pathway for the years 1750, 1950 and 2014
Mean simulated July solar noon-time fraction of radical termination at the surface that occurs through OH + NO2 (red), peroxyl-radical self-reactions (green) and aerosol uptake of HO2 (blue) for 1750, 1970 and 2014. Enlargements of North America, Europe and Asia are shown in the Supplementary Information.
Map of the photochemical ozone-control regime for the years 1750, 1950 and 2014
Mean simulated July solar noon-time photochemical ozone-control regime for 1750, 1970 and 2014. Enlargements of North America, Europe and Asia are shown in the Supplementary Information.
Map of the percentage increase in ozone concentration when uptake of HO2 is switched off for 1750, 1970 and 2014
Percentage increase in monthly mean solar 15:00 lt O3 concentration for July when heterogeneous uptake of HO2 on aerosols is switched off for 1750, 1970 and 2014.
Map of fractional NOx emissions decrease required to offset a 50% reduction in aerosol emissions
Fractional decrease in 2014 NOx emissions that is necessary to offset the increase in surface, monthly mean and solar 15:00 lt O3 caused by a 50% reduction in emissions of aerosol precursors (anthropogenic SO2, black carbon, organic carbon, dust and all biomass-burning emissions).
Atmospheric ozone (O3) is a pollutant produced through chemical chain reactions where volatile organic compounds (VOCs), carbon monoxide and methane are oxidized in the presence of oxides of nitrogen (NOx). For decades, the controlling chain termination step has been used to separate regions into either ‘NOx limited’ (peroxyl-radical self-reactions dominate) or ‘VOC limited’ (hydroxyl radical (OH) + nitrogen dioxide (NO2) reaction dominates). The controlling regime would then guide policies for reducing emissions and so O3 concentrations. Using a chemical transport model, we show that a third ‘aerosol inhibited’ regime exists, where reactive uptake of hydroperoxyl radicals (HO2) onto aerosol particles dominates. In 1970, 2% of the Northern Hemisphere population lived in an aerosol-inhibited regime, but by 2014 this had increased to 21%; 60% more than lived in a VOC-limited regime. Aerosol-inhibited chemistry suppressed surface O3 concentrations in North America and Europe in the 1970s and is currently suppressing surface O3 over Asia. This third photochemical O3 regime leads to potential trade-off tensions between reducing particle pollution in Asia (a key current health policy and priority) and increasing surface O3, should O3 precursors emissions not be reduced in tandem. Global chemical transport simulations reveal an ozone photochemistry regime where the uptake of hydroperoxyl radicals onto aerosol particles dominates ozone production.
Surface climate following springtime Arctic ozone depletion
a–l, Composites of SLP (a,d,g,j), surface temperature (b,e,h,k) and precipitation (c,f,i,l) anomalies in observations (MERRA2; N = 10) (a–c), WACCM INT-O3 (N = 50) (d–f), CLIM-3D (N = 50) (g–i) and CLIM-2D (N = 50) (j–l) after ozone minima in the 25% of winters with most extreme ozone loss (average over the 30 days after the ozone minimum date). Stippling shows significance on a 4.6% level (2σ) following a bootstrapping test. The following springs are included in the observations (a–c): 2020, 2011, 2005, 2002, 2000, 1997, 1996, 1995, 1993, 1990.
The AO index following winters with extreme ozone loss
The box plot shows the distribution of the mean AO index (20–90° N) at 1,000 hPa in the 30 days following the ozone minimum for MERRA2 (red) and WACCM (grey) INT-O3, CLIM-O3 and CLIM-2D. Triangles and numbers indicate the mean AO index in the 30 days after the ozone minimum date averaged over the 25% most extreme winters. The upper and lower edges of the boxes show the upper and lower quartile; the whiskers represent the maximum and minimum values of the respective distribution. For MERRA2, individual data points for each ozone minimum are shown. For a discussion of the robustness of the AO response, refer to section 2 of the Supplementary Information.
Simulation set-up and ozone feedback mechanism
a, Set-up of INT-3D, CLIM-3D and CLIM-2D. INT-3D treats ozone chemistry fully interactively; that is, the calculated ozone field has a direct feedback on the atmosphere via the model radiation schemes. By contrast, the CLIM experiments do not use interactively calculated ozone in the radiation module. Instead, the radiation module uses an ozone climatology, which has been derived from INT-3D runs with interactive ozone of the same model. b, Ozone feedback mechanism in the aftermath of strong springtime Arctic ozone loss. Shown are impacts of ozone depletion on short-wave heating, temperature (T) and wind speed (U) in the lower stratosphere and subsequent impacts on surface AO and upper stratospheric temperature.
Influence of ozone depletion on stratosphere–troposphere coupling
a–c, Composites of NAM indices (20–90° N) around the ozone minima in WACCM INT-3D (a), CLIM-3D (b) and CLIM-2D (c). Day zero indicates the date with the largest extent of the ozone minima (ozone minimum date’). Stippling shows significance on a 4.6% (2σ) level following a bootstrapping test (Methods).
Impact of ozone feedbacks on short-wave and dynamical heating
a–d, Differences of polar cap (60–90° N) temperature (a), ozone (b), short-wave heating (c) and dynamical heating (d) anomalies between INT-3D and CLIM-3D around the ozone minima in WACCM. Day zero indicates the date with the largest extent of the ozone minima (‘ozone minimum date’). Contour lines in the temperature plot show temperature anomalies in CLIM-2D around the ozone minima with a contour interval of 1.5 K. Stippling shows significance on a 4.6% (2σ) level following a bootstrapping test.
Large-scale chemical depletion of ozone due to anthropogenic emissions occurs over Antarctica as well as, to a lesser degree, the Arctic. Surface climate predictability in the Northern Hemisphere might be improved due to a previously proposed, albeit uncertain, link to springtime ozone depletion in the Arctic. Here we use observations and targeted chemistry–climate experiments from two models to isolate the surface impacts of ozone depletion from complex downward dynamical influences. We find that springtime stratospheric ozone depletion is consistently followed by surface temperature and precipitation anomalies with signs consistent with a positive Arctic Oscillation, namely, warm and dry conditions over southern Europe and Eurasia and moistening over northern Europe. Notably, we show that these anomalies, affecting large portions of the Northern Hemisphere, are driven substantially by the loss of stratospheric ozone. This is due to ozone depletion leading to a reduction in short-wave radiation absorption, when in turn causing persistent negative temperature anomalies in the lower stratosphere and a delayed break-up of the polar vortex. These results indicate that the inclusion of interactive ozone chemistry in atmospheric models can considerably improve the predictability of Northern Hemisphere surface climate on seasonal timescales. Ozone depletion in the Arctic stratosphere consistently disrupts surface temperature and precipitation patterns across the Northern Hemisphere, according to atmospheric chemistry–climate modelling and observations.
For decades, ozone pollution mitigation efforts relied on two chemical regimes. A global modelling analysis has revealed a third regime involving aerosols that would help with the concurrent control of both ozone and particulate pollution.
Observed lunar highlands crustal properties
a,b, GRAIL-derived lunar crustal porosity³² (a) and N(20) (b) exhibit an inverse relationship. White circles identify undated basins, and black circles identify dated basins. Regions excluded from the analysis are masked out in white and include PKT (thorium concentrations ≥3.5 PPM (ref. ⁶¹)), SP-A (assumed to be a 2,028-km-diameter circular basin¹⁰, centred at 191° E, 53° S) and regions for which the density gradient³² is <5 kg m⁻³km ⁻¹. N(20) values (b) are derived following ref. ¹¹, using 500 km moving window, with the updated catalogue of lunar craters³⁶ and the locations for craters ≥150 km in diameter updated using GRAIL observations (Supplementary Table 3). We used an unbiased color scheme⁶². The map projection is plotted using a free Matlab software by M_PROJ programme⁶³.
Source data
Modelled lunar highlands crustal porosities
a, Present-day crustal porosity. b, Crustal porosity circa SP-A formation. Crustal porosities are derived using our model and depend only on the sizes and cratering record of the large (>200-km-diameter) lunar basins. Regions shown in white and the white and black circles in panel a are the same as those defined in Fig. 1. Basins interpreted to have formed around the time of SP-A and thus that contribute to the historic porosity shown in panel b are indicated by black circles. Regions where large young basins have formed, and potentially erased evidence of older basins, should not be trusted when modelling historic porosity and are outlined in white.
Source data
Comparison of the relationship between observed porosity and crater number density for the lunar highlands and our compaction model
Points represent lunar highlands; yellow region represents our compaction model. The slope of the shaded region relates to the efficiency of impacts and overburden pressure in compacting porosity. The width of the shaded region is controlled by the dependence of basin-generated porosity on basin size. Data are area-weighted average samples, in which the observed N(20) and porosity values are derived using a grid spacing of 290 km (ref. ³²). Yellow data points represent regions with anomalous grain densities, which may not reflect the average grain density of the underlying crust and thus may have erroneous ‘observed’ porosity values. Cyan data points are from regions up range of oblique impacts (Extended Data Fig. 3) and generally exhibit higher porosity values than our model predicts. Orange data points indicate the distal ejecta (3.0–3.5 radii) of Orientale Basin, which exhibits anomalously low porosity and crater number density values.
Source data
The formation and evolution of the terrestrial planets were shaped by a bombardment of large impactors in a cluttered early Solar System. However, various surface processes degrade impact craters, and the early impact history of the Moon and the ages of its ancient impact basins remain uncertain. Here we show that the porosity of the lunar crust, generated by the cumulative crustal processing of impacts, can be used to determine the Moon’s bombardment history. We use a numerical model constrained by gravity data to simulate the generation of porosity by basin-forming impacts and the subsequent removal by smaller impacts and overburden pressure. We find that, instead of steadily increasing over the history of the Moon, lunar crustal porosity was largely generated early in lunar evolution when most basins formed and, on average, has decreased after that time. Using the Moon as a proxy for the terrestrial planets, we find that the terrestrial planets experienced periods of high crustal porosity early in their evolution. Our modelled porosities also provide an independent constraint on the chronological sequence of basin-forming impacts. Our results suggest that the inner solar system was subject to double the number of smaller impacts producing craters exceeding 20 km in diameter than has been previously estimated from traditional crater-counting analyses, whereas the bombardment record for the lunar basins (>200 km in diameter) is complete. This implies a limited late delivery of volatiles and siderophile elements to the terrestrial planets by impactors.
