Ocean Fertilization: a Scientific Summary for Policy Makers

Intergovernmental Oceanographic Commission

Ocean Fertilization A scientiﬁc summary for policy makers Experimental ocean fertilization using ferrous sulphate on UK-German FeeP study, 2004

OCEANFERTILIZATION: action to deliberately increase planktonic production in the open ocean. Fertilization might be carried out over a range of scales for a variety of purposes; it can be achieved by directly adding nutrients, or increasing nutrient supply from deep water, or potentially by other means.

This report was commissioned by the Intergovernmental Oceano- Design: Eric Loddé graphic Commission (IOC), which is part of UNESCO. It was pre- pared with the assistance of the Surface Ocean Lower Atmosphere Image credits Study (SOLAS), an international programme that focuses research Front cover: effort on air-sea interactions and processes, sponsored by the In- Matt Walkington, NIWA; Glynn Gorick; NASA/SeaWiFS ternational Geosphere-Biosphere Programme (IGBP), the Scien- tiﬁc Committee on Oceanic Research (SCOR), the World Climate Inside front cover: Research Programme (WCRP) and the International Commission Philip Nightingale, Plymouth Marine Laboratory on Atmospheric Chemistry and Global Pollution (ICACGP). The drafting of this report beneﬁtted from advice by the secretariat of p 2 (Box 1): World Ocean Atlas (2005); US National the International Maritime Organization (IMO); discussions by the Oceanographic Data Center 2009 Intersessional Technical Working Group on Ocean Fertiliza- tion of the London Convention/London Protocol (LC/LP), in which p 4 (Fig 1) & p 5 (Box 2): Jack Cook, Woods Hole IOC participated; and IOC Member States’ comments, including Oceanographic Institution; also Oceanus rd those made at the 43 session of the IOC Executive Council. p 6 (Fig 2): Philip Boyd and NASA

Authors: Doug Wallace (IFM-GEOMAR, Germany), Cliff Law p 9 (Box 3): Philip Boyd, in Encyclopedia of Sustainablility (NIWA, New Zealand), Philip Boyd (University of Otago, New Science & Technology Zealand), Yves Collos (CNRS Université Montpellier, France), p 10 (Box 4) World Ocean Atlas (2009) Peter Croot (Plymouth Marine Laboratory, UK), Ken Denman (Fisheries and Oceans Canada), Phoebe Lam (WHOI, USA), Ulf p 11 (Fig 3): Richard Lampitt Riebesell (IFM-GEOMAR, Germany), Shigenobu Takeda (Nagasaki p 12 (Fig 4): Ken Denman University, Japan) and Phil Williamson (NERC, UK). For bibliographic purposes this document should be cited as: p 13 (Box 5): Doug Wallace DWR Wallace, CS Law, PW Boyd, Y Collos, P Croot, K Denman, p 14 (Fig 5) NASA/Jim Gower, IOS Canada PJ Lam, U Riebesell, S Takeda, & P Williamson: 2010. Ocean Fer- tilization. A Scientiﬁc Summary for Policy Makers. IOC/UNESCO, Paris (IOC/BRO/2010/2).

2 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

>1< OCEAN FERTILIZATION context andnd keykey messagesmess

Concernern over humanhuman-driven driven climate change London Protocol (LC/LP). (LC/LP) To assist that pro- and the lack of success in constraining green- cess, an overview of our scientiﬁc understand- house gas emissions have increased scientiﬁc ing is timely. The following headline messages and policy interest in geoengineering − deliber- are considered to represent the consensus view, ate interventions in the Earth’s climate system discussed in greater detail in the main text and that might moderate global warming. Proposed based on assessments of the published litera- approaches involve either removing carbon di- ture and extensive consultations:

oxide (CO2) from the atmosphere by biological or chemical means (to reduce the forcing of cli- t &YQFSJNFOUBM TNBMMTDBMF JSPO BEEJUJPOT UP mate change), or reﬂecting part of the sun’s en- high nutrient regions can greatly increase the ergy back into space (to counteract the forcing, biomass of phytoplankton and bacteria, and

by altering Earth’s radiation budget). the drawdown of CO2 in surface water. The scale of these effects depends on physical Here we consider the practicalities, opportunities and biological conditions, and the levels of and threats associated with one of the earliest other nutrients. proposed carbon-removal techniques: large- t #FDBVTFTDJFOUJmDTUVEJFTUPEBUFIBWFCFFO scale ocean fertilization, achieved by adding iron short-term and of relatively small scale, it is or other nutrients to surface waters, directly or in- not yet known how iron-based ocean fer- directly. The intention is to enhance microscopic tilization might affect zooplankton, fish and marine plant growth, on a scale large enough not seafloor biota, and the magnitude of carbon only to significantly increase the uptake of atmo- export to the deep ocean is still uncertain. spheric carbon by the ocean, but also to remove There is even less information on the effec- it from the atmosphere for long enough to provide tiveness and effects of fertilizing low nutrient global climatic benefit. This suggestion grew out regions, either directly or by using mixing de- of scientific ideas developed in the late 1980s, vices. No experimental studies have been based on analyses of natural, longterm climate carried out at the larger spatial and temporal changes (ice age cycles) and experiments that scales envisioned for commercial and geoen- provided new insights into the natural factors that gineering applications. limit ocean productivity, and thereby control the t -BSHFTDBMFGFSUJMJ[BUJPODPVMEIBWFVOJOUFOE- cycling of carbon between sea and sky. ed (and difficult to predict) impacts not only locally, e.g. risk of toxic algal blooms, but Proposals for large-scale application of ocean also far removed in space and time. Impact fertilization have been controversial, attracting assessments need to include the possibility scientific and public criticism. Upholding the of such ‘far-field’ effects on biological pro- precautionary principle, the Convention on Bio- ductivity, sub-surface oxygen levels, biogas logical Diversity (CBD) decided in 2008 that no production and ocean acidification. further ocean fertilization activities for whatever t 8IJMTUNPEFMTDBOCFEFWFMPQFEUPJNQSPWF purpose should be carried out in non-coastal predictions of both benefits and impacts, the waters until there was stronger scientific justi- totality of effects will be extremely difficult − fication, assessed through a global regulatory and costly − to directly verify, with implica- mechanism. tions for the confidence and cost-effectiveness of commercial-scale applications. Such a regulatory framework is now being de- t &TUJNBUFT PG UIF PWFSBMM FGmDJFODZ PG BUNP-

