Chemical Reference Materials: Setting the Standards for Ocean Science (2002)

Many of the analytes of interest for solid phase chemical reference materials are the same as those in seawater, but the need for and the preparation of reference materials for suspended particulate matter and sediments is quite different. The low concentrations of many seawater species and the presence of the salt matrix create particular difficulties for seawater analyses. However while sediments frequently have higher component concentrations than seawater, they also have more complicated matrices that may require unique analytical methods. A number of particulate inorganic and organic materials are employed as paleoceanographic proxies, tracers of terrestrial and marine input to the sea, measures of carbon export from the surface waters to the deep sea, and tracers of food-web processes. Some of the most important analytes are discussed below as they relate to important oceanographic research questions.

As discussed in Chapter 3, measurements of the major elements composing particulate organic matter are among the longest established, most fundamental, and broadly informative analyses in the marine sciences. This tradition is based on the fact that six elements (carbon, hydrogen,

oxygen, nitrogen, sulfur, and phosphorus) make up essentially all the mass of marine organic matter and largely govern the processes by which these elements are cycled in the ocean. The rationale for the analysis of specific elements is presented below.

Carbon is the most abundant element common to all organic substances (approximately 55 wt percent of average marine plankton) (Hedges et al., 2002) and has been used as the practical currency for measuring total organic matter concentrations in dissolved and particulate marine samples (MacKinnon, 1981). Carbon comprises two stable isotopes (¹²C and ¹³C), and one useful radioactive isotope (¹⁴C), which together provide indicators of source, reaction history, age, and dynamics of the organic substances in which they occur (Raymond and Bauer, 2001). Carbon constitutes 27 wt percent of CO₂, a potent greenhouse gas and driver of climate change (Siegenthaler and Sarmiento, 1993; Sarmiento et al., 1998). At present, about 10 percent (3.5 Gt, 1 Gt = 10¹⁵ g) of all carbon actively cycling at the Earth’s surface occurs in organic substances. The global rate of net photosynthesis (approximately 100 Gt C/yr) is sufficient to pass the entire mass of carbon currently in atmospheric CO₂ through living biomass in less than 10 years. Thus, any process or perturbation that affects the formation or remineralization rate of organic carbon can substantially influence the global carbon cycle and the potential for climate change within a human life span.

As mentioned above, the three isotopes of carbon are powerful tools for studying ocean processes, the carbon cycle and paleoceanography. ¹³C and ¹⁴C in biological materials and sediments are most commonly measured in bulk organic matter and in either bulk CaCO₃ or isolated foraminiferal shells. With the advent of isotope-ratio-monitoring gas chromatography mass spectrometry (irm-GCMS), sophisticated analytical isolation techniques, and accelerator mass spectrometry (AMS), carbon isotopes can now be measured in specific organic compounds isolated from sediments and particulate material.

In coastal areas, measurements of δ¹³C in bulk organic matter can help identify the origins of organic material in sediments. In general, material produced using the dominant C₃ photosynthetic pathway has a value of δ¹³C around ­−27 ‰ for terrestrial matter and around ­−20 ‰ for marine matter (Deines, 1980). The interpretation of such results is complicated because some plants use the C₄ photosynthetic pathway, which

produces terrestrial matter with a δ¹³C value near ­−12‰, and other plants use both pathways (CAM plants)¹ (Deines, 1980).

In the open ocean, δ¹³C values of particulate organic matter in the surface ocean vary greatly. Isotopic variations have been linked to changes in temperature and latitude (Freeman and Hayes, 1992; Goericke and Fry, 1994). Fractionation is also a function of the concentration of CO₂ in the aqueous form, the δ¹³C of the CO₂ (a function of temperature), and the fraction of the dissolved inorganic carbon pool taken up by an organism (Hayes, 2001; Laws et al., 2001). In addition to its use in studying organic matter, the measurement of δ¹³C in the CaCO₃ of foraminifera is helping to evaluate the role of the deep ocean in global climate change. Long-term changes (i.e., over glacial cycles) in the mean ocean δ¹³C and the isotopic relationship to nutrients over time can be similarly studied.

Radiocarbon measurements provide the accurate chronologies required to pinpoint causal events, either by determining sedimentation rates or by assigning an actual date to a specific event. AMS now provides the capability to measure radiocarbon routinely in small (0.5-1 mg) samples, making it possible to add chronologies to paleoceanographic studies of sediments as well as to use the natural ¹⁴C cycle as a tracer of the transfer of carbon among the different global carbon pools. Measurement of radiocarbon in bulk organic matter establishes time histories within sediment columns and the rates of sediment accumulation in carbonate-poor areas of the ocean. The presence or absence of bomb-produced radiocarbon in bulk particulate and sedimentary matter provides clues as to how recently the material was actively involved in the carbon cycle.

Calibrations of ∆¹⁴C in corals with U/Th series isotopes are extending the calendar chronology of ¹⁴C back 20,000 to 40,000 years, allowing researchers to better study the events causing glacial/interglacial cycles (Bard et al., 1999). Radiocarbon measurements of bulk CaCO₃ can help provide information on the role of carbonate dissolution within sediments in the global carbon cycle (Broecker et al., 1980; Keir and Michel, 1993; Martin et al., 2000). Measurement of ∆¹⁴C in coral bands provides a record of ∆¹⁴C in the surface ocean (Druffel, 1995) and can be used to track the oceanic uptake of the atmospheric bomb pulse described in Chapter 3. Studies of the uptake of the bomb pulse provide useful constraints to ocean mixing models (Toggweiler et al., 1989a,b). The very detailed records of seasonal and annual fluctuations of δ ¹⁴C in corals provide

tighter constraints on ocean mixing in global carbon cycle models (Guilderson et al., 1998, 2000). Isotopes in specific organic compounds will be discussed separately.

Nitrogen and phosphorus, in addition to being potentially limiting nutrients, provide an indication of the source and history of the organic materials in which they occur. Both elements are typically lost in preference to carbon during the digestion and bacterial degradation of plankton biomass, and therefore have potential use as indicators of substrate freshness. Because marine organic nitrogen occurs predominantly in protein, carbon to nitrogen (C:N) ratios can be indicative of the food quality of different organic mixtures. Since increased nitrogen richness is characteristic of marine mixtures, C:N ratios have also been used to discriminate plankton- from land-derived organic matter in coastal ocean zones. Such C:N-based assessment of organic source and food quality, however, must take into account that nitrogen can become elevated versus carbon in the course of biodegradation (Suess and Müller, 1980).

