New OCB Research :: Ocean Carbon & Biogeochemistry

Studying marine ecosystems and biogeochemical cycles in the face of environmental change

Archive for New OCB Research – Page 27

Subtropical gyre productivity sustained by lateral nutrient transport

Posted by Katherine Joyce

· Saturday, November 5th, 2016

Vertical processes are thought to dominate nutrient resupply across the ocean, however estimated vertical fluxes are insufficient to sustain observed net productivity in the thermally stratified subtropical gyres. A recent study by Letscher et al. (2016) published in Nature Geoscience used a global biogeochemical ocean model to quantify the importance of lateral transport and biological uptake of inorganic and organic forms of nitrogen and phosphorus to the euphotic zone over the low-latitude ocean. Lateral nutrient transport is a major contributor to subtropical nutrient budgets, supplying a third of the nitrogen and up to two-thirds of the phosphorus needed to sustain gyre productivity. Half of the annual lateral nutrient flux occurs during the stratified summer and fall months, helping to explain seasonal patterns of net community production at the time-series sites near Bermuda and Hawaii. Figure from Letscher et al. (2016).

Exploring sources of uncertainty in ocean carbon uptake projections

Posted by Katherine Joyce

· Tuesday, October 25th, 2016

Having absorbed ~30% of the carbon dioxide released to the atmosphere by human activities, the oceans play an important role in mitigating warming and other climate-related impacts of rising carbon dioxide levels. Predictions of future climate change thus require more accurate projections of ocean carbon uptake. Using two different model suites, a recent study by Lovenduski et al. (2016) published in Global Biogeochemical Cycles documents the relative contributions of internal climate variability, emissions scenario, and model structure to overall uncertainty in ocean carbon uptake predictions on both regional and global scales. Figure from Lovenduski et al. (2016).

Trace metal uptake and remineralization and their impact on upper ocean stoichiometry

Posted by mmaheigan

· Wednesday, July 6th, 2016

1. Stoichiometry of metals in the ocean

The close relationship between the stoichiometry of nutrients dissolved in the upper ocean and the planktonic organisms that grow in these waters has long been recognized (1, 2). The stoichiometry of 106 C:16 N:1 P first summarized by Redfield has become a fundamental concept of marine biogeochemistry, with numerous studies using the ratio as a benchmark to assess ecosystem function. Decades after the work of Redfield, with the implementation of trace metal-clean techniques, oceanographers produced the first meaningful measurements of dissolved trace metals in the open ocean (3-5), and they found that many of the bioactive metals such as Fe, Zn, Ni, Cu and Cd are also depleted in surface waters and enriched at depth, similar to the macronutrients. Such nutrient-like behavior supported not only a growing understanding of the physiological roles that these metals play in phytoplankton physiology (6), but it also indicated that biological uptake and sub-surface remineralization were important processes for controlling the distributions of these bioactive metals in the ocean. Thus, the biogeochemical behavior of the micronutrient metals is in many ways analogous to that of the macronutrients N, P and Si.

In the open ocean far from coastal and shelf influences, dissolved concentrations of bioactive metals increase with depth at relatively consistent ratios to macronutrients (5), and these metal:nutrient remineralization ratios have been used to approximate the composition of sinking biogenic material and euphotic zone phytoplankton (7, 8). These ‘extended Redfield ratios’ have been compared to average compositions of marine phytoplankton species grown in culture (9-12), and the general agreement between these approaches further supports the importance of biological uptake and subsequent remineralization of trace metals in the upper ocean as key processes impacting trace metal geochemistry. Average metal:nutrient stoichiometries for phytoplankton have also been compared to dissolved stoichiometries in the ambient water, and relationships between these fractions have been used to estimate nutrient limitation and deficiency in the ocean (13). Thus, there is significant interest in controls on upper ocean metal stoichiometries, as well as the relationships between cellular/biological, particulate and dissolved fractions.

Analogous to macronutrients, there are also relationships between metal stoichiometries in phytoplankton and those in deeper waters of the ocean. Departures from these relationships are likely to provide insights into the internal biogeochemical cycling of metals in the ocean. Morel and Hudson (7) noted differences in the extended stoichiometries of plankton and the water column and concluded that they must reflect the relative efficiency of remineralization of the elements, as well as the propensity of elements to be scavenged onto sinking particles in the sub-surface ocean. Similarly, the rapid remineralization of trace metals from sinking plankton was addressed in seminal work by Collier and Edmond (14). Using carefully collected data on surface plankton material, and with more computational rigor than (7), they compared surface particle stoichiometries to deep water dissolved stoichiometries and calculated the relative remineralization of plankton-associated elements in sinking biogenic material. They noted significant differences among the behaviors of biogenic metals such as Cd, Ni and Fe due to their scavenging and remineralization behaviors. More recently, Morel (15) mused about these processes and their relationships to cellular biochemistry and evolution of phytoplankton physiology and ocean biogeochemistry.

Through the GEOTRACES program, the data to test and extend these early, relatively simple box models and stoichiometric comparisons are now available. Metal concentrations and stoichiometries for phytoplankton, bulk and size-fractionated particulate material, and co-located dissolved species have been measured in the North Atlantic and South Pacific Oceans thus far. Combined with data for non-bioactive metals such as Ti and Th, these data also provide the opportunity to discern the behavior and contributions of lithogenic vs. biogenic matter, as well as the processes of remineralization and scavenging.

2. Processes affecting dissolved and particulate stoichiometries of trace metals

Vertical profiles of dissolved macronutrients show characteristic depletion at the surface and enrichment at depth due to remineralization, and dissolved micronutrients often show the same behavior. However, the internal cycling of metals in the ocean is expected to differ from that of macronutrients for a few salient reasons. Some metals such as Fe are significantly less soluble than macronutrients and are prone to abiotic adsorption onto particulate surfaces (16). This process is driven by thermodynamics, and the accompanying process of desorption also occurs; the net observed process is typically called ‘scavenging’ (Fig. 1). Scavenging in the deep ocean causes concentrations of less soluble metals such as Fe and Al to decrease along the path of thermohaline circulation, in contrast to macronutrients and more soluble metals that may mimic macronutrient behavior such as Cd and Zn (17). In the absence of significant lateral nutrient inputs, the balance of scavenging and remineralization will influence the resulting vertical profiles of dissolved elements (18).

Another key difference between macronutrients and metals is the importance of abiotic particulate fractions such as lithogenic (e.g., aeolian dust and sediment) and authigenic (e.g., Fe- and Mn-oxyhydroxide) phases. While biogenic phases are almost universally produced at the surface and remineralized with depth, abiotic phases can exhibit very different and dynamic internal cycles (19). Dust events, lateral transport and poorly constrained scavenging processes can both deliver and remove specific metals alongside biological processes. Lithogenic phases are generally denser and more refractory than biogenic particles and detritus and are thought to sink more rapidly and remineralize more slowly and at greater depth (Fig. 1; 20). Lithogenic particles may also (re)scavenge metals differently than biogenic material. Efforts to examine these processes in sinking material have been extremely limited to date, with only a few studies examining metals in trace metal-clean sediment traps (21, 22). However, recently published datasets from the GEOTRACES program are shedding new light on the multiple facets of metal partitioning and how they affect subsurface remineralization and scavenging.

A comparison of metal:phosphorus ratios in the upper ocean illuminates some of these processes. Figure 2 displays Cd:P, Fe:P, Co:P and Ni:P ratios in particles in the upper 100m, 100-300m, and 300-1,000m of the water column in the middle of the North Atlantic basin. Particulate material is sub-divided into ratios for phytoplankton cells and non-lithogenic particles (corrected for lithogenic minerals using Ti; 19). Also plotted are dissolved remineralization ratios (that is, the slope of a linear regression between the dissolved metal and phosphate) for these upper ocean depth ranges. The close coupling of Cd and P biogeochemistry has long been recognized (4), and indeed we observe very close agreement (within a factor of about 2) between dissolved Cd:P remineralization and Cd:P in surface ocean particles, as well subsurface particles. Clearly these elements are remineralizing from sinking particles at similar rates. Such comparisons of particulate and dissolved constituents need to carefully consider the different residence times of these fractions and the likelihood for lateral inputs. Here, we have chosen to focus on stations from the mid-North Atlantic gyre, where the upper 700m of the water column consists primarily of a single water mass (23).

In contrast, the remineralization of Fe and P are quickly decoupled in the water column (Fig. 2). Between 100 and 300m, typically the depth of most rapid regeneration of sinking organic material, labile particulate Fe:P has more than doubled from that in surface waters, and the Fe:P ratio of remineralized dissolved elements (0.98 mmol/mol) is more than 10-fold below that of the labile material that is sinking into these waters. Looking deeper into the water column, Fe and P continue to decouple in labile (i.e., non-lithogenic) particulates, with Fe:P of 300-1,000m particles increasing 10-fold and the dissolved remineralization ratio being nearly 1,000-fold lower (0.35 mmol/mol). Additionally, organic ligands play an important role in stabilizing dissolved Fe (24), so dissolved Fe and P ratios may be further decoupled by biological processes impacting the production and fate of these ligands (20).

A strength of GEOTRACES datasets is their wide coverage of the periodic table, and additional insights can be gained from looking at the behaviors of other bioactive trace metals that are also incorporated into sinking biogenic material. Co:P ratios in particles and remineralized dissolved fractions in the water column follow the same trend as Fe, but the decoupling of Co and P is much more subtle than with Fe, presumably due to differences in ligand coordination and Co co-oxidation with Mn (25, 26). Dissolved Co:P remineralization ratios at 100-300m generally match those found in phytoplankton and drop only 3-fold below 300m. Similarly, labile particulate Co:P ratios don’t change between 0-100m and 100-300m, also indicating that Co and P remineralize in tandem in the upper 300m. Below 300m, labile particulate Co:P increases approximately 3-fold (in contrast with Fe:P, which increases 12-fold), and this depth effect matches the effect in dissolved remineralization ratios. Thus, even though Fe and Co are considered hybrid metals that display both biological uptake and scavenging, there are clear differences in the behaviors of these metals. Nickel provides yet another perspective on the coupling of metals and P. Dissolved remineralization ratios in both subsurface depth ranges closely resemble surface ocean labile particles, supporting the biological coupling of Ni and P (5). However, residual labile particulate Ni:P increases 2- to 4-fold in successive depth ranges, indicating that remineralization is rather decoupled. Given that Ni seems to be associated with both organic material and opal frustules in diatoms (27), it may be that Ni and P are remineralized from particulate organic matter in tandem, but some Ni remains associated with sinking biogenic silica in the ocean.

