The fractional solubility of aerosol-derived trace elements deposited to the ocean surface is a key parameter of many marine biogeochemical models. Despite this, it is currently poorly constrained, in part due to the complex interplay between the various processes that govern the solubilisation of aerosol trace elements. In this study, we used a sequential two-stage leach to investigate the regional variability in fractional solubility of a suite of aerosol trace elements (Al, Ti, Fe, Mn, Co, Ni, Cu, Zn, Cd, and Pb) from samples collected during three GEOTRACES cruises to the North Atlantic Ocean (GA01, GA03-2010, and GA03-2011). We present aerosol trace element solubility data from two sequential leaches that provide a “solubility window”, covering a conservative lower limit to an upper limit, the maximum potentially soluble fraction, and discuss why this upper limit of solubility could be used as a proxy for the bioavailable fraction in some regions.
Regardless of the leaching solution used in this study (mild versus strong
leach), the most heavily loaded samples generally had the lowest solubility.
However, there were exceptions. Manganese fractional solubility was
relatively uniform across the full range of atmospheric loading (32
Aerosol trace element (TE) solubility is a key parameter of many
biogeochemical models, but it is poorly constrained, e.g. Fe solubility
estimates range from 0.001 to 90 % (Aguilar-Islas et al., 2010; Baker et
al., 2016). The fractional solubility (herein referred to
as “solubility”) of aerosol TEs is defined in terms of the amount of a TE
in solution from any given leach that passes through a filter (usually
< 0.45 or 0.2
There have been a number of studies that have focused on the role of aerosol TEs on biogeochemical cycles in the North Atlantic (e.g. Sarthou et al., 2003; Baker et al., 2013; Buck et al., 2010; Ussher et al., 2013; Powell et al., 2015). More recently, the GEOTRACES programme has produced a number of aerosol datasets, which has stimulated further discussion on the use of these data to look for trends that link TE solubility and aerosol source (e.g. Baker et al., 2016; Jickells et al., 2016). Elemental ratios, enrichment factors, and air mass back trajectory (AMBT) simulations have long been used as a first approximation of aerosol source, and there are many studies that employ multivariate statistical analyses for aerosol source apportionment (e.g. Chueinta et al., 2000; Laing et al., 2015). In addition, more studies are making use of stable isotope ratios to investigate aerosol provenance. Some of these methods are well-established and have a relatively long history of use in this purpose, such as Pb isotopes (e.g. Maring et al., 1987) and Sr and Nd isotopes (e.g. Skonieczny et al., 2011; Scheuvens et al., 2013, and references therein), and data from investigations of novel isotope systems are increasing. For example, Fe isotopes show promise as a way to differentiate between anthropogenic and mineral dust aerosols (Conway et al., 2018). In contrast, Cd isotopes may not be a suitable tool for aerosol source apportionment (Bridgestock et al., 2017).
As the soluble fractions of aerosol TEs are thought to be the most readily
bioavailable forms (Shaked and Lis, 2009), the leachable (soluble) fraction
is used as a first approximation of the bioavailable fraction. Therefore,
experimental conditions should mimic natural conditions as closely as
possible, while yielding reproducible results. Ideally, the leach protocol
used fits both these criteria. However, that is not always strictly possible
for reasons such as access to the leach medium of choice, availability of
analytical instrumentation, and cost. Currently, however, there is no
standardised aerosol leaching protocol, but it is recognised that this
should be a priority for future studies (Baker et al., 2016). Some
commonly used leach media are ultra high-purity (UHP) water (18.2 M
To investigate the regional variation in the solubility of key TEs in the
North Atlantic, aerosol samples were collected during the US GEOTRACES GA03
campaigns in 2010 and 2011, and the French GEOTRACES GA01 campaign in 2014
(
In this study a two-stage leach protocol was followed. the first leach employed was the “instantaneous” leach described by Buck et al. (2006), which is a flow-through method in which the leach medium is in contact with the aerosols for 10–30 s. It can be conducted using UHP water or seawater. The advantages of using UHP water are that UHP water is a reproducible medium (allowing for inter-lab comparisons) that can easily be analysed by inductively coupled plasma mass spectrometry (ICP-MS) for many elements simultaneously without the need for time-consuming sample handling steps such as separation techniques and drying down then re-dissolving the residue. Leaches with UHP water can be conducted at sea or in the home laboratory. If fresh seawater is used the leaches must be undertaken at sea.