Neither nitrogen nor phosphorus has a long-lived naturally occurring radioisotope, and phosphorus has only one stable isotope. However, ¹⁵N to ¹⁴N ratios can be used under appropriate circumstances to detect nitrogen fixation, denitrification, cumulative uptake of upwelled nutrients, and passage of nitrogen up trophic levels (Goericke et al., 1994). Such stable isotope analyses have the potential to become a powerful tool in ecosystem studies that exploit the isotopic fractionation associated with the transfer of carbon and nitrogen between trophic levels (Gannes et al., 1997). Organic sulfur accounts for less than 1 wt percent of average marine plankton, is not a limiting nutrient in the ocean, and is difficult to quantify in marine samples due to interference by seawater sulfate.

Hydrogen and oxygen, although major components of marine organic matter with multiple stable isotopes, have remained largely “orphan elements” with respect to biogeochemical studies. In large part, this analytical avoidance results from the challenge of measuring these elements in a sea of potentially interfering water in living cells, hygroscopic salts, and hydrous minerals. Current estimates of the elemental composition of marine plankton biomass based on measured biochemical compositions (Anderson, 1995) and spectroscopic analysis (Hedges et al., 2002) indicate a compositional formula near C₁₀₆H₁₇₇O₃₇N₁₇S_0.4. This study did not consider phosphorus (or its associated oxygen), as it undergoes no

change in redox state during photosynthesis and remineralization and thus does not affect respiration demand. This revised formula corresponds to much lower hydrogen (8 wt percent) and oxygen (26 wt percent) contents than given in the Redfield-Ketchum-Richards equation (see Chapter 3) and requires approximately 10 percent more oxygen for complete respiration. Because the polysaccharides, proteins, and lipids that compose marine biomass have uniquely different elemental composition, their weight percent in plankton (and their remains) could be directly determined if the content of oxygen and hydrogen were measured in addition to carbon and nitrogen. The hydrogen content of organic matter is also of broad interest because it is positively related to the petroleum formation potential of marine sediments (Pederson et al., 1992; Gélinas et al., 2001a) and rocks (Demaison and Moore, 1980). Accurate measurements of the major elements composing marine organic matter offer a wealth of useful biogeochemical information, if appropriate methods can be developed and compared for widely available reference materials.

The multitude of organic compounds present in the environment makes it impossible to discuss each compound, or even compound class. A general rationale for the matrix approach to preparation of reference materials is presented below, as well as a specific discussion of several classes of particular importance.

The last 50 years have seen a steadily increasing fraction of ocean measurements directed toward organic substances. This growing emphasis reflects the importance of organic molecules as the fundamental components of all living organisms and as key agents for the transfer and storage of the partially degraded remains of plants and animals in ocean waters and sediments. Organic molecules occur in a multitude of structures that are chemically unique and often persistent, and which carry a wealth of information about their biological sources and subsequent reaction histories. Almost all the organic matter in the ocean occurs as the incompletely degraded nonliving remains of marine organisms—living biomass is relatively minor. On average, organic substances account for about one in thirty carbons dissolved in seawater, four of five carbons sinking in particulate debris from the surface to the interior ocean, and one in five of all carbons preserved in marine sediments.

The synthesis, destruction, and vertical transport of organic matter in the marine environment all have important implications for the produc-

tivity of higher trophic levels (e.g., micronekton and fish) in coastal, pelagic, and benthic habitats. The major components of the marine planktonic food web include bacteria, Archaea, phytoplankton, and zooplankton. Knowledge of plankton community structure and its variability is therefore essential for evaluating the predictive ability of complex food web models and assessing ecological response to climate change (e.g., the El Niño Southern Oscillation, the Pacific Decadal Oscillation, and the North Atlantic Oscillation). Historically, the determination of the biomass and diversity of the microbial component (bacteria, Archaea, and phytoplankton) within planktonic food webs posed a difficult problem for oceanographers. A range of uncertainties, including the selectivity of various culturing methods, overlapping size distributions, poor preservation properties, and the subjectivity of microscopic methods have encouraged the development of lipid biomarker approaches for characterizing microbial communities.

Lipids are major constituents of all living cells and include a wide range of functional and storage biomolecules, such as chlorophylls, porphyrins, carotenoids, hopanoids, sterols, fatty alcohols, diterpenes, ubiquinones, fatty acids, and waxes. This enormous structural diversity reflects, in part, phylogenetic relationships, and consequently, specific lipid compounds are frequently employed as biomarkers (Table 4.1). Lipid biomarkers have varying degrees of taxonomic specificity. The carotenoids and sterols of microalgae, for example, are typically used to provide taxonomic distinction at the class level. Individual fatty acids, hydrocarbons and sterols reflect their different plankton sources (Wakeham and Lee, 1993) and help distinguish terrestrial from marine origins. Bacteria add diagnostic branched-chain fatty acids to particles they colonize (de Baar et al., 1983; Wakeham and Canuel, 1988).

Chlorophyll a (Chl a) functions as the primary light harvesting pigment in marine oxygenic phototrophs. Even though the C:Chl a ratio of photoautotrophic cells varies considerably as a function of environmental conditions and growth rate (Laws et al., 1983), measurements of Chl a have been used extensively to estimate the biomass of photoautotrophic microorganisms in the sea.

Cells are typically concentrated by filtration and extracted into an organic solvent (usually acetone) after which, pigments are detected by fluorescence or absorption spectroscopy, sometimes after chromatographic separation (Bidigare and Trees, 2000). The application of HPLC to phytoplankton pigment analysis has lowered the uncertainty in the measurement of Chl a and accessory carotenoids, since compounds are physically separated and individually quantified.

Pigment distribution is useful for quantitative assessment of phytoplankton community composition, phytoplankton growth rate and

zooplankton grazing activity. Global maps of Chl a distribution are now available daily from satellite-based ocean color sensors such as SeaWiFS and MODIS. These high resolution (less than or equal to 1 km) Chl a images have provided new insight into the effect of physical and climate driven forcing on phytoplankton biomass and productivity at regional-to-global scales (Chavez et al., 1999; Behrenfeld et al., 2001; Seki et al., 2001).