AHA and its recent expansion
a, The difference (Δ) between 1980–2007 and 1950–1979 in mean winter sea-level pressure in hPa (blue–red contours, calculated using HadSLP2 observations), terrestrial precipitation mm month–1 (brown–green contours, calculated using GPCC precipitation) and moisture transport in g kg–1 m s–1 (grey arrows). b, The difference in mean winter sea-level pressure between 1980–2007 and 1950–1979 (blue–red contours) as well as the region considered when calculating AHA (red dashed box) and the winter sea-level pressure climatology in this region (black contours: low-pressure contours are dashed; high pressures are solid). The centre of the Azores High is indicated with an ‘H’ while the centre of the corresponding Icelandic Low is indicated with an ‘L’; the star above central Portugal indicates the location of the Buraca Gloriosa cave.
AHA in observations and simulations since 1850
a, Time series of AHA as reported by ERA-20C (blue), NOAA–CIRES 20CR (purple) and HadSLP2 (green) and as simulated by the 13 full-forcing members of the LME (light yellow). The winters with largest 10% of Azores High areas are indicated with crosses (ERA-20C), Xs (HadSLP2), stars (NOAA–CIRES 20CR) and circles (LME). b, Kernel density estimate for the AHA over the available period for reanalysis/observational data, as well as the kernel density estimate for the full-forcing members of the LME between 1850 and 2005. c, The number of winters with extremely large AHA that occur in a 25 yr window surrounding each year. The number of winters with extremely large AHA simulated by individual full-forcing members of the LME is shown in light yellow; the ensemble average is in bold yellow.
AHA and number of extreme events over the past millennium
a, Time series of AHA and number of winters with extremely large Azores Highs in a 100 yr window centred at each year for the ensemble mean (red line), individual ensemble members (grey lines) and individual extreme years (red triangles). b, Distributions of AHA during the industrial era (red) and the pre-industrial era (grey). c, Distribution of extremely large Azores High frequency (per 100 years) from Monte Carlo sampling from each ensemble member in grey. The red vertical lines show the number of extremely large Azores High events that occurred in the last 100 years of each ensemble member; their mean is in bold and ±1 s.d. is shaded. d, A 100 yr bin average comparison of AHA and a Suess effect-adjusted stalagmite carbon isotope record of hydroclimate from Buraca Gloriosa cave, Portugal⁴¹. The model and the isotope reconstruction show AHA size and related secular aridity in recent time to be unprecedented in the past 900 years. e, Ensemble mean frequency of winters with extremely large Azores Highs in a 100 yr window centred at each year for members with full forcing (black with grey shading for ±1 s.d.), GHGs only (green), volcanic (red) and solar (yellow); thick lines are ensemble mean; thin lines are individual forcing ensemble members.
Hydroclimate during winters with extremely large Azores High
a, Anomalies of precipitation (contours) and moisture transport in g kg–1 m s–1 (vectors) during winters with extremely large Azores High. Anomalies are calculated as the average difference between conditions during extremely large Azores High events and all other winter conditions. b, SLP (contours) and moisture transport anomalies during extremely large Azores High events. Anomaly patterns were calculated using the LME. The star above central Portugal indicates the location of Buraca Gloriosa cave.
The Azores High is a persistent atmospheric high-pressure ridge over the North Atlantic surrounded by anticyclonic winds that steer rain-bearing weather systems and modulate the oceanic moisture transport to Europe. The areal extent of the Azores High thereby affects precipitation across western Europe, especially during winter. Here we use observations and ensemble climate model simulations to show that winters with an extremely large Azores High are significantly more common in the industrial era (since ce 1850) than in pre-industrial times, resulting in anomalously dry conditions across the western Mediterranean, including the Iberian Peninsula. Simulations of the past millennium indicate that the industrial-era expansion of the Azores High is unprecedented throughout the past millennium (since ce 850), consistent with precipitation proxy evidence from Portugal. Azores High expansion emerges after ce 1850 and strengthens into the twentieth century, consistent with anthropogenically driven warming. The Azores High over the North Atlantic has expanded due to anthropogenic climate change, disrupting precipitation patterns in western Europe, according to climate modelling and precipitation proxy records spanning the past millennium.
Carbon fixation rates and metagenomic potentials across the aquifer
a, Rates of carbon fixation. Outer error bars depict one standard deviation; inner grey bars delineate standard error of the mean. Rates for well H51 are derived from non-labelled controls (Supplementary Information). Letters denote the results of ANOVA and post hoc Tukey tests (excluding H51). b, Relative importance of the predicted carbon fixation pathways. c, Electron donor sources in each well. Values are averages from the triplicate 0.2 μm filtered fraction metagenome samples. d, Mean dissolved oxygen (DO) concentrations in groundwater collected in summer months (May–September 2010–2018); error bars depict standard deviation. e, Redox-potential measurements from identical time points; error bars depict standard deviation. While groundwater from H52 and H43 exhibited anoxic or hypoxic conditions, the positive redox potentials were due to oligotrophic conditions and mean nitrate concentrations of ~4 mg l–1.
Comparison of carbon fixation rates within groundwater and the marine euphotic zone
Violin plots depicting the distribution of carbon fixation rates measured in oligotrophic marine surface waters and groundwater. HOTs, Hawaiian Oceanographic Timeseries (1999, cruises 101–110); BATs, data from 1988 to 2016 for the Bermuda Atlantic Timeseries; Liang, a collated dataset compiled by Liang et al.²⁶; GW, the range of groundwater samples shown in Fig. 1.
Phylogeny, relative abundances and transcriptional activities of putative chemolithoautotrophic MAGs
Approximately maximum-likelihood phylogenetic tree based on concatenated single-copy protein alignments for all bacterial MAGs considered. Branches are coloured according to the predicted carbon fixation pathway, and the matching leaf is indicated by a point. Bar charts present average normalized metagenomic coverages within each well from triplicate 0.2 μm filtered fractions. Pie charts show the coverage of mRNA transcripts recruited, normalized to gene number and library size. Asterisks denote MAGs discussed in greater detail in Supplementary Information. The tree is rooted using Patescibacteria (CPR) as an outgroup, indicated by the collapsed grey leaf in the upper left. Node numbers: (1) c_Nitrospiria, (2) c_Thermodesulfovibrionia, (3) o_Nitrospirales, (4) f_Nitrospiraceae, (5) g_Nitrospira_D, (6) g_Nitrospira, (7) p_Nitrospinota, (8) c_Gammaproteobacteria, (9) o_Acidiferrobacterales, (10) f_Sulfurifustaceae, (11) g_SM1-46, (12) f_UBA6901, (13) o_Burkholderiales, (14) f_Nitrosomonadaceae, (15) g_Nitrosomonas, (16) f_SG8-41, (17) f_SG8-39, (18) c_Brocadiae, (19) o_Brocadiales, (20) f_Brocadiaceae and (21) f_Scalinduaceae.
Metabolic reconstructions of dominant putatively chemolithoautotrophic MAGs
Bar charts below each metabolic model summarize the average normalized coverage across each sample, scaled proportionally. Values within the bar chart indicate the sum normalized coverage for each MAG. Balloon plots depict normalized transcript coverages for genes affiliated with each pathway. If multiple copies were present, only the most active copy was plotted. The text information over each panel includes the predicted taxonomy (a, Nitrospiria; b, Sulfurifustaceae; c, Brocadiaceae), MAG identifier and estimated % completion/% redundancy. DNRA, dissimilatory nitrate reduction to ammonia.
The terrestrial subsurface contains nearly all of Earth’s freshwater reserves and harbours the majority of our planet’s total prokaryotic biomass. Although genetic surveys suggest these organisms rely on in situ carbon fixation, rather than the photosynthetically derived organic carbon transported from surface environments, direct measurements of carbon fixation in the subsurface are absent. Using an ultra-low level 14C-labelling technique, we estimate in situ carbon fixation rates in a carbonate aquifer. We find these rates are similar to those measured in oligotrophic marine surface waters and up to six-fold greater than those observed in the lower euphotic zone. Our empirical carbon fixation rates agree with nitrification rate data. Metagenomic analyses reveal abundant putative chemolithoautotrophic members of an uncharacterized order of Nitrospiria that may be behind the carbon fixation. On the basis of our determined carbon fixation rates, we conservatively extrapolate global primary production in carbonate groundwaters (10% of global reserves) to be 0.11 Pg carbon per year. These rates fall within the range found for oligotrophic marine surface waters, indicating a substantial contribution of in situ primary production to subsurface ecosystem processes. We further suggest that, just as phototrophs are for marine biogeochemical cycling, such subsurface carbon fixation is potentially foundational to subsurface trophic webs. Direct measurements of carbon fixation rates in groundwater suggest a substantial contribution of in situ primary production to subsurface ecosystem processes.
Early Cenozoic tectonic and magmatic evolution of the northeast Atlantic
a, Map of the present day northeast Atlantic region showing the distribution of Palaeocene–Eocene lava flows and intrusives, with dated volcanics denoted by coloured symbols and the locations of major hydrothermal vent complexes in the Vøring and Møre Basins¹⁵ and offshore Northeast Greenland⁵⁴ shown as stars (SDRs = seaward-dipping reflectors; COB = continent–ocean boundary). b, Plate tectonic reconstruction showing nascent ridge systems developing along the Labrador Sea and northeast Atlantic at 55 Ma; the perimeter to brown shaded regions shows continent-ocean boundaries, and brown lines signify subduction zones (Methods). c, Ages of the volcanic sections discussed (Up. = Upper; VFF = Vandfaldsdalen Fm, E. = East, Greenl. = Greenland), defined by radiometric dates1,2, magnetostratigraphy and nannofossil zonation22,25,37 and corresponding carbon and oxygen isotope records showing the PETM isotope excursions (solid and faint lines show 1 Myr and 20 kyr locally weighted functions, respectively)⁵⁵. d, Seafloor production rates for the Labrador Sea and northeast Atlantic, derived from GPlates (Methods), shown alongside the timing of Eocene hyperthermals (PETM and Eocene Thermal Maximum (ETM) 2 and 3). e, Palaeolongitude of Greenland⁷, indicating the onset of ocean crustal production in the Northeast Atlantic at 56 Ma. Panel a is adapted with permission from ref. ³, Springer Nature Limited.
