Mapping Spatial Variability in Marsh Redox Conditions Using Biomass Stable Isotopic Compositions

Carol Kendall, Steven Silva, Daniel Steinitz, Erika Wise, Cecily Chang  (USGS, Menlo Park CA);  Jerry Stober, Phyllis Meyer  (USEPA, Athens GA)

Abstract Zones frequently dominated by particular redox reactions appear to be labeled by the C, S, and N isotopic compositions of local organisms.  Isotopic compositions of biomass may prove to be more cost-effective and reliable indicators of prevailing environmental conditions that favor methylmercury production than other parameters currently being considered because biomass isotopic compositions are much more difficult to perturb than the more transient concentrations of aqueous species.

Study Design As part of a collaboration between the USGS and the USEPA, periphyton, mosquitofish (gambusia), and sediment samples were collected during September 1996 at over 100 REMAP marsh sites throughout the Everglades to assess the local and regional ranges in their C, N, and S stable isotope ratios.  These organisms were chosen because periphyton communities play an important role in the ecosystem, and the ubiquitous mosquitofish might be a useful indicator species. Archived periphyton samples collected in May 96 were also analyzed for isotopes to assess possible seasonal changes.  The periphyton samples were composites of floating mat samples; samples were vapor acidified before analysis to remove carbonate.  Five gambusia were collected at each site; they were filleted and analyzed individually. The sediment samples were composites of material collected in the top 0-10 cm below the floc layer; they were vapor acidified before analysis to remove carbonate.

Theoretical Basis Biochemical reactions such as denitrification and methyl mercury (MeHg) production in the shallow subsurface sediments in aquatic systems and watersheds can have profound effects on surface-water quality.  These processes are controlled by redox conditions in the sediments or water column and, hence, may be strongly affected by transient changes in oxygen, nutrient, and trace element levels caused by seasonal changes in hydrology and anthropogenic inputs. Because of the ephemeral nature of chemical signals produced by these reactions, it is often difficult to assess the spatial and temporal extent of environmental conditions that favor these reactions, or the degree to which these reactions affect ambient water quality, by conventional chemical or isotopic measurements of dissolved species. Environmental Indicators:  We are attempting to use the 15N, 13C, and 34S of biota in marshes and canals as (1) indicators of local environmental conditions that may impact water quality, and (2) indicators foodweb structure (see below). The theoretical basis of this study is that the isotopic compositions of non-fixing, primary producing plants reflect to a large extent the isotopic compositions of the dissolved N, C, and S in the environment that is being utilized by the biota, as modified by various possible fractionating mechanisms in the plants. The isotopic compositions of the primary producers are then reflected by higher level organisms such as invertebrates and fish, as modified by mixed diets, trophic enrichments, and the larger foraging areas and lifetimes of the higher level organisms. Hence, under favorable conditions, the biomass isotopic compositions can reflect (and integrate) the extent of such processes as denitrification, sulfate reduction, and methane oxidation in the sediments and water column that affect the isotopic compositions of the dissolved species.  Fig. 1 (below) is a schematic for how the isotopic compositions reflect various biogeochemical processes and nutrient sources in the ecosystem.

Figure 1 (Click on image for larger version.)

Foodweb Structure:  The traditional method of foodweb investigation focused on the determination of gut contents (literally, "who ate what"), and is still used today.  More recently, stable C, N, and S isotope analyses of plants and animals have been used to establish relative trophic (predator) levels among various organisms because at each ascending trophic level (from prey to predator), there is an increase in the 13C content (13C value) and 15N content (15N value) of the organism due to selective metabolic loss of 12C and 14N during food assimilation and growth.  Thus, an organism is typically enriched in 13C and 15N relative to its diet by 1 to 3 parts-per-thousand (). In other words, "you are what you eat plus a little bit". This "little bit" is called a trophic fractionation. There appears to be little or no enrichment in 34S with increasing trophic level. The 15N, 13C, and 34S of aquatic plants (e.g., phytoplankton and algal mats) and animals (e.g., invertebrates and fish) reflect the isotopic compositions of the dissolved N (in nitrate and ammonium), C (in bicarbonate), and S (in sulfate) in the environment as modified by mixed diets and trophic enrichments (about 2-3  per trophic level for 15N and 0-1   per trophic level for 13C). This conceptual model is illustrated in Fig. 2 (below).