3. Additional tools to explore and differentiate remineralization processes

The GEOTRACES program has welcomed the application of new analytical approaches that further enable us to study the cycling of metals in the ocean. Spectroscopy and quantitative imaging methods using synchrotron radiation have become more common in the past decade (28), and these allow us to analytically distinguish the behaviors of different fractions of particle assemblages. During the FeCycle II project, a GEOTRACES process study, the fate of Fe was tracked during a spring diatom bloom (29). Diatom cells from the dominant bloom species (Asterionellopsis glacialis) were collected in surface waters and from trace-metal clean sediment traps at 100m and 200m in the 48h following the decline of the bloom. Synchrotron x-ray fluorescence (SXRF) analyses of individual cells showed that constituent elements were lost from sinking cells at notably different rates (Fig. 3). Phosphorus was rapidly released from sinking cells, with mean P quotas decreasing 55% and 73% from surface values by 100m and 200m, respectively (30). However, only 25% of cellular Fe was lost from cells sinking through the upper 200m, while 61% of cellular Ni was remineralized. This supports the story told by the bulk biogeochemical data from the North Atlantic: Ni is remineralized largely to a similar degree as P, while Fe is lost more slowly from sinking biogenic material.

Application of microanalytical techniques such as SXRF can be combined with bulk approaches to further advance understanding of subsurface metal remineralization and cycling. In FeCycle II, Fe:P of sinking A. glacialis cells increased, on average, only 2.3-fold in the upper 200m, while Fe:P in bulk particulate matter increased more than 13-fold (30). This indicates that the behavior of sinking cells was not representative of the full particle assemblage. Iron and P were likely more completely decoupled in sinking fecal pellets and detrital material (which appears to have contributed significantly to the particulate Fe pool during FeCycle II; 31) than in intact sinking cells. Further application of this approach will allow us to not only distinguish between the behavior of biogenic and lithogenic fractions (Fig. 1), but potentially also between detrital particles. By considering metals such as Mn that are prone to oxidation and scavenging in the subsurface ocean (32, 33), it may also be possible to separate abiotic scavenging from net biological remineralization (Fig. 1). Additionally, 2D (and potentially 3D) mapping of elements within cells and particles also provides information about the spatial and potentially chemical associations of elements with particles (30, 34).

The GEOTRACES program is generating unprecedented data, both in terms of quality and quantity, regarding the cycling of bioactive trace metals in the ocean. Syntheses of these data, and integration of insights from novel microanalytical tools, as well as transcriptomic and proteomic approaches, are resulting in substantial advances in our understanding of metal biogeochemistry. No longer are we limited to a few painstakingly collected dissolved metal profiles. There is now painstakingly collected full-depth coverage of most ocean basins, including in many cases dissolved and particulate fractions of nearly all biogenic elements, enabling testing of early hypotheses about trace metal cycling and parameterization of these processes into next-generation ocean biogeochemical models.

Authors

Benjamin S. Twining, Daniel C. Ohnemus, Renee L. Torrie (Bigelow Laboratory for Ocean Sciences)

Acknowledgments

This work was funded by NSF grant OCE-1232814 to BST. RLT was funded by NSF REU grant 1460861 to Bigelow Laboratory for Ocean Sciences.

References

1. A. C. Redfield, in James Johnstone Memorial Volume, R. J. Daniel, Ed. (Liverpool University Press, 1934), pp. 176-192.
2. A. C. Redfield, Amer. Scientist 46, 205-221 (1958).
3. E. A. Boyle, J. M. Edmond, Nature 253, 107-109 (1975).
4. E. A. Boyle, F. Sclater, J. M. Edmond, Nature 263, 42-44 (1976).
5. K. W. Bruland, Earth Plan. Sci. Lett. 47, 176-198 (1980).
6. J. R. R. Frausto da Silva, R. J. P. Williams, The Biological Chemistry of the Elements: the inorganic chemistry of life. (Oxford University Press, Oxford, ed. 2nd, 2001), pp. 575.
7. F. M. M. Morel, R. J. M. Hudson, in Chemical Processes in Lakes, W. Stumm, Ed. (John Wiley & Sons, New York, 1985), pp. 251-281.
8. K. W. Bruland, J. R. Donat, D. A. Hutchins, Limnol. Oceanogr. 36, 1555-1577 (1991).
9. T. Y. Ho et al., J. Phycol. 39, 1145-1159 (2003).
10. W. G. Sunda, Marine Chem. 57, 169-172 (1997).
11. W. G. Sunda, S. A. Huntsman, Limnol. Oceanogr. 40, 132-137 (1995).
12. W. G. Sunda, S. A. Huntsman, Limnol. Oceanogr. 40, 1404-1417 (1995).
13. C. M. Moore et al., Nature Geosci. 6, 701-710 (2013).
14. R. Collier, J. Edmond, Prog. Oceanogr. 13, 113-199 (1984).
15. F. M. M. Morel, Geobiol. 6, 318-324 (2008).
16. M. Whitfield, D. R. Turner, in Aquatic Surface Chemistry: Chemical Processes at the Particle-Water Interface, W. Stumm, Ed. (John Wiley & Sons, Inc., 1987), pp. 457-493.
17. K. W. Bruland, M. C. Lohan, in The Oceans and Marine Geochemistry: Treatise on Geochemistry, H. Elderfield, Ed. (Elsevier, Oxford, 2003), pp. 23-47.
18. P. W. Boyd, M. J. Ellwood, Nature Geosci. 3, 675-682 (2010).
19. D. C. Ohnemus, P. J. Lam, Cycling of lithogenic marine particles in the US GEOTRACES North Atlantic Transect. Deep-Sea Res. II 116, 282-302 (2015).
20. P. W. Boyd et al., Limnol. Oceanogr. 55, 1271-1288 (2010).
21. R. D. Frew et al., Glob. Biogeochem. Cycles 20, GB1S93, doi:10.1029/2005GB002558 (2006).
22. C. H. Lamborg, K. O. Buesseler, P. J. Lam, Deep-Sea Res. II 55, 1564-1577 (2008).
23. W. J. Jenkins et al., Deep-Sea Res. II 116, 6-20 (2015).
24. M. Gledhill, K. N. Buck, Frontiers Microbiol. 3, doi: 10.3389/fmicb.2012.00069 (2012).
25. J. W. Moffett, J. Ho, Geochim. Cosmochim. Acta 60, 3415-3424 (1996).
26. A. E. Noble et al., Limnol. Oceanogr. 57, 989-1010 (2012).
27. B. S. Twining et al., Glob. Biogeochem. Cycles 26, GB4001, doi:4010.1029/2011GB004233 (2012).
28. P. J. Lam et al., Prog. Oceanogr. 133, 32-42 (2015).
29. P. W. Boyd et al., Geophys. Res. Lett. 39, doi:10.1029/2012GL053448 (2012).
30. B. S. Twining et al., Limnol. Oceanogr. 59, 689-704 (2014).
31. A. L. King et al., Biogeosci. 9, 667-687 (2012).
32. J. P. Cowen, K. W. Bruland, Deep-Sea Res. 32, 253-272 (1985).
33. D. C. Ohnemus et al., Limnol. Oceanogr. In press (2016).
34. J. Nuester, S. Vogt, B. S. Twining, J. Phycol. 48, 626-634 (2012). 35. B. S. Twining et al., Prog. Oceanogr. 137, 261-283 (2015).
36. M. Hatta et al., Deep-Sea Res. II 2015, 117-129 (2015).
37. S. Roshan, J. Wu, Glob. Biogeochem. Cycles 29, 2082-2094 (2015).
38. E. Mawji et al. Marine Chem. 177, 1-8 (2015).

Marine particles: Distribution, composition, and role in scavenging of TEIs

Posted by mmaheigan

· Sunday, July 3rd, 2016

GEOTRACES and particles in the ocean

GEOTRACES is an international program to study the global marine biogeochemical cycles of trace elements and their isotopes (TEIs). The program’s guiding mission is to “identify processes and quantify fluxes that control the distributions of key TEIs in the ocean” (1).

Particles represent a key parameter for the GEOTRACES program because of their role as sources, sinks, and in the internal cycling of so many TEIs (1, 2). Particles in the ocean fall into two classes: 1. Those that have sources external to the system such as lithogenic material carried by atmospheric transport, river, or lateral transport from continental margin sediments; and 2. those that are produced internally in the system, primarily by biological production, but also by authigenic mineral precipitation (2).

External particle sources such as mineral dust deposition and sediment resuspension act as sources of dissolved TEIs when they partially dissolve in seawater. Conversely, dissolved TEIs are removed by active biological uptake or passive adsorption onto particles surfaces, followed by particle removal by aggregation and sinking. Indeed, the biological and abiotic interactions of dissolved TEIs with particles determine the residence time of a dissolved TEI.

In most open ocean basins away from ocean floor boundaries, external sources of particles are dwarfed by the much greater biological production and destruction of particles. Particle cycling in most open ocean basins is thus dominated by the biological pump, the processes by which suspended particles are produced by photosynthesis in the euphotic zone at the surface, and are then abiotically or biologically aggregated into larger particles that can sink into the abyss (3).

As particulate organic carbon (POC) cycles through processes such as aggregation, disaggregation, remineralization, and sinking (collectively referred to here as particle dynamics), other particle phases are swept along for the ride, including other major components such as biologically precipitated minerals (especially CaCO₃and opal), as well as lithogenic and authigenic particles, and scavenged TEIs adsorbed to the surfaces of other particles (Fig. 1).

In this article, I will briefly review the role of particle composition on the scavenging of TEIs.

Scavenging: A two-step removal process

Most adsorption of TEIs likely occurs onto small, suspended particles, which are usually more abundant, have more available surface area, and have a longer residence time in the water column than large, sinking particles. For TEIs to be removed from the water column, the suspended particles must then be aggregated into larger, sinking particles. There are thus two distinct steps for the removal of a dissolved TEI by scavenging: 1) adsorption onto suspended particle surfaces, followed by 2) removal via the aggregation of suspended particles onto larger particles that sink out of the water column. Fig. 1 shows a very simple schematic illustrating these basic processes. The adsorption step is governed by the affinity of a TEI for a particular particle surface, and the removal step is governed by the particle dynamics that package suspended particles into large, sinking aggregates, and are the focus of studies of the biological pump. The removal of TEIs by scavenging thus intimately links one of OCB’s scientific goals, the understanding of the biological carbon pump, to GEOTRACES’s mission to identify processes and quantify fluxes that control the distributions of key TEIs in the ocean.

