MANGROVE LEAVES.— The chemical analyses reported here show large changes in elemental and isotopic compositions of R. mangle leaves along the Shark River estuary, Florida. Changing nutrient supplies across the Shark River estuary may control much of this chemical variation. The Shark River estuary is in a relatively pristine location within Everglades National Park, and remote from direct, water-borne human inputs. The upstream freshwater Everglades marsh is highly P-limited, and ammonium can accumulate to 10-30 micromolar levels (Rudnick et al., 1999; B. Fry, unpubl. results). Freshwater flushing through the Shark estuary can thus be a source of N, while studies in nearby Florida Bay show that offshore waters supply P in this region (Fourqurean et al., 1992; Rudnick et al., 1999; Boyer et al., 1999). Long-term fertility of the Shark River estuarine system may result partially from crossing N and P gradients, and such gradients may influence the chemical composition of mangrove swamp soils and plants as well as the microbial colonization of detrital materials (Fell and Master, 1980; Molina, 2000). However, recent experiments show that decomposition of red mangrove leaves may be governed more by leaf compound chemistry than by availability of external N and P nutrients (Feller et al. 1999).

Here we offer some tentative explanations of changing mangrove leaf chemistries across the Shark River system. The increase in leaf N contents at the two most upstream stations may reflect relative N sufficiency from freshwater inputs. S results may indicate a more complex control of leaf chemistries, with maximal sulfur uptake occurring further downstream in the mid-estuarine zone. Enhanced concentrations of leaf S may result from increased nutrient levels and production of soil sulfides that are used by rooted estuarine macrophytes such as mangroves (Fry et al., 1981; Okada and Sasaki, 1995), and increased soil sulfide has been documented along upestuary-transects in the Shark River by Chen and Twilley (1999). Also, increased leaf S content possibly could reflect salt stress in mangroves. Plant incorporation of water from root zones is thought to modulate mangrove carbon isotopic compositions, with more salt stress resulting in higher leaf ¹³C (Lin and Sternberg, 1992a,b). Maximal green leaf S concentrations were observed in mid-estuarine plants with highest ¹³C values (Fig. 2, Fig.4), perhaps indicating increased S uptake accompanying salt stress.

The transition from yellow leaves senescing on trees to decaying black and orange leaves in the water involves many chemical changes. Submerged decaying leaves gained N and lost S relative to yellow leaves, with a general decrease in ¹⁵N values and increase in ³⁴S values. It seems likely that a combination of leaching and microbial colonization, processes previously documented in studies of decaying mangroves from the Shark River area (Heald, 1970; Pelegri et al., 1997; Pelegri and Twilley, 1998), are responsible for these net changes. In spite of these considerable changes, important aspects of the original cross-estuarine gradients in isotopic composition were retained, especially lower ¹⁵N in upestuary submerged leaves, and low ³⁴S in mid-estuary submerged leaves. This suggests dominance of the chemical source signal from the original mangrove organic material rather than dominance of microbial colonists in the decaying leaves. Longer-term decomposition time series or extraction of microbial biomarkers are needed to more clearly establish isotopic signals associated with the colonizing microbial flora.

FILTER FEEDERS.— The food resources used by filter feeders in mid-estuary are likely tied to increased inputs from local mangrove swamps. The mid-estuarine region is that part of the estuary receiving most influence from mangrove swamps, as judged by lowest ¹³C values of shells in this region (Table 1). These low values typically indicate higher respiratory inputs of CO₂ from marshes and swamps where tidal drainage provides abundant CO₂ from belowground respiration to adjacent channels (Cai et al., 1999). Another chemical indicator of mangrove swamp drainage is low pH, which also reached a minimum in mid-estuary (Table 1). While we have not calculated a landscape-level forest open water ratio for various segments of the estuary (Cai et al., 1999), the shell ¹³C and pH chemical indicators suggest increased importance of mangrove drainage in mid-estuary, and would be consistent with an increased mid-estuarine input of detrital mangrove plant material.

We used a mixing model based on S isotopes to evaluate mangrove use by consumers (Fig. 7). But we also considered a simpler possibility that filter feeders were using only phytoplankton which has a moderate standing stock in estuaries of this area (median values for water column chlorophyll a are 2-3 µg L^-1; Boyer et al., 1997). We rejected a phytoplankton-only explanation for the following reasons related to S isotopic compositions.