Given that UHP water and rainwater have broadly similar pH (
The second sequential leach was employed in order to estimate an upper limit of TE solubility and provide a “solubility window”, but it was also an estimate of the maximum bioavailable fraction during the residence time of aerosol particles in the euphotic zone. We used the 25 % acetic acid leach with hydroxylamine hydrochloride described by Berger et al. (2008). The pH of this leach (pH 2.1) is just below that of zooplankton or fish digestive tracts and the reducing agent mimics the low-oxygen environments inside faecal pellets and marine snow aggregates. Indeed, Schmidt et al. (2016) have demonstrated that lithogenic Fe is mobilised in the gut passage of krill, resulting in 3-fold higher Fe content in the muscle and 5-fold higher Fe content of the faecal pellets of specimens close to lithogenic source material compared to those from offshore.
Aerosol samples (
The GEOTRACES GA01 and GA03 cruise tracks (GA01, GA03-2010, and
GA03-2011). In total, 57 aerosol samples (GA01
To avoid contamination from the ship's stack exhaust, aerosol sampling was
controlled with respect to wind sector and wind speed using an anemometer
interfaced with a data logger (CR800, Campbell Scientific). The samplers were
programmed to run when the wind was
The total digestion method of Morton et al. (2013) was used for the
determination of total aerosol TE loadings (Al, Ti, Mn, Fe, Co, Ni, Cu, Zn,
Cd, Pb). The W41 filter discs were digested in tightly capped 15 mL
Teflon PFA vials (Savillex). Firstly, 1000
In this study, we used a two-step sequential leach to investigate regional
variation in aerosol sources, TE fractional solubility and bioavailability.
We discuss the results from (1) an “instantaneous” leach (Buck et al.,
2006), which provides a lower-limit estimate of the most labile TE fraction
(analogous to the initial rapid release of TEs into raindrops and the
surface mixed layer of the ocean), followed by (2) a more protracted leach
using 25 % acetic acid (with the reducing agent, hydroxylamine
hydrochloride, and heat, 10 min at 90
The first step, the “instantaneous” leach, was conducted under a Class-100
laminar flow hood. In this technique, 100 mL of UHP water (> 18 M
The UHP water fractional solubility was calculated using Eq. (1):
Following the instantaneous UHP water leach, the filter was transferred to a
15 mL centrifuge tube, and the second leach was undertaken, using 5 mL of 25 % (4.4 M) ultra pure acetic acid, with 0.02 M hydroxylamine hydrochloride
as the reducing agent (Berger et al., 2008). After a 10 min heating step (90
As all samples in this study were leached first using the UHP water
instantaneous leach, followed by a sequential leach with 25 % acetic
acid, the overall solubility in 25% acetic acid was calculated using Eq. (2):
Before the UHP water leachate was acidified, a 10 mL aliquot was taken from
each leach sample for the determination of the soluble major anions. The
aliquot was immediately frozen for storage. The anions, Cl
AMBT simulations were generated using the
publicly available NOAA Air Resources Laboratory Hybrid Single-Particle
Lagrangian Integrated Trajectory (HYSPLIT) model, using the GDAS meteorology
(Stein et al., 2015; Rolph, 2017). The 5-day AMBT simulations were used to
describe five regional categories, based on the predominant trajectories for
the air masses. The simulations and further details of these categories can
be found in Wozniak et al. (2013, 2014) and Shelley et al. (2015, 2017).
Briefly, for cruise GA03 air masses were characterised as European, North
American, North African, or Marine (no or minimal interaction with major
continental land masses within the 5-day simulation period). For cruise
GA01, all the samples were classified as High Latitude dust (originating
north of 50
AMBT simulations are frequently used to identify the origin and/or flow path of air masses, from which a first approximation of aerosol provenance (e.g. deserts, urban regions, or biomass burning) is made. Although they are useful tools in oceanographic studies, AMBTs used alone do have limitations. Perhaps the most significant of these is that they are unable to quantify the contribution of different aerosol types or the entrainment of particles along the flow path of the air mass. Indeed, within the five categories described in this study multiple sources are likely to have contributed to the composition of the bulk aerosol of each category. This study is likely to be particularly sensitive to this as the sampling site was not static (i.e. sampling occurred along three different transects), and air masses can, and do, take different pathways within a general wind direction. Consequently, AMBTs are not adequately discriminating for aerosol source apportionment. However, we have organised the data using the AMBT categories because the objective of this study was to look for trends in solubility at a regional level, as well as for consistency with our earlier published work from the North Atlantic (Wozniak et al., 2013, 2014; Shelley et al., 2015, 2017).