Figure 2 (Click on image for larger version.)

Results The five gambusia from each site showed little variability. Fig. 3 (below) shows histograms of the standard deviations of the average Delta 15N and Delta 13C values of the sets of 5 or 10 gambusia collected at each site. The average 1 sigma variations in 15N and 13C were 0.4 and 0.7 , respectively.  The good agreement in their compositions indicates that gambusia collected at the same place exhibit a well-mixed, consistent diet, over the life-time of the gambusia (despite the large range in their observed gut contents).

Figure 3 (Click on image for larger version.)

The 15N and 13C values for gambusia, periphyton, and sediment samples are shown in Fig. 4 (below). The average 15N values of gambusia and bulk periphyton are +8.4 and +0.9 , respectively; the gambusia and periphyton 15N values at each site are well correlated (r2 = 0.6). This +7.5  difference suggests that gambusia are 2-3 trophic units above bulk periphyton.  The average 13C values of gambusia and bulk periphyton are -27.5 and -26.1 , respectively.  This -1.4  difference (gambusia has a lower 13C than periphyton) is anomalous. If bulk periphyton is the dominant food source to the gambusia foodchain, the 2-3 trophic unit difference observed with 15N (i.e., +7.5 ) should be correlated with a 0 to +3  increase in 13C from bulk periphyton to gambusia.

Figure 4 (Click on image for larger version.)

Possible explanations for this puzzling discrepancy between the 13C values of gambusia and bulk periphyton include:  (1) gambusia (and the organisms they eat) generally do not derive significant C from periphyton mats, (2) gambusia (and the organisms they eat) primarily derive C from some fraction of the periphyton mat (perhaps diatoms or bacteria) that has a 13C value that is 0-3  lower than the bulk periphyton (i.e., what the gambusia eats might not be the same as what it assimilates into biomass), or (3) there is a weird "reverse" fractionation during assimilation of some types of algae by herbivores (perhaps because of poor food quality). This puzzle is a topic of active research; at some sites, we have strong evidence for possibilities #1 and/or #2. To further explore possibility #1, we are sampling alternative C sources and assessing possible variability in the mats by picking them apart and subsampling. In addition, collaborations with Paul McCormick (SFWMD) have generated an interesting set of data; algae were collected from sets of growth plates set at 3 sites at site U3 (in WCA2A) for 9 different 1-week time periods.  The large range in isotope and elemental values over time and space as different organisms colonized them indicated that it might be difficult to resolve which types of algae are actually being assimilated by local organisms. Recent attempts to use compound-specific stable isotope techniques to trace individual molecules up the foodchain will provide more insight into #2, and planned controlled growth experiments (in the lab and in mesocosms) will help resolve possibility #3. Over 50% of the sites have gambusia with 13C values that are lower than the corresponding periphyton 13C values. The distribution of these sites is shown in Fig. 5 (below).  Sites are divided into 3 categories, based on the relative 13C values of gambusia and bulk periphyton collected at the same site:  (1) sites where the gambusia 13C values are higher than the periphyton 13C values (blue), consistent with theory, (2) sites where the gambusia 13C values are roughly equivalent to periphyton 13C values (green), and (3) sites where the gambusia 13C values are considerably lower than the periphyton 13C values  (yellow), inconsistent with theory.  This model represents a "testable hypothesis" for the spatial distribution of sites dominated by algal-based foodwebs vs sites dominated by detrital-based or other types of foodwebs.  Sampling locations are shown as "dots."

Fig. 5 (Click on image for larger version.)

Interestingly, this plot (Fig. 5) shows strong, cohesive, spatial patterns (instead of looking like random data), suggesting that the patterns reflect some "real" property of the ecosystem. When this "foodweb-base model" is compared with data collected at nearby USGS sites, we find a fair amount of agreement.  For example, U3, a site where a large dataset collected by the USGS (and others) strongly suggests that periphyton is an important C source to local foodwebs, plots in an area on this map where the REMAP samples indicate that periphyton 13C values are "consistent" with being an important food source. And site 3A-15 and some WCA3A sites where detrital material appears more important, fall in the "inconsistent" area on Fig. 5. Hence, there is some reason for optimism that this spatial plot might provide a way to extrapolate foodweb relations determined at well-studied USGS sites to the larger ecosystem. In any event, the discrepancy between the 13C values of bulk periphyton and gambusia for the REMAP sites is providing a useful perspective on spatial variability in how the marsh systems "work."

