香京julia种子在线播放

The four locations studied are sites 658 (21°N, 19°W) and 659 (18°N, 21°W) from ODP expedition 108, located on top of the Cape Verde Plateau near the northwest African continental margin, and sites 1062 (28°N, 74°W) and 1063 (33°N, 57°W) from the sediment drifts of the western North Atlantic Ocean as part of ODP Leg 172. The sites were selected because they span a wide portion of the Northern Atlantic Ocean ( Figure 1 ). The marine sedimentary records of North African dust export offer the advantage of continuous sedimentation (Tiedemann et al., 1989; Giosan et al., 2002). These sites range from proximal to distal relative to the Sahara Dust Corridor source region. The average time resolution of the four records is about 4 kyr/sample covering the last glacial period to present (over the last 120,000 years) (Supporting Information).

Calculations of dust deposition (g m^-2 y^-1) were obtained from the average of three models (Mahowald et al., 1999; Ginoux et al., 2001; Jickells et al., 2005) ( Supplementary Table S7 ) to represent our best approximations of dust delivery.

Grain size distribution (GSD) in marine sediments are often used as a measure of dust delivery (Blott and Pye, 2001; Harrison et al., 2001; Baker et al., 2006; Ooki et al., 2009). Samples were dried at 80°C. Dry sieving was carried out for 15 minutes using a tapping sieve shaker (RO-TAP) equipped with a set of stainless-steel sieves. Short sieving times prevented the formation of aggregates due to electrostatic interactions and interparticle cohesion. Each fraction was weighed and recorded. Particle size distribution is represented graphically using a cumulative distribution curve and the D value method (S. J. Blott and Pye, 2001) from GRADISTAT (Blott, 2000). Samples were divided into >45μm (bulk), 45 to 20μm, and<20μm size fractions.

We used a state-of-the-art sequential Fe extraction procedure modified from Poulton and Canfield (2005) to characterize the Fe phases present (Fe_Na-Ac, Fe_Dith, and Fe_Ox), including minerals that might have formed diagenetically from initially bioavailable Fe. All Fe extracts were analyzed by inductively coupled plasma-mass spectrometry (ICP-MS; Agilent 7500ce) with H₂ and He modes in the collision cell, diluted with trace-metal grade 2% HNO₃ to enhance Fe detection by reducing interferences, thereby improving the accuracy of analytical results.

A multi-acid digest was performed to determine total solid-phase iron (Fe_T) and aluminum (Al) concentrations. Dried samples were ashed at 550°C, and a standard three-step digestion was performed using trace metal grade HF, HNO₃, and HCl at 140°C. This way, the potential bioreactivity of the Fe can be expressed as a fraction of the total Fe pool. Final concentrations were determined using the same ICP-MS. Reference standards SDO-1 (Devonian Ohio Shale) and SCO-1 (Cody Shale) were digested and analyzed in parallel with the sample extractions and yielded errors of less than<4%.

Splits from the multi-acid digest were used to measure the Fe isotope composition of the bulk sample (δ ⁵⁶Fe_T). Iron isotopes were analyzed at all four sites, resulting in total of 38 isotopic analyses ( Supplementary Table S11 ) Anion exchange resin and a standard ion chromatography protocol were used for Fe separation to eliminate matrix effects (Skulan et al., 2002; Arnold et al., 2004). Column yields were carefully monitored using the Ferrozine colorimetric method with UV-Vis spectrophotometry (λ = 562 nm) (Viollier et al., 2000) before and after chromatographic purification, ensuring only samples with yields ≥95% were used for isotopic analysis. Isotopic compositions were measured on a Neptune Thermo Scientific MC-ICP-MS (Multiple Collector-Inductively Coupled Plasma-Mass Spectrometer) at Rutgers University, applying the method of Arnold et al. (2004). Samples were introduced as a 1 ppm solution using a cyclonic spray chamber. Mass bias during the analysis was corrected using a Cu elemental spike, and standard reference material IRMM-014 served as a bracketing standard between each sample for accurate mass bias correction. The blank procedure involved spiking the blanks post-column separation to prevent overestimation of the procedure blank from incomplete spike recovery, ensuring accurate accounting of any blank contributions and preventing inaccuracies in the final results. The iron isotope composition (δ⁵⁶Fe) is defined as follows:

where the δ⁵⁶Fe is reported relative to IRMM-014 reference material. The measured Fe isotope composition of IRMM-014 is δ ⁵⁶Fe_T =-0.09‰ on this scale, with an internal precision of ±0.06‰ (2σ).