Particle concentration and composition: horizontal and vertical variations

Particles collected in the ocean are a heterogeneous mixture of biogenic, lithogenic, and authigenic (precipitated in-situ) components. The relative proportions of these different components vary geographically and with depth. Fig. 2 shows the distribution of total particle concentration from GA03, the U.S. GEOTRACES North Atlantic Zonal Transect cruise in 2010/2011, as well as the changing composition of small (<51 mm) particles at three stations along the transect (4). Particle concentrations are highest at the surface and at the margins, where biological production is highest. It is clear that particulate organic matter (POM) dominates particle composition in the upper 100 m, making up more than 70% of the suspended particle mass at all three stations. The balance in the upper 100 m is mostly made of other biogenic components such as CaCO₃and opal, with a small contribution from lithogenic particles directly under the Saharan dust plume. At all stations, the inorganic components (everything except for POM) become relatively more important with depth as POM is remineralized. In the eastern half of the basin, lithogenic particles make up the largest fraction of particle mass, accounting for >50% of particle mass below 1500 m. In the western half of the basin, further from the Saharan dust source, lithogenic particles are not as important, and CaCO₃makes up the largest fraction (~50%) of particle mass between 500 – 3000 m. A special case is found in a station over the Mid-Atlantic Ridge, where iron oxyhydroxides from the hydrothermal plume make up ~50% of the particle mass. Iron and manganese oxyhydroxides are rarely dominant components of particle mass, except in special situations such as hydrothermal plumes, but may exert a particularly large influence on TEI adsorption (5, 6 ).

Studies suggest that particle composition may affect both the affinity of dissolved TEIs for adsorbing onto particle surfaces (2), and the vertical flux of particles from the water column (7-9). Horizontal and vertical changes in particle composition thus allow us to test hypotheses of the importance of particle composition on both steps in the scavenging of TEIs.

Effect of particle composition on adsorption of TEIs

The affinity of TEIs to particles has typically been characterized by a partition coefficient, Kd, which is calculated empirically as:

Prior to the GEOTRACES program, the effect of particle composition on TEI adsorption affinity had been studied in the field using sediments and sinking particles collected in sediment traps. The affinity of trace metals to marine sediments of different compositions varied: Some trace metals (Cs, Be, Sn, and Fe) had a higher affinity to sediments dominated by aluminosilicate clay minerals, and others (Ba, Cd, Zn, Mn, and Co) had a higher affinity to sediments enriched in Mn oxyhydroxides (10). In the water column, correlations between the partition coefficient of ²³⁰Th and particle composition in sediment trap particles from around the world have variously implied that the scavenging efficiency of ²³⁰Th is controlled by CaCO₃(11, 12), lithogenic material (13, 14), and/or Mn oxyhydroxides (15). Studies that span strong opal gradients across the Polar Front in the Southern Ocean show higher partition coefficients for ²³¹Pa scavenging in areas of high opal content (11, 16). ²³¹Pa is generally not as particle-reactive as ²³⁰Th in the open ocean, but is often removed with equal efficiency as ²³⁰Th in near-margin areas (e.g., 17), presumably because opal is more important in margin settings. The Arctic, on the other hand, displays the opposite ²³⁰Th/²³¹Pa removal signal, with ²³¹Pa removal less efficient relative to ²³⁰Th at the margins compared to the open ocean (18).

Since TEIs adsorb primarily onto suspended particles rather than sinking particles, studying the correlations between partition coefficients and suspended particles may resolve some of the discrepancies observed in the sediment trap studies (c.f., 2).

The GEOTRACES GA03 North Atlantic Zonal Transect has provided the first opportunity to investigate the correlation between partition coefficients of various TEIs and the particle composition of suspended particles in the ocean. Thus far, this has been done for ²³⁰Th and ²³¹Pa partition coefficients, with Mn and Fe oxyhydroxides emerging as key controlling phases and opal having no controlling effect (5). The North Atlantic is very opal-poor (Fig. 2), so particles collected from more diatom-rich regions are needed to examine the potential of opal as a controlling phase. Other studies are underway to study the particle affinities of Hg (19), Po (20), and Pb (6) on this same North Atlantic transect. Subsequent U.S. GEOTRACES sections (GP16—Eastern Tropical South Pacific Zonal Transect and GN01—Western Arctic) will also have full ocean depth size-fractionated particle concentration and composition, allowing us to examine samples from different biogeochemical provinces, and hopefully expanding the range of particle compositions.

TEIs as tracers of scavenging rates and particle dynamics

The unprecedented data sets from GEOTRACES are also allowing us to estimate adsorption and desorption rate constants (Fig. 1) from inverse modeling of the observations of dissolved and particulate TEIs and particle concentrations (19, 21, 22). This gives us a kinetic view of the scavenging process to complement the empirically-derived partition coefficients, which are often viewed as representing equilibrium constants. Applying inverse modeling approaches to observations of the distributions of size-fractionated particles and particulate TEIs can also allow us to estimate rates of particle remineralization, aggregation, disaggregation, and sinking (21, 23). This approach requires only that a conceptual model relating the suspended and sinking particle size fractions be applied to observations of particle mass and particulate TEIs, and does not require knowledge about which specific physical or biological processes are responsible for particle transformations. For example, Fig. 1 illustrates a simple conceptual model in which a pool of suspended particles can be lost to the dissolved phase through remineralization, or by aggregation into sinking particles; conversely, sinking particles can sink, or can be disaggregated back into suspended particles. By assuming that particulate TEIs are simply part of the overall particle pool (e.g., a coating on organic particles in the case of radiogenic TEIs such as ²³⁰Th or as part of a lithogenic particle in the case of a TEI such as Ti) and thus are subject to the same rates of particle transformations as the major phases such as POC, we can apply the same conceptual model to observations of particle mass and to observations of particulate TEI to better constrain the rates of these transformations (23). As some of these rates such as aggregation and disaggregation are notoriously difficult to measure directly, these inverse approaches offer a way forward to quantify these important processes.

Particle composition and the biological pump

In addition to its effect on scavenging efficiency, particle composition has also been implicated as an important factor in the strength and efficiency of the biological pump. Several meta-analyses of global deep (>1000 m) sediment trap data showed strong correlations between POC flux and mineral flux (7-9), leading to the development of the “ballast hypothesis.” The mechanisms to explain the correlations, which are still being debated (24, 25), range from mineral protection of POC (8), mineral contribution to particle excess density (7), scavenging of mineral particles by POC (26), and minerals as proxies for particle packaging, POC lability, and ecosystem structure (9, 27-29).

Although the GA03 dataset is based on size-fractionated particle samples collected by in-situ filtration rather than sinking particles collected by sediment traps, we can nonetheless examine whether there is a correlation between POC and ballast minerals in small or large particle size fractions. We found that POC concentration in large (>51 mm) particles was not consistently correlated with any of the potential ballast minerals CaCO3, opal, and lithogenic particles (4).The lack of strong correlations within this regional dataset is consistent with the idea that ballast mineral correlations with POC may only emerge in global datasets that combine different biogeochemical provinces (25).

Outlook

The GEOTRACES program is not only rapidly expanding global observations of dissolved TEIs, but it is also the latest major program to systematically sample particle distributions since JGOFS and GEOSECS (2). These particle measurements are not only helping us understand the processes controlling TEI distributions, but the TEI measurements can also be used as tracers for quantifying key processes of particle cycling. Both GEOTRACES and OCB can benefit from the insights gained in each program.

Author

Phoebe J. Lam (Department of Ocean Sciences, University of California, Santa Cruz)

References

1. GEOTRACES, Scientific Committee on Oceanic Research, Ed. (Baltimore, Maryland, 2006).
2. C. Jeandel et al., Progress in Oceanography 133, 6 (4//, 2015).
3. C. L. De La Rocha, in Treatise on Geochemistry, H. Elderfield, K. K. Turekian, Eds. (Elsevier, 2003), vol. 6: The Oceans and Marine Geochemistry, pp. 83-111.
4. P. J. Lam et al., Deep Sea Research Part II: Topical Studies in Oceanography 116, 303 (6//, 2015).
5. C. T. Hayes et al., Marine Chemistry 170, 49 (3/20/, 2015).
6. E. A. Boyle et al., paper presented at the 2016 Ocean Sciences Meeting, New Orleans, LA, USA, 2016.
7. C. Klaas, D. E. Archer, Global Biogeochemical Cycles 16, 1116 (Dec 5, 2002).
8. R. A. Armstrong et al., Deep-Sea Research Part II-Topical Studies in Oceanography 49, 219 (2002).
9. R. François et al., Global Biogeochemical Cycles 16, (Oct-Nov, 2002).
10 . L. S. Balistrieri, J. W. Murray, Geochimica Et Cosmochimica Acta 48, 921 (1984).
11. Z. Chase et al., Earth and Planetary Science Letters 204, 215 (Nov 30, 2002).
12. Z. Chase et al., Deep-Sea Research Part II-Topical Studies in Oceanography 50, 739 (2003).
13. S. D. Luo, T. L. Ku, Earth and Planetary Science Letters 220, 201 (Mar, 2004).
14. M. Roy-Barman et al., Earth and Planetary Science Letters 286, 526 (2009).
15. M. Roy-Barman et al., Earth and Planetary Science Letters 240, 681 (2005).
16. H. J. Walter et al., Earth and Planetary Science Letters 149, 85 (1997).
17. R. F. Anderson, M. P. Bacon, P. G. Brewer, Earth and Planetary Science Letters 66, 73 (1983).
18. H. N. Edmonds et al., Earth and Planetary Science Letters 227, 155 (Oct, 2004).
19. C. H. Lamborg et al., Philos T R Soc A, (accepted).
20. Y. Tang et al., paper presented at the 2016 Ocean Sciences Meeting, New Orleans, LA, USA, 2016.
21. O. Marchal, P. J. Lam, Geochimica Et Cosmochimica Acta 90, 126 (2012).
22. P. Lerner et al., Deep Sea Research Part I: Oceanographic Research Papers 113, 57 (7//, 2016).
23. P. J. Lam, O. Marchal, Annual Review of Marine Science 7, 159 (2015).

24. P. Boyd, T. Trull, Progress in Oceanography 72, 276 (2007).
25. J. D. Wilson, S. Barker et al., Global Biogeochemical Cycles 26, GB4011 (2012).
26. U. Passow, Geochemistry Geophysics Geosystems 5, (Apr 6, 2004).
27. P. J. Lam et al., Global Biogeochem. Cycles 25, GB3009 (2011).
28. S. A. Henson et al., Global Biogeochem. Cycles 26, GB1028 (2012).
29. S. Z. Rosengard et al., Biogeosciences 12, 3953 (2015).
30. D. C. Ohnemus, P. J. Lam, Deep Sea Research Part II: Topical Studies in Oceanography 116, 283 (6//, 2015).