What kinds of environmental information will we get from the isotopic compositions? POM (particulate organic matter) isotopes: POM source information:  macrophyte detritus vs in situ production in the water column local biogeochemical reactions in the water column and/or sediments. Macrophytes (e.g., cattails, sawgrass, eleocharis) isotopes: reflect the C/N/S isotopes of atmospheric CO2,(and CO3), NO3 (and/or NH4), and SO4 in the ambient water (N-fixing plants utilize atmospheric N2 instead) local biogeochemical reactions in the sediments (where the roots uptake nutrients). Algae isotopes: reflect the C/N/S isotopes of the CO3, NO3 (and/or NH4), and SO4 in the ambient water (N-fixing algae utilize atmospheric N2 instead) local biogeochemical reactions in the water column. Invertebrate (e.g., shrimp, snails, insects) isotopes: integrative assessment of the short-term, site-specific variations in food sources trophic level information. Fish isotopes: integrative assessment of long-term, large-scale, variations in food sources (because fish can migrate in and out of the marshes) trophic level information.

Spatial Patterns in Isotopic Compositions In the Everglades, both periphyton and mosquitofish show wide ranges in isotopic composition throughout the marshes, with about a 10  range in  values for each isotope and species (Figs. 6-8). However, the spatial contour plots of the isotopic compositions display intriguing regional patterns that show little change with time or species. This is strong evidence that the environmental patterns reflected by the primary producers are being incorporated up the foodweb. The isotopic values show a general correspondence with peat thickness, with the lowest 13C values and highest 34S values along the major axis of the flow system. Areas with high MeHg contents commonly have high 15N and 34S values; these are areas where sulfate reduction and denitrification are dominant processes in the sediments. Zones frequently dominated by particular redox reactions appear to be "labeled" by the 13C, 15N, and especially the 34S of local organisms.

Delta 13C of Periphyton	Delta 13C of Periphyton	Delta 13C of Gambusia

Figure 6 (Click on image for larger version.)

Carbon Isotopes: Fig. 6 (above) shows the spatial distribution of 13C values in periphyton mat samples collected in May vs September.  The sample locations are noted as "dots."  Because of dry conditions in May, periphyton mats were not present at many sites in the NW and SE (note the areas marked as "dry"). There is a general correspondence between the 13C distributions in May vs September, with the lowest (most negative) 13C values found along the major NE-SW flow axis where the peats are generally thickest. One overly simplistic explanation for the lower 13C values where peats are thickest (i.e., where the peats contain the most reduced C) is because this is where methane production (followed by oxidation, which produces lower 13C values -- see Fig. 1), is likely to dominate. The 1   lower periphyton 13C in May vs September is consistent with the idea that redox chemistry may be an important control on 13C values because it is likely that the system is more reducing when the waters are shallower and more clogged with rotting vegetation. However, as shown in Fig. 1, there are a number of processes that can affect the 13C of organisms. Fig. 6 shows that the spatial distribution of 13C values in periphyton is very similar to the 13C patterns seen in gambusia.  Although Fig. 4 shows that the relationship between the 13C values of gambusia and periphyton is neither constant nor simple, on the "gross" scale of Fig. 6, both organisms appear to reflect a similar spatial distribution of environmental biogeochemical factors.

Delta 15N of Periphyton	Delta 15N of Periphyton	Delta 15N of Gambusia

Figure 7 (Click on image for larger version.)

Nitrogen Isotopes: Fig. 7 (above) shows the spatial distribution of 15N values in periphyton mat samples collected in May vs September. Although some of the 15N "hotspots" remain constant over time, in general there is little correspondence between the seasons. As shown in Fig. 1, high 15N values are characteristic of denitrification in anoxic environments; the fact that the high values are found to the south and not adjacent to the EAA indicates that these patterns are less caused by source differences than by subsequent biogeochemical cycling. Fig. 7 shows quite good agreement between the spatial distributions of high 15N values of periphyton and gambusia.  As shown in Fig. 4, the actual corresponding 15N values at any site are about 7  different because of the 2-3 trophic levels between periphyton and gambusia. The strong spatial correspondence of high values in gambusia and periphyton suggests that denitrification (or some other process producing high 15N values) has caused the entire foodweb at these sites to be isotopically "labeled."

