Stable Isotopic Analyses of Selected Narragansett Bay Molecules

.......................................................................................................................... 2 INTRODUCTION .......................................................................................................... 4 PREVIOUS STABLE ISOTOPIC INVESTIGATIONS OF MOLLUSC SHELLS .... 5 ENVIRONMENTAL SETTING .................................................................................... 7 Narragansett Bay ..................................................................................................... 7 MERL Mesocosms ....... .. ................................................................... ..... ................ ... 8 METHODS ...................................................................................................................... 9 018() Water & o13C Dissolved Inorganic Carbon Samples ............. .... ...... .......... 9 Molluscan Shell Samples ...................................................................................... 12 Predicted 018() Water & Hydrographic Data ..................................... .. .. ........... 13 Predicted 018() Calcite & Aragonite ................................................................... 15 RESULTS ....................................................................................................................... 16 018() Water & o13C Dissolved Inorganic Carbon ......... .................... ................. 16 Shell Data .......... ........................... ...... .................................................................. .... 18 Pitar morrhuana. ................................................................................................ 18 Mytilus edulis .................................................................................................. .. 19 Nucula annulata ................................................................................................ 21 DISCUSSION ............................................................................................................... 22 Oxygen Isotopes ........ ................ .................................. .......... .............. .................. 22

were incrementally sampled in order to obtain detailed isotopic profiles of the shell records. In this thesis I compare the isotope results from the shells to the observed shell growth patterns and to carbon and oxygen isotopic measurements of the water in Narragansett Bay. In addition, to develop a better understanding of ecological influences on growth rate, we set out to assess the suitability of Pitar morrhuana, Mytilus edulis and M. mercenaria as monitors of environmental conditions in the bay.
The isotopic analyses, in conjunction with the detailed data available from the MERL mesocosms, have enabled us to accurately interpret the 818() and 813C ii signals in the shells of these bivalves in terms of the seasonal timing of growth for Pitar and Mytilus and the known variations in the chemical and physical properties of the water in which they grew. The data indicate that the shells of both Pitar (aragonite) and Mytilus (calcite-outer layer only) appear to be forming at oxygen isotopic equilibrium with the ambient water and that the 018() profiles from these specimens are primarily controlled by variations in water temperature. The oxygen isotopes also indicate that in Narragansett Bay Mytilus has a longer growing season than Pitar by approximately one month at either end of the growing season.
The 0 18 0 analyses of shell samples of M. mercenaria, in conjunction with the water isotopic measurements, indicate that the shell of this species does not form in isotopic equilibrium with the surrounding sea-water. The measured o 1 8a of the shell aragonite is typically 1.5%o depleted relative to the values predicted with the aragonite-water isotopic fractionation equation of .
Although the isotopic fractionation is a problem for estimating absolute temperatures using the observed shell values, the oxygen isotopic profiles can be interpreted in terms of relative (high-low) seasonal temperature variations. The amplitude and cycles of o 1 8a variations in the outer shell layer enable one to determine the seasonal timing of growth of M. mercenaria living in Narragansett Bay. These observations are verified by the analyses of samples taken from the outer shell edge of specimens collected during different times of the year. The growth history of the specimens, as revealed by the oxygen isotopes, indicates that the longest growing seasons occur during the years of fastest shell growth and it appears that with increasing age the growing season is restricted to warmer temperatures.
The influence of productivity on the o13C compositions of the Pitar and the Mytilus shell carbonate and possible associations with the carbon cycling in the iii MERL tanks and Narragansett Bay were also noted. The different eutrophication levels of the mesocosms apparently had a profound affect upon the 813C composition of the Nucula shells. Nucula specimens from tanks with higher nutrient loadings (8x, 16x, and 32x) had much heavier shell 813C values than specimens from the normal nutrient level (control) mesocosms. Apparently much of the lighter carbon (C-12) was drawn out of the water column (and stored in the organic matter) due to the higher primary productivity in the tanks with increased nutrient loading.
The 8 13 C measurements of the both the shell edge and the incremental samples from the M . mercenaria specimens do not exhibit any obvious seasonal trends. The 8 13 C values obtained are much lighter than predicted, indicating that a portion of the carbon isotopic signal is probably derived from the uptake of isotopically light metabolic carbon during shell formation. This isotopically light metabolic carbon is most likely from the normal respiration occurring within the organism.       and Nucula annulata Grown in an Experimental Mesocosm. Annual temperature cycle in Narragansett Bay from 5/81to10/83 (weekly averages).   winter.
Oxygen and carbon isotope shell profiles from M. mercenaria specimen collected at station 2A in Narragansett Bay.   The data indicate that the aragonite shell of Pitar appears to be forming at oxygen isotopic equilibrium with the ambient water (as defined by the aragonite-water-temperature equilibrium equation of  and that the 018() profile from this specimen is primarily controlled by temperature changes. These data indicate that in Narragansett Bay Mytilus has a longer growing season than Pitar by approximately one month at either end of the growing season. Mytilus apparently ceases shell formation when the water temperature dips below about 4-6°C, whereas the critical temperature for Pitar is approximately 10 °C. Both species grow through the spring, summer and fall.