Using GEOTRACES data to appraise iron cycling as represented within global ocean models

Posted by mmaheigan

· Thursday, June 23rd, 2016

We rely on global ocean models to predict how climate change might affect the evolution of ocean productivity, acidification, and deoxygenation (1). Such platforms are also used to test hypotheses regarding the controls on ocean biogeochemical cycling and to understand past change (both on historical and geologic timescales). Ocean biogeochemistry models began with relatively simple formulations of a carbon export flux that involved restoring to observed phosphate distributions, but have more recently evolved into complex multi-element representations of the ocean. In line with our understanding that the trace micronutrient iron (Fe) limits phytoplankton productivity over large areas of the world ocean (2), most global models that aim to project future change also explicitly represent the Fe cycle.

Datasets regarding the major limiting nutrients (nitrate, phosphate, and silicate) have been available as gridded ‘climatologies’ since the early 1990s (3). This has greatly facilitated the development and evaluation of modelled distributions over the past two decades. However, over this period there has been little comprehensive evaluation of how different models represent the ocean Fe cycle. Over recent years, there has been a marked increase in the availability of iron measurements in the ocean (4), largely driven by the international GEOTRACES effort to conduct full depth, basin-scale surveys. This led us to initiate the first-ever attempt to critically compare a range of global ocean iron models against the largest global datasets, as well as against the newly emerging ocean section data (5).

The Fe Model Intercomparison Project (FeMIP) sought to be as inclusive as possible in this first step and therefore did not seek to standardize the underlying ocean circulation or external inputs. Instead, we simply asked each of the thirteen models to provide their best representation of dissolved iron in three dimensions at monthly resolution. We then compared these models against each other, a global iron database of over 20,000 observations and against five unique basin-scale sections from the GEOTRACES intermediate data product 2014 (IDP2014) (6).

Firstly, it is apparent that even when the underlying iron cycles of the different models are evaluated, a substantial degree of inter-model discord exists. The total iron input varies from around 2 to 200 Gmol yr-1 across the thirteen models (Fig. 1a). Even for ‘well known’ sources like atmospheric deposition, the inter-model variability is around an order of magnitude. On the other hand, the average concentrations of dissolved Fe between the models is much less variable and ranges from 0.35 to 0.81 nmol L-1 (Fig. 1b), or an average of 0.58±0.14 nmol L-1. This apparent constancy reflects an initial view of the ocean Fe cycle in which interior Fe concentrations were held at a quasi-constant value of 0.6 nmol L-1 assuming a constant concentration of Fe-binding ligands (7). Thus the FeMIP models are balancing widely varying Fe input fluxes against relatively constant overall Fe concentrations by tuning the Fe scavenging rate, which is a crucial but poorly known parameter. This results in residence times for Fe that range from <5 to >500 years across the FeMIP models (Fig 1c). This difference is important, as it represents substantial inter-model deviation concerning the timescales over which the different models respond to a perturbation in Fe supply.

When compared statistically against the global dataset, similar levels of variability arise. Some models display correlation coefficients of >0.5, whereas others are slightly anti-correlated. When the FeMIP models are compared against the five GEOTRACES sections, it becomes apparent that those models that represent the newly emerging features of the iron cycle perform much better. For instance, having Fe scavenging rates that vary in space and time, including variable Fe:carbon (C) stoichiometry, multiple Fe sources, and representing ligand concentrations in a dynamic manner, all act to improve the representation of different observed features in the models. Importantly, the IDP2014 provided the opportunity to demonstrate that the issues at hand were specific to Fe, since the models could represent the observed distributions of major nutrients with a much greater degree of skill (5).

The next stage of FeMIP will be a deeper comparison of the processes themselves. Of particular interest is whether GEOTRACES datasets can provide broader assessments of the rates of Fe scavenging – e.g., using other particle-reactive, non-biological tracers such as thorium (8, 9). Equally, the emerging database of GEOTRACES process studies provides an important opportunity to appraise the way different models represent biological iron cycling and in particular, the often-observed importance of recycled and remineralized sources of Fe (10, 11). Finally, the new GEOTRACES intermediate data product 2017 will also facilitate further evaluation of models, providing new section data from the Atlantic, Pacific and Arctic Oceans.

Author

Alessandro Tagliabue (Dept. of Earth, Ocean and Ecological Sciences, School of Environmental Sciences, University of Liverpool)

References

1. L. Bopp et al., Biogeosci. 10(10), 6225-6245, doi:10.5194/bg-10-6225-2013 (2013).
2. C. M. Moore et al., Nature Geosci., doi:10.1038/ngeo1765 (2013).
3. S. Levitus et al., Prog. Oceanogr. 31(3), 245-273, doi:10.1016/0079-6611(93)90003-v (1993).
4. A. Tagliabue et al., Biogeosci. 9(6), 2333-2349, doi:10.5194/bg-9-2333-2012 (2012).
5. A. Tagliabue, A. et al., Glob. Biogeochem. Cycles, doi:10.1002/2015gb005289 (2016).
6. E. Mawji et al., Marine Chem. 177, 1-8. doi:10.1016/j. marchem.2015.04.005 (2015).
7. K. S. Johnson, R. M. Gordon, K. H. Coale, Marine Chem. 57(3-4), 137-161, doi:10.1016/s0304-4203(97)00043-1 (1997).
8. C. T. Hayes et al., Geochim. Cosmochim. Acta 169, 1-16, doi:10.1016/j.gca.2015.07.019 (2015).
9. N. Rogan et al., Geophys. Res. Lett. 43(6), 2732-2740. doi:10.1002/2016gl067905 (2016).
10. P. W. Boyd et al., Glob. Biogeochem. Cycles 29(7), 1028-1043, doi:10.1002/2014gb005014 (2015).
11. R. F. Strzepek et al., Glob. Biogeochem. Cycles 19(4), GB4S26, doi:10.1029/2005gb002490 (2005).

Exploring molecular methods for assessing trace element bioavailability in phytoplankton

Posted by mmaheigan

· Saturday, June 4th, 2016

This summer OCB and GEOTRACES are co-sponsoring a synthesis workshop on the biogeochemical cycling of trace elements in the ocean. The overall goal of the workshop is to bring together expertise from across the field of oceanography to take advantage of the growing datasets of trace elements in the ocean and explore biological-chemical-physical underpinnings of trace element cycling within the ocean. One of the three main themes that will be addressed at the workshop is “biological uptake and trace element bioavailability.” Part of this theme will include a discussion of how molecular markers have been used to address questions of trace element bioavailability in the past and the exciting future for continued efforts in this area given the growing molecular toolkit.

As a starting point to open the door to this broad discussion topic, let us turn our attention to how molecular methods have been used to evaluate iron (Fe) bioavailability to certain phytoplankton groups. Marine phytoplankton play a key role in the global carbon cycle by performing a significant fraction of global primary production (1). Since John Martin’s groundbreaking work introduced the concept that Fe is a limiting nutrient for phytoplankton growth (2), numerous studies have shown that insufficient Fe limits primary productivity in the major high-nitrate, low-chlorophyll (HNLC) regions of the ocean, and that Fe availability can regulate phytoplankton processes in many other oceanic settings (3, 4). Fe is a critical micronutrient required by phytoplankton for a Exploring molecular methods for assessing trace element bioavailability in phytoplankton Dreux Chappell (Old Dominion Univ.) multitude of cellular tasks, including electron transfer in photosynthesis and respiration, as well as macronutrient acquisition and assimilation (5). Studies of Fe limitation in marine environments beyond the traditional HNLC regions suggest that Fe limitation may be driven not simply by low Fe concentrations, but a combination of low Fe bioavailability coupled with high macronutrient supply (e.g., 6, 7). In an admittedly overly simplified summary, the different pools of Fe present in the ocean are defined based on filter pore-size cut-offs and chemical interactions with organic compounds (3). Knowledge of the distribution of these various forms of Fe in the oceans has increased dramatically in recent years thanks to coordinated sampling efforts like GEOTRACES.

While the bioavailability of Fe to phytoplankton is believed to be different for the various Fe pools, there is still no clear consensus as to which, if any, pools of Fe are always bioavailable and which, if any, are completely unavailable (8, 9). Additionally, it is known that all phytoplankton are not created equal with respect to their ability to persist under low-Fe conditions (10) and access different Fe pools (9). These factors make it difficult to accurately predict how changing concentrations of the different pools of Fe may impact phytoplankton productivity in a changing ocean. One way to address questions about biological availability is to use phytoplankton themselves as in situ indicators of Fe stress. An approach that appears promising in addressing these questions of bioavailability of Fe to individual phytoplankton species is the development of species-specific molecular markers of Fe limitation such as those that have been developed for the oceanic diatom Thalassiosira oceanica (11), and for the two main groups of the nitrogen-fixing cyanobacterial genus Trichodesmium (12). These assays in particular follow the expression of genes that encode flavodoxin, a non-Fe-containing protein that phytoplankton are known to substitute for the Fe-containing ferrodoxin protein to maintain photosynthetic electron transport under Fe-limiting conditions (13). They focus on gene expression analysis rather than protein analysis, as many phytoplankton have multiple genes that encode for flavodoxin proteins and, at least in diatoms, not all gene copies are sensitive to Fe (14).

The Trichodesmium assays were calibrated using cultures grown with six different concentrations of Fe in the media. In laboratory cultivation experiments, gene expression was shown to be inversely proportional to Fe present in the media and expression was downregulated when Fe was fed back to Fe-limited cultures (12). The Trichodesmium thiebautii assay was further used to evaluate field populations from open ocean samples collected globally (15), providing insights into Fe bioavailability to wild populations of Trichodesmium. Overall, there was an inverse correlation between gene expression and total dissolved Fe concentrations (Fig. 1). Comparing the results with the laboratory calibration led to the conclusion that most of the dissolved Fe, including organically bound Fe, was available to T. thiebautii. An intriguing result from this study was that one sample collected in the plume of the Amazon River had significantly higher gene expression than would be expected based on the measured dissolved Fe at that site (Fig. 1; 15). These findings suggest that there is a fraction of the dissolved Fe in the Amazon River plume that is not bioavailable to T. thiebautii, the clade of Trichodesmium that is more abundant and active in the open ocean (15, 16).