Sedimentary total carbon (TC) was analyzed by combustion using an Eltra CS-500 carbon-sulfur analyzer. Total inorganic carbon (TIC) was determined by acidification of a split of the sample. Total organic carbon (TOC) content was calculated as the difference between TC and TIC. The Eltra limestone geostandards AR4007 and AR4011 were analyzed routinely, with values falling within reported ranges and deviating less than<5%. Geo-reference standards AR4007 (carbon = 7.58%) and AR4011 (carbon = 8.91%) were used for analytical calibration and quality control. Calcium carbonate concentrations (CaCO₃), as weight percent (wt. %), were calculated from the measured TIC content assuming that all evolved CO₂ was derived from the dissolution of CaCO₃:

Standard CaCO₃ (>99.9% calcium carbonate, Fisher Scientific) was used during individual batches of analyses to confirm accuracy and instrument performance before, during, and after each run (with reproducibility better than 3%). No correction was made for the presence of other carbonate minerals.

The analysis of variance (ANOVA), the F-test, and t-tests were conducted to determine whether there is a significant difference of bioavailable Fe distribution between proximal and distal sites and to assess any dependence of Fe_HR/Fe_T with grain size (Montgomery et al., 2009; Haynes, 2013). The 95% confidence interval was used in all analyses (see Supplementary Information for grouping rationale and sensitivity analyses).

Aerosol optical depth (AOD) data are from the MIRS—Multi-angle Imaging SpectroRadiometer (https://misr.jpl.nasa.gov/getData/accessData/). These data provide a benchmark for calibrating models and interpreting sediment records that span long-term geological timescales. The averaged concentration of AOD between 2009-2019 over the North Atlantic Ocean was utilized to provide an estimate of atmospheric African dust deposition to the North Atlantic Ocean. Modern AOD data help bridge temporal scales, offering historical perspectives on changes in dust patterns and aerosol loading, while sediment cores offer long-term insights into dust deposition and iron (Fe) content over glacial-interglacial cycles.

To model the relationship between Fe flux and primary productivity dynamics, we used the ecoGEnIE paleoclimate model, an extension of cGEnIE—a carbon-centric, Grid Enabled Integrated Earth system model featuring comprehensive marine biogeochemical uptake (Ward et al., 2018). ecoGEnIE incorporates a scheme for plankton ecology (ECOGEM) with a size-dependent control on the plankton biogeochemical function (Ward et al., 2018). This addition allows for a better representation of biodiversity, including ecosystem shifts in response to environmental forcing. ecoGEnIE provides dynamic simulations of nutrient usage in response to availability. For our purposes, we used cGEnIE/ecoGEnIE default configurations (Ridgwell and Hargreaves, 2007).

Iron isotope data (δ⁵⁶Fe) can be a powerful way to constrain the Fe sources to the oceans over Earth history (Beard et al., 2003; Waeles et al., 2007; Owens et al., 2012; Conway et al., 2019). The average bulk δ ⁵⁶Fe values for samples from all four sites (-0.05 ± 0.02‰; 2-standard deviations) is consistent with continentally derived dust ( Figure 1 , Supplementary Table S1 ) given the similarity to the δ ⁵⁶Fe composition of most silicate rocks (Beard et al., 2003; Dauphas and Rouxel, 2006; Waeles et al., 2007; Conway and John, 2014). The Fe_T/Al ratio of 0.55 ± 0.02 also overlaps with the mean Fe_T/Al ratio for terrigenous sediments and average continental crust (0.55 ± 0.11; Martinez et al., 2007), suggesting no significant hydrothermal contribution, which typically has a δ⁵⁶Fe similar to crustal values but has a Fe_T/Al >2.00 (Clarkson et al., 2014; Raiswell et al., 2018a). Our average Al/Ti ratio of 18.9 ± 0.78 is also consistent with an aeolian origin for our samples (Yarincik et al., 2000).