Biomarker distributions can also provide important information on microbial metabolism. For example, the presence of Chl a, bacteriochlorophyll a, or trans-retinal in a culture or environmental sample would indicate microbial growth driven by oxygenic photosynthesis, anoxygenic photosynthesis, or photo-heterotrophy, respectively (Takaichi et al., 1990, 1991; Béjà et al., 2000; Kolber et al., 2000). Diagenetic products for many of the lipid biomarkers have been characterized structurally and used as “molecular fossils” to study ancient rocks, sediments, and petroleum (Lee and Wakeham, 1988; Wakeham and Lee, 1989; Peters and Moldowan, 1993; Kenig et al., 1994; Summons et al., 1999; Moldowan et al., 1996). These studies have demonstrated that biomarkers have utility for tracing the evolution of life in the geologic record, recreating the structure of ancient microbial communities, and determining which organisms contribute to hydrocarbon deposits.

The haptophyte microalga Emiliania huxleyi produces biomarkers in the form of long-chain (C₃₇, C₃₈, and C₃₉) alkenones (Brassell, 1993). Alkenones are well preserved in marine sediments and their molecular distributions and isotopic composition have been used to infer paleotemperatures (Brassell, 1993) and pCO₂²values (Jasper et al., 1994), respectively. Unsaturation patterns in the alkenone series are related to the growth temperature of the haptophyte algae that produce these compounds (Brassell et al., 1986; Prahl and Wakeham, 1987), and hold great promise as indicators of absolute ocean paleotemperature.

A variety of molecular indicators of the freshness (and perhaps nutritional quality) of organic matter exists. For example, certain labile phytoplankton constituents, such as polyunsaturated fatty acids, are readily degraded in the environment or in herbivore guts, and are thus depleted in more degraded particles (de Baar et al., 1983; Wakeham and Canuel, 1988). Preferential loss of labile algal fatty acids resulting in the enrichment of more stable components in the products of heterotrophic metabolism has been observed in both field studies and laboratory feeding experiments (Prahl et al., 1985; Wakeham and Canuel, 1988; Harvey et al.,

1987). A resistant organic matrix may protect otherwise labile lipids. For example, the resistant waxy coating common to land plants appears to physically shield terrestrial biomarkers, which are then selectively preserved, as opposed to their marine-derived counterparts in abyssal sediments (Gagosian et al., 1983; Wakeham et al., 1984; Volkman , 1986). The freshness of organic matter can also be influenced by an intimately associated mineral or organic matrix (Gordon and Millero, 1985; Hedges and Keil, 1995; Mayer, 1994).

Amino acids are structural components of proteins and compose the largest reservoir of organic nitrogen in most organisms. They make up a major fraction of characterized carbon in marine particulate matter and are useful indicators of decomposition and transport in the marine environment. Their natural occurrence in marine plankton (Degens and Mopper, 1976; Tanoue et al., 1982; Cowie and Hedges, 1996), suspended and sinking particles (Lee and Cronin, 1982, 1984; Ittekkot et al., 1984a, b; Cowie and Hedges, 1992; Cowie et al., 1992), and sediments (Henrichs and Farrington, 1984; Henrichs et al., 1984; Dauwe and Middleburg, 1998) is frequently used to elucidate the biogeochemical behavior of marine organic matter, particularly the diagenetic state. Chromatographically measurable carbohydrates constitute a maximum of 20 to 40 percent of plankton carbon, 13 percent of sinking particulate organic carbon (POC), 15 percent of suspended POC, and 13 percent of sedimentary carbon (Hernes et al., 1996). Amino acids and carbohydrates can be used to elucidate the fate of organic carbon and to infer past marine sources and oceanic conditions.

Although generally not as species-specific as lipids and pigments, compositional differences among carbohydrates and amino acids can also reflect biological sources. For example, some siliceous diatoms are characterized by elevated concentrations of glycine and fucose while organisms with carbonate tests often exhibit unusually high concentrations of aspartic acid and arabinose (Hecky et al., 1973; Carter and Mitterer, 1978; Ittekkot et al., 1984a,b). Although both plankton and bacteria are characterized by high relative abundances of ribose and fucose (Cowie and Hedges, 1984), O-methyl sugars and uronic acids may provide a means of discriminating between these two biological sources (Mopper and Larsson, 1978; Bergamaschi et al., 1999). In addition, muramic acid can serve as a specific marker for bacteria (Lee et al., 1983).

High relative concentrations of nitrogen and carbon in the form of amino acids (Whelan, 1977; Lee and Cronin, 1984; Cowie and Hedges, 1992) and carbohydrates (Cowie et al., 1992; Cowie and Hedges, 1992,

1984) indicate relatively undegraded organic matter. Conversely, high relative concentrations of characteristic diagenetic products, such as the non-protein amino acids ornithine and β-alanine can indicate the presence of bacterially degraded material (Lee and Cronin, 1982, 1984; Ittekkot et al., 1984a,b). The combined use of multiple amino acid- and carbohydrate-based diagenetic indicators provides a sensitive and consistent means of comparing the state of organic matter in aquatic environments (Hedges et al., 1999). Preservation of certain amino acids and sugars occurs in both siliceous and carbonate tests. These can be sensitively detected in part because amino acid composition of fresh materials is so uniform.

Although carbon isotopic measurements of bulk organic material can be very informative, much more information can be obtained from the isotopic characterization of specific compound classes or individual organic compounds isolated from bulk material. Such compound-specific isotope analyses (CSIA) can target individual molecules known to be associated primarily with specific processes (e.g., photosynthesis) or with specific sources.

The stable isotopic compositions of algal organic matter can provide important insight into the environmental conditions under which carbon and nitrogen fixation occur (Hayes et al., 1990; Goericke et al., 1994). Determination of δ¹³C and δ¹⁵N provides valuable tools for assessing the rates of ancient biological processes such as phytoplankton growth and NO₃ uptake (Altabet and Francois, 1994; Bidigare et al., 1999). Application of stable isotopes to the modern ocean and sedimentary record, however, is confounded by the presence of non-algal carbon and nitrogen biomass. To resolve this problem, CSIA has been employed as a means of estimating the δ¹³C and δ¹⁵N of modern and ancient phytoplankton populations (Jasper et al., 1994; Bidigare et al., 1997, 1999; Sachs and Repeta, 1999).