The T. oceanica assay is also highly sensitive to Fe with high gene expression in cultures that were Fe-limited, a rapid reduction of gene expression following an Fe pulse to Fe-limited cultures, and no induction of expression by macronutrient limitation (11). Using this method on field samples from the northeast Pacific Ocean, T. oceanica flavodoxin expression was found to be highest in samples with low measured dissolve Fe and vice versa (11). Two notable exceptions to this trend were samples collected along the shallow shelf of Haida Gwaii, stations 26 and 27, which showed anomalously high expression of both genes despite high measured dissolved Fe (Fig. 2), suggesting that something about the dissolved Fe in these shallow coastal stations made it unavailable to the T. oceanica in these waters. It should be noted that T. oceanica is an oceanic diatom species, so both sets of findings suggest that there is a fraction of dissolved Fe from a terrestrially influenced water sample that was not bioavailable to a species of oceanic phytoplankton.

While these two datasets are intriguing, they are limited in scope and admittedly raise more questions than they answer. There are a variety of questions that stem from these results, including what, if anything, is different about the dissolved Fe at these stations? Was this coastal/terrestrially sourced Fe unavailable because only oceanic phytoplankton, which rarely encounter dissolved Fe from terrestrial sources, were queried? Did coastal phytoplankton simply outcompete the oceanic phytoplankton for access to the dissolved Fe perhaps because they have different or more efficient Fe uptake mechanisms? Would an assay targeting coastal phytoplankton reveal the same results? We know that coastal diatoms have significantly higher Fe requirements (10). What if coastal and oceanic phytoplankton also differ in their abilities to access Fe from different sources? A recent study in the Sea of Okhotsk yielded a significant correlation between a bulk diatom community indicator of Fe stress and dissolved Fe with increasing distance from the mouth of the Amur River, suggesting that the riverine Fe was bioavailable to the community that was dominated by coastal diatoms (17).

A number of recent advances in molecular microbial oceanography are making it increasingly possible to start answering these types of questions on a broader scale. A major sequencing effort that was completed in 2014, the Marine Microbial Eukaryote Transcriptome Sequencing Project (MMETSP), added a wealth of data on the functional genetic diversity of marine microeukaryotes (18). The project resulted in over 650 publicly available transcriptomes from over 250 genera of marine microeukaryotes. Even before all the transcriptomes from the project were released, data mining of a limited portion of the MMETSP publically available dataset yielded valuable information as to the varied Fe management strategies utilized by different species of marine diatoms, creating a list of additional potential molecular markers for evaluating Fe nutritional status of diatoms in the field (19). Another data mining technique that has proven useful in identifying coordinated transcriptional responses in diatoms has involved the application of clustering algorithms to evaluate publically available microarray data for two of the more commonly studied diatom species, Phaeodactylum tricornutum and Thalassiosira pseudonana, grown under a wide variety of conditions (20). New insights into the diatom Fe stress response and Fe uptake mechanisms have also been gained through combining physiological experiments with genetic knockdowns of previously uncharacterized Fe responsive genes (21). These new molecular advances are providing a suite of new potential targets for querying the physiological status of phytoplankton present throughout the global ocean. At the OCB/GEOTRACES synthesis workshop on the biogeochemical cycling of trace elements in the ocean, the conversation will include a discussion of how to effectively combine these new analyses with the growing datasets of bioactive trace elements to answer questions regarding the biological availability of different trace element pools. While this mini-review has focused on Fe, additional trace elements will be discussed at the meeting in August.

Author

Dreux Chappell (Old Dominion University)

References

1. C. B. Field et al., Science 281, 237-240 (1998).
2. J. H. Martin, S. Fitzwater, Nature 331, 341-343 (1988).
3. P. W. Boyd, M. J. Ellwood, Nature Geoscience 3, 675-682 (2010).
4. P. W. Boyd et al., Science 315, 612-617 (2007).
5. F. M. M. Morel, N. M. Price, Science 300, 944-947 (2003).
6. K. W. Bruland, E. L. Rue, G. J. Smith, Limnol. Oceanogr. 46, 1661-1674 (2001).
7. A. L. King, K. N. Buck, K. A. Barbeau, Marine Chem. 128-129, 1-12 (2012).
8. E. Breitbarth et al., Biogeosci. 7, 1075-1097 (2010).
9. Y. Shaked, H. Lis, Frontiers Microbiol. 3 (2012).
10. W. G. Sunda, S. A. Huntsman, Marine Chem. 50, 189-206 (1995).
11. P. D. Chappell et al., ISME Journal 9, 592-602 (2015).
12. P. D. Chappell, E. A. Webb, Envir. Microbiol. 12, 13-27 (2010).
13. J. LaRoche et al., J. Phycol. 31, 520-530 (1995).
14. L. P. Whitney et al., Frontiers Microbiol. 2 (2011).
15. P. D. Chappell et al., ISME Journal 6, 1728-1739 (2012).
16. A. M. Hynes, Massachusetts Institute of Technology (MIT) (2009).
17. K. Suzuki et al., Biogeosci. 11, 2503-2517 (2014).
18. P. J. Keeling et al., Plos Biol. 12 (2014).
19. R. D. Groussman, M. S. Parker, E. V. Armbrust, PloS One 10 (2015).
20. J. Ashworth et al., Marine Genom. 26, 21-28 (2016).
21. J. Morrissey et al., Curr. Biol. 25, 364-371 (2015).

Biogeochemical cycling of organic iron-binding ligands: Insights from GEOTRACES data in the Atlantic Ocean

Posted by mmaheigan

· Friday, June 3rd, 2016

Iron is a limiting nutrient for phytoplankton in nearly half of the global surface ocean, and much attention has been paid to the biogeochemical cycling of iron in seawater since suitable trace metal-clean sampling and analysis procedures were developed (1). The organic complexation of dissolved iron, in particular, has emerged as an inherent feature of iron chemistry in the oceans (2-5), and iron speciation measurements are increasingly incorporated into field studies (Fig. 1). The integration of organic iron-binding ligands into biogeochemical models improves their ability to reproduce global dissolved iron distributions (6), and changes in ligand concentrations in the Southern Ocean can have a more pronounced impact on atmospheric CO₂in model studies than changes in iron supply terms from hydrothermal and dust sources (7). Indeed, field distribution measurements, targeted experimental studies and modeling efforts over the last twenty years have left little doubt that organic ligands are a critical factor in the global biogeochemical cycling of iron. Here we highlight some of the features of iron-binding ligand distributions in the Atlantic from the unprecedented basin-scale datasets coming out of the GEOTRACES program and attempt to elucidate some of the sources and sinks of iron-binding organic ligands in the oceans.

Dissolved iron speciation

The speciation of dissolved iron (Fe), which describes the chemical forms or species of iron in a filtered (typically <0.2 μm) sample, includes both inorganic (Fe´) and organic (FeL) components. In the oceans, iron speciation studies consistently report that nearly all dissolved iron (>99.9%) is organically complexed (8). This is perhaps not surprising given that the inorganic speciation of iron in oxygenated seawater is dominated by hydrolysis reactions leading to iron precipitation and very low inorganic iron solubility (~0.1 nmol L^-1) under most ocean conditions (9).

Studies of organic iron speciation in seawater use an electrochemical technique, competitive ligand exchange-adsorptive cathodic stripping voltammetry (CLE-ACSV), to determine the concentrations and conditional stability constants of iron-binding organic ligands. This is accomplished by titrating iron-binding ligands in a sample with additions of iron and competing against any natural iron-ligand complexes with an added well-characterized ‘competitive’ ligand, which forms an electroactive complex with iron that can be measured at the surface of a hanging mercury drop electrode. Titrations often depict no measurable iron bound to the competitive ligand for the first few iron additions, reflecting the presence of excess strong iron-binding ligands in most seawater samples (8). These ligands are described as ligand classes, L₁, L₂, L₃, L₄, defined by the conditional stability constants determined by CLE-ACSV, with L₁and L₂-type ligands the strongest iron-binding organic ligands

Iron speciation in the Atlantic: Observations from recent GEOTRACES efforts

The iron speciation datasets emerging from the GEOTRACES program allow a first look at basin-scale distributions of iron-binding ligands in the oceans. Iron speciation datasets from GEOTRACES Sections GA02 (10) and GA03 (11) from the Dutch and U.S. GEOTRACES programs, respectively, document the ubiquitous nature of iron-binding ligands in the Atlantic basin. In particular, both datasets evince the presence of strong, L₁-type ligands throughout the water column and no discernible trend with depth in the conditional stability constants for these ligands (10, 11). These observations support the emerging picture from many other field studies of a strong iron-binding ligand pool that is not necessarily restricted to the surface ocean or euphotic zone (Fig. 2) (8).

Elevated dissolved iron and aluminum concentrations in surface waters across the GA03 zonal section, particularly near the center of the basin, demonstrate the widespread contribution of dust deposition to the iron inventory in the Atlantic (12). Iron isotope studies indicate that 71-87% of the dissolved iron along the entire GEOTRACES GA03 section was attributable to dust (13). Water-soluble organic matter characterized from the surfaces of aerosols collected on GA03 exhibited structural differences between aerosol sources that were consistent with their iron solubilities (14), and organic complexation of some of the leached iron was observed in seawater leaches of these aerosols (15). Recent studies using model ligands highlight the particular importance of stronger iron-binding ligands in the stabilization of iron leached from natural aerosols (16).

In the Atlantic GEOTRACES sections, the highest concentrations of ligands in excess of dissolved iron ([L]-[Fe], or L´) were often measured at the surface, where dissolved iron concentrations were low (10, 11). The overall complexation capacity for iron, which is a function of both ligand concentration and the conditional stability constant, also tended to be high in the upper water column. Antarctic Intermediate Water (AAIW) stands out in both datasets as exhibiting higher complexation capacity for dissolved iron than the surrounding water masses (10, 11). These waters originate from highly productive surface waters, and elevated ligand concentrations subducted with these water masses may be the result of higher strong iron-binding ligand concentrations commonly observed in and around chlorophyll maxima and in incubation experiments of iron-stressed diatom communities (see (8) and references therein). In a compilation of three iron speciation datasets that extend from the Arctic (17) down through the Western Atlantic (10) and into the Antarctic (18), higher ligand concentrations were reported at high latitudes relative to low latitudes, with the strongest (highest conditional stability constants) excess ligands measured in the Antarctic (10), and larger excesses of weaker ligands in the Arctic (10, 17).