veloped, through the London Convention and spheric CO2 uptake in response to iron-based

1 ocean fertilization have decreased greatly (by 5 bly by including comparison with several oth- – 20 times) over the past 20 years. Although erwise similar but non-fertilized regions; and uncertainties still remain, the amount of carbon iii) continue over appropriate time and space that might be taken out of circulation through scales, potentially over several years and cov- this technique on a long-term basis (decades ering many thousand square kilometres. to centuries) would seem small in comparison to fossil-fuel emissions. Fertilization achieved This document focuses on scientific issues. through artificial upwelling is inherently less ef- Whilst socio-economic, ethical and legal consid- ficient for sequestration. erations are also highly important, they are not t .POJUPSJOHNVTUCFBOFTTFOUJBMDPNQPOFOU given equivalent attention here. Where estimates of any large-scale fertilization activity, both to of likelihood or certainty/uncertainty are given, check claims of carbon sequestration (for in- they are intended to be equivalent to definitions tended geoengineering benefit) and to assess used by the Intergovernmental Panel on Climate ecological impacts. Monitoring will need to: Change; however, there has been no formal pro- i) include a wide range of sensitive parameters; cess to quantify risks and probabilities. ii) take into account natural variability, prefera-

Limitation of oceanic biological production in high and low nutrient regions 1 box

Average levels of available nitrogen (as nitrate, left) and phosphorus (as phosphate, right) in the surface ocean

Biological production in the ocean usually There are also large areas of the surface refers to growth of planktonic (drifting) micro- ocean – shown above in red, yellow and green organisms that fix carbon by photosynthesis. – where N and P levels remain well above their This requires light and a range of essential limiting concentrations year-round. In these elements or nutrients. Since carbon (C), ni- high nutrient regions, the concentration of trogen (N) and phosphorus (P) are required in iron (Fe) can instead be limiting. Since phyto- relatively large amounts, they are known as plankton need around a thousand times less macro-nutrients. Fe than either N or P, it is known as a micronutrient. The amount of biomass produced in the sunlit, upper ocean is controlled by the availability of Addition of limiting nutrient(s) to an ecosystem the scarcest nutrient. In low nutrient regions can have a fertilizing effect. If limitation is by a – shown above in light purple – N or P is the micronutrient, such as iron, much less needs limiting macro-nutrient. Such areas are ef- to be added to stimulate plant growth. fectively biological deserts, since their surface waters receive very low (re-)supply of N and P, In some low nutrient regions, limitation by N mostly by slow mixing with deeper, nutrient- can be overcome by specialised microorgan- rich water. In other regions, macro-nutrient isms that can use dissolved nitrogen gas in supply, and plant biomass, may be larger but seawater. Fertilization with iron and/or phos- with a strong seasonal cycle, e.g. with mixing phate may then increase the abundance of caused by winter storms. these N-fixing organisms.

2 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

>2< WHY FERTILIZE the ocean?an?

For scientiﬁc research plied continuously or semi-continuously to To date, 13 small-scale fertilization studies many millions of square kilometres for de- have been performed in the open ocean. cades. The aim would be to increase the

They have each affected a few hundred oceanic sequestration of CO2 — its storage square kilometres for a few weeks, on a in the ocean interior — in sufficient quantity similar scale (and with similar consequenc- and for a sufficient time period to make a es) to natural blooms of phytoplankton. The climatically-significant reduction in the in-

main purpose of these studies has been to crease of atmospheric CO2. This would re- improve scientific understanding of nutrient quire verification and also confirmation that limitation, a factor closely connected to ma- there would be no deleterious unintended rine ecosystem structure, productivity and side effects. Trials to test the viability of such resource exploitation, and the global cycling ideas would need to be at the scale of thou- of carbon and other key elements. A major sands of square kilometres; they have yet to achievement has been the conclusive dem- be attempted. onstration that the supply of a micronutrient, iron − that constitutes 35% of the mass of For fishery enhancement the Earth as a whole − controls biological Increases in ocean productivity following large- production in high nutrient regions of the scale ocean fertilization might provide addi- ocean (Box 1). tional benefits from a human perspective, since growth enhancement of fish stocks might re- For deliberate carbon sult, increasing the yield of exploitable fisheries. sequestration If this were the main objective, the fertilization The oceans will, over thousands of years, application would be on a regional, rather than

take up almost all of the CO2 that will be global basis, with a clear need to demonstrate released through the burning of fossil fuels. commercial cost-effectiveness. However, the Ocean fertilization for the purpose of geoen- science is still highly uncertain, the supposed gineering aims to enhance the rate of ocean beneﬁts have yet to be demonstrated, and

uptake of atmospheric CO2 in order to slow ‘ownership’ issues for open ocean ﬁshery en- down climate change. This could (in theory) hancement have yet to be resolved. be achieved by large-scale fertilization, ap-

3 3 >3< HOW IS THE OCEAN FERTILIZED

and how could CO2 be sequestered?