The potential Fe bioavailability can be explored by speciation studies, which are broadly linked to Fe phases in the sample (Shi et al., 2009). The concentrations of Fe extracted using sodium acetate (Fe_Na-Ac) comprised 0.5-2.9 wt.% of Fe_T ( Supplementary Tables S2a-d, S3a-d ). Although Poulton and Canfield (2005) used Fe_Na-Ac to remove carbonate-bound Fe, their data show that minor amounts are also removed from iron-bearing silicates. We confirmed this possibility by extracting a range of Fe-bearing silicates with sodium acetate ( Supplementary Table S4 ), suggesting that acetate removes a minor fraction of Fe from such minerals—potentially from weakly bound surface sites (Heron et al., 1994; Raiswell et al., 2018b). This silicate-associated pool has been documented to be readily bioavailable (Shoenfelt et al., 2017) because Fe(II)-rich silicate minerals can enhance diatom growth as well as photosynthetic efficiency. However, this silicate-bound Fe fraction is very small relative to the other potentially bioavailable pools we now discuss. A dithionite Fe extraction (Fe_Dith), dominantly representing Fe (oxyhydr)oxide minerals, removed 8.6 to 21.6 wt.% of Fe_T ( Supplementary Tables S2a-d, S3a-d ). Previous studies have considered Fe_Dith to be an effective measure of potentially bioavailable iron from aeolian dust particles (Fan et al., 2006; Baker and Croot, 2010). An oxalate Fe extraction (Fe_Ox) mainly targets Fe in magnetite and ranges from 5.4-12.1 wt.% of Fe_T ( Supplementary Tables S2a-d, S3a-d ). If delivered unaltered to the ocean, rather than forming during diagenesis, magnetite would likely represent refractory (insoluble) iron in surface seawaters.

On the broadest scales, an overall increase in atmospheric dust deposition occurs during glacial intervals due to an expansion in the source areas (Mahowald et al., 1999) and stronger winds (McGee et al., 2010) that consequently influence the supply of bioavailable Fe to the open ocean. However, at low latitudes, there is no significant glacial-interglacial trend in dust input (Maher et al., 2010). Iron bioavailability in marine systems is linked to complexation with prokaryotic-released organic compounds, such as siderophores, polycarboxylate ligands (Barbeau, 2006; Shaked and Lis, 2012), and saccharides (Hassler et al., 2011). Thus, a fraction of this delivered Fe pool should be bioavailable. Our analyses of Fe_T, Fe_HR/Fe_T, and total organic carbon (TOC) at each site have limited to no statistically relevant stratigraphic variation over the last glacial period to the present (p>0.05) ( Figure 2 , Supplementary Tables S10a-d ). Further, the TOC contents argue against productivity that is enhanced in one interval relative to the other (glacial versus interglacial). This combination of data suggests that there is little or no temporal variation at low latitudes on glacial-interglacial timescales. Importantly, however, past studies have suggested that downwind locations may be impacted biologically by the enhanced solubility of delivered Fe, such as the Bahamas and Amazon region (Shinn et al., 2000; Muhs et al., 2007; Bristow et al., 2010; Muhs et al., 2012; Prospero and Mayol-Bracero, 2013; Swart et al., 2014; Yu et al., 2015). Consistent with this possibility, our data document a transport-dependent process that would also be relevant to other regions, such as the south Atlantic, and time intervals that are characterized by Fe limitation. Moreover, our data suggest atmospheric pathways to enhanced reactivity that are relatively constant, at least at low latitude, under glacial versus interglacial global climatic regimes (see discussions below). The glacial-interglacial uniformity we observe implies consistency in sourcing and processing during transport despite temporal differences in weathering relationships in the source regions and wind patterns, among other varying controls (McGee et al., 2013). Additional work is necessary to demonstrate whether the observed relationship is the same for all source regions and latitudes. Importantly, however, this surprising result is likely to influence future climate models for this region and potentially beyond.