More powerful approaches for aerosol source apportionment consider the physicochemical composition of the aerosols, either as the bulk aerosol or as individual particles. There have been a number of field campaigns (e.g. DABEX, DODO, SAMUM, and AMMA) and individual studies which have provided a wealth of information about the physicochemical composition of African dust before, during, and after long-range transport (e.g. Johansen et al., 2000; Johnson et al., 2008; McConnell et al., 2008; Petzold et al., 2009; Marticorena et al., 2010; Trapp et al., 2010; Formenti et al., 2011). These studies and satellite data have identified the key dust source regions in North Africa (Prospero et al., 2002). Chemical composition data for other aerosol end members which supply aerosols to the North Atlantic are not as extensive, but some examples of individual studies and field campaigns can be found in Table S2. In addition, campaigns in the Atlantic Ocean which have sampled marine aerosols (e.g. Atlantic Meridional Transect, CLIVAR, GEOTRACES) have identified aerosol sources from characteristic groups of elements and elemental ratios (e.g. high concentrations of lithogenic elements are characteristic of a mineral dust; K is a tracer of biomass burning; and correlations between V and Ni are diagnostic of emissions from marine shipping; Baker et al., 2006a; Sippula et al., 2014; Baker and Jickells, 2017), organic compounds (e.g. Wozniak et al., 2013, 2014, 2015), and/or stable isotopic signatures (Scheuvens et al., 2013, and references therein).
Although atmospheric inputs to the ocean are episodic and exhibit a
seasonality in the tropical and subtropical North Atlantic that is largely
driven by the migration of the intertropical convergence zone (Prospero et
al., 1981; Adams et al., 2012; Doherty et al., 2014), North African/Saharan
mineral dust dominated the aerosol composition in the GA03 study region
(Conway and John, 2014; Shelley et al., 2015; Conway et al., 2018). Other
aerosol sources in Europe and North America and sea salt also contributed to
the bulk aerosol to varying extents. In contrast to GA03, the GA01 transect
was located north of the extent of the Saharan dust plume
(
Total aerosol Fe and Al (ng m
Total Fe and Al were strongly correlated (
As the aerosol source has a direct bearing on the type and composition of
aerosols, determining the source could provide useful data that might be
used to predict the fractional solubility of aerosol TEs. As positive matrix
factorisation (PMF) can be used for source apportionment, we used the US
Environmental Protection Agency's EPA PMF model (v. 5.0) with the total TE
concentration data to look for trends in the data. However, the GA01 and
GA03 dataset is relatively small (
Elemental mass ratios from the 10 most heavily loaded GA03 North African
aerosols were averaged to derive a value for the “North African” ratio
depicted by the dashed horizontal line in Fig. 3a–i. Aluminium was used to
normalise the data (Fig. 3; Table S2) and was chosen instead of Ti, another
proxy for mineral dust, due to the presence of some anomalously high
Ti
Elemental mass ratios (normalised to Al) of total (black circles)
and UHP water-soluble (white triangles) TEs. The UCC elemental ratio
(Rudnick and Gao, 2003) is indicated by the solid horizontal line, and the
elemental ratio in North African sourced aerosols (Shelley et al., 2015) is
indicated by the dashed horizontal line on each plot. The red vertical lines
separate the aerosol source regions, which are labelled in panel
For North African dust there does not appear to be a discernible source-dependent trend in Fe
Aerosols from the more northerly section, GA01, were largely outside the
influence of the Saharan dust plume (Shelley et al., 2017) and are all
classified as High Latitude in this study (Fig. 3). For this group of
samples, there were also sub-groups of Fe
A second group of GA01 samples (G7, G9, G11, and G12) had Fe
While there is evidence for anthropogenic source(s) of aerosol Fe to the
North Atlantic (Conway et al., 2018), which is more soluble than Fe
associated with mineral dust (Sedwick et al., 2007; Sholkovitz et al., 2009,
2012), North African mineral dust dominates the supply of Fe to much of the
study region (Baker et al., 2013; Shelley et al., 2015, 2017; Conway et al.,
2018). In addition to the samples classified as European and North
American, elevated Fe
The Marine and the High Latitude samples had the widest range in Fe
For the anthropogenically derived TEs, Ni, Cu, Zn, Cd, and Pb (Fig. 3e–i)
and for at least some of samples of the mixed-source TEs (i.e. with crustal