Researchers are now investigating the potential for dating targeted compound classes and individual compounds to understand the true chronology of sediments. Pearson et al. (2000, 2001) measured the radiocarbon content of individual sterols and other lipids isolated from sediments. While most of the compounds were derived from marine photosynthesis or subsequent remineralization, data from the archeal isoprenoids were consistent with chemoautotrophic growth below the euphotic zone. Eglinton et al. (1997) found significant variation in the age of individual compounds isolated from the same sediment, indicating that older components either have a different inorganic carbon source or are stored for appreciably longer time periods in “upstream” reservoirs.

Measurements of radionuclides and metals in marine sediments and particulate matter are conducted for a variety of purposes, including the determination of sedimentation rates, trace metal and radionuclide fluxes through the water column, enrichment of metals in specific phases of the sediments, and examination of new sedimentary phases produced after sediment deposition. Such studies address fundamental questions concerning the chronology of deep-sea and near-shore sedimentary deposits, removal mechanisms and cycling of metals in the ocean, and diagenesis within deep-sea sediments.

Numerous radionuclides have been applied to marine sedimentary problems. These are generally grouped into cosmic-ray produced (cosmogenic) nuclides (¹⁴C, ¹⁰Be, ⁷Be, ²⁶Al), nuclear bomb-produced (fallout) nuclides (¹³⁷Cs, plutonium isotopes, ²⁴¹Am), and naturally occurring nuclides ultimately derived from the decay of ²³⁸U, ²³⁵U, and ²³²Th parents.

²³⁰Th and ²³¹Pa are ubiquitous components of recently deposited deep-sea sediments because they are produced uniformly throughout the ocean from the decay of dissolved uranium isotopes and they are actively collected onto sinking particles. The distribution with depth of these nuclides in deep-sea sediments may be modeled to estimate rates of sedimentation extending over the past 200 to 300 thousand years. These techniques complement ¹⁴C dating methods that only extend to about 40 thousand years before the present.

Other techniques are used for shorter time scales, including the measurement of the ²²⁶Ra:Ba ratio in barite extracted from sediments (Paytan et al., 1996). This technique has a time scale of about five thousand years. Alternately, assessments of rapid sedimentation and bioturbation on time scales of days to centuries require shorter half-life nuclides such as ²¹⁰Pb, ²²⁸Th, ²³⁴Th, and ²²²Rn.

Several major matrix types are found in marine particles and sediments. Marine organisms surround themselves with tough polymeric organic cell walls and/or with opal or calcium carbonate tests. These contrasting matrices respond differently to various analytical methods. In sediments, the remains of these organisms combine with clay minerals to form a heterogeneous mixture. In this section, the influence of these matrices on analyte quantification are discussed.

The organic carbon, hydrogen, and nitrogen content of marine samples is usually measured by quantitative combustion of a weighed amount of solid (or liquid) material into volatile gases whose concentration can then be measured in a CHN analyzer, equipped with a non-specific thermal-conductivity detector. Typically, carbon is measured as CO₂, nitrogen as N₂ (after reduction of nitrogen oxides), and hydrogen as H₂O. In the simple case of dry, purely organic samples, this method is fast, accurate, precise, and requires only 1 mg or less of sample. Unfortunately, dry, purely organic materials are almost never obtained in marine samples because many common types of phytoplankton and zooplankton secrete calcium carbonate (e.g., coccoliths and foraminifera) or opal (e.g., diatoms and radiolarians) tests. Even samples of living plankton taken directly from surface ocean waters characteristically contain 20 to 60 weight percent of mineral matter (Parsons et al., 1977). Due to the relative lability of the organic versus the mineral components of plankton remains, the weight percent of mineral matter in marine particles increases with depth and degradation, reaching typical values near 90 wt percent in particles raining to the ocean floor and more than 99 wt percent for those particulate materials that make up marine sediments (e.g., Wakeham et al., 1997).

The minerals that make up the increasingly predominant inorganic fraction include:

• hydrous aluminosilicate clays of largely terrigenous origin usually predominating along continental margins and in open-ocean red clay deposits,

• calcite and aragonite remains that are most abundant in carbonate oozes accumulating on ocean topographic highs such as mid-ocean ridges, and

• opal oozes formed largely from diatom debris accumulating beneath highly productive surface ocean waters in upwelling zones along continental margins and at equatorial and high latitudes.

Because samples of suspended, sinking, and sedimentary marine particles often contain appreciable amounts of clay minerals, carbonate, and opal, any quantification method for organic elements must avoid interference from these mineral types.

Organic carbon is by far the most commonly quantified organic element in marine samples, for which CaCO₃ is the most typically encountered interference. This problem occurs because carbonates decompose upon heating, releasing CO₂, the same gas that CHN analyzers measure after the oxidation of organic matter. Early attempts to distinguish these two sources by selectively breaking down CaCO₃ at a lower temperature (400 to 650 ºC) than is used to oxidize organic matter (greater than 1000 ºC) failed because the temperature ranges over which CO₂ is generated by these two degradation processes overlap substantially (Froelich, 1980; Yamamuro and Kayanne, 1995). The only practical approach to analyzing organic and inorganic carbon separately has been to acidify samples at low temperature so that inorganic carbon can be quantitatively evolved as CO₂ while organic carbon remains behind relatively unaltered in the solid sample. For example, Weliky et al. (1983) directly quantified inorganic carbon by adding hydrochloric acid (HCl) to a known mass of sediment in a closed container and measuring the released CO₂. Although organic carbon can then be determined by the difference between total carbon (measured by combusting an untreated sample at high temperature) and inorganic carbon, this difference approach is imprecise for small amounts of organic carbon in carbonate-rich samples.

Most methods for directly measuring POC involve the use of a non-oxidizing acid (e.g., HCl, H₃PO₄, H₂SO₃) to remove carbonate prior to quantifying the organic carbon using high temperature combustion. Early methods of this type often involved preparative treatment of solid samples in large volumes of acid solution, after which the remaining solids were separated (by filtration or centrifugation), rinsed, dried, and weighed for high temperature combustion. Both organic matter and non-carbonate mineral matter could be dissolved and lost with the discarded treatment water (Froelich, 1980; Hedges and Stern, 1984), however. Such losses occur to varying extents depending on sample type and are especially pronounced for carbonate oozes (Yamamuro and Kayanne, 1995).