Source data
Palaeocene–Eocene volcanostratigraphy and geochemistry of the proto-northeast Atlantic ridge
a, Simplified log of the Rockall ‘Phase 1’ sequence¹⁹ showing lithologies, Mg# (that is, 100 × molecular MgO/(MgO + FeO), where FeO is assumed to be 0.9FeOT) and ϵNd (Extended Data Fig. 1). Please note the scale break at 56.6 Ma; the vertical dashed lines in panels (a) to (c) show mean (x̄) values for variables over the intervals shown. b,c, Simplified log of the Faroes Basalt formations²⁶ (b), with Mg# and (Eu/Yb)n (chondrite-normalized⁵⁶); Mg# data are from ref. ²⁶ and (Eu/Yb)n are from refs. 27,32 (Extended Data Fig. 2); note the sharp transition to high Mg# (and enriched rare earth element contents) at ca. 56.1 Ma¹, which is also observed in east Greenland (Milne Land Formation)²⁶ MORB, mid-ocean ridge basalt; Fm., formation. (c). d, (La/Yb)n versus (Eu/Yb)n of the Faroes and Hold with Hope (HwH) lavas (chondrite-normalized⁵⁶) and modelled non-modal batch melting of a lherzolitic mantle source, adopted from ref. ³³, showing different degrees of melting of a garnet lherzolite (green, blue and red curves). e, (Sm/Yb)n versus (Ce/Sm)n and a rare earth element melting model (Methods), showing percentage melt along the top and the relative proportions of garnet- and spinel-lherzolites from 100% gt-lherzolite (red curve) to 100% sp-lherzolite (green curve). Both models indicate that the Faroes Middle Lava Formation (that is, high-Mg# basalts in the lower 500 m of the Middle Lavas; see b), which erupted immediately before and during the PETM, experienced the highest degrees of melting of a mantle source containing ≳10% garnet.
Source data
Simulations of volcanic carbon release during the PETM
Results are plotted as cumulative distribution functions (CDFs). The grey lines show the estimated carbon output from ridge volcanism and LIPs alone; with S1 and S2 showing low (0.6 km³ per year) and high (2.4 km³ per year) LIP volcanic productivity (prod.) scenarios⁴, respectively (Extended Data Fig. 4 and Methods). The coloured lines show the effects of adding 4% to 8% carbonated (c-) SCLM melt along the incipient ridge during breakup. The grey vertical bars denote the carbon output necessary to drive and sustain PETM warming estimated by Gutjahr et al.¹¹ (labelled G; 10,200–12,200 Gt C) and Haynes and Hönisch¹³ (labelled H; 14,900 Gt C).
Deep carbon mobilization and release in the North Atlantic during the PETM
The model stages are shown. a, Thermomechanical weakening of the SCLM by the Iceland plume. b, Thermal removal, delamination and lateral advection of metasomatized SCLM (shown as the white stippled pattern) by edge-driven convection in the asthenospheric mantle. c, Carbonate-rich domains are entrained into the central melt zone beneath the evolving rift, where they are thermodynamically unstable and involved in transient decompression melting just before continental breakup. d, Massive outgassing of CO2 at Earth's surface from the northeast Atlantic rift and associated volcanic and tectono–magmatic systems. Credit: Gary Hincks.
Plume magmatism and continental breakup led to the opening of the northeast Atlantic Ocean during the globally warm early Cenozoic. This warmth culminated in a transient (170 thousand year, kyr) hyperthermal event associated with a large, if poorly constrained, emission of carbon called the Palaeocene–Eocene Thermal Maximum (PETM) 56 million years ago (Ma). Methane from hydrothermal vents in the coeval North Atlantic Igneous Province (NAIP) has been proposed as the trigger, though isotopic constraints from deep sea sediments have instead implicated direct volcanic carbon dioxide (CO2) emissions. Here we calculate that background levels of volcanic outgassing from mid-ocean ridges and large igneous provinces yield only one-fifth of the carbon required to trigger the hyperthermal. However, geochemical analyses of volcanic sequences spanning the rift-to-drift phase of the NAIP indicate a sudden ~220 kyr-long intensification of magmatic activity coincident with the PETM. This was likely driven by thinning and enhanced decompression melting of the sub-continental lithospheric mantle, which critically contained a high proportion of carbon-rich metasomatic carbonates. Melting models and coupled tectonic–geochemical simulations indicate that >104 gigatons of subcrustal carbon was mobilized into the ocean and atmosphere sufficiently rapidly to explain the scale and pace of the PETM. A change in the style of rifting in the North Atlantic led to carbon fluxes from subcrustal melting that helped trigger the Palaeocene–Eocene Thermal Maximum, according to geochemical analyses of volcanic sequences as well as melting and tectonic modelling.
Unrest outcome as a function of the inflow rate in mafic calderas
a, Duration of unrest and corresponding inflow rate, Q, as a function of unrest outcome. The grey area highlights the zone with transitional rates (5 ± 4 × 10⁻² km³ yr⁻¹). The vertical dashed grey line separates the unrest episodes lasting <1 year from those lasting >1 year. Data are reported with the confidence limits (error bars). b, Histogram of frequency of the injected rates. Tr. Q, L. Tz. and U. Tz. refer to transitional, lower transitional and upper transitional rates, respectively. Background colours: blue=Low Q, grey=Transitional Q, red=High Q.
Source data
Inflow rate, magma overpressure and related critical time
a, Maximum overpressure depending on inflow rate and volume of the magma reservoir. Grey area indicates ΔPcrit = 10 ± 4 MPa and dashed line indicates ΔPcrit = 10 MPa. Data below the grey area lie in the viscous domain, promoting magma storage and plain unrest. Data above are in the elastic domain promoting nucleating unrest. b, See caption Fig. 1b. c, Time required to reach the critical overpressure to nucleate a magma-filled fracture in the elastic domain, assuming ΔPcrit=10 MPa and E = 30 GPa. The thicker line marks t = 1 year.
Forecasting eruption is the ultimate challenge for volcanology. While there has been some success in forecasting eruptions hours to days beforehand, reliable forecasting on a longer timescale remains elusive. Here we show that magma inflow rate, derived from surface deformation, is an indicator of the probability of magma transfer towards the surface, and thus eruption, for basaltic calderas. Inflow rates ≥0.1 km3 yr−1 promote magma propagation and eruption within 1 year in all assessed case studies, whereas rates <0.01 km3 yr−1 do not lead to magma propagation in 89% of cases. We explain these behaviours with a viscoelastic model where the relaxation timescale controls whether the critical overpressure for dyke propagation is reached or not. Therefore, while surface deformation alone is a weak precursor of eruption, estimating magma inflow rates at basaltic calderas provides improved forecasting, substantially enhancing our capacity of forecasting weeks to months ahead of a possible eruption. Using magma inflow rate improves eruption forecasting on timescales of weeks to months for basaltic caldera systems, compared with using surface deformation alone, according to analysis of 45 unrest case studies and viscoelastic modelling.
Temporal evolution of the experiment with COH = 600 H/10⁶ Si
a, Viscosity of upper mantle and marker composition of crust. White lines denote isotherms up to 1,300 °C. b, Deformation mechanism in the uppermost mantle and composition of crust. Red, diffusion creep; white, dislocation creep; blue contours indicate grain size; grey and yellow colours, continental crust marker composition; green, oceanic crust marker composition; x, width; y, depth.
Grain sizes in the mantle at variable water content
a, Vertical profile at x = 990 km after 5 Myr. See location in Fig. 1a. b, Temporal evolution of grain size within lithospheric shear zones at y = 50 km. c, Temporal evolution of grain size between 200 and 410 km depth.
Percentage of finite strain accumulation in a lithospheric shear zone
Strain due to diffusion creep (blue) or dislocation creep (white) after 20 Myr of divergence. Red, contour of η = 1021.5 Pa s indicating thickness of the elastic lithosphere. a, COH = 50 H/10⁶ Si. b, COH = 175 H/10⁶ Si. c, COH = 600 H/10⁶ Si. d, COH = 2,500 H/10⁶ Si.
Strength of the lithosphere
a, Temporal evolution of the laterally averaged integrated strength (boundary force) of pure dislocation and grain-size-dependent composite diffusion–dislocation-creep experiments. b, Laterally (x = 990–1,000 km) averaged lithospheric strength profiles after 5 Myr. For colour code, see a. c, Strength and grain-size profile along lithospheric shear zone at 5 Myr. Location of profile indicated in Fig. 1b.
Variation in the effective strength of the lithosphere allows for active plate tectonics and is permitted by different deformation mechanisms operating in the crust and upper mantle. The dominant mechanisms are debated, but geodynamic models often employ grain-size-independent mechanisms or evaluate a single grain size. However, observations from nature and rock deformation experiments suggest a transition to grain-size-dependent mechanisms due to a reduction in grain size can cause lithospheric weakening. Here, we employ a two-dimensional thermo-mechanical numerical model of the upper mantle to investigate the nature of deformation and grain-size evolution in a continental rift setting, on the basis of a recent growth law for polycrystalline olivine. We find that the average olivine grain size is greater in the asthenospheric mantle (centimetre-scale grains) than at the crust–mantle boundary (millimetre-scale grains). This grain-size distribution could result in dislocation creep being the dominant deformation mechanism in the upper mantle. However, we suggest that along lithospheric-scale shear zones, a reduction in grain sizes due to localized deformation causes a transition to diffusion creep as the dominant deformation mechanism, causing weakening of the lithosphere and facilitating the initiation of continental rifting.
Map of the seismic survey area near the MAR in the equatorial Atlantic Ocean
The dashed black line refers to the MAR and the triangles show the locations of the OBSs used for this study. The bold black line passing through the OBSs shows the part of shot profile used here, and the two pink bars indicate the boundaries of the seismic images shown in Fig. 2. The thin black line and numbers indicate the distance from the MAR. The contour labelling 10 indicates the 10 Myr lithospheric age²⁶. The bathymetry was taken from ref. ¹⁴. See the inset map for the location of the study area.
Seismic P-wave velocity models of the oceanic crust
a–c, FWI²⁸ velocity model starting from the tomographic seismic velocity model¹⁴ (a), the velocity anomaly (the difference between the velocity models from FWI and tomography; b and the vertical velocity gradient (the derivative of velocity model from FWI with respect to depth; c. Black triangles in a mark the OBS locations. The velocity contours are from 5 to 7 km s⁻¹ with an increment of 0.2 km s⁻¹. The coloured parts in b and c start from the basement (the top of layer 2). The lithospheric age in c is calculated using a spreading rate of 16 mm yr⁻¹ (ref. ²⁶). The horizontal distance starts with 0 km at the MAR.
Observed and synthetic seismic waveform data from OBS 57
a, The field data (black) and the synthetic waveform (red) using the tomographic model (Extended Data Fig. 1). b, The field data (black) and the synthetic waveform (red) using the true-amplitude FWI model (Fig. 2). Seismic waveforms from more OBSs can be found in Extended Data Figs. 3 and 4. A reduced travel time of 7 km s⁻¹ velocity was applied to both the field and synthetic data.