For ³⁴S, phytoplankton can be expected to have near-constant values across estuaries, reflecting the dominance role of seawater sulfate in buffering S isotopic compositions of estuarine waters. For 28 mM seawater sulfate and typical freshwater sulfate concentrations < 0.2 mM, weighted averages strongly favor expression of the seawater sulfate isotopic signal (+21) in all but the lowest salinity (< 1 PSU) estuarine waters (Fry, in press). Even though estuarine systems are known for extensive sulfate reduction and production of sulfides, the high concentration of sulfate buffers changes in sulfate isotope values even at low salinities (e.g., see Fig. 7 in Stribling et al., 1998). Plants generally have ³⁴S values slightly lower than sulfate sulfur sources by an average of -1.5 (Trust and Fry 1992), so that phytoplankton ³⁴S values should be near +19.5 across most of the Shark River estuary. This basic expectation is met in some estuaries, e.g., barnacle ³⁴S is fairly high and constant in open water transects from Barataria Bay, Louisiana, where barnacle values remain at 13.5-17 throughout their range of occurrence (from about 5-25 PSU; A. Gace and B. Fry, unpubl. results). Barnacle values lower than 19.5 in Barataria Bay could be due in part to consumption of resuspended benthic algae. Benthic algae often have lower ³⁴S values than phytoplankton, with reported average values near 10 for benthic microalgae (Wainright et al., 2000; Moncreiff and Sullivan, 2000), likely reflecting some use of sedimentary sulfides. The mid-estuarine depression in filter feeder ³⁴S to values as low as -5 in the Shark River (Fig. 6, bottom) thus contrasts to results found for open bays and to results expected for reliance on phytoplankton. For this reason, we considered it unlikely that filter feeders were using only phytoplankton in the Shark River estuary.

The simplest mixing model explaining the S isotope results was a two-source model for mangroves and phytoplankton (Fig. 7). Once percent contributions for the two sources were calculated based on the S data, carbon isotope values of phytoplankton (Table 3) could be estimated as a check on the reasonableness of this simple model. Many of the calculated phytoplankton carbon isotope values in Table 3 are low relative to -18 to -32 values observed in other estuarine studies (Coffin et al., 1994; Chanton and Lewis, 1999), but are generally near the measured values of the consumers (Table 3) and may reflect large fractionations associated with high free CO₂(aq) concentrations that characterize the Shark River estuary (Table 1). Large fractionations and low carbon isotope values for phytoplankton are expected with high CO₂(aq) levels (Goericke et al., 1994; Fry, 1996; Popp et al., 1999a), but there was a limit expected for such fractionations that provided a useful constraint in the modeling exercises, namely that phytoplankton carbon isotope values could not be lower than -49. The -49 lower limit for C isotopes in Shark River phytoplankton is based on the following summation of several isotope steps important for photosynthesis: a maximum enzymatic fractionation of 25-29 vs. dissolved CO₂ (Goericke et al., 1994; Popp et al., 1998), a 7.4 lower isotopic value for dissolved CO₂ in equilibrium with dissolved inorganic carbon (DIC) at 30°C (Mook et al., 1974), and finally the assumption that isotopic compositions of barnacle and mussel shell materials listed in Table 1 approximate DIC isotopic compositions (Mook and Vogel, 1968; Killingley and Lutcavage, 1983; Morse and Mackenzie, 1990).

While the solution shown in Figure 7 was thus possible given reasonable assumptions about S and C isotopes in estuarine phytoplankton, it was also unsatisfying in some ways. Especially, detailed calculation showed that S isotope values of green rather than yellow or detrital leaves had to be used to avoid negative contributions from phytoplankton or very low (< -49) phytoplankton C isotope values. Using green mangroves as nutritional inputs could be rationalized if original leaf proteins became incorporated into food webs during initial leaching phases of mangrove leaf decomposition. However, colonization by bacterial and fungal communities is usually associated with further detritus decomposition (Newell et al., 1987), and these processes can result in incorporation and immobilization of S into microbial and leaf material that increases S isotopic compositions (Currin et al., 1995) above initial values in source green leaves. Calculation showed that using higher mangrove S isotope values expected after S immobilization (Currin et al., 1995) and generally measured for detrital leaves vs. green leaves (Table 2) resulted in unrealistic negative number solutions in the two-source models. These calculations indicate that future studies may wish to consider multiple food sources beyond phytoplankton and mangroves, and, in order to balance inputs of detrital mangroves with relatively high S isotope values > 0 in mid-estuary, these food sources would need to have low (< 0) S isotope values.