Nutrients are supplied naturally to the sur- ents and CO2 that were previously used for face ocean from external sources (rivers, plant growth in the upper, sunlit waters (Fig submarine volcanoes and seeps, glacial ice 1). About a quarter of the nutrient release and atmospheric dust) and also internally, takes place in the sub-surface ocean, as a through nutrient recycling in the surface, result of sinking downward of biological ma- mid- and deep ocean. The recycling involves terial, mostly as small particles; this export the decomposition of dead marine plants, of carbon from the upper ocean is referred animals and microbes, releasing the nutri- to as the ‘biological pump’.

Fig 1. Processes involved in biological production, decomposition and nutrient cycling in the open ocean. Interactive version at www. whoi.edu/oceanus/viewFlash. do?ﬁleid=30687&id=23452&a id=35609

4 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

Artiﬁcial fertilization techniques 2

Iron in seawater is mostly in an insoluble Phosphorus addition experiments have used form which precipitates and sinks out of the concentrated phosphoric acid mixed with so- surface ocean rapidly. For fertilization experi- dium bicarbonate, or direct addition of anhy- ments, iron has been added as iron sulphate drous monosodium phosphate. The solutions box

(FeSO4∙7H2O) which is a common agricul- are pumped into surface waters behind a tural fertilizer and relatively soluble. The iron moving vessel. sulphate is dissolved in acidified seawater,

and pumped into the ocean behind a moving Nitrogen: addition of urea (NH2)2CO has been vessel. The acidic solution is neutralised rap- commercially-proposed, either as a liquid idly upon mixing with ambient seawater and mixed with phosphate solution and seawater the iron is transformed chemically into its in- and pumped into the ocean or as spherical soluble form, more rapidly in warmer waters. grains spread over the ocean surface. Commercial fertilization activities might add chemical complexing agents to keep iron in solution for longer.

Artiﬁcial upwelling: floating pipes (right) have been proposed, incorporating one-way valves that exploit wave energy or oceanic temperature and salinity gradients to bring deeper water to the near-surface. Typical di- mensions suggested for the pipes are ~10 m diameter with lengths of 100–300 m or longer. Networks of pipes, either free-floating or tethered to the seafloor, could be distributed across regions with low surface nutrient concentrations.

Most ocean fertilization approaches (by small- the atmosphere via air-sea gas exchange. In

scale experiments and by models) have to order that any additional CO2 uptake from date focused on increasing the external sup- the atmosphere can subsequently be consid- ply of nutrients. However, acceleration of the ered to be sequestered, it should be stored internal recycling of nutrients is also being at least below the depth to which seasonal explored, using artiﬁcial upwelling to bring mixing occurs, and generally, the deeper the to the surface naturally nutrient-rich deeper better (Box 3). In contrast, artiﬁcial upwell- waters (Box 2), or by using optical devices to ing not only pumps nutrients upwards, but

increase light penetration. also the CO2 released from previous cycles of production/export and sinking/ decompo- There is an important distinction between fer- sition. Although some net uptake of carbon tilization with external or recycled nutrients. may be possible, e.g. if nitrogen-ﬁxation is

An increase of the external supply of nutri- stimulated, the drawdown of CO2 from the ents to surface waters can, potentially, re- atmosphere by artiﬁcial upwelling is inher-

duce their concentration of dissolved CO2 − ently limited.

hence increasing ocean uptake of CO2 from

5 5 >4< WHAT HAPPENS when the ocean is fertilized?

Iron addition t #BDUFSJBM CJPNBTT JODSFBTFE EVSJOH NPTU PG the experiments (by 2-15 times). A transient The bullets points below summarise findings from increase in the stocks of small grazers, micro- the 13 small-scale, iron addition experiments car- zooplankton, was also reported from some ex- ried out to date by independent researchers (Fig periments. 2). These studies initially fertilized patches of sur- t 5IF EVSBUJPO PG UIF FYQFSJNFOUT XBT VTV- face ocean in high nutrient regions over the range ally too short to allow larger zooplankton to 40 - 300 square kilometres. Two pilot studies us- respond. However, grazing increased in two ing iron have also been carried out by commercial experiments with high pre-existing stocks of organisations, on a similar scale. Full-scale dem- medium-sized zooplankton (copepods), and onstrations or deployments for geoengineering or played a major role in controlling the develop- fishery enhancement would, however, need to be ment of these blooms. very much larger, involving fertilization of around t 5IFSF JT BT ZFU OP JOGPSNBUJPO GSPN FYQFSJ- 10,000 square kilometres. mental studies on responses further up the food chain (e.g. by fish). t -FWFMT PG UIF QMBOU QJHNFOU DIMPSPQIZMM JO- creased in all experiments, by 2-25 times, with associated increases in carbon fixation. Some of the artificially-induced blooms of phytoplankton were visible to satellite-based ocean colour sensors. t 1IZUPQMBOLUPOSFTQPOEFEUPUIFJSPOBEEJUJPO by an increase in photosynthetic efficiency and by altered rates of nutrient uptake. t &GGFDUTPOQIZUPQMBOLUPOQSPEVDUJPOBOECJP- mass were greater in shallower surface mixed layers due to the more confined depth range and, consequently, higher average light intensity experienced by the fertilized plankton. Re- sponse was more rapid in warmer waters. Fig 2. Sites of the 13 iron fertilization experiments (red), two commercial trials using iron (pink) and two phosphate addition studies (white) t *ONPTUPGUIFFYQFSJNFOUT UIFEPNJOBOUQIZ- toplankton group changed, with a shift in com- carried out to date, on map of satellite-based ocean primary production munity composition from smaller groups (cy- (yellow/green, high; dark blue, low). anobacteria), via medium-sized phytoplankton (haptophytes), to larger diatoms. Phosphorus addition experiments t "MUIPVHI EJBUPNT VTVBMMZ EPNJOBUFE TQFDJFT composition after iron addition, the most abun- There have been two small-scale field studies in- dant diatom species varied between locations volving P-additions, both in low nutrient waters. In and experiments. This may reflect regional spe- the Eastern Mediterranean, the experiment result- cies differences of initial ‘seed’ populations as ed in rapid increases in bacterial production and well as competition under a range of ocean zooplankton biomass, and a moderate increase conditions. in rates of nitrogen-fixation. However, there was