Once this problem became clear, later methods for organic carbon analysis used procedures in which only volatile substances were allowed to leave the sample, and any treatment water was retained. Acidification was accomplished using either HCl vapor (Hedges and Stern, 1984) or liquid acids added to watertight containers (Verado et al., 1990; Nieuwenhuize et al., 1994). Even in the absence of appreciable carbonate, open ocean clays often contain such low concentrations of organic carbon (approximately 0.1 to 0.3 wt percent) that precise measurement is difficult because of the small carbon blank intrinsic to all CHN measurements (Nieuwenhuize et al., 1994). In theory, organic and inorganic carbon can

now be measured accurately for all major sample matrix types. Whether this is in fact the case across the oceanographic community is unknown because no intercomparisons have been reported for different mineral matrix types.

Many of the same matrix effects mentioned above affect stable and radiocarbon isotopes. An additional constraint is imposed on the analysis of organic carbon for δ¹³C measurements because the CO₂ generated during combustion must be pure enough to introduce to an isotope ratio mass spectrometer. This has not presented a problem in practice because the small amount required for stable isotope analysis permits the isolation of very clean CO₂. The size of the sample required for AMS analysis (40-80 µmol C) for radiocarbon measurements presents an even greater analytical challenge. With some methods (e.g., H₂SO₃ dissolution), it is very difficult to determine the point at which all of the carbonate has been removed from a large sample (Verado et al., 1990). Other methods leave residual salts that produce large amounts of undesirable gases upon sample combustion. In addition, many of the salts are hygroscopic and the decarbonated samples must be cleaned prior to combustion. In practice, many laboratories resort to acidification, followed by rinsing. The resulting removal of some of the organic matter may affect the radiocarbon age measured on sediments.

Nitrogen presents its own unique analytical challenges in solid marine samples. First and foremost, all forms of inorganic and organic nitrogen are converted in CHN analyzers to N₂ gas, which thus represents total, rather than strictly organic nitrogen. Although this is not a major problem for most organic-rich samples, open ocean clays can have concentrations of organic carbon substantially less than 1 weight percent (Suess and Müller, 1980). Thus with such low fractions of organic matter, a major fraction of the total nitrogen in these clays is derived from fixed ammonia rather than organic nitrogen. Acid treatments, used to remove inorganic carbon prior to CHN analysis, have a variable effect on measured nitrogen contents, further complicating analysis. A second complication is the variable effect on measured nitrogen content that is caused by the acid treatment frequently used to remove inorganic carbon prior to CHN analysis. Although this complication is not always observed (Hedges and Stern, 1984), comparisons of total nitrogen content across

different sediment types are preferably done on untreated samples (Gélinas et al., 2001b).

CHN analysis of most marine particle samples has typically ignored hydrogen because at temperatures over 1000ºC generated in CHN analyzers, both opal (SiO₂.nH₂O) and hydrous aluminosilicates decompose with the evolution of water. For most marine sediments, resulting yields of mineral-derived water greatly exceed those generated by organic matter oxidation, making organic hydrogen analyses impossible. Acidification of CaCO₃ with HCl yields CaCl₂.nH₂O, an extremely hygroscopic salt that releases copious amounts of water upon heating, further complicating analysis. Finally, many types of organic matter hold water tenaciously, which elevates organic hydrogen values unless the sample is heated to high temperatures, which can cause biochemical decomposition. These complications in directly measuring organic hydrogen in marine materials is unfortunate given the importance of this element as an indicator of biochemical composition, biological oxygen demand, and petroleum-forming potential.

Oxygen lies somewhere between organic carbon and hydrogen with respect to its analytical difficulty. In samples that are largely organic (e.g., marine macrophytes and mineral-poor plankton), organic oxygen content can be estimated by subtracting the mass of carbon, hydrogen, nitrogen, and “ash” from the total sample mass. This mass difference approach, however, assumes the other three organic elements can be quantified accurately which is often in question (as detailed in previous discussion). In addition, “ash” is operationally defined as the mass of mineral matter that remains after all the organic component has been oxidized away by holding the sample under air in a furnace at a high temperature (400 to 600 ºC). Residual ash is assumed to have the same mass as the original inorganic matter in the sample, so that this value can be subtracted from the initial total sample mass to estimate the initial organic mass. However, severe heating during ashing can cause inorganic components of the sample to either lose (e.g., by evolving CO₂, SO₂, and H₂O) or gain (e.g., by adding oxygen to iron and sulfur) mass, making accurate ash corrections difficult for mineral-rich samples. An alternate, and potentially much more accurate approach, is to measure organic oxygen evolved pyrolytically as CO at a temperature near 1000 ºC. This is typically done in an oxygen-free gas stream within a modified CHN analyzer, where the

evolved oxygen compounds are converted over an activated carbon bed to CO, which is then measured with a thermal conductivity detector. In practice, this direct oxygen method is seldom used. When it is, precision is typically poorer than for carbon, hydrogen, and nitrogen (Cheng-Tung et al., 1996).

Measuring the type and amount of organic substances in marine systems involves major challenges, some of which are unique to large, covalently-bonded molecules. Primary among these analytical hurdles is that essentially all marine biomass, seawater, or sediment samples contain thousands of different molecules that cannot all be analyzed together. Even the most concerted efforts to individually quantify the biochemicals in marine samples have accounted for only 80 percent of the total organic matter in living plankton, and less than 25 percent of the molecules in seawater and sediments (Wakeham et al., 1997). Such inventories remain incomplete in large part because each biochemical class (e.g., amino acids, sugars, lipids, and pigments) requires a different analytical procedure. As a result, a comprehensive survey across all the quantitatively important molecular types is logistically impractical.

The challenge of specific molecular analysis is compounded by the constraint that structurally definitive measurements of individual organic compounds are still primarily limited to relatively small (less than 1000 atomic mass units [amu]) organic molecules that can be chromatographically separated and quantified against commercially available chemical standards. In contrast to lipids, most organic substances in ocean samples exist in large molecules (e.g., proteins and polysaccharides) that must be chemically broken down into small structural units (e.g., amino acids and sugars) in order to be extracted from the sample matrix, identified, and quantified. Even in the simple case of a pure macromolecule, chemical treatments (e.g., hydrolysis or oxidation) severe enough to efficiently release small structural units often alter these desired products before the parent macromolecule can be completely broken down. Quantifying multiple compound types in organic mixtures is even more difficult because the reaction conditions for maximum conversion to measurable structural units seldom overlap, and the freed structural units can react with each other. Hydrolysis-resistant macrobiomolecules such as lignins, algaenans, and plant chars require extremely severe breakdown reactions such as pyrolysis and nitric acid oxidation, which typically yield only a fraction of the parent material in greatly altered form.