Nucula has been shown to record mainly spring values of 018() in its shell, although this interpretation is based on whole shell analyses.
The influence of productivity on the o13C composition of shell carbonate and possible associations with the carbon cycling in the MERL tanks and Narragansett Bay was also noted. The Pitar specimen exhibited low amplitude cycles of o13C that seem to correlate with the observed phosphate cycling in the bay. The carbon isotopic results from the detailed sampling of Pitar and Mytilus indicates that both specimens are utilizing a depleted source of carbon for a 2 fraction of the total carbon incorporated into the carbonate. We suggest that in addition to the ambient dissolved inorganic carbon, these molluscs are incorporating metabolic carbon during carbonate precipitation. The effect of eutrophication on the shell carbonate of Nucula grown in tanks with nutrient addition was reflected in elevated values of 813C in the treated tanks.

INTRODUCTION
Mollusc shells contain a potentially detailed record of both the life history and the environmental changes experienced by the mollusc during growth, and have been used to interpret past environmental conditions based on chemical analyses of the shell. Recently the results of stable isotopic analyses of closely spaced, incremental samples taken along the axis of radial growth on mollusc shells have been used to draw inferences about environmental variablility (e.g., . The results of these studies demonstrate the utility of oxygen isotopes as a tool for estimating growth rate, age, and seasonality of shell growth, as well as their potential utility for addressing such diverse problems as history of coastal upwelling, environmental reconstruction, and interpretation of archeological   , was done on shell samples from molluscs grown in controlled tanks. In these initial experiments the shells of living molluscs were purposely damaged by drilling or notching, and the subsequent calcite shell repair was examined isotopically, in conjunction with isotopic analyses of water samples from the same tanks. This work demonstrated the empirical relationship between the lBQ/16() ratio of the shell carbonate and the temperature of the water, if the 018() value of the water was known. The original calcite-temperature (or paleotemperature) equation of  was modified slightly by . The earlier paleotemperature equations relating the equilibrium oxygen isotope fractionation as a function of temperature were for the calcite polymorph of calcium carbonate. Most molluscs, however, form shells of aragonite and some are bimineralic ). Grossman (1982) and Ku (1981, 1986) have calculated equilibrium equations for the biogenic formation of the aragonite polymorph, based on molluscan and foraminiferal samples.
The utilization of modern mass spectrometers, requiring much smaller sample sizes, has enabled us to investigate the stable isotopic records of molluscan shells using detailed sampling methods. The more recent isotopic work involving molluscs generally involves closely spaced sampling of the outer shell in order to obtain a high-resolution isotopic record. These studies have revealed seasonal temperature and growth patterns using oxygen isotope compositions Krantz et al., 1987), and have helped determine the initiation and/ or cessation of seasonal growth. The investigation of  used sequential isotopic samples to document the annual periodicity of the major shell increments observed in Spisula solidissima, as proposed earlier by Jones (1980).
Other studies have demonstrated that coastal upwelling changes can be related to the carbon isotopic composition of the water as recorded in the molluscan shell Berger, 1979, Arthur et al., 1983). The oxygen isotopes of detailed 6 shell samples from archeological shell middens have also been used to determine the season of shell collection (Shackleton, 1973. Work involving paleo-salinity reconstructions using the oxygen isotopic composition of mollusc shells as related to changes in salinity, rather than temperature, have been done in some areas where a wide geographic distribution of molluscs of the same age can be found (Eisma et al., 1976, Rye and.

Narragansett Bay
Narragansett Bay, Rhode Island, (Figure 1) is a weakly stratified, temperate estuary with strong tidal mixing and relatively low fresh water input . Salinity variations in Narragansett Bay are relatively minor and are predominantly controlled by variations in the fresh water input. The tidal input into the bay is typically 250 times that of the fresh water . The annual precipitation (averaging 104.8 cm per year, from 1905-1981) to the drainage basin feeding Narragansett Bay is relatively constant, but the annual runoff is seasonal, with the greatest runoff usually occurring during January to April.
Salinities decrease from south (lower Bay) to north (upper bay) and from the bottom to the surface. The surface salinities in the lower part of Narragansett Bay fall within a fairly narrow range of 29%0 to 33%0 . The vertical salinity stratification in the lower bay is generally less than 3%o during most of the year. The average bay (whole bay) to ocean salinity difference is typically 2.80%0,    to mimic the magnitude and timing of the seasonal cycles of temperature, nutrients, production, and benthic respiration that occur in Narragansett Bay. During normal mesocosm operation the tanks are routinely monitored for temperature, salinity, seawater flow, light, pH, alkalinity, fluorescence, dissolved oxygen, total C02, total nutrients, and chlorophyll-~ levels.