6 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

a slight decrease in phytoplankton biomass and Fate of the added nutrients chlorophyll (in contrast to a predicted increase). The fate of externally-added nutrients depends Similar effects on bacteria and phytoplankton on their chemical nature. Several experiments were observed off NW Africa when phosphate with iron required re-fertilization because the was added alone and with iron. These results added iron rapidly ‘disappeared’, either through are not yet fully explained; they suggest alter- formation of organic complexes or through ad- native food-chain pathways and/or additional sorption onto particles which sank. Thus added complex limitations operating in low nutrient iron can be lost from surface waters before it is systems subject to P limitation. used by plankton, and much may be removed from the ocean permanently through burial of Artiﬁcial upwelling particles in sediments. In the case of fertilization with phosphate or nitrogen, the added nu- Technologically-robust designs for ‘ocean pipes’ trients are expected to be incorporated rapidly would be needed to operate in the way envis- into biomass, to be subsequently recycled and aged for artiﬁcial upwelling systems. Those de- released through decomposition in surface or veloped to date have delivered pumping rates subsurface waters, with relatively little being lost of 45 m3 per hour, but for less than a day − too to sediments. short for the expected biological and biogeo-

chemical responses to be observed. Modelling CO2 drawdown and carbon export studies have been undertaken, but with major uncertainties concerning ecosystem response; Increases in phytoplankton biomass due to ex- in particular, whether induced upwelling of water perimental fertilization have been accompanied

with high P levels might stimulate nitrogen-ﬁxa- by reductions in CO2 levels in surface water, pro-

tion, with potential for net CO2 drawdown. Over- moting CO2 drawdown from the atmosphere by

all, it seems more likely that artificial upwelling gas exchange. The amount of CO2 drawdown will become a tool to study marine ecosystem has varied greatly between studies, depending responses to nutrient perturbations and chang- on: the amount of nutrient added; whether other es in mixing regimes, rather than a cost-effective factors limited the biomass increase; the nutri- measure to counteract climate change. ent-carbon ratio of the enhanced biomass; the extent to which there were additional removal Nutrient depletion and co-limitation processes for the added nutrients; conditions following fertilization at the air-sea interface (e.g. wind speed, wave characteristics); the depth of the surface mixed The addition of a limiting nutrient will, ultimately, layer; and the time that fertilized waters remained result in another factor becoming limiting. In the in direct contact with the atmosphere. Most ex- case of iron additions to high nutrient regions, periments did not continue for a sufficiently long macronutrients such as silicate (required by time period to follow the decline of the stimulat- diatoms) and nitrate (required by all phytoplank- ed phytoplankton bloom and associated carbon ton) subsequently became depleted. In several export. Two studies did report increased carbon experiments, the diatom bloom either crashed export, but of different proportions. within two weeks of fertilization or, in one case, did not develop at all − due to a lack of silicon. Unexpected responses Light can be an additional limiting factor, es- pecially in polar regions, due to season, cloud The experiments to date show that the biological cover, deep mixing and self-shading caused by and chemical responses to nutrient fertilization phytoplankton themselves. For phosphate ad- are variable and difficult to predict. Examples dition experiments in low nutrient regions, the include the unexpected decrease in chlorophyll biological response was probably limited by ni- levels in response to phosphate addition in the trogen availability. Mediterranean; and the observation that mark- edly different phytoplankton communities and total biomass resulted from two iron addition experiments conducted a year apart at the same site in the north west Pacific Ocean.

7 7 >5< ARE THERE UNINTENDED impacts of ocean fertilization?