The bioavailable Fe supply in sediments as a function of grain size distribution (GSD) during glacial-interglacial periods could have important impacts on marine primary productivity (Mahowald et al., 2014). Therefore, we determined the GSD at all four sites by dry sieving ( Supplementary Figure S1 , Supplementary Table S5 ). A statistical one sample t-Test was carried out for all four sites to test the variability of GSD during glacial and interglacial periods, (n=128; Figure 2 and Supplementary Table S9 ). The p-values for proximal sites 658 and 659 are 0.88 and 0.76, respectively, and 0.72 and 0.19 for distal sites 1062 and 1063, respectively. The end member p-values suggest that variability in GSD over glacial-interglacial timescale is not statistically significant for any of the four sites, consistent with previous findings that the flux of low-latitude Saharan dust does not vary significantly over these timescales (Maher et al., 2010; Skonieczny et al., 2019).

Previous work has suggested that Fe bioavailability is grain size-dependent, primarily due to the greater surface-area-to-volume relationship of small grains of atmospheric dust (Mahowald et al., 2018). Greater relative surface areas for small-sized particles could support proportionally larger surface alteration during transport and associated coatings of soluble Fe. To test for statistically significant differences of bioavailable Fe distribution for GSD in our drill-core sediment samples, Fe_HR/Fe_T (n=394) were compared using an analysis of variance (ANOVA) and a t-test (see Supplementary Information for grain size rationale and sensitivity analyses). We found no statistically significant differences in Fe_HR/Fe_T with grain size at a given location ( Supplementary Tables S8a-d ). For example, the ANOVA for site 658 shows no significant dependence (p<0.05), with an f-critical value of 2.80e^-4 and a p-value of 1.0 among the various grain sizes. The ANOVA data for site 659 have an f-critical value of 1.13 and a p-value of 0.33, which is also not significant at p<0.05. Sites 1062 and 1063 also show no significant dependence at p<0.05, with f-critical values of 1.12 and 0.99 and p-values of 0.33 and 0.37 for sites 1062 and 1063, respectively. The negligible variability in potentially bioavailable Fe distribution as a function of GSD in these sediments from each site is likely due, at least in part, to the particles having experienced many series of aggregation and disaggregation as they settled through the water column (Bacon et al., 1985)—thus the larger particles in the sediments are mostly aggregates of finer original materials (Anderson et al., 2016) but initially were larger but through disaggregation and then aggregation have been modified. Consistent with the possibility, our the Fe_HR/Fe_T ratios of larger grains are similar to those of smaller particles i.e. showing no significant locality relation between GSD and Fe geochemistry ( Supplementary Figure S3 ). Importantly, the results from these cores do not suggest there is a statistical difference in grain size variations related to in Fe reactivity among grains of differing primary size.

North African dust is carried great distances over thousands of kilometers, as would be true for any ocean basin. Because atmospheric transport is a size-selective process (Pye, 1989), proximal and distal Fe dust can be distinguished by the grain-size distributions related to distance from the source, whereby dust particle size decreases with increasing distance (Mahowald et al., 2014). Average particle size distributions for our samples show a decrease of ~20% in grain size ( Supplementary Table S5 ) and a corresponding increase of ~30% in surface area ( Supplementary Table S6 ) from proximal to distal sites. Small particles have a longer lifetime in the atmosphere and thus experience enhanced atmospheric processing (chemical and/or photochemical), likely leading to an increase in the solubility (Fe_Sol) of Fe_HR (Spokes et al., 1994; Desboeufs et al., 2001; Shi et al., 2009). This suggests these grains are disaggregated primary grains where the finer distal grains (as suggested by grain size analyses) have traveled farther - increasing processing time. Thus, these finer distal grains have high surface are-to-volume ratios, which also favors the extent of processing.

