New Database to explore the Findings from large-scale Ocean Iron enrichment experiments

Some of the largest scientific manipulation experiments conducted 
on our planet have enriched broad swaths of the surface ocean with iron. Surface 
ocean signatures of these iron enrichment experiments have covered areas up to 
> 1,000 km2 and have been conspicuous from space. Twelve of these multidisciplinary 
studies have been conducted since the early 1990s in three specific ocean regions— 
the Southern Ocean, and equatorial and sub-Arctic areas of the Pacific Ocean— 
where plant nutrients are perennially high (termed high nutrient low chlorophyll, 
or HNLC). In addition, a combined phosphorus and iron enrichment experiment 
was conducted in the oligotrophic North Atlantic Ocean. Together, these studies 
represent a unique set of physical, chemical, optical, biological, and ecological data. 
The richness of these data sets is captured in an open-access relational database at 
the Biological and Chemical Oceanography Data Management Office (BCO_DMO; 
http://osprey.bco-dmo.org/program.cfm?flag=viewp&id=10&sortby=program). It 
is a product of Working Group 131 (The Legacy of in situ Iron Enrichment: Data 
Compilation and Modeling; http://www.scor-int.org/Working_Groups/wg131.htm) 
of the Scientific Committee on Oceanic Research. The purpose of this article is to 
make the wider community aware of this resource. It also presents the merits and 
provides examples of the utility of this database for exploring emerging topics in 
oceanography, such as the links between ecosystem processes and biogeochemical 
cycles; the feasibility and many side effects of oceanic geoengineering; and how 
understanding the coupling among physical, chemical, and biological processes at the 
mesoscale can inform the emerging field of submesoscale biogeochemistry.

the interglacials) in atmospheric carbon dioxide concentrations across several ice ages (Sigman and Boyle, 2000;Watson et al., 2000).However, to test the Iron Hypothesis, it was clear that the community would have to design a better experimental approach.
The design of an in situ iron enrichment experiment (mesoscale, i.e., roughly 10 km length scale) required the addition of iron salts dissolved in acidified seawater (not iron filings, as is often reported in the press) and the simultaneous release of a conservative tracer to track the iron-enriched waters (Watson et al., 1991).The tracer chosen was SF 6 (sulfur hexafluoride), which had been used successfully to investigate the physical dynamics of the upper ocean (Ledwell et al., 1993).
In 1993, the first in situ mesoscale iron enrichment-IronEx I-took place in the HNLC waters of the Equatorial Pacific (Martin et al., 1994; see Table 1 for definitions of experiment acronyms).
IronEx I was inconclusive because the experiment's iron-enriched and SF 6labeled 50 km 2 "patch" of water was subducted under a less-dense water body, prematurely ending the study after only a few days (Martin et al., 1994).
Nevertheless, it revealed the power of the in situ iron enrichment approach and hinted that Martin's hypothesis (1990) was probably correct.IronEx II, conducted in the same region, confirmed

INtRODuCtION
For the last 80 years, oceanographers have been puzzled about why certain oceanic regions are characterized by low stocks of phytoplankton despite a perennial excess of nutrients such as nitrate (De Baar, 1994).For example, Hart (1934) observed this paradox in the Southern Ocean, and it is now commonly referred to as the HNLC (high nutrient low chlorophyll) condition (Chisholm and Morel, 1991).There has been considerable progress since the late 1980s in revisiting the Southern Ocean paradox.Martin and Fitzwater (1988) revealed that adding iron to the resident phytoplankton in the HNLC waters of the sub-Arctic Northeast Pacific resulted in a striking increase in phytoplankton stocks to levels equivalent to a bloom, and a concomitant decrease in nutrient concentrations.However, others, such as Banse (1991), offered a "nonferrous" alternative explanation for such trends, questioning whether, given the smallvolume (i.e., less than 25 L) incubation vessels used in these experiments, artifactual exclusion of zooplankton grazers, rather than iron enrichment, was shifting the balance toward boosting phytoplankton growth and hence stocks.
John Martin used shipboard experimental findings, including those described in Martin and Fitzwater (1988), to develop the Iron Hypothesis (Martin, 1990), which drew widespread attention to the potential links between changes in iron supply to the ocean in the geological past and consequent modulation of biological productivity.Changes to marine productivity led to alteration of the ocean's carbon cycle, which contributed as much as 30% to roughly 80 µatm shifts (i.e., from 280 to 200 µatm toward these hints, demonstrating that supplying iron can result in a diatom bloom and that iron plays a fundamental role in many ecological and biogeochemical functions (Coale et al., 1996).
The success of IronEx II paved the way for considering a mesoscale enrichment experiment in the challenging environment of the Southern Ocean (Frost, 1996), which has the largest inventory of unused nutrients in surface waters of any HNLC region (Figure 1).Experiments, commencing with SOIREE (Boyd et al., 2000) and followed by EisenEx (Gervais et al., 2002), SOFeX (Coale et al., 2004), EIFEX (Smetacek et al., 2012), and most recently, LOHAFEX (Nature Geoscience editorial, 2009), confirmed that, as in the HNLC waters of the Equatorial Pacific, increased iron supply dramatically alters phytoplankton dynamics, food web structure, and biogeochemical cycles in the Southern Ocean.These major conclusions were also confirmed by the mesoscale experiments SEEDS I (Tsuda et al., 2003), SEEDS II (Tsuda et al., 2007), and SERIES (Boyd et al., 2004) carried out in the third major HNLC region, the sub-Arctic Pacific.
In other regions of the ocean, iron and nutrient colimitation has been studied in a joint mesoscale phosphate and iron enrichment experiment, FeeP, in the oligotrophic North Atlantic Ocean (Dixon, 2008).Boyd et al. (2007) compiled the major Philip W. Boyd (pboyd@chemistry.otago.ac.nz)This joint Indo-German experiment was ordered not to sail from Capetown by the German government, as it was deemed in contravention of a de facto moratorium on such large-scale manipulation experiments (Schiermeier, 2009).However,