Changes to the surface ocean between DMS and climate is relatively weak. ecosystem Several other trace gases have been observed to have altered concentrations after The iron fertilization experiments conducted fertilization, with potential implications for at- to date are not known to have resulted in mospheric ozone concentrations. The over- harmful algal blooms. However, shipboard all significance of such effects is currently experiments in the north west Pacific sug- unclear. gest that diatom species that produce the toxin domoic acid might increase in abun- Far-ﬁeld effects dance in response to iron fertilization, and their rate of toxin production might also be Far-field effects, hundreds or thousands raised. This possibility requires further in- of kilometres from the fertilization site and vestigation. ‘Non-deliberate’ ocean fertiliza- occurring months, years or decades after- tion with nitrogen-containing urea, through wards, include potential impacts on subsur- sewage, is known to favour the growth of face waters and sediments into which the cyanobacteria and dinoflagellates, including fertilized biomass sinks. For small-scale, toxic species. short-term experimental studies such effects are almost certainly trivial and non-measur- As already indicated, fertilization experiments able, but they are likely to become significant have been of insufficient duration and spa- if large-scale, longterm fertilization is carried tial scale to reveal changes at higher levels out. Prediction and assessment of far-field within the food chain. Thus any suggestions impacts requires information on biomass of either positive or negative impacts on fish production and sinking as well as on the stocks remain speculative. circulation and mixing of both the fertilized surface waters and the subsurface waters Production of climate-relevant beneath the fertilized location; such informa- gases in the surface ocean tion can then be used in complex models which simulate ocean circulation, biology Ocean fertilization has been observed to in- and chemistry. However, model predictions crease the surface water concentrations of of far-field effects will be extremely difficult a range of climate-relevant gases associ- to verify with direct observations because ated with phytoplankton growth. Of these, of the large spatial and time-scales involved the best studied is dimethylsulphide (DMS) (Section 7). which, after emission to the atmosphere, might influence climate via the formation of An important far-field consequence of large- particles that promote cloud formation. Most scale fertilization with limiting nutrients (e.g. iron fertilization experiments have shown in- with iron in a high nutrient region) involves creased DMS production. Results have been the depletion of other non-limiting nutrients, extrapolated to suggest that fertilization of such as nitrate or phosphate. This depletion 2% of the Southern Ocean could decrease can, in turn, reduce the productivity of re- temperatures by ~2°C in that region. How- mote regions downstream of the fertilization ever a fertilization study in the sub-Arctic Pa- location, particularly where natural sources cific observed a DMS decrease, and recent of the fertilizing nutrient are available (e.g. modeling analyses indicate that the linkage iron from shelf sea sediments or atmospheric

8 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

The importance of transport and timescales 3 box

Vertical and horizontal transport processes over a range of timescales affect the fate of biologically-ﬁxed carbon in the ocean

A key characteristic of the oceanic ecosystem Oceanic mixing also causes impacts to is transport over long distances associated spread, so that fertilization of a relatively small with mixing, sinking of particles (on a timescale area could, to some degree, ultimately impact of weeks to months), and ocean circulation. A vast regions of the ocean. There can be long consequence is that changes at one place in time delays as well as large distances sepa- the surface ocean can impact deeper water a rating large-scale fertilization and its impacts, few kilometers away in the vertical and thou- with associated difficulties for the attribution sands of kilometers away in the horizontal. of impacts or verification of effects.

dust). This potential far-field impact has been In an analogous way, any additional CO2 referred to as ‘nutrient robbing’. Thus it is taken up locally due to the fertilization can possible that fertilization of an open ocean potentially ‘rob’ regions downstream of their

location in international waters could reduce CO2 uptake capacity due to the reduced, productivity around islands and countries far-field, biological production. This must be

not involved with the fertilization activity. considered in determining the overall CO2 Models have examined the scale of such ef- sequestration efficiency of any fertilization fects and, for scenarios involving large-scale (Section 6). fertilization over long periods, large reductions in far-field productivity are indicated. Subsurface oxygen decrease These reductions could have significant consequences, including a re-distribution or Decomposition of any fertilization-enhanced overall decrease in fish production. biomass will decrease oxygen levels in the sub-surface ocean, with impacts that may The other side of the coin to ‘nutrient rob- be local or remote, depending on the re- bing’ in the surface ocean is that increased gional circulation, and could lead to critical nutrient levels in deep ocean waters (due thresholds or tipping points being crossed to decomposition of the biomass that was (Box 4). Mid-water oxygen depletion has not increased by fertilization) may enhance the been reported for fertilization experiments productivity of ecosystems in other remote conducted to date due to their limited scale regions, where these waters are eventually and duration, but additional oxygen demand returned to the surface ocean by upwelling is an inevitable consequence of enhanced or mixing. downward carbon export. Decreased oxygen

9 9 ‘Tipping points’ relevant to ocean fertilization 4 box

Global distribution of oxygen at 350 m. Red/purple areas show oxygen minimum zones

There are at least two critical thresholds or far-field oxygen concentrations below these ‘tipping points’ relevant to ocean fertilization threshold concentrations in regions that are impacts: removed from close contact with the atmosphere via mixing. Oxygen. The abundance of dissolved oxygen in the oceanic water column and sediments Carbonate concentration. The tendency of is a key control for life in the sea as well as carbonate minerals to dissolve in seawater, for an array of chemical processes, including including the carbonate shells of both living nutrient recycling. Subsurface waters, not and dead marine organisms, is governed by in direct contact with the atmosphere, have a critical concentration of the carbonate ion 2 reduced oxygen levels representing the bal- (CO3 -) as well as by temperature and pres-

ance between oxygen supply by ocean cir- sure. Release of CO2 to subsurface seawa- culation and the cumulative demand due to ter during decomposition of organic carbon respiration processes. Critical threshold con- reduces pH (acidification) and carbonate ion centrations of oxygen are process-depen- concentration. Increased organic carbon sup- dent, but are greater than zero and generally ply to the deep ocean could, therefore, alter

in the range 5-40 μmol O2 per litre. Increased the depths and locations where these critical organic carbon supply due to large-scale carbonate concentrations are reached in the ocean fertilization could, potentially, drive ocean interior.