Finally, but critically, the matrix of a sample can have a profound effect on measurements of individual organic molecules within it. Such matrix effects can take many forms and be complexly interrelated. Organic substances in seawater and sinking or sedimentary particles are intimately associated with the much greater amounts of inorganic materials such as sea salt and mineral grains present in the marine environment. It is necessary, therefore, that the organic moiety either be isolated before analysis, or that it be broken down within (and extracted from) these matrices. Unfortunately, physical or chemical separation of organic substances from inorganic matrices of marine samples is almost never complete and often involves combinations of severe reagents (e.g., HF:HCl or NaOH:HCl) that can lead to appreciable chemical alteration. Thus, many types of useful characterizations (e.g., infrared, Raman, or NMR spectroscopy) that are possible for purely organic samples are infeasible for the same dilute components of natural mixtures. Matrix effects that can occur when the target analyte is generated within and extracted from an intact sample include:

• reaction of added chemical agents with the matrix instead of with the desired analyte precursor,

• physical shielding of the analyte or its precursor from the reagent or extraction solvent,

• generation of chemicals from the sample that react with the target analyte,

• uptake of the released analyte by the matrix, and

• coextraction of a matrix component that subsequently interferes with analysis of the target material.

Analysis of specific compounds in marine samples is currently constrained to a restricted number of organic components within a complicated mixture of organic matter of widely varying composition. This restriction is similar in many ways to the problem of determining the “speciation” (exact chemical form) of inorganic elements. Although it may be relatively straightforward to quantify the total amount of a component element (e.g., organic carbon) in a sample, it is extremely challenging to define its actual chemical forms and proximate environments as they existed at the time the sample was taken. However, speciation of an element or molecule is often the main determinant of both its reactivity and information potential. Even though molecular-level organic analyses are incomplete and capture only a fraction of the total information available in the intact sample, they nevertheless can provide unique and invaluable insights in both the form of qualitative (e.g., ratios and homo-

logue fingerprints) and of quantitative parameters (e.g., amounts, ages, and fluxes).

Given this wealth of potential information, the many types of molecules that can be analyzed, and the relatively short time (approximately 50 years) that a small number of biogeochemists have studied the ocean, it is understandable that a diverse array of analytical methods have evolved around individual types of organic compounds. The current state of the art in marine organic geochemistry is that each lab group typically uses its own home-grown method for a given compound class. Often, but not always, the procedure is traceable to a rigorous methods paper in which analyte identities have been confirmed and their recoveries from a limited set of sample matrices have been systematically maximized. Seldom, however, is the analytical method rigorously tested for all the matrix types to which it is eventually applied, nor are the consequences of this adaptation of a published procedure always studied. Because of the many individual complex steps by which the target molecules are generated, extracted, isolated, derivatized, and eventually quantified, the overall outcome of an organic analysis can be very sensitive to small differences in laboratory techniques and available equipment. Studies of analytical drift over time within individual labs, as well as rigorous intercomparisons among different labs using contrasting analytical procedures for the same compound types are rare. The overall result of this analytical discontinuity is that it can be extremely difficult to bring data from multiple sources together in the study of extended environmental and temporal trends that are increasingly critical at this time of unprecedented global change. Analytical results from different organic measurements, however, could be effectively related by routine comparisons to reference materials that represent the matrices encountered in the environmental samples being analyzed (e.g., Fig. 2.1).

The analysis of pigment biomarkers in microbial cultures and marine particles is relatively straightforward, with the exception of pigments in armored dinoflagellates, heavily silicified diatoms and heavily walled green algae, which can be notoriously difficult to extract (Wright et al., 1997). The extraction of lipid biomarkers from certain sediment matrices is even more problematic. This is especially true for carbonate sediments, where extraction efficiencies and molecular distributions can vary significantly and can depend upon the type of extraction techniques used (e.g., wet or dry solvent extraction and acid or base hydrolysis, followed by solvent extraction).

Amino acids are generally considered to be labile and easily analyzed in most marine matrices. However, recent studies have shown that a significant percentage of the proteinaceous component of marine particulate matter is not accessible using traditional methods (Hedges et al., 2001). The actual source of the uncharacterized fraction is not known, but at least part of it is undoubtedly associated with mineral matrices. For example, the acidic amino acids—aspartic and glutamic acid—are enriched in carbonate skeletons (Degens and Mapper, 1976) and are preferentially adsorbed onto carbonates (Carter and Mitterer, 1978). Organic molecules serve as a template for the secretion of all biogenic minerals (Lowenstam and Weiner, 1989). Usually rich in glycoprotein, this organic matter can be occluded in the skeletal matrix, which protects it from degradation unless the mineral is dissolved (King, 1974; Robbins and Brew, 1990). Sediments with high biomineral concentrations may contain a large fraction of organic carbon in skeletal matrices (Ingalls et al., in press). Organic matter preserved in biominerals currently serves as a proxy for numerous paleoenvironmental studies. Some examples are: the calculation of historic surface water pCO₂ (Maslin et al., 1996), estimates of δ¹⁵N-based paleoproductivity (Sigman et al., 1999) and species identification (King, 1974; Robbins and Brew, 1990; Collins et al., 1991; Endo et al., 1995).

Molecular-level carbohydrate analyses are challenging for several reasons. First, acid hydrolysis releases simple sugars which are prone to dehydrate (or otherwise degrade) before the parent polysaccharide can be completely depolymerized. In addition, hydrolysis and isolation conditions for neutral carbohydrates are often inappropriate for acidic and basic sugars (Benner, 2002), which may also interact differently with matrix minerals due to their charges. Finally, sugars also can condense with amines, including mineral-derived ammonia, to form highly altered nitrogenous products that escape measurement. A recent study of mineral-rich plankton tow materials (Hedges et al., 2001) indicated that only 25 percent of carbohydrate-like material (as determined by ¹³C NMR) was measurable chromatographically as neutral sugars.