Modelled layering in the lower oceanic crust
a, 1D velocity–depth profiles at the distance of 182 km from the MAR, derived from tomography (blue) and FWI (black). Vp is the P-wave velocity of seismic waves. The thin dashed black line is the interpreted boundary of layer 2A/2B from the FWI model; the bold dashed blue and black lines indicate the boundaries of layer 2/3 from the tomographic and the FWI models, respectively. We invert the velocity model from FWI down to a maximum of 9 km depth (~5.4 km from basement at distance 182 km), because of potential Moho reflection interference for greater depths (Supplementary Fig. 22). b, A schematic diagram illustrating the oceanic crustal structure at the MAR and away from the ridge axis. The black balls in layer 2A and stripes in layer 2B refer to pillow lava and basaltic dykes, respectively. The blue ellipsoids refer to hydrothermal alteration above the roof of the AML that could contribute to the top low-velocity layer. The red vertically elongated ellipsoids suggest magma from mantle upwelling. The light brown layers in the lower crust refer to the low-velocity layers from FWI. The imaged layers in the lower crust may comprise multiple thinner layers as indicated by the darker brown patches, but are beyond data resolution.
Oceanic crust forms at mid-ocean spreading centres through a combination of magmatic and tectonic processes, with the magmatic processes creating two distinct layers: the upper and the lower crust. While the upper crust is known to form from lava flows and basaltic dykes based on geophysical and drilling results, the formation of the gabbroic lower crust is still debated. Here we perform a full waveform inversion of wide-angle seismic data from relatively young (7–12-Myr-old) crust formed at the slow-spreading Mid-Atlantic Ridge. The seismic velocity model reveals alternating, 400–500 m thick, high- and low-velocity layers with ±200 m s ⁻¹ velocity variations, below ~2 km from the oceanic basement. The uppermost low-velocity layer is consistent with hydrothermal alteration, defining the base of extensive hydrothermal circulation near the ridge axis. The underlying layering supports that the lower crust is formed through the intrusion of melt as sills at different depths, which cool and crystallize in situ. The layering extends up to 5–15 km distance along the seismic profile, covering 300,000–800,000 years, suggesting that this form of lower crustal accretion is a stable process.
Map of Amundsen Sea Embayment, Antarctica, showing sites mentioned in the text
Our study locations are shown in bold text. Textured grey represents the current ice-sheet surface (from Reference Elevation Model of Antarctica³⁵). Shaded grey areas outlined in blue represent ice shelves. Dashed blue lines indicate former ice flow directions through Pine Island Trough (PIT). Bathymetry (blue and white shades) is from GEBCO2019 global dataset³⁶. The basemap was created from the SCAR Antarctic Digital Database.
Satellite imagery and photographs of study sites in the Amundsen Sea Embayment
a–c, Left panel shows Edwards Islands (a), Lindsey Islands (b) and Schaefer Islands (c). Dashed lines in photos on right panels denote raised marine beaches from which shell and bone samples were collected. Photos are from sites corresponding to the red squares in adjacent imagery. Yellow stars in a denote location of additional sampled islands (Photos provided in Supplementary Figs. 2, 3 and 6). Credit: a–c (left), WorldView-2/DigitalGlobe, a Maxar Company; a–c (right), Scott Braddock.
RSL reconstruction and comparison with GIA models for the Amundsen Sea Embayment, Antarctica
a, RSL curve based on radiocarbon ages of shells from raised beaches. All samples are presented as calibrated radiocarbon ages for shells and bones with horizontal bars indicating 2-sigma errors. The solid black line with associated bootstrap confidence intervals is an interpolation of the shell data (see Supplementary Information) using the method of ref. ³⁷. The horizontal dashed black line represents the proposed regional marine limit (elevation of highest sampled beach) at 19 m. Elevation errors (Methods) associated with radiocarbon samples are represented by vertical bars on each data point and range from 0.3 to 0.7 m. Vertical error bars for Lindsey Island samples (0.3 m) are smaller than symbols at this scale. b, Comparison of RSL data with GIA model predictions derived using two different ice-history models (ICE-6G_C27,32 and W1228,33). For each ice-history model, two curves are shown; these represent RSL change assuming either a strong (solid lines, upper mantle viscosity = 5 × 10²⁰ Pa s) or a weak (dashed lines, upper mantle viscosity = 5 × 10¹⁹ Pa s) Earth model. The lithosphere thickness is set at 71 km (ref. ³⁴).
The rapidly retreating Thwaites and Pine Island glaciers together dominate present-day ice loss from the West Antarctic Ice Sheet and are implicated in runaway deglaciation scenarios. Knowledge of whether these glaciers were substantially smaller in the mid-Holocene and subsequently recovered to their present extents is important for assessing whether current ice recession is irreversible. Here we reconstruct relative sea-level change from radiocarbon-dated raised beaches at sites immediately seawards of these glaciers, allowing us to examine the response of the earth to loading and unloading of ice in the Amundsen Sea region. We find that relative sea level fell steadily over the past 5.5 kyr without rate changes that would characterize large-scale ice re-expansion. Moreover, current bedrock uplift rates are an order of magnitude greater than the rate of long-term relative sea-level fall, suggesting a change in regional crustal unloading and implying that the present deglaciation may be unprecedented in the past ~5.5 kyr. While we cannot preclude minor grounding-line fluctuations, our data are explained most easily by early Holocene deglaciation followed by relatively stable ice positions until recent times and imply that Thwaites and Pine Island glaciers have not been substantially smaller than present during the past 5.5 kyr.
Global daily emissions changes in 2019 and 2020
a, Daily global CO2 emissions trends in 2019 and 2020. Real emissions data are shown as solid black lines in the form of a seven-day running mean. The dotted line represents simulated baselines in 2020 (see Methods for more details). b, Global daily CO2 reduction in 2020 compared with 2019 and the global daily new death cases of COVID-19 in 2020. The green area shows the seven-day running mean of daily CO2 emissions change in 2020 compared with 2019. The daily new death numbers of COVID-19 (ref. ³⁸) are shown in blue areas and are used as indicators for the progress of the COVID-19 pandemic.
Country-specific daily CO2 emissions in 2019 and 2020
The thick solid lines show the seven-day running mean of daily CO2 emissions in 2020, and the results in 2019 are shown by thin solid lines. The simulated baselines, which combine the daily variation from 2019 with historical sector-specific and country-specific emissions trends, are shown as dashed lines. The ranges of uncertainty (95% confidence interval; 2-sigma errors) associated with the simulated baselines and daily estimates in 2020 are shown in light and dark shaded areas, respectively.
Sector-specific daily CO2 emissions in 2019 and 2020
a, Global daily CO2 emissions by sector in 2019 and 2020 in the form of a seven-day running mean. b, The contribution of each sector to the total emissions change in 2020 compared with 2019. ‘International bunkers’ includes the emissions from the international aviation sector and the international shipping sector.
Correlation matrixes of five indicators during 1 March~31 May 2020 and 1 October~31 December 2020
a, 1 March–31 May 2020. b, 1 October–31 December 2020. CO2, daily CO2 changes in 2020 compared with the simulated baseline emissions, which combine the daily emissions patterns in 2019 and historical sectoral trends; D, the daily new death cases of COVID-19³⁸; SI, the stringency index of government responses to COVID-19³⁹; GR, the duration spent at places of residence⁴⁰; E, the daily changes of power demand in 2020 compared with the same day in 2019. These five indicators are the averages of the United States, India, the United Kingdom, France, Germany, Italy, Spain, Russia, Brazil and Japan. The daily averages of the five indicators during these two periods are shown in Extended Data Table 1. Gradient colors indicate negative strong relationship (red) through no relationship (yellow) to positive strong relationship (green).
Day-to-day changes in CO2 emissions from human activities, in particular fossil-fuel combustion and cement production, reflect a complex balance of influences from seasonality, working days, weather and, most recently, the COVID-19 pandemic. Here, we provide a daily CO2 emissions dataset for the whole year of 2020, calculated from inventory and near-real-time activity data. We find a global reduction of 6.3% (2,232 MtCO2) in CO2 emissions compared with 2019. The drop in daily emissions during the first part of the year resulted from reduced global economic activity due to the pandemic lockdowns, including a large decrease in emissions from the transportation sector. However, daily CO2 emissions gradually recovered towards 2019 levels from late April with the partial reopening of economic activity. Subsequent waves of lockdowns in late 2020 continued to cause smaller CO2 reductions, primarily in western countries. The extraordinary fall in emissions during 2020 is similar in magnitude to the sustained annual emissions reductions necessary to limit global warming at 1.5 °C. This underscores the magnitude and speed at which the energy transition needs to advance. Observed daily changes in CO2 emissions from across the globe reveal the sectors and countries where pandemic-related emissions declines were most pronounced in 2020.
The evolution of land plants during the Palaeozoic Era transformed Earth’s biosphere 1. Because the Earth's surface and interior are linked by tectonic processes, the linked evolution of the biosphere and sedimentary rocks should be recorded as a near-contemporary shift in the composition of the continental crust. To test this hypothesis, we assessed the isotopic signatures of zircon formed at subduction zones where marine sediments are transported into the mantle 2,3, thereby recording interactions between surface environments and the deep Earth. Using oxygen and lutetium-hafnium isotopes of magmatic zircon that respectively track surface weathering (time-independent) 4 and radiogenic decay (time-dependent) 5, we find a correlation in the composition of continental crust after 430 Myr ago, which is coeval with the onset of enhanced complexity and stability in sedimentary systems related to the evolution of vascular plants. The expansion of terrestrial vegetation brought channelled sand-bed and meandering rivers, muddy floodplains, and thicker soils, lengthening the duration of weathering before final marine deposition 6,7. Collectively, our results suggest that the evolution of vascular plants coupled the degree of weathering and timescales of sediment routing to depositional basins where they were subsequently subducted and melted. The late Palaeozoic isotopic shift of zircon indicates that the greening of the continents was recorded in the deep Earth.
Unrest episodes observed in basaltic systems indicate magma influx rates may be key to generating long-term eruption forecasts. The findings predict that, if a critical flow rate is surpassed, a volcano will erupt within a year.
Global warming-induced melting and thawing of the cryosphere are severely altering the volume and timing of water supplied from High Mountain Asia, adversely affecting downstream food and energy systems that are relied on by billions of people. The construction of more reservoirs designed to regulate streamflow and produce hydropower is a critical part of strategies for adapting to these changes. However, these projects are vulnerable to a complex set of interacting processes that are destabilizing landscapes throughout the region. Ranging in severity and the pace of change, these processes include glacial retreat and detachments, permafrost thaw and associated landslides, rock–ice avalanches, debris flows and outburst floods from glacial lakes and landslide-dammed lakes. The result is large amounts of sediment being mobilized that can fill up reservoirs, cause dam failure and degrade power turbines. Here we recommend forward-looking design and maintenance measures and sustainable sediment management solutions that can help transition towards climate change-resilient dams and reservoirs in High Mountain Asia, in large part based on improved monitoring and prediction of compound and cascading hazards.