levels close to the site of fertilization might and oxygen minimum zones could, however, precondition subsurface waters so that they cause increased frequency and intensity of cross a critical threshold during subsequent near-shore hypoxia and, as a consequence, transport through the ocean interior (e.g. to- significant mortality of marine organisms. wards oxygen minimum zones). Important within-ocean nutrient recycling processes might also be altered. The chang- Early studies using highly-simplified ‘box es of subsurface oxygen concentrations are models’ predicted that large volumes of the dependent on the location as well as the subsurface ocean would become anoxic as scale of the fertilization in relation to ocean a consequence of large-scale and continu- circulation patterns and existing oxygen dis- ous fertilization. More sophisticated models, tributions, and can only be assessed using based on more likely fertilization scenarios, complex models. These models have inher- predict a less dramatic scenario involving ent limitations in their ability to represent growth of the extent of low-oxygen regions existing oxygen distributions and hence pre- rather than oceanic anoxia. Fertilization- dictions of change in oxygen levels must be induced oxygen depletion of mid-depth wa- considered uncertain. ters that supply certain upwelling systems

10 release to the atmosphere of relatively small amounts of these gases could offset the de-

sired effects of CO2 sequestration. Methane is considered the lower risk, since most of this gas naturally produced within the ocean is used as an energy source by other ma-

rine microbes and converted to CO2 before A B reaching the atmosphere. Diversity The ocean is, however, an important source OligotrophicFlux of particulate organic carbon to seafloor Eutrophic of N2O and any enhanced production is likely Fig 3. The greatest seafloor biodiversity occurs when organic car- to be emitted to the atmosphere. The far-field bon export from the upper ocean is midway between very productive impact of large-scale fertilization has been (eutrophic) and very unproductive (oligotrophic) conditions. The addi- simulated by models. If fertilization takes tional biomass stimulated by large-scale ocean fertilization could there- place over waters that are already low in oxy-

fore increase biodiversity if initial state was at A, or decrease it if at B. gen (e.g. the tropics), the N2O yield could be large, with an estimated 40 - 70% offset of

the beneﬁts of CO2 reduction after 100 years. Effects on seaﬂoor ecosystems The offsetting would be much lower (~10%) for fertilization of waters underlain with higher The effect of large-scale ocean fertilization oxygen concentrations, such as in the South- on seafloor ecosystems depends critically ern Ocean. Assessments of overall climate on the water depth where the fertilization forcing depend critically on the accuracy of takes place and the sinking speeds of the ocean circulation models, the representation particulate biomass produced. In deep wa- of oxygen in these models, and our limited

ters, a large proportion of any enhanced car- knowledge of N2O yield during biomass de-

bon flux will be decomposed before reaching composition. Only minor increases in N2O the sea floor. The enhanced carbon flux to production have been observed during iron the seafloor is likely to increase the amount addition experiments; at this scale only tran- of seafloor biomass, as long as oxygen is not sient and highly dispersed effects are likely, depleted; this might have either a positive or without ecological or climatic signiﬁcance. negative effect on seafloor biodiversity, depending on its background state (Fig 3). Ocean acidiﬁcation

Production of climate-relevant If large-scale fertilization were to lead to

gases and greenhouse gas substantive additional CO2 sequestration at ‘offsetting’ depth, this would increase the acidiﬁcation of ocean interior waters. Such changes would Decomposition of sinking biomass can pro- alter the depth at which carbonate biominer- duce the long-lived, greenhouse gases ni- als start to dissolve (Box 4), potentially re-

trous oxide (N2O) and methane (CH4), with stricting the habitat of deep-ocean organisms global warming potentials 320 times and 20 that build shells and other structures out of

times greater than CO2 respectively. Thus the these biominerals, e.g. deep-sea corals.

11 11 >6< HOW EFFICIENT IS LARGE-SCALE OCEAN FERTILIZATION for sequestering atmospheric carbon?

Efficiency with addition of external and can only be addressed with models. Such nutrients models have undergone steady development so that estimates of the atmospheric uptake efficiency Twenty years ago, fertilization of surface waters are still changing as new processes are investi- with iron looked like a highly efficient process for gated and more realistic models are implemented. stimulating export of large amounts of carbon, via Early models, based on very simple treatments of sinking particles, to the deep ocean where it would nutrient uptake, suggested atmospheric uptake be isolated from the atmosphere for 100 - 1000 efficiencies of less than 10-40% whereas more re- years. This early view was based on the calcula- cent models suggest higher efficiencies (70-90%), tion that 1 tonne of added iron might sequester at least for fertilization of tropical waters. Clearly more than 100,000 tonnes of carbon, i.e. a carbon this is an area of continued uncertainty which export ratio (Box 5) greater than 100,000:1. greatly impacts estimates of the overall sequestration efficiency. However, the one experimental fertilization carried out to date that gave detailed data on carbon ex- However, even using the highest estimates for port indicated a much lower estimates of this ef- both carbon export ratios and atmospheric uptake ficiency, at less than 5,000:1. This could be due efficiencies, the overall potential for ocean fertiliza-

to rapid grazing or decomposition of the enhanced tion to remove CO2 from the atmosphere is rela- phytoplankton growth. An additional factor, ob- tively small. Thus recent calculations of cumulative served in other studies, was the rapid loss (of up sequestration for massive fertilization effort over to 75%) of the added iron, by its precipitation and 100 years are in the range 25-75 Gt (gigatonnes) of scavenging onto particles before it could be uti- carbon (Fig 4), in comparison to cumulative emis- lized for phytoplankton growth. Improved delivery sions of around 1,500 Gt carbon from fossil fuel mechanisms for iron, such as the use of chemical burning for the same period under business-as- complexing agents, could improve this efﬁciency, usual scenarios. but with cost implications.