Validation of Ocean Biogeochemistry Model simulations
A wide range of modeling experiments have been conducted in the last 15 years.
They include studies to better understand the regional biogeochemistry of HNLC regions, such as the Equatorial Pacific (Chai et al., 2007); global simulations that included mesoscale iron-enriched "patch" experiments (Gnanadesikan et al., 2003;Sarmiento et al., 2010); and site-specific, more-detailed models to investigate iron-mediated processes, such as carbon flow through ecosystems (Denman et al., 2006), carbon export and physical mixing (Boyd et al., 2002), and modes of controls on dimethyl sulfide (DMS) dynamics (Le Clainche et al., 2006).algal growth rates (Boyd, 2002), silicate concentrations (Coale et al., 2004), and resident phytoplankton (Silver et al., 2010).The availability of model simulations to test the skill of models across these disparate regions is essential if we are to be able to derive a set of underlying principles with which to demarcate global and regional drivers of the functioning of HNLC regions.
understanding the Interplay of Biogeochemical Cycles and ecosystem Food Webs The phytoflagellates increase in abundance until there is a shift in the microzooplankton community from small heterotrophic flagellates to larger ciliates.
The ciliates graze on the phytoflagellates, resulting in a release of DMSP that is converted to DMS (Turner et al., 2004).(Legendre and Rivkin, 2005;Boyd et al., 2010).