Aftershocks of M ≥9.0 megathrust ruptures since 1960
a,b, 2011 Tohoku: M ≥3.0 shocks from JMA; slip from ref. ¹⁹. c,d, 1964 Prince William Sound: M ≥4.5 shocks from ref. ⁴³ for first 21 months, and from the Advanced National Seismic System (ANSS) afterwards; slip from ref. ⁴⁴. e,f, 2004 Sumatra: M ≥4.5 shocks from ANSS, with M ≥7.0 events labelled; slip from ref. ⁴⁵. g,h, 1960 Valdivia: M ≥4.5 shocks from ANSS; slip from ref. ²⁴, with three isolated patches of slip below 75 km, probably numerical artefacts, not shown. Seismicity on the rupture surface shut down within 5 yr, whereas seismicity in the surrounding corona lasts up to 50 yr. In each case, we plot seismicity near or above the magnitude of completeness for the period shown.
Change in seismicity rate beginning 5 yr after the Tohoku M 9 earthquake
The seismicity 5–10 yr after the mainshock is compared with 13 yr before the M 9 (11 March 2016–11 March 2021 / 1 January 1998–11 March 2011 14:45), for <150 km depth and a 20 km smoothing radius. a, Observed rate change. The ‘core’, in which the seismicity has shut down, collocates with the rupture19,46. The surrounding corona has been the site of 22 M ≥6.7 shocks since the M 9 struck. Aftershock zones and swarms during the pre-mainshock period are masked where coefficient of variation (COV) ≥ 3, as explained in Methods. Active faults are green. b, The rate–state Coulomb model resembles the observed seismicity-rate changes, with a spatial regression coefficient of 0.61 and a slope of 0.67.
Source data
Cross-sections of seismicity-rate and stress change for the Tohoku M 9 earthquake
a, Seismicity-rate change as in Fig. 2a, showing that the core and corona extend through the lithosphere. b, Coulomb stress change resolved onto aftershock focal mechanisms (side projections, with the most recent shocks plotted on top). Most nodal planes were brought closer to failure by the mainshock, activating these faults, probably causing the corona aftershocks in a. c, Background mechanisms are brought farther from failure, inactivating these faults. The diversity of receiver faults leads to heterogeneity of the stress transfer. Assumed fault friction is 0.4, with mechanisms coloured by the most positively stressed nodal plane. If the outer rise shocks were shallower, as suggested by an ocean bottom seismometer network⁴⁷, they would be even more strongly promoted.
Source data
Modelled response of seismicity to a megathrust earthquake
In the corona (red curve), the seismicity rate jumps and then decays to the background rate over ~40 yr. In the core (dark-blue curve), the rate also jumps, but decays within a few years below its pre-mainshock level, where it remains for centuries. The stress-change s.d. values are used to represent the heterogeneity of the imparted stress. ta is 20 yr, Aσ is 0.5 bar and the curves are means of Monte Carlo simulations. A larger stress drop in the core (light-blue curve) causes a more rapid shutdown, perhaps explaining the 1960 M 9.5 aftershocks.
Contemporary gaps in seismicity could represent prehistoric megathrust earthquakes
a, 1952 M ~8.8 Kamchatka⁴⁸ (the authors’ preferred ‘JASmod7’ model shown). b, 1762 M ~8.8 Arakan⁴⁹. c, 1906 M ~8.8 Ecuador⁵⁰. d, 1868 M ~9.0 Arica, Peru–Chile⁵¹ with several holes (arrows) seen along the modelled uniform-slip rupture. e, 1700 M ~9.0 Cascadia⁵². f, 1944 M 8.1 Tonankai and 1946 M 8.3 Nankai⁵³. ANSS seismicity.
Megathrust earthquakes release and transfer stress that has accumulated over hundreds of years, leading to large aftershocks that can be highly destructive. Understanding the spatiotemporal pattern of megathrust aftershocks is key to mitigating the seismic hazard. However, conflicting observations show aftershocks concentrated either along the rupture surface itself, along its periphery or well beyond it, and they can persist for a few years to decades. Here we present aftershock data following the four largest megathrust earthquakes since 1960, focusing on the change in seismicity rate following the best-recorded 2011 Tohoku earthquake, which shows an initially high aftershock rate on the rupture surface that quickly shuts down, while a zone up to ten times larger forms a ring of enhanced seismicity around it. We find that the aftershock pattern of Tohoku and the three other megathrusts can be explained by rate and state Coulomb stress transfer. We suggest that the shutdown in seismicity in the rupture zone may persist for centuries, leaving seismicity gaps that can be used to identify prehistoric megathrust events. In contrast, the seismicity of the surrounding area decays over 4–6 decades, increasing the seismic hazard after a megathrust earthquake.
Low-latitude clouds control the existence and absence of waterbelt states in the GCMs CAM and ICON
a–c, Bifurcation diagrams of global-mean ice edge versus atmospheric CO2 concentrations for CAM in its default configuration and with clouds made transparent to radiation in a narrow tropical region (pCOOKIE) (a), the default configuration of ICON (b) and ICON WBF (c). Filled symbols show stable states. Circles show simulations initialized from ice-free conditions, squares show simulations initialized from stable waterbelt states and diamonds represent simulations initialized from transient waterbelt states. Unfilled diamonds mark slowly drifting simulations that remain in a waterbelt-like state for at least 40 years, with arrows indicating the drift of the ice edge. Lines are drawn as best guesses of equilibrium states, with solid lines indicating stable and dashed lines indicating unstable states. T\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\mathcal{T}}}}$$\end{document}, W1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{{\mathcal{W}}}}}^{1}$$\end{document} and W2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{{\mathcal{W}}}}}^{2}$$\end{document} here label the bifurcation points (nose points) corresponding to the CO2 thresholds referred to in the text. Bifurcation points mark the unstable transitions between temperate and waterbelt/snowball climate (T\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{\mathcal{T}}}}$$\end{document}), between waterbelt and snowball climate (W1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{{\mathcal{W}}}}}^{1}$$\end{document}) and between waterbelt and temperate climate (W2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{{\mathcal{W}}}}}^{2}$$\end{document}).
Differences in clouds and their SWCRE as obtained from the GCMs CAM and ICON as well as ICON WBF
a–c, Zonal-mean annual-mean planetary albedo α for all-sky (solid) and clear-sky (dashed) conditions (a), total cloud cover (b) and SWCRE (c) averaged over all stable waterbelt states for each model. The grey band in a indicates the range of global-mean ice-edge latitudes for all stable equilibrium states found in CAM, ICON and ICON WBF simulations. d,e Zonal-mean cloud cover together with 273 K, 235 K and 192 K isotherms in CAM (d) and ICON (e) for simulations with comparable global-mean ice cover (10,000 ppmv CO2 in CAM; 4,063 ppmv CO2 in ICON). The orange dotted box in d shows region of CAM pCOOKIE modification for the Northern Hemisphere. Purple contours in e show the cloud-cover difference between ICON WBF (at 6,000 ppmv CO2) and ICON (contour interval of 3%; positive differences in solid). f, LCF from simulations shown in d and e. The range of temperatures for which liquid and ice are equally prevalent for 26 Coupled Model Intercomparison Project phase 5 (CMIP5) GCMs is shown in grey. The red line shows the combined observational range of ground-based LIDAR and aircraft measurements²¹.
Analysis of the waterbelt regime in a one-dimensional EBM
a, Bifurcation diagrams of ice-edge latitude versus CO2-radiative forcing F calculated from the EBM with parameters estimated from the GCMs. b, Climate-sensitivity parameter λ determined from the EBM as a function of the ratio of planetary albedo over ice-free ocean αo to bare sea-ice albedo αi,b and ice-edge latitude. c, Map in αo–αi,b space showing domains of stable and accessible waterbelt states (white) and stable but unaccessible waterbelt states (light grey). The black dashed lines indicate the margins within which the waterbelt regime can exist. The red dotted box indicates the range of plausible values for αo and αi,b. The black circle marks albedo values for CAM.
Geological evidence of active tropical glaciers reaching sea level during the Neoproterozoic (1,000–541 Ma), suggesting a global ocean completely covered in ice, was the key observation in the development of the hard Snowball Earth hypothesis. These conditions are hard to reconcile with the survival of complex marine life through Snowball Earth glaciations, which led to alternative waterbelt scenarios where a large-scale refugium was present in the form of a narrow ice-free strip in the tropical ocean. Here we assess whether a waterbelt scenario maintained by snow-free dark sea ice at low latitudes is plausible using simulations from two climate models run with a variety of cloud treatments in combination with an energy-balance model. Our simulations show that waterbelt states are not a robust and naturally emerging feature of Neoproterozoic climate. Intense shortwave reflection by mixed-phase clouds, in addition to a low albedo of bare sea ice, is needed for geologically relevant waterbelt states. Given the large uncertainty in mixed-phase clouds and their interaction with radiation, our results strongly question the idea that waterbelt scenarios can explain the Neoproterozoic geology. Hence, Neoproterozoic life has probably faced the harsh conditions of a hard Snowball Earth.
Time series of proxy data across the inception interval
a, Global sea-level change³, top-of-atmosphere insolation at 65° N at June solstice¹ and CO2 concentration⁵¹. b–e, Data from North Atlantic sediment cores. Coloured lines in b–e depict SST (red), IRD (black), abundance of Neogloboquadrina pachyderma sinistral (N. pach. (s); blue), a planktonic foraminifera indicator species for polar waters and its δ¹³C variations (δ¹³C N. pach. (s); green). Positive/negative δ13C N. pach. (s) indicates dominance of Arctic water (ArW)/Atlantic water (AtlW), respectively. Panels depict data from the western (cores CH69-K09 and EW9302-JPC2; ref. ⁶) (b), northeastern (core MD99-2304; refs. 6–8) (c), eastern (core MD95-2010; refs. 7,8) (d) and east-central (core ODP 980; refs. 6,10) (e) North Atlantic, respectively. Proxy-data sites are indicated in the top right corner. Discontinuous lines in b–d are due to missing data in the proxy datasets. Maps generated using Cartopy with Natural Earth shapefiles.