Benthic microalgae and sulfur bacteria could be potentially such food resources. Benthic microalgae have been assessed as important foods in several shallow systems (Currin et al., 1995; Newell et al., 1995; Page, 1997; Moncreiff and Sullivan, 2001), and via resuspension, could become foods for filter feeders in channels. Some of the nutrition attributed to phytoplankton in the two-source model of Figure 7 could be due to benthic algae that typically have somewhat lower S and higher C isotope values than do phytoplankton (Currin et al., 1995; Wainright et al., 2000). Recent reports are that benthic microalgae can sometimes have S isotope values < 0 (C. Currin, pers. comm.). White colorless sulfur bacteria and photosynthetic sulfur bacteria were also potential food sources with low S isotope values. Sulfur bacteria were observed in shallow creek systems of the Shark River estuary, and are common in many intertidal systems (Jorgensen, 1982) where they have been hypothesized to form an important food resource (Howarth, 1984). Such sulfur bacteria can have low C and S isotope values (Peterson et al., 1980; Peterson et al., 1986; Conway et al., 1994), and some increased consumption of such bacteria by mussels vs. barnacles could potentially account for the lower S isotope values of the mussels. This idea is Consistent with the observation that mussels commonly consume bacteria and foods of smaller average size than barnacles (Zobell and Landon, 1937; Barnes, 1959; Riisgard, 1988). Although previous study has not shown an important food web role for such sulfur bacteria in estuarine systems (Peterson et al., 1985, 1986), an important role for sulfur bacteria in the Shark River would increase the mangrove-related nutritional support of filter feeders shown in Figure 7.

A general conclusion in considering these multiple sources concerns the differentiation of leaf protein in fresh detritus vs. aged detritus that develops a different, microbial-influenced chemistry over several months after leaves fall from mangrove trees. To the extent that aged detritus with high ³⁴S rather than fresh detritus with low ³⁴S is important, sulfur isotope mass balance requires that benthic algal, sulfur bacteria, or other food sources with low ³⁴S must also be important for filter feeder nutrition. In this case, results of the parsimonious two-source mixing model represented by Figure 7 would need to be modified to reflect contributions of these sources. Current data could not exclude the influences of these other sources, and it is probably more appropriate to regard results of Figure 7 as reflecting a broader mixture of 'microalgae' (phytoplankton + benthic algae) and 'marsh materials' (mangrove detritus plus various microbial decomposers and sulfur bacteria associated with mangroves), rather than as a simple binary mixture of phytoplankton and mangroves.

Using C rather than S isotopes as a basis for mixing models in mangrove systems may be preferable in future studies. Because the S and N isotopes can change substantially with detritus age (Zieman et al., 1984; Currin et al., 1995; Caraco et al., 1998; this study), tracing detrital dynamics with S and N isotopes may generally prove less satisfying than studies conducted with C isotopes, because C isotopes in mangroves and other aquatic macrophytes do not change greatly with age (Zieman et al., 1984; Fenton and Ritz, 1988; Rao et al., 1994; this study). Also, detailed studies showing fractionations of C, N and S isotopes during animal growth on mangrove detritus are needed to complete a mixing model analysis like that of Figure 7. Stable isotope studies related to mangrove systems thus far have not specifically used mangrove plant detritus in laboratory food trials (Dittel et al., 1997; Wiedemeyer, 1997). Isopods collected from mangrove wood in this study showed respective CNS isotope differences of +3.7, -1.0 and +7.8 vs. wood (Table 2), indicating that deviations from simple 'you are what you eat' isotope rules are possible in food webs involving aged mangrove materials.