The atmospheric uptake efﬁciency (Box 5) based

on the CO2 drawdown measured during these short-duration experiments was only 2 - 20%. These may be lower bound estimates to this ef-

ﬁciency given that uptake of CO2 is likely to have continued for a period of time after measurements ended. On the other hand, ~50 % of the exported biomass is likely to decompose above a depth of 500m. In several of the high nutrient oceanic regions that might be considered for fertilization, water mixing in wintertime extends to at least this

depth so that much of the CO2 from the exported biomass would return to the atmosphere within a year of fertilization. Fig 4. Model-based estimates of the effectiveness of carbon How long exported carbon remains sequestered sequestration (cumulative drawdown over 100 yr) for large-scale, strongly affects the atmospheric uptake efﬁciency iron-based ocean fertilization. Dates relate to year of publication.

12 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

Sequestration efﬁciency 5 The overall efficiency of ocean fertilization The carbon export ratio is controlled by

as a means to sequester atmospheric CO2 nutrient loss processes, the carbon:nutrient is the product of two difficult-to-estimate ratio in fertilized biomass, and the proportion factors: 1) how much additional (net) car- of biomass resulting from fertilization which box bon is exported from surface waters into sinks into the deep ocean. the deep ocean for a given addition of nutrient (the carbon export ratio), and 2) what The atmospheric uptake efﬁciency de- proportion of the additional carbon export pends on factors such as wind and waves is, ultimately, resupplied by carbon taken which determine the rate of air-sea gas ex- up from the atmosphere (the atmospheric change and the depth to which exported uptake efficiency). Some sources of inef- carbon sinks before being decomposed (with ficiency are depicted schematically as red higher efficiency at greater depths). arrows in the figure below, the thicker red arrows indicating inefficiencies that occur The efficiency of sequestration over dec- relatively rapidly, and the thinner red arrows adal to century timescales depends also on those that may take years or decades. whether the fertilizing nutrient is recycled or lost from the ocean.

Fertilization CO2 uptake CO2 reeleaasse

NNuuttrrieenntt Organic Inntteerrnnal Nutrients CO2 loossss biomass supppply Reccyycclliinng RReeccyycclliinng Near-surface Upwweelling mixxiinngg Sinking

NNuuttrrieenntt RReeccyycclliinng Organic Recycclling Nutrients CO2 loossss particles

Intermediate depth waters

Nutrients CO2

Deep ocean and sediments

Principle processes and inefﬁciencies involved with fertilization for carbon sequestration. sequestration Blue arrows repre- repre- sent the intended sequestration pathways whereas red arrows represent pathways by which the efﬁciency of sequestration is reduced.

13 13 Carbon export efficiency with CO2 sequestration potential for longer peri- artificial upwelling ods depends on what happens when artifi-

cially CO2 enriched deep waters are eventually The proposed enhancement of the biological returned to the ocean surface. This in turn pump by artiﬁcial upwelling is less efﬁcient for depends on the nature of the nutrient used

CO2 sequestration. Initial modeling has indicat- for fertilization. If the nutrient is re-released to ed that global deployment of pipes could result deep waters via decomposition in the same in large changes to biological production and proportion to carbon as used for growth, then export of carbon, but relatively small changes to the added nutrient can be considered to be

the air-sea CO2 uptake. This is because most of recycled. When such recycled nutrient is up- the additional exported carbon is decomposed welled, it can fuel another cycle of growth, and recycled close to the surface (<500 m). Al- carbon uptake and sinking so that the original ternative scenarios, yet to be investigated, could extra carbon remains in the ocean. However, involve manipulation of the nutrient supply rate, if the fertilizing nutrient is removed permanent- or stimulation of nitrogen-ﬁxing organisms or or- ly from the ocean by burial in sediments (the ganisms that can sink deep into the ocean. likely fate of added iron), then the nutrient is

unavailable when the CO2-enriched deep wa- Long-term (century-scale) ter is brought to the surface again by upwell-

sequestration ing processes — and much of the extra CO2 Most model simulations for large-scale fertil- drawdown resulting from the initial fertilization ization are for periods of 10-100 years. The will be returned to the atmosphere.

Alaska Canada

Fig 5. Satellite image of the phytoplankton bloom stimulated by the SERIES iron fertilization experiment in the North East Paciﬁc (circled). Black areas are cloud cover. The red/orange colours south of Alaska and in other coastal areas are natural blooms. This SeaWiFS image was acquired 19 days after initial addition of iron (on 29 July 2002); ﬁve days later, the patch was barely visible.