Delta 34S of Periphyton	Delta 34S of Periphyton	Delta 34S of Gambusia

Figure 8 (Click on image for larger version.)

Sulfur Isotopes: Fig. 8 (above) show the spatial distribution of 34S values.  There is good correspondence between the 34S values in periphyton in May vs September, and excellent correspondence between the 34S values of periphyton vs gambusia. As indicated in Fig. 1, 34S values greater than about +20  are almost certainly caused by sulfate reduction.  Areas near the EAA have values similar to fertilizer S (Orem et al., 1999c), and values in the center of WCA1 have values typical for rain sulfate. The high 34S values are oriented roughly parallel to L67, from WCA-2BS to Shark River.  Hence, not only are the environmental isotope patterns persistent over time, but they have been incorporated up the foodchain from periphyton to gambusia.  There is a slight enrichment in the gambusia (fractionation? differences in diet?) relative to the periphyton, and the May algae 34S values are also a little higher than the September values, consistent with a greater "redox" signal in the dry season as seen in the C isotopes. When the biomass 34S values along several likely water flowpaths were compared with the REMAP sulfate concentrations (Stober et al., 1998) measured at the same sites and times, there was absolutely no hint of the kind of inverse relation of 34S vs sulfate that would be expected for progressive sulfate reduction in the water column as the waters slowly move SW (as observed in the canal 34S data in Orem et al., 1998). This is moderately good refutation of the idea that the 34S values are the result of progressive water-column processes during movement of the sulfate "plume" southward, and supports the hypothesis that the biomass 34S values reflect sulfate reduction (and other S cycling) in the local sediments (and maybe in the periphyton).

Biomass Isotopes as Indicators of Redox Conditions:  The compositions and spatial distributions of the C, N, and S isotopic compositions suggest that the isotope values reflect spatial variability in reducing conditions in the marshes that favor methane production, sulfate reduction, and (perhaps) denitrification.  The isotopic compositions of aquatic plants appear to integrate the more variable water-column isotopic compositions produced by redox reactions (and other factors) in the ecosystem, and these same patterns are incorporated throughout the food chain.  These compositions are relatively stable over time because the biomass remains in the system and is actively recycled without significantly affecting the isotopic compositions of the residual material.  Therefore, zones frequently dominated by particular redox reactions appear to be labeled by the C, S, and N isotopic compositions of local organisms. The general similarities in the spatial distribution of 15N, 13C, and 34S values for periphyton and gambusia suggest that the isotopic compositions of the entire foodweb are being affected by the biogeochemical processes described above. Hence, in order to interpret differences in isotopic composition of organisms in terms of trophic levels, one must first "remove" the isotopic effects of these biogeochemical processes. One way to do this is by normalizing (by subtraction) all the organism isotope data to the isotopic composition of some widely-distributed "indicator species" that appears to have a relatively constant diet over time.  Since gambusia collected at the same place and time (Fig. 3) show a small range in 15N and 13C values, they are a suitable candidate for a "normalizing agent."  In a sense, normalizing organism data to the 15N and 13C values of gambusia removes the effect of the spatial patterns seen in Figs. 6-7, making it easier to interpret the "residuals" as possible trophic differences.

Delta 13C of Sediments	Delta 15N of Sediments	Delta 34S of Sediments

Figure 9 (Click on image for larger version.)

Sediment Isotopes:  The sediment samples (Fig. 9, above) were analyzed as a way of ascertaining whether the isotope patterns seen in the biomass are a function of long-term, in-situ, sulfate reduction in the marsh sediments (that presumably would affect local biogeochemical conditions enough to isotopically "label" local biota), or whether the biomass patterns are largely due to water-column processes.  Surprisingly, there is little correlation between the patterns seen in the periphyton and gambusia and those in the sediments. The 13C values are very consistent across the wet parts of the marsh, but get rapidly heavier to the west (diagenetic fractionation due to increased oxidation? different source material to the cypress-area peat?). The sediment 15N values show an odd E-W trending band of heavy values south of the Tamiami Trail. The sediment 34S values in the sediments are significantly different from the patterns in the biomass, mainly because the sediments lack the extremely prominent NE-SW band of very high 34S values (seen in the biota) that parallels the L67 canal and is caused by sulfate reduction. So why aren't we seeing the same patterns in the sediments (Fig. 9) as in the biomass? One possibility is that the isotope patterns seen in the algae and fish samples are simply not present in the macrophytes that are preserved as peat, and since the algae and fish are not major contributors to the formation of peat, their isotope patterns are simply not preserved. Preliminary data from USGS sites suggests that macrophytes do have a broad range of 34S values at some sites but it is not yet clear whether these values correspond with the 34S values of periphyton and gambusia.  However, a more complete set of macrophyte samples (from REMAP II) will need to be analyzed to address this question.  Another explanation is that the isotope patterns, especially the very striking 34S patterns seen in the current biota samples, reflect recent changes in environmental conditions that affect the biogeochemistry of the marshes, caused by increases in sulfate (and nutrients) derived from the EAA. The difference in the 34S patterns in modern biota and older sediments (mostly >50 years ??) is strong evidence for a significant increase in sulfate reduction in the marshes in the last few decades, probably in response to increased S loading from the EAA.