Our measurements show a systematic decrease in Fe_HR/Fe_T from proximal to distal sites ( Figure 3A ). Importantly, suggestions of lower total Fe concentrations at proximal sites ( Figure 3B ) mostly reflect increased carbonate dilution at those locations. Detailed insight into Fe behavior is provided by our Fe_Dith data, which show the greatest decrease, from 21.5 wt.% and 15.5 wt.% of Fe_T at proximal sites 658 and 659, respectively, to 9.0 wt.% at both distal sites (1062 and 1063; Supplementary Tables S2a-d and S3a-d ). We suggest that our Fe_HR/Fe_T ratios decrease with increasing transport due to enhanced Fe solubility via atmospheric reactions (Oakes et al., 2012), leading to dissolution in seawater and likely uptake of the bioavailable Fe by primary producers in the surface ocean at these distal sites. Bioavailable Fe delivery and utilization from primary producers has been observed in this general region of the Atlantic previously (Borchardt et al., 2019) and has been suggested for transport of Fe to oligotrophic lakes in Spain (Bhattachan et al., 2016). Modern dust transport from the Saharan desert to the Atlantic shows seasonal differences. In the boreal summer, the trade winds typically carry dust from northern Africa to the Caribbean, while in winter, transport shifts south toward the Amazon Basin (Bakker et al., 2019). This relationship indicates that the central and western Sahara are significant dust sources in the productive summer, while the Sahel region contributes more dust to the Caribbean during the winter. ( Figure 4A ). Furthermore, the aerosol optical depth (AOD) decreases progressively from the Sahara Desert to the open ocean. Continued research on seasonal dust transport variation is crucial for a comprehensive understanding of these dynamics and the seasonal impacts, particularly since these may evolve due to climate change. The ecoGEnIe model suggests significant increases in Fe flux ( Figure 4B ) in these distal locations, underscoring the significant role of dust deposition. This scenario, based on limited data, is thought provoking and highlights an area of research that should be explored. Evaluating our experiments, we confirm that lower Fe concentrations may indeed reflect higher carbonate dilution in these specific locations. However, full quantification of the loss of soluble Fe and the role of transport in Fe bioavailability will require more information about the North African dust source region (Huneeus et al., 2011; McGee et al., 2013), post-depositional processing of the Fe (Bressac and Guieu, 2013; Meskhidze et al., 2019), biogeochemical conditions of the surface seawater (Boyd and Ellwood, 2010), and atmospheric processes (Baker and Croot, 2010).

Further, it is important to consider alternative interpretations of our data. For example, the Fe content of analyzed dust samples could decrease downwind of North Africa, as described by Zhang et al. (2015), due to selective, progressive loss of heavier hematite-rich grains through gravitational settling during transport in the atmosphere. In truth, many of our observations are consistent with this possibility, and it may play a role, but there are other observations that are less consistent. First, it is not clear that the trend observed in Fe deposition in Zhang et al. (2015) is an expression of hematite availability. Other Fe phases, in particular Fe (oxyhydr)oxides, are likely a substantial part of the Fe pool in the dust—both as original constitutes from the source region and as products of atmospheric reactions. Importantly, these phases are approximately half as dense as crystalline hematite and would be decidedly less vulnerable to differential settling effects during transport. Moreover, soil hematite, the likely source of dust, would also be less dense than well-crystallized hematite, and these Fe phases are likely to be only a small part of the total grain mass (discrete Fe oxide grains are rare; Poulton and Raiswell, 2005). Further, hematite and other Fe(III) phases are often present as coatings on aluminosilicate grains that tend, if anything, to have lower densities than other silicates.

Our detailed speciation provides additional insight. Specifically, we see the same distal trends in our data from the acetate extraction, which does not target hematite but instead extracts more reactive, less dense Fe (oxyhydr)oxide phases. Finally, we also do not observe a difference in Fe chemistry as a function of grain size, in contrast to expected transport-related physical sorting that is controlled by grain size and/or density. These observations do not preclude other important processes, but they do leave us with our interpretation as the most parsimonious explanation of the full range of observations.