and pollution sources; e.g. Mn and Co in Fig. 3b and d), there is some degree
of source-dependence in the elemental ratios, with some significant increases
from the UCC mass ratios in the total (Shelley et al., 2015) and UHP
water-soluble fractions (Fig. 3). The higher ratios of the UHP water-soluble
fraction compared to the total indicates that these TEs are more labile than
Al. In addition, studies that have investigated the size distribution of
aerosols have found that anthropogenically derived TEs tend to be associated
with fine-mode aerosols (< 1
The UHP water-soluble fraction of aerosol Fe and Al determined for all the
North Atlantic GA01 and GA03 samples varied by 2 orders of magnitude (Fig. 4a:
Fe
The inverse relationship between total aerosol loading and fractional
solubility has previously been reported for Fe (Sholkovitz et al., 2009,
2012; Jickells et al., 2016) and Al (Jickells et al., 2016). Jickells et al. (2016)
compiled solubility data from the North Atlantic and found that the
general trend between Fe and Al solubility and atmospheric loading was
robust over the range of atmospheric loadings found in the North Atlantic,
regardless of the leach protocol employed. In this study, both the UHP
soluble and 25 % acetic acid soluble fractions of Fe and Al (Fig. 4a and b) were related to atmospheric loading, i.e. the highest loaded North
African samples had the lowest solubility. The possible exception to this
trend is the fraction of Al that dissolved from North African aerosols
following the 25 % acetic acid leach (Fig. 4b). However, it could simply
be that we are observing scatter in our data that is smoothed out in the
larger dataset (
Aluminium, Ti, and Fe show very similar behaviour in Fig. 4a (sharply
decreasing solubility as loading increases). Cobalt, Ni, Cu, Zn, and Pb
solubilities decrease less strongly as loading increases, whereas Mn and Cd
show no clear trend. For the acetic acid leaches (Fig. 4b), Ti follows the
same trend as the UHP water leach (Fig. 4a), while Al and Fe plateau at 8–10 % solubility. The other TEs (Mn, Co, Ni, Cu, Zn, Cd, and Pb) all show
almost no trend with loading. The absence of an inverse trend between
solubility and loading has previously been noted for Mn (Jickells et al.,
2016). For Co the inverse relationship between UHP water solubility and
loading was not observed when using the 25 % acetic acid leach, most
likely because Co may be associated with the Mn and Fe oxides that are
easily reduced using this leach. For Zn and Cd, although their average
fractional solubilities (37
All 10 TEs from the five different categories were less soluble in UHP
water than 25 % acetic acid (Fig. 5). This is not a surprising finding
given the lower pH of acetic acid compared with UHP water, acetate being a
bidentate ligand, the longer contact time of the aerosols with the leach
solution, the addition of the hydroxylamine reducing agent, and that the
fractional solubility of TEs in 25 % acetic acid was calculated using
Eq. (2) (which sums the UHP water and 25 % acetic acid leach
concentrations). In addition, there is some degree of source-dependent
variability in the relative proportions of each TE that is released by the
two leaches. In general, as with the leaches with UHP water, the North
African aerosols were distinctly less soluble in 25 % acetic acid compared
with aerosols from the other source regions (Fig. 5). Figure 5 highlights
the distinction between the lithogenic elements, Al, Fe, and Ti, which have
uniformly low solubility in UHP water (mostly < 20 %) and
extremely low solubility in North African aerosols (< 1 %), and
the anthropogenic, pollution-dominated elements, Ni, Cu, Zn, Cd, and Pb, which
have solubility up to 100 %. Manganese and Co have both lithogenic and
anthropogenic sources and so are classified as “mixed source” and have
intermediate solubilities. Like all the TEs reported here, Mn solubility in
UHP water was significantly less (
Seawater leaches were conducted on a subset of samples (GA03-2011) to
investigate the suitability of seawater as the leach medium in the
instantaneous leach (Fig. 6). During this study, Fe solubility in seawater
was lower than in UHP water (Fig. 6c). This phenomenon has previously been
observed in atmospheric aerosols from the North Atlantic Ocean (Buck et al.,
2010). For Fe, only a few samples of North American and Marine provenance
conformed to the relationship described by the equation proposed by Buck et al. (2010), with most of our data plotting above the regression line of the
Buck et al. (2010) study (Fig. 6c), indicating that our data were relatively
more soluble in UHP water compared to seawater than in this earlier study.