Measurement of radionuclides in marine solids may be accomplished by either destructive or nondestructive techniques. The nondestructive

technique involves direct measurement of gamma rays emitted during radionuclide decay. The interaction of these gamma rays with the sample is a function of the energy of the gamma-ray and the composition of the sample; low energy gamma rays are strongly absorbed by sedimentary minerals. For example, ²¹⁰Pb is often measured by direct counting of low energy gamma rays (46.5 keV) that are partially absorbed by minerals within the sample matrix. The fraction of the gamma rays absorbed by the sample is a function of the sample matrix. There are methods to account for self-absorption, but there are no widely distributed reference materials to test these corrections. Similar problems exist for measuring gamma rays from other radionuclides.

Measurement of radionuclides by destructive analyses can involve combustion, fusion, leaching, or acid-dissolution of the samples. The choice of technique depends on both the radionuclide of interest and the sample matrix. Techniques developed for one matrix may not work for other matrices. In addition, many radionuclides in sediments exist both as surface-bound components adsorbed by particles sinking through the water column, and also as a structural component within physically intact mineral grains. These detrital structural components generate a background of supported activity that does not change with time. Thus, to use nuclides such as ²³⁰Th and ²³¹Pa for dating, the radioactivity derived from uranium decay in the solid (supported activity) must be subtracted from that derived from adsorbed ²³⁰Th and ²³¹Pa (unsupported activity). Different analytical techniques for these physically dissimilar phases may lead to different corrections for supported activity, which has yet to be separated with convincing conclusions.

For the ²²⁶Ra:Ba ratio in barite technique, the results depend on the isolation of a pure barite sample and the assumption that the separation techniques do not alter the ²²⁶Ra:Ba ratio of the barite. Applying different techniques to reference sediments could strengthen this technique.

A selected list of reference materials (sediments as well as biological tissues) distributed by several Canadian, U.S., and E.U. sources shows a wide range of solid samples that could be used for comparative analysis of major organic elements (Table 4.2). These materials are widely available and have been analyzed for at least some constituents. In addition, these materials are homogeneous and can be expected to exhibit stable compositions over time. All of the thirty or so listed reference materials,

however, lack precise measurements of the major bio-organic element compositions of the materials. The trace organic components that are certified, such as PAH, PCBs, pesticides, and hydrocarbons, do not constrain the concentrations of the organic elements. These samples do have potential for use as consensus reference materials, but only after elemental data are provided by different labs for selected samples using contrasting and well-defined analytical procedures.

The matrices and sources of the sediments listed in Table 4.2 are sometimes unclear. Those that are known are highly weighted toward clastic (quartz- and aluminosilicate-rich) marine sediments from coastal environments. Some of these reference materials, such as MESS-3 (NRC-Canada), MAG-1 (USGS) and the Arabian Sea and Pacific Ocean samples (IAEA 315, and 368), could provide excellent examples of clastic marine sediment representing the main repositories of organic matter in the ocean (Hedges and Keil, 1995). The listed materials fail to include both open-ocean opal and carbonate oozes, as well as pelagic red clays.

The biological reference materials listed in Table 4.2 represent a small subset of all the biomass samples available for analysis, though it is representative of the emphasis placed on animals versus macrophytes. Reference materials for plankton are particularly limited, although samples of the freshwater green alga, Chlorella, are available for trace element analysis from the Slovak Institute of Radioecology and Japan’s National Institute of Environmental Studies. Notably, no reference materials appear to be available for marine phytoplankton of any type. This dearth of marine biological material for elemental analysis is unfortunate because it fails to represent the opal and carbonate matrices that can greatly complicate carbon, oxygen, and hydrogen analyses.

Organic and carbonate reference materials for δ¹³C and ∆¹⁴C are both available and widely used for isotopic analysis. The stable carbon isotope reference materials currently available are useful for instrument calibration and limited methods verification. These materials are available as powders for carbonates and in a wide range of forms for organic materials (e.g., oil, graphite, etc.) (Table 4.3). There is one certified primary standard for radiocarbon (NIST SRM 4990C) and a number of additional reference materials are available from the International Atomic Energy Agency (IAEA) in Vienna, and from Marian Scott at the University of Glasgow (Table 4.4, Appendix E). The FIRI samples represent a study in progress by an international group of radiocarbon researchers (Bryant et al., 2001). Samples from these sources are available primarily as discrete organic or carbonate materials and do not represent materials (such as

sediment) that have small amounts of organic matter present in a complex mineral matrix.

A wide range of high quality, non-certified carotenoid and chlorophyll chemical standards are commercially available (e.g., Sigma-Aldrich, DHI, and Roth). The availability of a mixed pigment reference standard and biological matrix reference materials would improve analytical performance in individual laboratories, facilitate method and laboratory in-

ter-comparisons, and help resolve inter-method variability during oceanographic expeditions. A variety of high quality, non-certified organic chemical standards are commercially available, e.g., from Alltech (fatty acids), Pierce Chemical Co. (amino acids), Sigma-Aldrich (phytol and a wide range of fatty acids and amino acids), Steraloids, Inc. (a wide range of marine and terrestrial sterols), and Chiron AS (diagenetic products of key lipid biomarkers, including deuterated internal standards).

Several organizations (e.g., NIST, NRC-Canada, and IAEA) provide sediment reference materials containing radionuclides, many of which are only certified for artificial radionuclides (¹³⁷Cs, ⁹⁰Sr, ²⁴¹Am, and ²³⁹Pu). Certain specific radionuclides have no certified natural matrix materials, including ocean, lake, and river sediments. Although these sediments are certified for a few naturally occurring and artificial radionuclides, the extent of radioactive equilibrium of the uranium and thorium decay series in these environmental materials is not provided. NIST currently offers an ocean sediment Standard Reference Material³ (SRM 4357) in

which ²³⁰Th, ²²⁶Ra, ²³²Th, ²²⁸Th, and numerous artificial radionuclides are certified, and which give non-certified activities of uranium isotopes, ²¹⁰Pb, ²²⁸Ra, and additional artificial radionuclides. This reference material is a blend of sediments collected from both the Chesapeake Bay and from the seafloor off of the British Nuclear Fuels Sellafield facility in the United Kingdom. The IAEA supplies reference materials in the form of a number of marine and terrestrial sediments, soils, and ores, some of which are certified for long-lived members of the uranium and thorium decay series, as well as trace elements. The IAEA reference materials are prepared in limited quantity and replaced frequently.