Map of regions and major currents in the Northern Hemisphere high latitudes
Ocean regions and primary Arctic ocean gateways are denoted by black text, and land regions are denoted by green text; Spitsbergen is indicated by a green star, and the CAA is highlighted by orange shading. Arrows show ocean currents with Pacific (orange), Atlantic (red) and Arctic (blue) provenance. Dashed circles in the Labrador Sea, the Irminger Sea and the Nordic Seas indicate the approximate locations of North Atlantic deep convection regions that drive the meridional overturning circulation. The background map is showing topography and bathymetry as represented on the 4 × 4 km ice-sheet model grid (Methods). This grid has 2,592 × 2,592 grid cells and matches the resolution of the default ice-sheet model grid currently supported by CESM2 (outlined by the red box around Greenland).
Climate response to 116 ka insolation and greenhouse gas forcing and to the closing of the CAA ocean gateways
a–c, The difference between the 116 ka simulation with open CAA gateways and the pre-industrial control simulation for boreal summer (June–July–August (JJA)) surface temperature anomaly (filled circles indicate proxy-data sites in Fig. 1) (a), JJA freshwater flux anomaly (shading) and 50% sea-ice margin in the pre-industrial (black contours) and the 116 ka simulation with open CAA gateways (red contours), respectively (b) and JJA sea surface salinity anomaly (shading) and annual mean mixed-layer depth anomaly in contours (solid/dashed lines denote positive/negative mixed-layer depth anomalies, respectively) (c). d–f, The corresponding response to closed CAA gateways with respect to the 116 ka simulation with open CAA gateways; contours in e indicate the 50% JJA sea-ice margin in the 116 ka simulation with open (red contours) and closed (blue contours) CAA ocean gateways, respectively. Stippling indicates differences that are not significant at the 95% level. Maps generated using Cartopy with Natural Earth shapefiles.
Simulated ice-sheet thickness in the simulation with open and closed ocean gateways in the CAA, respectively
a, Simulated ice-sheet thickness in the 116 ka simulation with open CAA ocean gateways. b, More-detailed view of Scandinavia (region indicated by black box in a). Note that there is no ice growth in Scandinavia in this simulation. c, Simulated ice-sheet thickness in the 116 ka simulation with closed CAA ocean gateways. d, More-detailed view of Scandinavia (region indicated by black box in c). The Greenland Ice Sheet is fully interactive in these simulations; however, the initial Greenland extent is indicated by grey shading for plotting purposes. The background map is showing topography and bathymetry as represented on the 4 × 4 km ice-sheet model grid.
The last glacial cycle began around 116,000 years before present during a period with low incoming solar radiation in Northern Hemisphere summer. Following the glacial inception in North America, the marine sediment record depicts a weakening of the high-latitude ocean overturning circulation and a multi-millennial eastward progression of glaciation across the North Atlantic basin. Modelling studies have shown that reduced solar radiation can initiate inception in North America and Siberia; however, the proximity to the temperate North Atlantic typically precludes ice growth in Scandinavia. Using a coupled Earth-system–ice-sheet model, we show that ice forming in North America may help facilitate glacial expansion in Scandinavia. As large coherent ice masses form and start filling the ocean gateways in the Canadian Archipelago, the transport of comparatively fresh North Pacific and Arctic water through the archipelago is diverted east of Greenland, resulting in a freshening of North Atlantic deep convection regions, sea-ice expansion and a substantial cooling that is sufficient to trigger glacial inception in Scandinavia. This mechanism may also help explain the Younger Dryas cold reversal and the rapid regrowth of the Scandinavian Ice Sheet following several warm events in the last glacial period.
Maps of IME detection in the tropical Pacific
The main map represents the 2002–2018 average Chl (from MODIS data) in the study region (30° S–30° N). IME regions are contoured in different shades of red depending on the Chl increase near islands relative to a REF region (maximum value in mean conditions: 558% observed for the Galapagos Islands). a–f, The climatological month with the highest IME Chl enrichment in selected regions: the Gilbert Islands (Kiribati) in May (a), the Hawaiian Islands in September (b), the Galapagos Islands in January (c), the Solomon Islands in July (d), New Caledonia, Vanuatu, Fiji and Tonga in February (e) and the Marquesas Islands in May (f). Note the differing colour scales. Extended Data Fig. 2 provides full maps of climatological months. Maps displayed using m_map (
Maps of island impacts on PHYSAT phenoclasses
The maps are indicative of impacts on phytoplankton community composition and biodiversity. For each island, the maximum value observed over the seasonal cycle is displayed (only computed for IME regions containing at least 100 PHYSAT data points). a, Phenoclass Bray–Curtis dissimilarity between IME and REF regions, which is indicative of differences in phenoclass composition. b, Phenoclass richness increases in the IME region relative to the REF region (richness was normalized to 100 data points). c, Summary of island impacts on phenoclasses, defined according to the median value for phenoclass Bray–Curtis dissimilarity (composition) and for richness increase, where ‘impacts’ are defined as values above the median. Phenoclass richness increases are correlated with absolute latitude (r = 0.18) whereas phenoclass Bray–Curtis dissimilarity is anticorrelated (r = −0.24). Despite their opposite relation to latitude, phenoclass richness increases and Bray–Curtis dissimilarity are actually weakly correlated (r = 0.30). Maps displayed using m_map (
Island impacts on phytoplankton community structure as depicted by PHYSAT phenoclasses
Statistics are based on paired climatological IME/REF regions with a minimum of 100 PHYSAT data points. Box plots report the median (centre line), 95% confidence interval around the median (notches equal to ±1.58 times the interquartile range divided by the square root of the sample size), the interquartile range (box), and 1.5 times the interquartile range (whiskers). Data points are displayed as crosses (outliers beyond whiskers as dots). Symbols show whether distributions are statistically different according to a Mann–Whitney U test (=, nonsignificant; + and − indicate how IME medians relate to REF medians). a, Left: phenoclass Bray–Curtis dissimilarity between IME and REF regions compared with data points randomly taken across IME/REF regions. Right: occurrence frequency of dominant phenoclasses (frequency > 0.02) in the IME (red) and REF (blue) regions, sorted by label: Prochlorococcus (Pros), Synechococcus (Syn) and nanoplankton (Nano). b, Left: phenoclass richness is significantly higher in IME than in REF regions (9.6% difference in median), partly linked to higher Chl (12.4% difference in the medians). Right: summary of Chl and phenoclass richness characteristics for 1,000 random permutations independently taken within IME and REF regions (N = 712) such that their Chl distribution is equivalent (P > 0.05). Chl may be insignificantly higher within IME or REF subsets (as described by the difference in medians) because IME/REF pairs are decoupled (see an example in Extended Data Fig. 4). The box plots were generated using the IoSR Matlab toolbox (
In the relatively unproductive waters of the tropical ocean, islands can enhance phytoplankton biomass and create hotspots of productivity and biodiversity that sustain upper trophic levels, including fish that are crucial to the survival of islands’ inhabitants. This phenomenon, termed the island mass effect 66 years ago, has been widely described. However, most studies focused on individual islands, and very few documented phytoplankton community composition. Consequently, basin-scale impacts on phytoplankton biomass, primary production and biodiversity remain largely unknown. Here we systematically identify enriched waters near islands from satellite chlorophyll concentrations (a proxy for phytoplankton biomass) to analyse the island mass effect for all tropical Pacific islands on a climatological basis. We find enrichments near 99% of islands, impacting 3% of the tropical Pacific Ocean. We quantify local and basin-scale increases in chlorophyll and primary production by contrasting island-enriched waters with nearby waters. We also reveal a significant impact on phytoplankton community structure and biodiversity that is identifiable in anomalies in the ocean colour signal. Our results suggest that, in addition to strong local biogeochemical impacts, islands may have even stronger and farther-reaching ecological impacts.
An overview of monitoring locations and timing of talik development
a, Distribution of study sites showing where talik formation has been observed (orange), where no talik has been observed and permafrost remains unaffected (blue), where no permafrost is present within the upper 5 m (red) and where a talik was pre-existing (brown). The blue circle with a black dot indicates the location of the SL#2 site, and the BC site is denoted by the orange circle with the number 22 in Fig. 1a, where PT, PF and talik thickness were modelled. FDD and TDD values shown in Fig. 2a–c were calculated using climate records from Fairbanks, Nome and Kotzebue airports, indicated by solid black circles. Permafrost extent forms the base layer of the map⁴⁶. b, Photo of the Council (COT) borehole site in the Seward Peninsula region. c, Photo of the Kugaruk Cabin site in the Selawik region. d, Photo of the Bonanza Creek borehole site in the Interior Alaska region. e, Table showing all sites where talik was observed, the observation period and years of permafrost unaffected by talik (blue), new talik formation (orange), talik refreezing (green) and no data (grey). Map in a reproduced from ref. ⁴⁶, Institute of Northern Engineering. Credit: b–d, V. Romanovsky. See Supplementary Table 1 for more information on sites classified as ‘permafrost, no-talik’, ‘pre-existing talik’ and ‘no permafrost’.
FDD, TDD and snow-depth values at key sites
a–c, TDD (red circles) and FDD (blue circles) values for Fairbanks International Airport (a), Nome (b) and Kotzebue airport (c) (locations shown in Fig. 1a). TDD values were calculated by summing all positive values during the summer, which we defined as 15 May to 31 October. FDD values were calculated by summing all negative values between 1 November and 14 May. A comparison of long-term (1950–2018) and recent (2010–2018) FDD and TDD values can be seen in Supplementary Table 2. d–f, Observed snow-depth values from three monitoring sites: Bonanza Creek in the Fairbanks region (d), Pilgrim Hot Springs in the Seward Peninsula region (e) and Selawik (f). The dotted line indicates when talik formation became widespread (the winter of 2017–2018). While Bonanza Creek and Pilgrim Hot Springs are sites of talik formation, the Selawik site is not a site of talik formation but is provided as a representative site for sites of talik formation in the Selawik region (Fig. 1).
Source data
Example of talik formation at six sites from across the study region
a, Kuzitrin River, in the Seward Peninsula region in northwest Alaska. b, Council (COT), in the Seward Peninsula region. c, Kugaruk Cabin, in the Selawik region in northwest Alaska. d, Bonanza Creek, in the Fairbanks region. e, Fox, in the Fairbanks region. f, Healy, in the Alaska Range foothills, Interior Alaska. Yellow boxes represent periods of talik formation and follow the legend in Fig. 4. Red lines indicate that the temperature sensor was in the active layer, or seasonally frozen layer, and a blue line indicates that the sensor was within permafrost. The x axis indicates the calendar year, beginning in January. To illustrate talik development more clearly, not all sensor depths are shown. * indicates the deepest sensor is shown.