One possibility is that the higher aerosol Fe loadings we observed during
GA03-2011 (this study, maximum
Solubility of Al, Ti, Mn, Fe, Co, Ni, Cu, Zn, Cd, and Pb following a
UHP water leach (UHP water, black circles, calculated using Eq. 1) and a
sequential leach of 25 % acetic acid (HAc, grey squares, calculated using
Eq. 2). The red vertical dashed lines represent the different aerosol source
categories, as labelled in panel
Comparison of TE solubility following instantaneous leaches using
UHP water or locally collected, filtered seawater. The solid line is the 1
Hierarchical cluster analysis of
Manganese is the only TE that had a slope close to unity (0.98; Fig. 6b),
suggesting that solubility estimates were not impacted by the choice of
leach medium used. This is consistent with other studies that have found
that Mn solubility is less sensitive to the choice of leach media or to
aerosol provenance than other TEs (Baker et al., 2006b; Jickells et al.,
2016). Due to the large variability in the dataset, there was no
significant difference between Mn solubility in UHP water or seawater (32
Lead was the only TE with all slopes differing significantly from 1.0 and
the only TE where the solubility in seawater was higher than in UHP water
for virtually every sample (Fig. 6i). As for Pb, most of the Co data fall
below the 1
As the PMF analysis was only able to identify two significant factors
accounting for the total aerosol TE concentrations, another multivariate
approach was taken. Hierarchical cluster analysis (Ward's method, Euclidian
distance) was performed using the R statistical package (v. 3.3.0; R Core
Team, 2016) to look for trends in the data that might reveal the various
aerosol sources. Hierarchical cluster analysis was performed on (1) log-transformed total aerosol TE plus NO
Figure 7a shows two main branches to the dendrogram of the total TE
concentration data. One branch groups all the North African and European
samples and two North American samples (N2 and N4) together, and the other
branch groups all other samples together. Samples closest to each other are
the most similar to each other, and those joined in the same groups share
similar characteristics. Therefore, in this analysis, the North African
samples are grouped together, as are the High Latitude samples. All but
three North African samples form a distinct sub-group. The three remaining
North African samples (A8, A9, and A11) share more characteristics with the
European samples, lending support for mixing of aerosols from the two
regions. Counterintuitively, the two European samples with the lowest Fe
Although there are differences between Fig. 7a (total TEs) and b (25 % acetic acid fractional solubility; Eq. 2), the general trend of an inverse relationship between TE atmospheric loading and fractional solubility holds, as the North African samples with the highest concentrations and lowest fractional solubilities appear on the left in Fig. 7a and on the right in Fig. 7b. In terms of fractional solubility, the North African samples form a distinct cluster, but this cluster is made up of two sub-groups: one collected during GA03-2010 and one during GA03-2011. The samples collected from near Greenland and the Labrador Sea are also distinct from the other GA01 samples (again with the exception of G15) and also distinct from all other samples. The European samples, all other GA01 samples, and three North American samples form a loose cluster. The remaining North American samples and all the Marine samples form another loose cluster.
Plotting the data this way still does not allow us to identify the aerosol sources definitively, but it does allow us to visualise which samples have the most similar physicochemical characteristics and confirms the general trend of a relationship between aerosol loading and fractional solubility and, by extension, bioavailability, even though we have demonstrated that this relationship is not present for all TEs. This knowledge is then useful as a general rule of thumb in biogeochemical models, although clearly other factors also exert controls on aerosol TE solubility. For example, during their investigations of the GA03 aerosols, Wozniak et al. (2013, 2014, 2015) proposed a role for water-soluble organic carbon in controlling the solubility of Fe. Desboeufs et al. (2005) also found evidence for a link between total carbon and TE solubility in regions impacted by anthropogenic activity. Thus, the carbon content of aerosols is also implicated as a control on aerosol Fe solubility, but the relationship is frequently not linear.