14 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS

>7< MONITORING FOR veriﬁcationﬁcation and reversibility

Veriﬁcation Reversibility

If the objective of fertilization is to claim ‘credit’ There is a consensus within the scientific com- for enhanced sequestration of carbon then veri- munity that none of the small-scale iron fertiliza- fication must include measurement-based es- tion experiments conducted to date are likely to timates of the amount of carbon sequestered. have resulted in long term alteration of ocean Alternatively, if the objective is to increase the ecosystems. Thus the individual fertilizations amount of biomass at a particular trophic level of several hundred square kilometres of ocean of the ecosystem (e.g. of a harvestable marine surface, each with ~10 tonnes of iron sulphate, resource, such as fish), then the increase in bio- represent a scale comparable to natural bloom mass of the target species must be measured, to events, having effects limited to a few months. show that the desired effect has been achieved. However, the findings from small scale fertiliza- In both cases, verification requires: tion experiments cannot be directly scaled up to the much larger scales envisioned for commer- t NPOJUPSJOHPGDIBOHFTJOUIFEPXOXBSEDBS- cial and geoengineering applications. Purpose- bon export or fish biomass in both the fertil- ful fertilization on a scale large enough to cause ized areas and adjacent areas that were not a measurable change in atmospheric carbon fertilized but were otherwise similar dioxide concentration will also cause major al- t MPOHUFSN NPOUIT UP ZFBST BOE GBSmFME terations to the structure of regional planktonic monitoring to determine if there are subse- ecosystems, since large-scale sequestration of quent rebound effects that might offset some carbon requires a major shift in plankton com- of the initial change or might have negative munity composition. impacts. Would such an artificial change to a marine Monitoring must be sufficiently extensive to pro- ecosystem be reversible if it were later judged vide defensible verification that fertilization ob- to be deleterious? For comparison, a ‘regime jectives have been achieved without unaccept- shift’ associated with natural variability was able or unintended negative impacts. Verification documented in the subarctic North Pacific eco- should address far-field effects on the concen- system in 1977 with a return to more or less trations of oxygen and nitrous oxide (Section 5) the initial state observed in 1989. The biological as well as far-field reductions in surface nutrient indicators of the regime shift were more clearly levels that might decrease carbon sequestration obvious than the physical factors, which were and productivity elsewhere (‘nutrient robbing’ presumed to have been the causative factors.

and ‘CO2 sink robbing’). In general, we rarely understand the factors and mechanisms that cause large-scale, natural re- Effective monitoring of the short-term, near-field gime shifts within marine ecosystems. Hence it intended effects of large scale fertilization will it- is arguable that we have insufficient knowledge, self be costly. In the opinion of several scientists let alone technique, to purposefully manipulate who have been involved in past iron fertilization an ecosystem to reverse any large scale, long experiments, adequate verification cannot yet be term changes to ecosystems that might be have achieved with currently available observing ca- been initiated by deliberate ocean fertilization. pabilities.

15 15 >8< GOVERNANCE and policyicy

The United Nations General Assembly has en- sidered that more extensive and targeted field couraged States to support the further study work, and better models of marine processes, and enhance understanding of ocean fertilization were needed – whilst recognising that ocean fer- (Resolution 62/215; December 2007). Four UN tilization presents serious regulatory challenges, bodies and associated secretariats have ma- to avoid harm to the marine environment. jor interests in this topic: the Intergovernmental Oceanographic Commission of UNESCO (IOC), The IMO is addressing such challenges in its the Convention on Biological Diversity (CBD), role as Secretariat for the Convention on the the International Maritime Organization (IMO) Prevention of Marine Pollution by Dumping of and the UN Convention on Law of the Sea Wastes and Other Matter 1972 (the London (UNCLOS). Together they cover the spectrum Convention) and its 1996 Protocol (the London of marine science, marine conservation and Protocol; together known as LC/LP). In Octo- pollution regulation. ber 2008, the LC/LP Parties decided that: 1) given the present state of knowledge, ocean In response to concerns that large-scale ocean fertilization activities other than legitimate sci- fertilization might be attempted before its con- entific research should not be allowed; 2) they sequences were fully understood, the CBD re- would develop a potential legally binding reso- quested Parties, and urged other governments, lution or an amendment to the London Protocol to ensure that ocean fertilization activities do on ocean fertilization; and 3) they would also not take place until there is an adequate sci- develop a framework for assessing the com- entific basis on which to justify such activities. patibility of ocean fertilization experiments with This justification should include an assessment the London Convention and Protocol. The IMO of associated risks, and a global, transparent definition of ocean fertilization excluded “con- and effective control and regulatory mechanism ventional aquaculture, or mariculture, and the is in place for these activities, with the excep- creation of artificial reefs”. tion of small scale scientific research studies within coastal waters (Decision IX/16; May The IOC has considered issues relating to 2008). The ‘coastal waters’ exception was ocean fertilization at its 25th Assembly (June intended to recognise that territorial seas and 2009) and its 43rd Executive Council (June other maritime jurisdiction zones already gave 2010). The IOC has been closely involved in states the responsibility for conserving and the CBD and IMO discussions. IOC Member managing their own marine resources. States have agreed that the precautionary principle is fundamental to the regulation of ocean The CBD Secretariat subsequently published fertilization, and reasserted that IOC’s main a review of the impacts of ocean fertilization role is to respond to requests for scientific or on marine biodiversity, with its main conclu- technical information and advice from relevant sion being that sound and objectively verifiable bodies or Member States. The current review scientific data of such impacts are scarce. To provides an example of such contributions to provide such information, the CBD review con- the overall process.

16 OCEAN FERTILIZATION A SCIENTIFIC SUMMARY FOR POLICY MAKERS