Excess ligand concentrations in the Atlantic usually decreased with depth as dissolved iron concentrations increased, consistent with saturation of excess ligands with iron (10, 11). A north-south trend of decreasing ligands and excess ligands was reported in the GA02 Western Atlantic meridional section, which was clearly depicted in the samples collected from the North Atlantic Deep Water (NADW) along the section (10). In the GA03 zonal section of the North Atlantic (11), excess ligand concentrations in the water column were on the high end of the two datasets, consistent with the northern end of the GA02 meridional section (10). If anything, organic matter remineralization appeared to be a source of weaker L₃-type iron-binding ligands in the GA03 zonal section dataset (11).

Excess ligand concentrations in the North Atlantic exhibited local minima in the heart of the oxygen minimum zone west of Mauritania (11), possibly due to scavenging of ligand complexes on sinking particles (19), or elevated reduced iron(II) concentrations complexing the excess ligands (20). It is unclear how much of the iron-binding ligand pool measured by CLE-ACSV may also bind iron(II), or what chemical form of iron(II) is present in these samples, though some may be biogenic (21). One of the most pronounced features in the iron(II) data from GA03 is the exceedingly high iron(II) concentrations in the TAG hydrothermal plume samples, where elevated dissolved iron was ~80% colloidal-sized (0.02-0.2 μm size fraction) iron(II) species (20, 22). These iron(II) colloids are likely pyrite nanoparticles (23), which themselves may be stabilized by organic matter (24).

In the TAG plume samples collected along GA03, excess ligands were at a minimum in the highest iron samples of the plume, but the conditional stability constants of the excess ligands that were detected in these samples were among the highest in the dataset, leading to an elevated complexation capacity for iron around the vent. It is likely that some of the elevated dissolved iron in these samples was not exchangeable with the added competitive ligand during the voltammetry measurements, which would lead to an overestimation of ligand parameters (8). It is also possible that some of this observed increase in complexation capacity in the plume reflects a microbial response to the iron-enriched plume (25). Several studies have now reported varying degrees of organic complexation of iron in hydrothermal vent plumes (10, 11, 26, 27), and while the cycling of ligands in these systems remains unclear, organic stabilization must be a key factor in determining the chemical speciation and transport of iron from these vents into the deep sea (28-30).

The increasingly rich database of iron-binding ligand distributions from individual field studies and the GEOTRACES program show the widespread organic complexation of iron in the oceans. These datasets also display the inherent complexity of ligand cycling, since ligands are at the interface between the dynamic biogeochemical cycles of both trace metals and organic matter in seawater.

Identity of iron-binding ligands

Electrochemistry (CLE-ACSV) measures ligand concentrations and conditional stability constants from a combination of titration and competition, and is the basis of most of our insights to date into the sources, sinks, and cycling of organic iron-binding ligands in the oceans. Mass spectrometry-based techniques, on the other hand, are increasingly being employed to identify the chemical structures of natural iron-ligand complexes in the oceans. Results from both of these approaches appear to be converging on similar descriptions of the iron-binding ligand pool, one which comprises a mixture of defined biomolecules with high affinities for iron (e.g., siderophores, heme) and weaker iron-binding, ill-defined compounds with high chemical heterogeneity (e.g., humic substances, polysaccharides) (Fig. 3; (8)).

Siderophores are small iron-binding ligands widely produced by bacteria to acquire iron from the environment (31), including the marine environment (32, 33). Model siderophores are typically, though not exclusively, characterized as among the strongest L1-type ligands measured in CLE-ACSV (3, 8). Iron complexed by these discrete biomolecules can usually be chromatographically resolved, although isolation of these compounds from seawater is notoriously difficult. Ferrioxamines and amphibactins have been the most widely reported from the water column (Figure 3; (34-36)) and shipboard incubations (37), though marine bacteria cultures have produced a broader suite (32). The limited diversity and only picomolar concentrations of siderophores extracted from seawater compared to marine bacteria cultures likely reflects limitations in sampling, the extraction procedures available (8, 32), and detection of certain siderophore functional groups that are preferentially photodegraded in surface waters (38). Importantly, siderophores have been observed to persist throughout the water column below the euphotic zone (36) and are not restricted to low-iron waters (34). Additional discrete biomolecules, like heme or intracellular iron storage proteins, are also expected to contribute to the strong iron-binding ligand pool in seawater, though they may be prone to particle adsorption and aggregation processes, making them more likely to be found in the colloidal and particulate phases (8, 39). Because of their role as intracellular iron-binding ligands, these molecules are Science usually released to the extracellular environment as iron complexes, rather than as free ligands. Similarly, viruses may even constitute a component of colloidal organically complexed iron (40).

Unlike siderophores, humic substances and polysaccharides are complex molecules with high heterogeneity and complexity (Fig. 3), which cannot generally be resolved chromatographically, but represent a large component of the natural organic matter pool (41). Some of these molecules actually form electroactive complexes with iron and have been directly measured by electrochemistry in estuarine, coastal, and deep open ocean waters (42-44). Suwannee River Fulvic Acid (SRFA), a suspected component of the refractory dissolved organic matter pool (45), has been identified as a model ligand that can be used to reproduce the peak of natural electroactive iron complexes (46); exopolymeric substances can similarly form electroactive iron complexes (47). The conditional stability constants for exopolysaccharides and SRFA determined by CLE-ACSV generally fall under the L₂to L₄ligand class definitions (8).

Ligand processes at ocean interfaces (Fig. 4)

Photochemical degradation of natural iron-binding ligands is variable in field studies (48, 49), and may account for sea surface minima in ligand concentrations observed in some profiles (8). Experimental studies indicate that siderophore photolability depends on chemical structure and whether the siderophore is bound to iron (FeL) or not (L´) (38). Humic substances, on the other hand, are universally photoreactive by the nature of their molecular structure (50), though the iron-binding ability of their photoproducts is unknown. Dust deposition and rainfall may serve as ligand sources if depositing strong iron-binding ligands in addition to their iron loads (14, 51-54), or stimulating ligand production by surface microbial communities (55, 56), which may be critical for stabilizing atmospherically-derived iron in surface waters (16). Inorganic iron additions in mesoscale fertilization experiments have also been shown to stimulate ligand production in the fertilized waters (8, 57). Similarly, a microbial iron cycle fueled by hydrothermal iron inputs at the crust-ocean interface has recently been suggested (25) to support the organic stabilization and transport of dissolved iron in plumes extending remarkable distances from vent systems (28, 30).

Along the coastal margins, organically complexed iron is delivered to the coastal ocean from river plumes, estuaries, and shelf sediments, often along with excess weaker iron-binding ligands, including humic substances (8, 42, 44, 58). Elevated excess iron-binding ligands were observed in bottom waters of several of the GA03 stations in the North Atlantic (11) that were also local maxima in excess copper-binding ligands (59), indicating overlap in the ligand pool between these two bioactive elements (44).

Internal ligand cycling (Fig. 4)

The low solubility of inorganic iron, and the overwhelming organic complexation of the dissolved iron pool by a diverse suite of ligands has significant implications for iron bioavailability to marine phytoplankton (60, 61). The high biological demand for iron by phytoplankton and heterotrophic bacteria in turn supports a myriad of iron acquisition strategies, which are largely mediated by organic complexation (33, 61). High excess ligand concentrations in low-iron waters may potentially result from ligand production, iron uptake from a ligand complex, or both. Production of excess iron-binding ligands has been observed under a range of iron conditions, with iron additions in large-scale iron fertilization experiments (57), and in iron-stressed diatom communities (49, 62, 63), perhaps indicative of a community iron cycle including diatom-associated bacterial communities (64).

Grazing, viral lysis, and organic matter remineralization are likely important sources of weaker iron-binding ligands (e.g., humic-like substances, exopolysaccharides, or intracellular organic iron complexes like heme) to the ocean interior (8, 19). Mounting evidence points to bacterial production of the strongest ligands observed in seawater as an iron uptake strategy (8, 49, 57). Organic matter remineralization may similarly be a source of strong iron-binding ligands to the entire water column, given that siderophore production by heterotrophic bacteria is not necessarily restricted to the surface ocean and may be associated with ‘hot spots’ of sinking organic matter in the deep sea (65).

The extent to which iron-binding ligands are remineralized themselves is unknown. Excess ligand concentrations tended to decrease with depth in the Western North Atlantic as dissolved iron concentrations increased (10, 11). Decreasing total ligand concentrations in NADW samples were negatively correlated, albeit weakly, with apparent oxygen utilization (AOU) along the GA02 meridional section (10), implying some ligand remineralization during circulation since dilution was not expected to impact ligand concentrations. Gerringa et al. (2015) calculated a residence time on the order of 10³years for iron-binding ligands in the NADW, up to four times longer than that of dissolved iron, suggesting that particles must scavenge iron from strong organic complexes in the deep sea (10). Overall, the interactions between iron-binding ligands and sinking particles, whether lithogenic or biogenic in origin, are largely uncharacterized. These particles likely serve as both sources and sinks of iron-binding ligands (19) depending on the nature of the ligands, particles and biological communities involved.

Conclusions

Detailed large-scale datasets from the Atlantic Ocean have given us an unparalleled view of ligand cycling in this basin. These studies have enabled us to take a holistic look at ligand sources and sinks and internal cycling for the first time, and new paradigms have emerged. Biological contributions to the ligand pool are clear across nearly all ligand datasets. Although the direct connection between the organisms responsible for ligand production and the compounds they produce is still uncertain, marine microorganisms appear to be active producers of strong iron-binding ligands that influence iron cycling through the water column. Expansion of basin-scale datasets to the other basins and collaborative experimental studies to elucidate the mechanisms of ligand cycling behind the basin-scale distributions, some of which have been described here for the Atlantic, will improve under standing of the cycling of organic iron-binding ligands and inform global biogeochemical models of iron and carbon cycles.

Authors

Kristen N. Buck (Univ. of South Florida, College of Marine Science), Chelsea Bonnain (Univ. of South Florida, College of Marine Science), and Randelle M. Bundy (Woods Hole Oceanographic Institution)

Acknowledgments

The authors thank Kathy Barbeau for her helpful comments on the text. This work was funded in part by the National Science Foundation through an award to KNB, OCE-0927453.