Radiochemists require reference materials with matrices that reflect sediments with widely different composition. For example recently deposited deep-sea sediments contain ²³⁰Th and ²³¹Pa primarily as surface-adsorbed components, whereas sediments collected on (or close to) continental margins contain a higher proportion of structurally-bound components. No reference materials exist that can be used to test selective leaching procedures designed to identify how these radionuclides are incorporated into the matrix. Additionally, no reference materials exist with which to evaluate matrix-correction procedures in gamma-ray analyses or to compare analytical results of radionuclide measurements from different laboratories. To correct this problem, radiochemists need solid reference materials containing known amounts of the longer-lived radionuclides with an estimate of the degree of radioactive equilibrium through the series.

Reference materials that represent the primary deep-sea and coastal depositional environments and biological materials would solve many of the problems that radiochemists face in analysis of sediments from these settings. Radiochemists require reference materials comprising the primary end member sediment and biological types (calcium carbonate, opal, and red clay from the deep-sea and carbonate-rich, silicate-rich, and clay mineral-rich sediments from coastal environments and representative biological materials). Additional sediment reference material from a river delta would be valuable to test the release of radionuclides that occurs as riverine particles contact seawater.

Given that analyses of particulate and sedimentary elements and compounds are highly matrix-dependent, the committee recommends that reference materials be made available for ten important matrix types, covering three biological and seven sedimentary forms described below, which also cover a large variety of diagenetic states. Rather than recommend that these reference materials be certified for a wide variety of

constituents, the greatest value will be derived by producing stable, homogeneous materials that can be used to derive community consensus values for many specific analytes. Further certification could then be accomplished as needed.

The diversity of marine photosynthetic microalgae and bacteria is extensive, making it impractical to develop reference materials for all of the species currently in culture. For the three biological matrices, it is recommended that diatom (Thalassiosira pseudonana), dinoflagellate (Scrippsiella trochoidea), and haptophyte (Emiliania huxleyi) mass cultures be developed as reference materials in order to improve analytical performance and facilitate inter-laboratory comparison. These three organisms provide a wide range of oceanographically relevant mineral, trace metal, and organic analytes for the establishment of a comprehensive list of consensus values (Table 4.5). They represent three major matrices—opal, carbonate, and organic matter. They also include the two mineral phases thought to be important “ballast” materials for facilitating preservation and vertical export of POM to the sediments (Armstrong et al., 2002). Consensus analyte values (e.g., isotope ratios) can be obtained for both the inorganic (calcium carbonate, opal) and organic (POC, PN, and biomarkers) phases of these biological matrices. Biological reference materials can also be used for the preparation of both C_37:2 and C_37:3 alkenone standards (isolated from the E. huxleyi) and a mixed pigment standard containing a wide range of individual chlorophylls and carotenoids (prepared by mixing acetone extracts obtained from the three phytoplankton reference materials).

Recommended sedimentary reference materials include three separate carbonate-, opal- and clay mineral- rich open-ocean sediments; three coastal sediments containing the same three mineral types; and a deltaic sediment that has not been in contact with seawater. To include relatively young material, each of these seven samples should be taken from within the top 10 cm of the sedimentary column. Table 4.6 suggests possible locations to sample the types of material matrices of interest. The combination of both coastal and open-ocean sediments would provide a wide range of diagenetic states, with the previously described phytoplankton reference materials providing fresh organic matrices and open-ocean redclay material providing some of the most degraded sedimentary material that exists in the marine environment. The mineral types represented in

Table 4.6 (opal, carbonate and aluminosilicate) were chosen to provide end members for matrix analysis. These samples also could be blended to produce any mixture desired. Sampling sediments from well-studied depositional regions such as MANOP Site R in the North Pacific (for open-ocean red clay) and Florida Bay (for carbonate sediments) would offer a variety of supplemental information that is already available. In addition, the (Atchafalaya) river sediment would provide a useful link between those studying terrestrial and marine processes.

As riverine particulate matter encounters seawater, a variety of changes may occur to the terrestrial particles. Some surface-bound ions are released to solution as they exchange with dissolved species. Such exchange reactions are especially important in understanding the geochemistry of radium. Evaluation of these exchange reactions is often approximated by collecting sediment from the freshwater region of an estuary and exposing it to seawater. However, such experiments may contain artifacts due to a lack of control of the simulated conditions. A freshwater deltaic reference sediment could be employed to help refine these experiments and eliminate artifacts.

Some primary standards exist for the uranium to thorium series isotopes currently used extensively in the field. There is a need, however, to produce a certified reference material for uranium and thorium decay series isotopes with masses greater than ²²⁶Ra, using a natural material such as uraninite. Sediment dating/mixing studies would benefit from a marine sediment reference material containing ²¹⁰Pb. The committee recognizes that producing this material in different matrices would be difficult, though not impossible. Few samples contain background levels for the cosmogenic and bomb-produced radionuclides ¹⁰Be, ³⁶Cl, ²⁶Al,

²⁴¹Am, ⁹⁰Sr, ²³⁹Pu, ²⁴⁰Pu, ¹³⁷Cs, ¹²⁹I, or ¹⁴C, which also lack solution standards. NIST has an active program to address the development of all of these radionuclide standards except ¹⁴C.

Long-term needs for organic reference materials include ether-linked lipids, hopanoids, pheopigments, and mycosporine-like amino acids (MAA). Members of the Archaea possess diagnostic ether-linked isoprenoid lipids that are used in the construction of monolayer or bilayer cell membranes. These ether-linked lipids occur as glycerol diethers or diglycerol tetraethers. Hopanoids are chemical components found in bacterial cell membranes and are thought to play an important role in membrane stabilization. Ether-linked lipids and hopanoids are useful biomarkers for detecting the presence of Archaea and Bacteria in environmental samples. Pheopigments are magnesium-free chlorophyll degradation products that are useful as tracers of heterotrophic processes such as zooplankton grazing. MAAs are imino-carbonyl derivatives of mycosporine cyclohexenone. They are structurally diverse and widely distributed among marine cyanobacteria, microalgae, invertebrates and vertebrates. MAAs absorb strongly in the 310-360 nm waveband and are thought to protect marine organisms from the damaging effects of ultraviolet radiation. It is likely that the need for quantification of these organic compounds will grow in the future. Consequently, future matrix-based reference materials should be prepared from materials that contain these analytes.