PT, PF and talik thickness through time
a, Hindcast (1960–2019) and projected (2020–2050) PF and PT at the BC site, where talik development has occurred. b, Hindcast (1970–2019) and projected (2020–2050) PF and PT at the SL#2 site, where talik development has not yet been observed. Error bars for both sites are based on the coefficient of variance determined by a comparison between observed and modelled active-layer depths at BC. For both a and b, the black dotted line shows PF depth under low snow conditions, the black line shows PF under normal historical and normal projected snow conditions and the red line represents potential summer thaw. c, Conceptual diagram illustrating the relationship between PT, PF and talik development through time. d, Talik development through 1970–2100.
Source data
Talik formation has long been acknowledged as an important mechanism of permafrost degradation. Currently, a lack of in situ observations has left a critical gap in our understanding of how ongoing climate change may influence future sub-aerial talik formation in areas unaffected by water bodies or wildfire. Here we present in situ ground temperature measurements from undisturbed sub-aerial sites across the discontinuous permafrost zone of Alaska between 1999 and 2020. We find that novel taliks formed at 24 sites across the region, with widespread initiation occurring during the winter of 2018 due to higher air temperatures and above-average snowfall insulating the soil. Future projections under a high emissions scenario show that by 2030, talik formation will initiate across up to 70% of the discontinuous permafrost zone, regardless of snow conditions. By 2090, talik in areas of black spruce forest, and warmer ecosystems, may reach a thickness of 12 m. The establishment of widespread sub-aerial taliks has major implications for permafrost thaw, thermokarst development, carbon cycling, hydrological connectivity and engineering. Temperature observations from across Alaska show widespread talik formation in the discontinuous permafrost zone due to higher air temperatures and above-average snowfall in recent years.
Fluxes and stoichiometry of nutrient cycling across the trophic gradient
a,b, The N/P ratios of nutrient inflow and outflow (a) and in-lake enrichment and depletion (b). s.d., standard deviation. c, The N/P distributions of inflow and outflow across the trophic gradient. The dashed line in c represents the difference between the medians of inflow N/P and outflow N/P. d, The N/P distributions of in-lake enrichment and depletion. The dashed line in d represents the difference between the medians of in-lake enrichment N/P and in-lake depletion N/P. e–h, The distributions of fluxes of nutrient inflow and outflow (e,g) and in-lake enrichment and in-lake depletion (f,h) across the trophic gradient for N (e,f) and P (g,h). The dashed lines in e–h represent the medians of contributions of inflow, in-lake enrichment, outflow and in-lake depletion to total nutrient input (or output). The white dot in each violin plot represents the median, the thick black line represents the 25th and 75th percentiles and the thin black line represents the 10th and 90th percentiles.
Source data
Global pattern of preferential nutrient retention
Lakes located in quadrant A (8%) all have a negative net N and P retention. Lakes located in quadrant B (<1% of total) retain N but release P, while lakes located in quadrant D (10%) retain P but release N. Lakes located in quadrant C (81%) retain both N and P. Lakes located below the black dashed line (87.8%) tend to retain more P than N (ln(ENN/DEN) > ln(ENP/DEP)). ENN, in-lake N enrichment; DEN, in-lake N depletion; ENP, in-lake P enrichment; DEP, in-lake P depletion.
Source data
Imbalance of nutrient retention in global lakes
a,b, The global distribution of in-lake enrichment/depletion ratio in lakes for N (a) and P (b). Yellow dots represent lakes that positively retain N or P; blue dots represent lakes that export N or P to downstream ecosystems.
Source data
Classification of N and P cycling stoichiometry
Imbalanced anthropogenic inputs of nitrogen (N) and phosphorus (P) have significantly increased the ratio between N and P globally, degrading ecosystem productivity and environmental quality. Lakes represent a large global nutrient sink, modifying the flow of N and P in the environment. It remains unknown, however, the relative retention of these two nutrients in global lakes and their role in the imbalance of the nutrient cycles. Here we compare the ratio between P and N in inflows and outflows of more than 5,000 lakes globally using a combination of nutrient budget model and generalized linear model. We show that over 80% of global lakes positively retain both N and P, and almost 90% of the lakes show preferential retention of P. The greater retention of P over N leads to a strong elevation in the ratios between N and P in the lake outflow, exacerbating the imbalance of N and P cycles unexpectedly and potentially leading to biodiversity losses within lakes and algal blooms in downstream N-limited coastal zones. The management of N or P in controlling lake eutrophication has long been debated. Our results suggest that eutrophication management that prioritizes the reduction of P in lakes—which causes a further decrease in P in outflows—may unintentionally aggravate N/P imbalances in global ecosystems. Our results also highlight the importance of nutrient retention stoichiometry in global lake management to benefit watershed and regional biogeochemical cycles. Lakes preferentially retain phosphorous over nitrogen, amplifying the imbalance of nutrient cycles caused by anthropogenic inputs, according to analyses of more than 5,000 lakes globally.
Clinopyroxene H2O abundances plotted against major-element compositions
a,c, Data for both the Jijal (blue (garnet pyroxenite) and red (garnet diorite) circles) and Chilas (black circles). b,d, Data for Chilas only. a,b, Octahedrally coordinated Na in atoms per formula unit (a.p.f.u.). c,d, Mg# (molar Mg/(Mg+Fe)). Error bars are 2 s.e.m. Samples are individually labelled; see text for sample description.
Kohistan equilibrium melt H2O abundances compared to arc magma literature
a. Equilibrium melt H2O abundances plotted against melt Mg#. Melt Mg# inferred from clinopyroxene Mg# for both Chilas (black circles) and Jijal (blue (garnet pyroxenites) and red (garnet diorite) circles) cumulates. Mg# error bars denote highest and lowest measured clinopyroxene Mg# for each sample, H2O error bars are 2 s.e.m. Melt-inclusion literature data (white circles) are maximum water abundances for individual arc volcanoes⁵ converted from olivine forsterite content to melt Mg#. Thick vertical black line denotes Mg# of melts in equilibrium with mantle olivine Fo90 (Mg# 72). b. Fluorine and H2O abundances of arc melt inclusions compared with parental melts calculated for Kohistan. Horizontal line signifies mean H2O content of arc melts (dashed lines are 1 s.d.) as interpreted from melt inclusions⁵. Olivine-hosted melt-inclusion data for Fig. 2b from the literature53–56.
Chilas clinopyroxene H2O concentrations plotted against trace-element concentrations with various geochemical affinities
a, The REE Ce. b, The high-field-strength element Zr. c, The volatile halogen F. d–f, The fluid-mobile elements Sr (d), Ba (e) and Li (f). Concentrations are the mean of each sample (clinopyroxene cores only for H2O). Error bars are 2 s.e.m. White squares denote experimental partitioning data from the literature19,36. Black solid lines and curves denote best-fit slopes calculated using MATLAB functions poly1 (a–c) and power1 (d).
Crystallization models for Chilas and Jijal
a, Equilibrium crystallization model (white squares) for Chilas gabbronorites showing the evolution of Sr/Y ratios against H2O abundances in melts. Upon continued crystallization of plagioclase, Sr decreases as it is compatible in plagioclase, whereas Y increases, being incompatible in all crystallizing phases. Values along curves are percentage crystallization. Parental melt has an H2O concentration of 1.21 wt% and is the melt parental to C66. b. Fractional crystallization models⁴⁰ for Jijal cumulates showing the evolution of Sr/Y ratios against varying initial H2O abundances in melts. The fitted range of H2O contents in parental melts varies from 4 to 10 wt% H2O. Dashed blue and red lines are tie lines connecting identical extents of crystallization for different starting melt H2O abundances (Red: 4 and 5 wt.% H2O, blue: 6 and 10 wt.% H2O). Error bars are 2 s.e.m. For details of the two models, see Supplementary Information.
Jijal crystallization model
a, Jijal melt crystallization model based on published petrogenetic model⁴⁰. Equilibrium melt SiO2 content for each sample calculated on the basis of clinopyroxene Mg# using modelled LLD. Thick black line denotes 1 GPa H2O solubility for a basaltic melt². Silicic magmas tend to have higher water solubilities than basalt; however, constraints at relevant pressure and temperature conditions are lacking. Values next to each curve represent the range of H2O abundances in parental melts, with the lower bound constrained by the minimum amount of water needed to crystallize amphibole in the fractionation sequence⁴⁰, that is, 3.25 wt% H2O. See Supplementary Information for details. y-axis error bars are 2 s.e.m.; x-axis error bars are 2 s.e.m. of modelled melt SiO2 content calculated from the range of Cpx Mg# found in each sample. b, Probability density function of mineral-hosted arc melt inclusions from the GEOROC Database⁵⁷. Primitive melt inclusions (Mg# > 65) are shown as a solid back line (n = 171). Melt inclusions with Mg# of 35 to 65 (n = 879), similar to the Mg# calculated for Kohistan parental melts, are shown as a dashed black line. Primitive melt inclusions (solid black line) display two probability peaks at 0.3 and 2.3 wt% H2O with a maximum H2O of 4.8 wt%, whereas those with an Mg# of 35 to 65 show a single peak at 2.0 wt% H2O and a maximum of 7.5 wt% H2O. c, H2O solubility curve for basaltic melts at 1,200 °C² compared with calculated melt H2O abundances for Jijal cumulates based on thermobarometry (Supplementary Table 1). y-axis error bars are 2 s.e.m.; x-axis error bars are 1σ based on the Cpx-garnet barometer of ref. ⁵⁸.
Magmatic volatiles (for example, water) are abundant in arc melts and exert fundamental controls on magma evolution, eruption dynamics and the formation of economic ore deposits. To constrain the H2O content of arc magmas, most studies have relied on measuring extrusive products and mineral-hosted melt inclusions. However, these methods have inherent limitations that obfuscate the full range of H2O in arc magmas. Here, we report secondary-ion mass spectrometry measurements of volatile (H2O, F, P, S, Cl) abundances in lower-crustal cumulate minerals from the Kohistan palaeo-arc (northwestern Pakistan) and determine H2O abundances of melts from which the cumulates crystallized. Pyroxenes retained magmatic H2O abundances and record damp (less than 1 wt% H2O) to hydrous (up to 10 wt% H2O) primitive melts. Subsequent crystal fractionation led to formation of super-hydrous melts with approximately 12–20 wt% H2O, predicted petrologically yet virtually absent from the melt-inclusion record. Porphyry copper deposits are probably a natural eventuality of fluid exsolution from super-hydrous melts, corroborating a growing body of evidence. The water content of arc magmas in the lower crust can reach up to 20 wt% during crystallization, according to geochemical analyses of minerals from the Kohistan palaeo-arc, Pakistan, underscoring the role of water in porphyry deposits formation.