% carbonate on Periphyton	% carbonate on Periphyton

Figure 10 (Click on image for larger version.)

Calcite in Periphyton Mats:  The percent calcite in the periphyton mats (Fig. 10, above) was calculated from the changes in %C and 13C for acidified vs unacidified samples (all plant and sediment samples are normally acidified to remove calcite before analysis). These data were then plotted to show the spatial patterns in 13C for September and May 1996 periphyton. This calculation produced a cohesive set of data, showing a very prominent NE-SW trending area of low calcite extending across the relative pristine parts of WCA2A and WCA3A, parallel to L67 but considerably to the NW of it, with some evidence of higher % values in the southern Everglades in September compared to May. It is not yet clear what is controlling the distribution of calcite-precipitating algae: water chemistry, hydroperiod, or what.  Certainly high nutrient concentrations strongly affect the composition of algal mats.  Another possibility for the low-calcite zone NW of L67 is some kind of inhibition by organics -- either because degradation of organics (respiration) affects pCO2 (and hence calcite solubility) or because the organics affect calcite kinetics. Spatial Variation in Foodwebs:  The discrepancies between the 15N and 13C data suggest that different C sources are being utilized by different organisms, suggesting more than 1 foodchain at some locations or tremendous temporal/spatial variability in 13C. With the combination of 13C and 15N data, we can positively rule out some organisms as significant food sources to others (e.g., adult dragonflies). It is not yet clear whether there are different foodwebs in the N vs S Everglades. However, nutrient-impacted sites certainly have different isotopic compositions than more pristine sites, and we can see that some places have isotope values consistent with periphyton being an important C source (i.e., U3) and others where detrital materials appear to be very significant (i.e., ENR Cell 3).

Summary The compositions and spatial distributions of the C, N, and S isotopic compositions suggest that the values reflect spatial variability in reducing conditions in the marshes that favor methane production, sulfate reduction, and (perhaps) denitrification.  The isotopic compositions of aquatic plants appear to integrate the more variable water-column isotopic compositions produced by redox reactions (and other factors) in the ecosystem, and these same patterns are incorporated throughout the food chain.  These compositions are relatively stable over time because the biomass remains in the system and is actively recycled without significantly affecting the isotopic compositions of the residual material.  Therefore, zones frequently dominated by particular redox reactions appear to be labeled by the C, S, and N isotopic compositions of local organisms.  Isotopic compositions of biomass may prove to be more cost-effective and reliable indicators of prevailing environmental conditions that favor MeHg production than other parameters currently being considered because biomass isotopic compositions are much more difficult to perturb than the more transient concentrations of aqueous species (like sulfate or sulfide).  Hence, the spatial isotope patterns are likely to provide a valuable integration of long-term environmental conditions in the Everglades.

References Orem, W.H., Bates, A.L., Lerch, H.E., Corum, M., and Boylan, A. (1999c) Sulfur contamination in the Everglades and its relation to mercury methylation. USGS Open-File Report 99-181, U.S. Geological Survey Program on the South Florida Ecosystem, Proceedings of South Florida Restoration Science Forum, May 17-19, 1999, Boca Raton, FL, pp. 78-79. Stober, J., Scheit, D., Jones, R., Thornton, K., Gandy, L., Trexler, J., and Rathbun, S. (1998) South Florida Ecosystem Assessment Monitoring for Adaptive Management: Implications for Ecosystem Restoration, Final Technical Report - Phase I, EPA#904-R-98-002, vol. 1.