References

K. W. Bruland et al., Analytica Chimica Acta 105, 233-245 (1979).
M. Gledhill, C. M. G. van den Berg, Marine Chemistry 47, 41-54 (1994).
E. L. Rue, K. W. Bruland, Marine Chemistry 50, 117-138 (1995).
C. M. G. van den Berg, Marine Chemistry 50, 139-157 (1995).
J. Wu, G. W. Luther III, Marine Chemistry 50, 159-177 (1995).
A. Tagliabue et al., Global Biogeochemical Cycles 30, 149-174 (2016).
A. Tagliabue et al., Geophysical Research Letters 41, 920-926 (2014).
M. Gledhill, K. N. Buck, Frontiers in Microbiology: Microbiological Chemistry 3, Article 69 (2012).
X. Liu, F. J. Millero, Marine Chemistry 77, 43-54 (2002).
L. J. A. Gerringa et al., Marine Chemistry 177, 434-446 (2015).
K. N. Buck et al., Deep-Sea Research II 116, 152-165 (2015).
C. Measures et al., Deep-Sea Research II 116, 176-186 (2015).
T. M. Conway, S. G. John, Nature 511, 212-215 (2014).
A. S. Wozniak et al., Marine Chemistry 154, 24-33 (2013).
K. N. Buck, A. M. Aguilar-Islas, (unpublished).
M. P. Fishwick et al., Global Biogeochemical Cycles 28, 1235-1250 (2014).
C.-E. Thuróczy et al., Journal of Geophysical Research 116, C10009 (2011).
C.-E. Thuróczy et al., Deep-Sea Research II 58, 2695-2706 (2011).
P. W. Boyd et al., Limnology and Oceanography 55, 1271-1288 (2010).
P. N. Sedwick et al., Deep-Sea Research II 116, 166-175 (2015).
B. P. von der Heyden et al., Environmental Science & Technology Letters 1, 387-392 (2014).
J. N. Fitzsimmons et al., Deep-Sea Research II 116, 130-151 (2015).
M. Yücel et al., Nature Geoscience 4, 367-371 (2011).
B. M. Toner et al., Nature Geoscience 2, 197-201 (2009).
M. Li et al., Nature Communications 5, 3192-3199 (2013).
S. A. Bennett et al., Earth and Planetary Science Letters 270, 157- 167 (2008).
J. A. Hawkes et al., Earth and Planetary Science Letters 375, 280- 290 (2013).
J. N. Fitzsimmons et al., Proceedings of the National Academy of Sciences 111, 16654-16661 (2014).
S. G. Sander, A. Koschinsky, Nature Geoscience 4, 145-150 (2011). Science
J. A. Resing et al., Nature 523, 200-203 (2015).
J. B. Neilands, Journal of Biological Chemistry 270, 26723-26726 (1995).
J. M. Vraspir, A. Butler, Annual Review of Marine Science 1, 43-63 (2009).
M. Sandy, A. Butler, Chemical Reviews 109, 4580-4595 (2009).
E. Mawji et al., Environmental Science & Technology 42, 8675- 8680 (2008).
R. M. Boiteau, Massachusetts Institute of Technology-Woods Hole Oceanography Institution, (2016).
R. M. Bundy et al., Limnology and Oceanography, (in prep).
M. Gledhill et al., Marine Chemistry 88, 75-83 (2004).
K. Barbeau et al., Limnology and Oceanography 48, 1069-1078 (2003).
S. L. Hogle et al., Metallomics 6, 1107-1120 (2014).
C. Bonnain et al., Frontiers in Marine Science 3, article 82 (2016).
L. I. Aluwihare et al., Nature 387, 166-169 (1997).
R. M. Bundy et al., Marine Chemistry 173, 183-194 (2015).
L. M. Laglera, C. M. G. Van den Berg, Limnology and Oceanography 54, 610-619 (2009).
M. M. Abualhaija et al., Marine Chemistry 172, 46-56 (2015).
N. Arakawa, L. Aluwihare, Environmental Science & Technology 49, 4097-4105 (2015).
L. M. Laglera et al., Analytica Chimica Acta 599, 58-66 (2007).
C. S. Hassler et al., Proceedings of the National Academy of Sciences 108, 1076-1081 (2011).
K. Barbeau, Photochemistry and Photobiology 82, 1505-1516 (2006).
R. M. Bundy et al., Frontiers in Marine Science 3, Article 27 (2016).
P. G. Coble, Chemical Reviews 107, 402-418 (2007).
A. C. Saydam, H. Z. Senyuva, Geophysical Research Letters 29, 1524 (2002).
M. Cheize et al., Analytica Chimica Acta 736, 45-54 (2012).
J. N. Fitzsimmons et al., Marine Chemistry, (2015 in press).
J. D. Willey et al., Limnology and Oceanography 53, 1678-1684 (2008).
T. Wagener et al., Geophysical Research Letters 35, L16601 (2008).
L. J. A. Gerringa et al., Marine Chemistry 102, 276-290 (2006).
C. L. Adly et al., Limnology and Oceanography 60, 136-148 (2015).
M. E. Jones et al., Limnology and Oceanography 56, 1811-1823 (2011).
J. E. Jacquot, J. W. Moffett, Deep-Sea Research II 116, 187-207 (2015).
F. M. M. Morel et al., Limnology and Oceanography 53, 400-404 (2008).
Y. Shaked, H. Lis, Frontiers in Microbiology 3, Article 123 (2012).
K. N. Buck et al., Marine Chemistry 122, 148-159 (2010).
A. L. King et al., Marine Chemistry 128-129, 1-12 (2012).
S. A. Amin et al., Microbiology and Molecular Biology Reviews 76, 667-684 (2012).
F. Azam, F. Malfatti, Nature Reviews 5, 782-791 (2007).
P. W. Boyd, A. Tagliabue, Marine Chemistry 173, 52-66 (2015).

Mesodinium rubrum: An Old Bug Meets New Technology

Posted by mmaheigan

· Tuesday, April 12th, 2016

Blooms of red water associated with the remarkable ciliate Mesodinium rubrum have been observed at least since Darwin’s time (1). This ciliate retains the chloroplasts from ingested prey and is able to use them for photosynthesis (reviewed in 2). Recent studies have shown that the plastids can reproduce within the ciliate and that nuclei from the original algal prey remain
transcriptionally active (3). It is very likely that there are at least two different species of Mesodinium that perform this feat, the original M. rubrum and a recently described larger species, M. major (4). Both species have in common certain species of cryptophyte
algae as their preferred food, and hence are colored deep red by their prey’s phycoerythrin pigment and characteristic yellow fluorescence (Fig. 1). Mesodinium is believed to hold the ciliate swimming speed record, with short jumps of up to 1.2 cm s-1, and can change its position vertically in the water column to access nutrients (5). Along with rapid growth, its impressive motility probably contributes to the large aggregations obvious to the naked eye, in which concentrations of >106 cells l-1 have been observed (6 ) (Fig. 1). Even outside of bloom conditions, they are a regular component of estuarine and coastal plankton assemblages and can contribute significantly to primary productivity (7). However, as mixotrophs (organisms capable
of both photosynthesis and ingestion), they are undersampled and underappreciated by phytoplankton and zooplankton ecologists alike.

Red water has been reported in Long Island Sound on occasion by other observers. While Mesodinium was present in >80% of all samples examined in >10 years of monthly plankton monitoring data, no sample ever exceeded 2.6 x 104 cells l-1. In Fall 2012, Univ. Connecticut personnel servicing a moored array observed and sampled red water in western Long Island Sound (40.9°N 73.6°W). Microscopy and DNA sequencing confirmed that the bloom was due to Mesodinium (100% identical by small subunit rDNA to the larger M. major), and we subsequently reported on our efforts to document the bloom using satellite imagery (8). Here, we summarize those results and discuss the promise of new sensors for quantifying blooms of specific plankton groups by their pigment signatures, especially when coarsely resolved monitoring samples are inadequate.

Ocean color satellites provide a means to assess such red tides, but the standard chlorophyll products are inaccurate in the optically complex waters of Long Island Sound, which contain river runoff with colored dissolved organic matter (cDOM) and suspended sediments (9, 10) (Fig 2). Imagery from the MODIS sensor of fluorescence line height (Fig. 2A) indicated the presence of an unspecified bloom in Western Long Island Sound coincident with the bloom, but the spatial resolution (1-km pixels) did not allow us to gauge the bloom extent adequately, and the spectral bands of that sensor are not sufficient to discriminate the type of bloom.

Serendipitously, an image was available for the western Sound from the novel Hyperspectral Imager for the Coastal Ocean (HICO) instrument aboard the International Space Station. This sensor contains >100 channels in the visible and near infrared regions of the spectrum and hence has the capability to resolve multiple peaks and valleys due to fluorescence and absorbance of the chlorophylls and accessory pigments found in various phytoplankton groups. It also has the higher spatial resolution (110-m pixels) needed to quantify the extent of the bloom and variation in ciliate abundance within it. Because the red water we observed appeared (microscopically) to be almost exclusively due to Mesodinium, the HICO reflectance spectrum was an almost pure example of the in situ optical signature of this unique organism (i.e. an “endmember” in remote sensing terminology).

In addition to phycoerythrin, the cryptophyte chloroplasts that the ciliate retains contain chlorophyll-a, chlorophyll-c2, phycocyanin, and the carotenoid alloxanthin. The reflectance spectrum measured with the HICO sensor revealed features related to the fluorescence and absorption associated with these pigments that can be used as a spectral “fingerprint” of this specific organism (Fig. 3A). With reflectance measured across the full visible spectrum, small dips in the spectrum can be revealed with a 4th derivative analysis and related to the associated pigments (11) (Fig. 3B). In addition to absorbing green light, phycoerythrin also fluoresces yellow light (12) (Fig. 1B) and a peak in reflectance was observed at ~565 nm associated with this feature. This unique fluorescence feature allowed us to map the surface distribution of Mesodinium in Long Island Sound. Traditional ocean color satellites do not measure reflectance of light at this waveband, but yellow fluorescence (band depth at 565 nm) could be detected from the hyperspectral measurements of HICO and related to the relative amount of Mesodinium up to the measured 106 cells L-1 with distinctly red colored water (Fig. 4).

The fine-scale distribution of the HICO imagery reveals that Mesodinium was found in small 100-m patches along the sea surface rather than distributed throughout a single multi-kilometer patch as suggested by the 1-km MODIS imagery (Fig 2A). Such high spatial resolution from aircraft has been used to assess concentration mechanisms of blooms, including internal waves (13) and Langmuir circulation (14). Further research is underway to assess the observed patterns with hydrographic and air-sea processes local to this region. Understanding the spatial distribution may also lead to a better understanding of the environmental factors that lead to these episodic blooms of Mesodinium. Generally, Mesodinium is more abundant in lower salinity estuarine water, but the causes of bloom initiation and demise are not well known (15).