Given these caveats about the mixing model results of Figure 7, this study still arrives at somewhat different conclusions than those reached by Odum and Heald (1975) about the importance of mangrove systems for food web support of common consumers. Our conclusion is that microalgae (combined phytoplankton and benthic microalgae) provide the strongest food web support for the filter feeders in the Shark River system (Fig. 7). Odum (1970) used a low threshold to classify many animals of this region as detritivores, and thereby may have overestimated the general importance of the mangrove food resource. Filter feeders and other animals with a minimum of 20% detritus in their guts were classified as detritivores (Odum, 1970), but microalgal cells were usually present in guts of these same animals. Subsequent studies have questioned whether ready assimilation of small amounts of algal material might outweigh slow use of refractory mangrove material (Rodelli et al., 1984; Newell et al., 1995; Dittel et al., 1997; France, 1998), but a clear consensus on this issue has been lacking, especially because some animals can survive on a strictly mangrove diet (Malley, 1978; Poovachiranon et al., 1986; Profitt et al., 1993). The current study points to increased use of mangrove materials where inputs were likely strongest in mid-estuary, but agrees with other studies that show a strong microalgal basis for food webs in shallow estuarine systems (Currin et al., 1995; Newell et al., 1995; Moncreiff and Sullivan, 2000), However, this study also leaves open the possibility that benthic consumers from interior creek and basin forest systems may show greatly increased importance of mangrove nutritional support much beyond that documented here for filter feeders collected in main channels. High food web importance for mangroves was documented in earlier studies at interior sites in this region (Odum and Heald, 1975), and this higher importance may reflect decreased mangrove export due to low tidal amplitudes of south Florida (0.55 m mean amplitude; Twilley, 1985) and due to the interior location of the original North River sites upstream of an enclosed marine bay (Whitewater Bay, Fig. 1).

Much could be done to improve future isotope studies of detrital mangrove processing in the Shark River, for example more frequent sampling than performed in this one-time study, and consideration of other mangrove species that also contribute to detrital pools in the Shark River system (Twilley, 1985; Chen and Twilley, 1999). Comparative work thus far indicates Rhizophora spp. mangroves have low S isotope values relative to other mangrove species (Okada and Sasaki, 1995; Loneragan et al., 1997; Wiedemeyer, 1997). Future studies should characterize isotopic compositions of not only red mangroves (this study), but also white mangrove (Laguncularia racemosa) and black mangrove (Avicennia germinans) that are common in parts of the Shark River System, and further should also sample other potential organic matter sources such as benthic microalgae and sulfur bacteria. This characterization will require not only bulk isotope measurements such as those performed in this study, but also more sophisticated measurements of isotopic compositions of chlorophyll, lipid and protein biomarkers (Macko, 1994; Popp et al., 1999b; Sachs et al., 1999; Meziane and Tsuchiya, 2000). Careful sampling and sorting of representative source organic matter samples (Wainright et al., 2000), and some attention to gut content samples as 'sorted-by-the-organism' samples, is important for this biomarker work. Such biomarker work may be necessary in this and other systems to really understand the role of terrestrial detritus for aquatic consumers, a topic which continues to arouse controversy (Lewis et al., 2001; Hall et al., 2001). Finally, the isotope studies would also benefit from a wider ecosystem context in which productivities as well as standing stocks of various organic matter sources are measured. Such information is currently lacking for the Shark River system, and may be especially important for microbial production associated with S cycling and with metabolism of dissolved organic matter leaching from forests (Odum et al., 1982; Twilley, 1985).

A general result of this study also concerned interpretation of C isotope values that are most commonly measured to determine sources of organic matter ('isotope sourcery') in food web studies. We found that low ¹³C values near the -27 mangrove values were necessary but not sufficient to assign a high importance to mangroves in estuarine food web studies. Especially consideration of S isotope data in modeling exercises led to conclusions that other autotrophs (sulfur bacteria, benthic microalgae, phytoplankton) can potentially also have ¹³C values of -27 or lower in CO₂-rich mangrove swamps. These low ¹³C values are also often found in low salinity regions of estuaries (Matson and Brinson, 1990; Chanton and Lewis, 1999), and -27 ¹³C values reported for consumers in some interior mangrove waterways (Fry et al., 1999; Chong et al., 2001) might be considered first as indicating habitat use of mangrove swamps, rather than food web / reliance on mangroves per se (Bouillon et al., 2000).