Bralgah Crater and the surrounding uniform terrain
a, OCAMS PolyCam mosaic¹³ showing the uniform terrain (white border) northward of and surrounding the 70-m-diameter crater. The terrain northeast of the two rocks labelled 1 and 2 (−20°, 333° E and −28°, 337° E) is rougher and darker. The top of this image is just north of the equator, where elevations are lowest on Bennu. b, Tilt variation, a measure of surface roughness, showing the range of surface slopes within the local area. c, Surface roughness from OSIRIS-REx Laser Altimeter measurements. The colours represent the standard deviation of the radii of all laser altimeter measurements within 80 cm facets. The black dashed line marks a smooth area surrounding the crater. d, The b′/v band ratio map for 300° E to 0° and −60° to 60° latitude. Bralgah Crater and the surrounding terrain have higher ratios.
Relationships between the mass and velocity of ejecta from scaling relationships
a, Ejecta velocities increase with surface strength proportional to the velocity scaling constant, C3Y/ρ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_3\sqrt {Y/\rho }$$\end{document}. If strength is greater than a few pascals on Bennu, then most of the ejecta, which are launched near the crater edge, will either land far from the crater or escape (the dashed lines indicate these transitions). Variations due to the unknown density of the upper 7 m of regolith on Bennu represent uncertainty in the mass–velocity relationship, shown for a sand/fly-ash mixture²⁴. Variations in the microgravity crater-forming processes at the crater edge due to differing material properties are much larger and discussed in Methods. b, During a cratering event, most mass is ejected at lower speeds and near the crater edge. Using the size of Bralgah Crater, gravity-regime scaling, parameters derived from experiments (Extended Data Table 1) and the modification factor for velocity near the edge (Methods), this plot shows the fraction of total ejected mass that has speeds above the plotted values. The difference between the two curves, one using parameters for sand (upper curve) and one for glass microspheres (lower curve), is a proxy for the uncertainty in the mass–velocity relationship in the gravity regime; 10% or less of the ejected mass is ejected faster than 8 cm s–1.
Simulation results of ejecta leaving the rim of Bralgah Crater according to gravity scaling
a, Density map of the mass deposition. The north/south asymmetry is due to the regional slope, and the westward curve is due to Bennu’s rotation. b, Map of the ejecta collision velocity tangent to the surface as the ejecta re-impacts Bennu, overlain onto a map of the Bralgah Crater region. Ejecta that returns to the surface within a distance of one crater radius to the north lands with a velocity <3.5 cm s–1. Material ejected at speeds between 8 and 12 cm s–1 are omitted from the plots because their landed locations are widely dispersed over Bennu, and they make up less than 10% of the total ejected mass. Most material ejected at >12 cm s–1 does not return to Bennu.
The surface strength of small rubble-pile asteroids, which are aggregates of unconsolidated material under microgravity, is poorly constrained but critical to understanding surface evolution and geologic history of the asteroid. Here we use images of an impact ejecta deposit and downslope avalanche adjacent to a 70-m-diameter impact crater on the rubble-pile asteroid (101955) Bennu to constrain the asteroid’s surface properties. We infer that the ejecta deposited near the crater must have been mobilized with velocities less than Bennu’s escape velocity (20 cm s–1); such low velocities can be explained only if the effective strength of the local surface is exceedingly low, nominally ≤2 Pa. This value is four orders of magnitude below strength values commonly used for asteroid surfaces, but it is consistent with recent estimates of internal strength of rubble-pile asteroids and with the surface strength of another rubble-pile asteroid, Ryugu. We find a downslope avalanche indicating a surface composed of material readily mobilized by impacts and that has probably been renewed multiple times since Bennu’s initial assembly. Compared with stronger surfaces, very weak surfaces imply (1) more retention of material because of the low ejecta velocities and (2) lower crater-based age estimates—although the heterogeneous structure of rubble piles complicates interpretation.
Examples of different fracture morphologies and fracture mapping
a,b, Fracture networks. c, Through-going fracture. d, Fracture affecting only part of the hosting boulder. e, Layer-like fractures. f,g, Through-going fractures appearing to split the boulders. h,i, Fractures on hummocky boulders. The red poly-lines represent the mapped fractures and their segments. Extended Data Fig. 1 presents the same panels without the mapping poly-lines, allowing better visualization of the fractures.
Windrose histograms of the azimuthal direction of fractures and fracture segments
a,b,c, Fracture directions, represented by a dark shade of grey. d,e,f, The directions of the elementary segments that compose the fractures, represented by a light shade of grey. The diagrams are plotted for three different bands of latitude, ϕ: northern mid-latitudes (a,d); equatorial (b,e); southern mid-latitudes (c,f).
Source data
Cumulative distribution of the length of fractures and fracture segments
The dashed-dotted curves represent best-fit exponential functions of the form N=N1eβLx\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N={N}_{1}\mathrm{e}^{\beta {L}_\mathrm{x}}$$\end{document}, where Lx is the length of the fractures or of the fracture segments (abscissa), N is their cumulative number (ordinate) and β is a factor that scales the exponent of the function. We obtain N1 = 2,525 ± 24, β = −0.585 ± 0.003 for the fracture length distribution and N1 = 4,284 ± 12, β = −0.939 ± 0.003 for the fracture-segment length distribution. The thick dashed straight line is the best fit of a power-law function of type N=N0Lxγ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N={N}_{0}{L}_\mathrm{x}\!^{\gamma }$$\end{document} for 4 ≤ Lx ≤ 10 m, where γ represents the power-law exponent. We find N0 = 6,347 ± 20 and γ = −2.4 ± 0.4.
Source data
On asteroids, fractures develop due to stresses driven by diurnal temperature variations at spatial scales ranging from sub-millimetres to metres. However, the timescales of such rock fracturing by thermal fatigue are poorly constrained by observations. Here we analyse images of the asteroid (101955) Bennu obtained by the Origins, Spectral Interpretation, Resource Identification and Security-Regolith Explorer (OSIRIS-REx) mission and show that metre-scale fractures on the boulders exposed at the surface have a preferential meridional orientation, consistent with cracking induced by diurnal temperature variations. Using an analytical model of fracture propagation, we suggest that fractures the length of those on Bennu’s boulders can be produced in 104–105 years. This is a comparable or shorter timescale than mass movement processes that act to expose fresh surfaces and reorient boulders and any preferential direction signature. We propose that boulder surface fracturing happens rapidly compared with the lifetime in near-Earth space of Bennu and other carbonaceous asteroids. The damage due to this space-weathering process has consequences for the material properties of these asteroids, with implications for the preservation of the primordial signature acquired during the accretional phases in the protoplanetary disk of our solar system. Fractures on the asteroid Bennu imaged by the OSIRIS-REx spacecraft are consistent with cracking induced by diurnal temperature variations over geologically rapid timescales.
Current dust emissions obtained with the ECHAM6-HAM2-BIOCRUST model
a, Global dust emissions, considering the mean effect of biocrusts on TFV. b, Bar chart showing regional dust emissions obtained with the ECHAM6-HAM2-BIOCRUST model, considering the mean effect of biocrusts on TFV for the period 1990–2020. Biocrust (BSC), as compared with dust emissions obtained by the standard ECHAM6-HAM2.1 model (ECH) and the median of AeroCom (Acom) models for the different regions (indicated by different colours). Error bars in the BSC model indicate uncertainty range in our estimation (minimum and maximum effects of biocrusts on TFV) whereas error bars in Acom represent the range of values obtained by the different models considered in the AeroCom project.
Impact of biocrusts on current global dust cycling and aerosol radiative effect
a–d, Hypothetical change of total annual dust emission (a), total annual dust deposition (b), mean annual atmospheric dust burden (c) and mean aerosol net radiative effect at the top of the atmosphere (d) upon complete removal of biocrusts. All calculations refer to mean annual values for the period 1990–2020 and are based on the mean effects of biocrusts on TFV. Individual maps showing biocrust effects on dust deposition by sedimentation, wet deposition and dry deposition are presented in Extended Data Fig. 4, and individual maps showing biocrust effects on short-wave and long-wave radiative forcing at the top of the atmosphere are shown in Extended Data Fig. 6.
Effect of global change and the induced biocrust cover loss estimated by 2070 on future dust cycling
a–d, Change in future total annual dust emission (a), total annual dust deposition (b), mean annual atmospheric dust burden (c) and mean annual aerosol net radiative effect at the top of the atmosphere (d). Data calculated according to RCP 2.6. Individual maps showing biocrust effects on dust cycling according to RCP 4.5 and RCP 8.5 are shown in Extended Data Figs. 7 and 8, respectively.
Effect of biocrusts on current global dust cycling and future changes with expected biocrust cover loss
a, Current effect of biocrusts on mean annual global dust emission, dust deposition, atmospheric dust burden and net radiative effect at the top of the atmosphere (period 1990–2020). b, Expected effects of anthropogenically induced biocrust cover loss by 2070 on future mean annual global dust emission, dust deposition, atmospheric dust burden and net radiative effect at the top of the atmosphere (calculated according to RCP 2.6 and RCP 8.5). Increases (decreases) of values are shown in red (black) letters.
Biological soil crusts (biocrusts) cover ~12% of the global land surface. They are formed by an intimate association between soil particles, photoautotrophic and heterotrophic organisms, and they effectively stabilize the soil surface of drylands. Quantitative information on the impact of biocrusts on the global cycling and climate effects of aeolian dust, however, is not available. Here, we combine the currently limited experimental data with a global climate model to investigate the effects of biocrusts on regional and global dust cycling under current and future conditions. We estimate that biocrusts reduce the global atmospheric dust emissions by ~60%, preventing the release of ~0.7 Pg dust per year. Until 2070, biocrust coverage is expected to be severely reduced by climate change and land-use intensification. The biocrust loss will cause an increased dust burden, leading to a reduction of the global radiation budget of around 0.12 to 0.22 W m⁻², corresponding to about 50% of the total direct forcing of anthropogenic aerosols. This biocrust control on dust cycling and its climate impacts have important implications for human health, biogeochemical cycling and the functioning of the ecosystems, and thus should be considered in the modelling, mitigation and management of global change.
Top-cited authors
Sebastiaan Luyssaert
  • Vrije Universiteit Amsterdam
Philippe Ciais
  • Laboratoire des Sciences du Climat et l'Environnement
Pierre Friedlingstein
  • University of Exeter
Jaia Syvitski
  • University of Colorado Boulder
Liviu Giosan
  • Woods Hole Oceanographic Institution