Though now defunct, the HICO sensor should serve as a model for remote sensing in the coastal zone. With its high spectral and spatial resolution, images from HICO could be used to assess coastal processes, as highlighted here, but only at infrequent intervals. While possible with airborne technology, no existing or planned satellite sensor can sample at high spectral, spatial, and temporal resolution for adequate monitoring of the coastal zone. Providing near-daily coverage for much of the globe, the next generation NASA ocean color sensor, Pre-Aerosol, Cloud and ocean Ecosystems (PACE), is slated to have the unique hyperspectral capabilities to allow for better discrimination of marine blooms and habitats, but with a larger km-scale resolution. International sensors with new capabilities will also help to fill this gap (16). With new hyperspectral technology in space, autonomous and routine differentiation of phyto- and mixotrophic plankton blooms in surface waters may be possible and could provide an important tool for resource managers. Improved monitoring of bloom-forming plankton will also lead to more refined estimates of coastal primary productivity and mechanisms for their episodic growth and decline. If future sensors or sensor constellations combine high repeat sampling with the hyperspectral capabilities and high spatial resolution of HICO, we will be able to understand not only the composition and extent of blooms, but also the sub-mesoscale processes that drive their persistence and spatial structure.

Authors

Heidi Dierssen and George McManus (University of Connecticut)

Acknowledgments

We thank Kay Howard-Strobel, Senjie Lin, and the NOAA Phytoplankton Monitoring Network for images of the bloom and of Mesodinium. Dajun Qiu verified the genetic identity of the ciliate. Adam Chlus and Bo-Cai Gao contributed to the image processing. We also thank the HICO Science Team and NASA Ocean Biology Distributed Active Archive Center for providing satellite imagery.

References

Darwin, C., 1909. The Voyage of the Beagle, P.F. Collier.
Crawford, D. W., 1989. Mar. Ecol. Prog. Ser. Oldendorf 58, 161–174.
Johnson, M. D. et al., 2007. Nature 445, 426–428.
Garcia-Cuetos, L. et al., 2012. J. Eukaryot. Microbiol. 59, 374–400.
Crawford, D. W., T. Lindholm, 1997. Aquat. Microb. Ecol. 13, 267–274.
Taylor, F. J. R. et al., 1971. J. Fish. Board Can. 28, 391–407.
Smith, W. O., R. T. Barber, 1979. J. Phycol. 15, 27–33.
Dierssen, H. et al., 2015. Proc. Natl. Acad. Sci., doi:10.1073/pnas.1512538112.
Aurin, D. A., H. M. Dierssen, 2012. Remote Sens. Environ. 125, 181–197.
Aurin, D. A. et al., 2010. J. Geophys. Res. 115, 1–11.
Bidigare, R. R. et al., 1989. J. Mar. Res. 47, 323–341.
McManus, G. B., J. A. Fuhrman, 1986. J. Plankton Res. 8, 317–327.
Ryan, J. P. et al., 2005. Oceanography 18, 246–255.
Dierssen, H. M. et al., 2015. Remote Sens. Environ. 167, 247–258.
Herfort, L. et al., 2011. Estuar. Coast. Shelf Sci. 95, 440–446.
International Ocean Colour Coordinating Group (IOCCG). www.ioccg.org

New Satellites Paint a Portrait of Plankton Spatial Variability

Posted by mmaheigan

· Saturday, April 2nd, 2016

The newest generation of satellites reveals plankton variability changes in character from uniform to chaotic at different spatial scales, reviving a classic question in oceanography. How does plankton variability change at different spatial scales, and why?

New satellites, new insights

Satellite technologies can now collect images with resolution down to the scale of meters, presenting oceanographers data with unprecedented information about the fine-scale structure of plankton communities in the surface ocean. In August 2015, there was significant media attention after two of the world’s most advanced satellites, Landsat 8 and Sentinel-2, published images of a cyanobacteria (algal) bloom in the Baltic sea (Fig. 1). For scale, the images conveniently have boats in them (you really have to squint, or just zoom in – a little game of Where’s Waldo at sea).

While these images are beautiful in their own right, to an oceanographer they also illustrate the complexity of the biophysical interactions that drive plankton distributions. When we run computer models to simulate e.g., how plankton communities might respond to a changing climate, we can’t replicate all of this variability, so we typically represent an X km × Y km square of ocean with a single value (e.g., plankton concentration), which we consider as the average for that box; one peek at an image like this demonstrates that it’s difficult to justify this approach as doing full justice to the system it’s simulating. Similarly, when we take samples out in the field, we often fill bottles with seawater and assume that sample represents a X km × Y km area around it. This image suggests that taking a measurement off one side of the boat might give you a very different representation of that region than if you had taken it off the other side! These approaches are further complicated by studies indicating that the variability we see in these images persists at microscopic scales.

This is not meant to needlessly criticize these approaches; oceanography is a challenging science, and we do the best we can. Often, these approaches can yield wonderful insights. These images just draw attention to the fact that plankton spatial variability remains a fascinating and open problem in oceanography, which present-day technology puts us in good position to start addressing.

Characterizing variability

One way we can characterize such variability is by using a power spectral density (PSD), which allows us to quantify how much variability is contained at each scale in an image. Computing the PSD for each of the above images is a straightforward exercise, thanks to modern computational capabilities. To draw an analogy, we can also compute the PSD for a painting by each of Rothko and Pollock (Figs. 2a. and 2b., respectively); we might take the former to represent ’homogeneity’ and the latter to represent ’chaos’ (as Pollock’s paintings have been thought of for years). That is, imagine a satellite looks down on a plankton bloom and sees a rather gargantuan painting of each type; how do these paintings compare with observed blooms, in terms of spatial variability?

Methods

The PSD has been computed for the red band of the RGB image of the Rothko painting, a black and white conversion of the Pollock painting, and for the green band of each of the satellite images. Computing the PSD for other configurations did not change the result. The wavenumber k = 1 in this case corresponds to a wavelength λ ≈ 50 km. Wavenumbers have been rescaled to those of the Sentinel-2 image, and PSDs have been normalized to their L2 norm.

Comparing power spectral densities

When we computed the PSDs for these four images (Figs. 1a, b and 2a, b), we found remarkable consistency (almost identical PSDs) between the two satellite images (Figs. 1a and b), which were taken four days apart. This suggests that 1) the satellites are accurately and reproducibly capturing spatial bloom variability, and 2) bloom PSDs don’t change significantly from day to day. The PSDs from the satellite images matched the Pollock spectrum at smaller spatial scales (i.e. high wavenumber) and the Rothko spectrum at larger spatial scales (i.e. low wavenumber) (Fig. 3). This raises the question: why might this be happening? Also, at what scale does the ’Rothko-Pollock’ transition occur and why?

Significance

If the distribution of plankton was purely that of Brownian (random) motion, we’d expect a flatter PSD (i.e. a line with slope = -2). Another null hypothesis is that the distribution of plankton might be set passively by advection of oceanic currents. In this case, we’d expect plankton distributions to have the same signature as temperature, which also has a PSD slope of -2. However, these spectra (Fig. 3) have slopes that are steeper than -2 (closer to -2.5 or -3), so clearly there’s more afoot. The steeper slope of -3 at larger scales means that variability falls off faster as we look at smaller scales, i.e. something about the plankton distribution is ’homogenizing’ at larger scales. Then, the PSDs get shallower at wavelengths of ~1 km, indicating that something kicks in at sub-kilometer scales that introduces more variability. One way to think about this transition, which has been hypothesized since the 1970s (1), is that different processes can dominate at different spatial scales. The specifics of the 70s manner of thinking aren’t quite compatible with these data, but the general concept is plausible. Plankton grow in response to light and nutrient conditions, but also live in a turbulent environment. At large scales, growth occurs somewhat uniformly and is dominated by ambient light and nutrient conditions, whereas smaller-scale biophysical interactions can introduce an additional source of variability in plankton growth. Biophysical variability can occur in many ways, including small-scale horizontal motions that can stir plankton patches into filaments and small-scale vertical motions that can enhance growth locally by bringing up nutrients. In either case, these biophysical interactions are only observable at smaller scales.

Thus, at larger scales, the plankton will be distributed relatively homogeneously as uniform (light-/temperature-driven) growth wins out (. la Rothko), and at smaller scales, they will be distributed heterogeneously as advective processes come into play (à la Pollock). The spatial scale at which this transition occurs is controversial and depends on many factors, though was originally hypothesized to be ~1 km, which here appears plausible. See the vertical line in Fig. 3, which corresponds to a 1-km wavelength and appears to agree well with the scale of the observed transition from Rothko-type to Pollock-type behavior.

Another thing to note is that these cyanobacterial mats (Fig. 1) are very thin and form just at the ocean surface –zoom in and you can see how the boat tracks cut through them. Thus, these patterns may be representative of a different set of physical processes occurring only in the uppermost layer of the ocean.

While two satellite images of the same bloom may not be enough to verify the growth vs. turbulence hypothesis, ’Rothko-type’ versus ’Pollock-type’ behavior may not be quantitative enough descriptions to satisfy any oceanographer, and the equally-complex third dimension isn’t included in these pictures, there is still a clear message here. The spatial resolution available from the newest generation of satellites provides a novel opportunity to approach problems of scale in oceanography.

Author

B. B. Cael (MIT Earth, Atmosphere and Planetary Sciences, Woods Hole Oceanographic Institution)

Acknowledgments

It is a pleasure to thank Bror Jonsson, Mick Follows, Bryan Kaiser, and Amala Mahadevan for useful discussion of this topic.

References

Denman, K.L., T. Platt, 1976. J. Marine Res. 34, 593-601.

Funding for the Ocean Carbon & Biogeochemistry Project Office is provided by the National Science Foundation (NSF) and the National Aeronautics and Space Administration (NASA). The OCB Project Office is housed at the Woods Hole Oceanographic Institution.

Archive for New OCB Research – Page 27

1. Stoichiometry of metals in the ocean

2. Processes affecting dissolved and particulate stoichiometries of trace metals

3. Additional tools to explore and differentiate remineralization processes

Authors

Acknowledgments

References

GEOTRACES and particles in the ocean

Scavenging: A two-step removal process

Particle concentration and composition: horizontal and vertical variations

Effect of particle composition on adsorption of TEIs

TEIs as tracers of scavenging rates and particle dynamics

Particle composition and the biological pump

Outlook

Author

References

Author

References

Author

References

Dissolved iron speciation

Iron speciation in the Atlantic: Observations from recent GEOTRACES efforts

Identity of iron-binding ligands

Ligand processes at ocean interfaces (Fig. 4)

Internal ligand cycling (Fig. 4)

Conclusions

Authors

Acknowledgments

References

Acknowledgments

References

New satellites, new insights

Characterizing variability

Methods

Comparing power spectral densities

Significance

Author

Acknowledgments

References

Filter by Keyword