understanding the Issues and Implications of geoengineering
In the last five years, a growing literature has explored a range of potential geoengineering methods (Markels and Barber, 2001;Lovelock and Rapley, 2007;Boyd, 2008;Shepherd et al., 2009;White et al., 2010;Russell et al., 2012).
However, data from pilot studies or trials are scarce (Figure 1), and it is thus difficult to make progress on this important debate (Russell et al., 2012).There have been three ocean trials (Figure 1), only one of which has been published in the peer-reviewed literature.In situ mesoscale iron enrichments, although carried out for very different reasons than that of geoengineering (Strong et al., 2009a,b), nevertheless represent an important asset.These mesoscale, multidisciplinary experiments can provide insights into key outcomes of oceanic perturbations, such as their efficacy, side effects, safety, unknowns, and unanticipated results, that can be used to inform the ongoing debate about geoengineering methods (Boyd, 2008).Such data sets also highlight the limitations of even these largescale carefully monitored perturbations of the planet, as they could not provide sufficient data to fully inform the geoengineering debate (Watson et al., 2008).
a Primer for submesoscale Biogeochemistry studies A growing interest in the study of biogeochemistry at relatively small (0-10 km) scales (Lévy et al., 2012) has followed the significant findings about how ocean physics varies at the submesoscale (see references in Lévy et al., 2012).
However, to date, many studies have had limited success in linking submesoscale physics to chemistry, then to biology, and eventually to biogeochemical processes, such as export downward flux (Guidi et al., 2009;Resplandy et al., 2012).Issues   2 for examples of shifts in flora and fauna.Bloom termination is defined as the period during which a sustained decrease in bloom stocks is observed.export is defined as enhancement of the downward flux of particles in the upper 300 m.The temporal trends demonstrate that different locales have characteristic response times (red denotes tropical hNlC [high nutrient low chlorophyll] waters; green, subpolar hNlC waters; blue, polar hNlC waters).Data are available from the BCO-DMO relational database.
R e g u l a R I s s u e F e at u R e a New Database to explore the Findings from large-scale Ocean Iron enrichment experiments aBstR aCt.Some of the largest scientific manipulation experiments conducted on our planet have enriched broad swaths of the surface ocean with iron.Surface ocean signatures of these iron enrichment experiments have covered areas up to > 1,000 km 2 and have been conspicuous from space.Twelve of these multidisciplinary studies have been conducted since the early 1990s in three specific ocean regionsthe Southern Ocean, and equatorial and sub-Arctic areas of the Pacific Oceanwhere plant nutrients are perennially high (termed high nutrient low chlorophyll, or HNLC).In addition, a combined phosphorus and iron enrichment experiment was conducted in the oligotrophic North Atlantic Ocean.Together, these studies represent a unique set of physical, chemical, optical, biological, and ecological data.The richness of these data sets is captured in an open-access relational database at the Biological and Chemical Oceanography Data Management Office (BCO_DMO; http://osprey.bco-dmo.org/program.cfm?flag=viewp&id=10&sortby=program).It is a product of Working Group 131 (The Legacy of in situ Iron Enrichment: Data Compilation and Modeling; http://www.scor-int.org/Working_Groups/wg131.htm) of the Scientific Committee on Oceanic Research.The purpose of this article is to make the wider community aware of this resource.It also presents the merits and provides examples of the utility of this database for exploring emerging topics in oceanography, such as the links between ecosystem processes and biogeochemical cycles; the feasibility and many side effects of oceanic geoengineering; and how understanding the coupling among physical, chemical, and biological processes at the mesoscale can inform the emerging field of submesoscale biogeochemistry.B y P h I l I P W. B O y D , D O R O t h e e C .e .B a k k e R , a N D C y N t h I a C h a N D l e R Oceanography | Vol. 25, No. 4

table 1 .
The iron enrichment experiments in the sCOR Wg 131 data compilation effort are listed along with the FeeP and lOhaFeX projects.References to the experiments are in the text.= Biological and Chemical Oceanography Data Management Office BODC = British Oceanographic Data Centre eIFeX = european Iron Fertilization experiment eisenex = european Iron enrichment experiment FeeP = Phosphate and Iron addition experiment Ironex I = Iron experiment I Ironex II = Iron experiment II lOhaFeX = loha Iron eXperiment (where loha is hindi for iron) PaNgaea = Data Publisher for earth & environmental science sage = surface-Ocean lower-atmosphere studies air-sea gas exchange (experiment) seeDs I = sub-arctic-Pacific Iron experiment for ecosystem Dynamics study I seeDs II = sub-arctic Pacific Iron experiment for ecosystem Dynamics study II seRIes = sub-arctic ecosystem Response to Iron enrichment study sOFeX-N = southern Ocean Iron experiment -North sOFeX-s = southern Ocean Iron experiment -south sOIRee = southern Ocean Iron Release experiment findings of the many studies conducted in the three HNLC regions.This synthesis revealed that these data sets were a valuable and diverse resource for exploring the links between perturbation of iron supply and the consequent effects on the oceanic carbon cycle, for example, the ratio of photosynthetic carbon fixation per unit of iron added (Boyd et al., 2007).Moreover, such intentional iron enrichment of HNLC waters had additional, and, in some cases, unexpected influences on other biogeochemical cycles, including sulfur, silicon, and nitrogen (e.g., low iron supply increases the silicate demand by diatoms; Hutchins and Bruland, 1998).These mesoscale experiments were conducted over a more than 15-year period and were each characterized by detailed multidisciplinary studies (see Table 1 and S-Tables 1-3 in Boyd et al., 2007).Thus, each study provided data sets ranging from high-resolution underway sampling (e.g., photosynthetic competence [F v /F m ]; Behrenfeld et al., 1996) to specialized shipboard experiments, such as the interplay of iron and light on phytoplankton dynamics (Boyd et al., 2000).Taken together, although they were scattered across more than 100 publications, these in situ mesoscale experiments had the potential to supply a wealth of valuable data sets.Furthermore, many of the data sets were unpublished.In 2007 at their Bergen meeting, members of the Scientific Committee on Oceanic Research (SCOR) ratified support for a working group to rescue these data sets and to create a relational database that would be publicly available.This would become WG 131, The Legacy of in situ Iron Enrichment: Data Compilation and Modeling (http://www.scor-int.org/Working_Groups/wg131.htm).the Rel atIONal DataBase WG 131 members, in partnership with staff at the Biological and Chemical Oceanography Data Management Office (BCO-DMO) at the Woods Hole Oceanographic Institution, compiled data sets from in situ iron enrichment experiments.The resulting database includes detailed information, data, and supporting documentation from nine mesoscale experiments, which represent some of the most challenging and informative experiments undertaken in ocean science.The in situ iron enrichment projects database is part of a larger collection of ocean biogeochemistry data curated by BCO-DMO.The data from nine projects (IronEx I and II, SOIREE, SOFeX North and South, SERIES, SEEDS I and II, and SAGE [Harvey et al., 2011]) are available directly from the BCO-DMO catalog.Data from the European-led mesoscale experiments aboard R/V Polarstern (EisenEx, EIFEX, and soon to be added LOHAFEX) can be accessed via links to the PANGAEA information system (http://www.pangaea.de),hosted by the Alfred Wegener Institute for Polar and Marine Research in Bremerhaven and the Center for Marine Environmental Sciences at the University of Bremen, both in Germany.Finally, data from the FeeP study will be archived at the British Oceanographic Data Centre (http://www.