The ability of models to replicate subtleties in aerosol TE solubility may
prove critical in forecasting ecosystem impacts and responses. Due to the
magnitude of North African dust inputs to the North Atlantic region (very
high dust inputs result in a high soluble TE aerosol flux despite relatively
low fractional solubility), this is a particular challenge and is compounded
by additional unknowns such as how aerosol acidity will be impacted by the
combined effects of increasing industrialisation and urbanisation, as well as changes
in the magnitude of future mineral dust supply and biomass burning
(Knippertz et al., 2015; Weber et al., 2016). In other words, it is
important to accurately constrain aerosol TE solubility with high-quality
data in order to improve the predictive capacity of models. Clearly the
choice of leach media and protocol impacts the measured fractional
solubility. This is shown in both Figs. 4 and 5 and has a number of
implications with regard to modelling the impact of atmospheric deposition
on marine biogeochemistry. For example, for elements with generally low
solubility, such as Fe, the difference between 1 and 2 % solubility
is an increase of 100 %, meaning that only half the amount of dust is
needed to yield the same amount of dissolved Fe. To complicate matters
further, recent research has demonstrated that some diazotrophs are able to
directly access particulate Fe (Rubin et al., 2011). The significance of
this is that
There are implications for modelling the impact of atmospheric deposition
for other TEs. Although the lack of source-dependent differences in Mn
solubility in these aerosols makes modelling Mn solubility simpler, there
was still a difference in the fractional solubility calculated from the two
leaches (UHP water: 32
Given that the different leaching approaches access different fractions of aerosol TEs that can dissolve from aerosols at different rates (e.g. TEs loosely bound to surfaces and TEs that are associated with less reactive phases) (e.g. Koçak et al., 2007; Mackey et al., 2015), we need to conduct experiments that elucidate the relationship between the soluble and bioavailable fractions. In the meantime, we suggest that the 25 % acetic acid leach might be better to estimate the bioavailable fraction given that Fe (and perhaps other TEs) associated with lithogenic particles are directly available to micro-organisms in productive regions and regions with high dust inputs (Rubin et al., 2011) and that aerosol particles can be processed by zooplankton (Schmidt et al., 2016).
Aerosol TE solubility is usually determined using operationally defined methods, while biogeochemical models require robust relationships between two or more parameters that can be used to predict TE solubility in order to constrain the bioavailable fraction of aerosol TEs. In this study, we used a two-stage leach (UHP water followed by 25 % acetic acid with hydroxylamine hydrochloride) to investigate the fractional solubility of a suite of trace elements (Al, Ti, Mn, Fe, Co, Ni, Cu, Zn, Cd, Pb) from aerosols collected in the North Atlantic during three GEOTRACES research cruises (GA03-2010, GA03-2011, and GA01). Five regions were identified based on AMBT simulations; (i) North Africa, (ii) Europe, (iii) North America, (iv) High Latitude, and (v) Marine. However, the AMBTs were not able to sufficiently discriminate aerosol sources within these regions. Of these five categories, the North African aerosols were the most homogeneous in terms of their fractional solubility and elemental ratios. In contrast, samples from the most remote locations, the Marine and High Latitude aerosols, had the most spread in their fractional solubility and elemental ratios. Elemental ratios were discussed rather than enrichment factors normalised to UCC composition since earlier work highlighted that the UCC ratios are not representative of the North African mineral dust end member, which dominates aerosol supply in much of the study area.
We observed an inverse relationship between the fractional solubility of Al, Ti, Fe, Ni, Cu, and Pb and aerosol loading for all leach media (UHP water, filtered seawater, and 25 % acetic acid with hydroxylamine hydrochloride). However, Mn, Zn, and Cd fractional solubility appears to be independent of atmospheric loading. For Co, the inverse relationship between UHP water solubility and loading was not observed when using the 25 % acetic acid leach, most likely because Co may be associated with the Mn and Fe oxides that are easily reduced using the 25 % acetic acid leach. Further work is required to assess exactly which fraction is accessed by the various leach protocols in order to understand links between the soluble and bioavailable fractions.
Data are available at BCO-DMO (GA03;
The authors declare that they have no conflict of interest.
This article is part of the special issue “GEOVIDE, an international GEOTRACES study along the OVIDE section in the North Atlantic and in the Labrador Sea (GA01)”. It is not associated with a conference.
Many thanks to the captains and crews of the R/V