Figure 1 .
Figure1.locations of in situ mesoscale iron enrichment studies (white symbols), iron and/or phosphorous enrichments (green symbols), bloom studies in naturally high iron waters (red symbols), and oceanic geoengineering trials or pilot studies (yellow diamonds), including iron fertilization(Markels and Barber, 2001) and nutrient upwelling using ocean pipes(lovelock and Rapley, 2007;White et al., 2010).all are overlaid on a map of surface nitrate concentrations for the world ocean.Nitrate concentrations courtesy of the National Oceanographic Data Centre However, few studies have conducted model simulations for each of the three distinct HNLC regions-subpolar, polar, and tropical.With the availability in the relational database of detailed HNLC regional initial conditions-chemical, physical, optical, and biological-prior to enrichment with iron, modelers may be encouraged to undertake HNLC iron enrichment projects, as these data are valuable for both model initialization and later for validation (using different data sets from the relational database).Such a suite of model simulations would provide rigorous validation for models, as the environmental conditions that characterize each region vary considerably among polar, subpolar, and tropical sites, for example, in water temperature and hence Figure 2) that detail how iron-mediated increases in the abundances of particular phytoplankton groups in HNLC waters result in shifts in the dominant Moreover, such a holistic view of the interplay of bottom-up environmental versus top-down ecological shifts in controlling biota provides valuable insight into the extent, and on which timescales, ecosystems may be altered by another, more complex, environmental perturbation-climate change.This broad view will help us better understand how climate change-mediated shifts in ecosystem structure and function will feed back biogeochemically to climate change

Figure 2 .
Figure 2. an example of the indirect effects of iron enrichment (on day zero) on food web structure and the consequent effects on sulfur biogeochemistry.a faunistic shift to large microzooplankton (heterotrophic ciliates, denoted by hCIl, upper panel) occurs in response to iron-mediated increases in phytoflagellate abundances (denoted by PFlag, upper panel).The increase in phytoflagellate abundances leads to a buildup of DMsPp (particulate dimethylsulfonioproprionate, lower panel) concentrations, and then to grazer-mediated release of the DMsPp and conversion to the climate-reactive gas DMs (dimethyl sulfide, lower panel).Data are from the sOIRee experiment, and are available from the BCO-DMO relational database.

F
Figure3. a summary of the timescales of biological responses (log time scale, 0 denotes iron release), from photosynthetic to biogeochemical, following purposeful mesoscale iron enrichment.F v /F m is phytoplankton photosynthetic competence; NPP denotes Net Primary Production.see Figure2for examples of shifts in flora and fauna.Bloom termination is defined as the period during which a sustained decrease in bloom stocks is observed.export is defined as enhancement of the downward flux of particles in the upper 300 m.The temporal trends demonstrate that different locales have characteristic response times (red denotes tropical hNlC [high nutrient low chlorophyll] waters; green, subpolar hNlC waters; blue, polar hNlC waters).Data are available from the BCO-DMO relational database.
is Professor of Ocean Biogeochemistry, NIWA Centre of Chemical and Physical Oceanography, Department of Chemistry, University of Otago, Dunedin, New Zealand.Dorothee C.E. Bakker is Research Officer, School of EnvironmentalSciences, University of East Anglia, Norwich Research Park, Norwich, UK.Cynthia Chandler is Information Systems Associate, Biological and Chemical Oceanography Data Management Office, Woods Hole Oceanographic Institution, Woods Hole, MA, USA.