8 A sampling bias may be present when comparing water samples taken from the header tank and extrapolating the results to the other MERL tanks or the lower bay. Such a bias is possible because of the shallow depth of the sea-water intake for the MERL tanks and the difference in turnover times between the header tank and the other mesocosm tanks. At a depth of less than 3 meters, the intake only samples the surface water, which is warmer and lower in salinity (and hence lower values of 818()) than the water near the bottom. The short turnover time of the water in the header tank (approximately 15 minutes) results in a sample of surface water near the intake averaged over only that time interval.
Because of the much longer turnover time of the mesocosms tanks (27 days The dissolved inorganic carbon was extracted by acidification of the water sample with 100% phosphoric acid followed by stripping of the C(h from the sample using helium carrier gas, as described by R. Fairbanks (pers comm., 1986).
The extraction line consisted of a stripping vessel equipped with a sintered glass frit and a replaceable serum injection port, two water separation traps (using a "slush" of isopropyl alcohol mixed with liquid nitrogen), a liquid nitrogen trap to collect the C(h, and a metal bellows pump to recirculate the carrier gas. Samples were prepared as follows: the entire extraction line was evacuated with a rough pump, approximately 1 ml of acid was then added to the stripping vessel and pumped down, helium was let into the line to roughly 1 atmosphere pressure and the sample (usually about 10 ml in size) was injected using a glass syringe into the stripping vessel. The helium was pumped through the line for 20 to 25 10 minutes, after which the trapped CC)i was purified and isolated for later analysis on the mass spectrometer. The best results were obtained when the phosphoric acid was pumped down just prior to a sample extraction. The storage of the samples prior to extraction was also important. Freezing the samples immediately after collection proved to be the most reliable method of storage.
Unless otherwise noted, all oxygen and carbon isotopic ratios are given in the conventional delta (B) notation as the per mil (%0) enrichment or depletion of l8Q, or 13C, relative to the Pee Dee Belemnite (PDB) standard  where: B = (Rs/Rref-1) x 1000; Rs= isotopic ratio of sample gas and Rref = isotopic ratio of the reference gas, all reported vs. PDB. The measured precision of duplicate water samples (free of contamination) using this extraction technique was ± 0.1 %0 .
The B18() water samples were prepared following the method described by T. F. Anderson (pers. comm., 1987): a 1 gram sample of water is placed (using a glass syringe and Tygon tubing) in the rounded bottom of a 10 mm borosilicate tube. The water sample is connected to a vacuum line (using a Cajon Ultra-Torr fitting) and frozen with an isopropyl alcohol-liquid nitrogen trap (IP A trap). The water sample is then pumped down using a rotary vacuum pump to remove atmosphere and dissolved gases. The water sample is isolated from the pump, thawed under vacuum, refrozen and pumped down again. Approximately 50-100 micromoles of pure, dry C02 gas is frozen with the sample using liquid nitrogen. The sample tube is flame sealed and placed in a 25°C water bath for a minimum of 8 hours to allow for complete equilibration between the H20 and C02. The equilibrated sample is reattached to the vacuum line and frozen with an IP A trap. The C02 is then trapped out with liquid nitrogen and purified for analysis on the mass spectrometer. All B180water values are reported in the delta (B) notation relative to Standard Mean Ocean Water, or SMOW . The SMOW values were corrected to PDB values (PDB = SMOW + 0.22%0) for use in the carbonate-temperature fractionation equations of  and    and Frithsen et al.,(19B5), with different mesocosm tanks subjected to: zero nutrient addition (control tanks), 2x, 4x, Bx, 16x and 32x the normal nutrient loading of Narragansett Bay. The shells of Pitar and Mytilus that we selected for detailed isotopic investigation were from the same mesocosm control tank (T-0), and the shells of Nucula were from the control tank, the Bx, 16x and 32x tanks. Pitar morrhuana is considered an interface feeder, Nucula annulata a subsurface deposit feeder, and Mytilus edulis is an epifaunal filter feeder.
The interiors of the molluscan shells were carefully cleaned of organic matter, then dried at room temperature. The right valves from a specimen of Mytilus edulis (length = 63 mm) and Pitar morrhuana (length = 42 mm) were sectioned along the radial growth axis (from the umbo to the ventral margin) using a diamond wafer blade on a variable speed sectioning saw. The periostracum of each shell was scraped or filed off prior to sampling. The individual samples for isotopic analysis were taken from the outer shell layer using a high speed drill and a 0.5 mm diameter carbide dental burr . Each powdered sample was ground off parallel to the growth bands in successive fashion across the external shell surface.
Samples for isotopic analyses were roasted under vacuum at 390°C for one hour and then reacted in 100% orthophosphoric acid at a constant temperature of so°C to generate C()i gas. Gas samples were purified using low temperature isopropyl-liquid nitrogen baths to separate out the residual water and the resulting C02 gas was analyzed on a V. G. Micromass 602-D isotope ratio mass spectrometer. The isotopic data are given in the conventional delta (8) Table 3, Appendix I) by calculating the average of the amount of fractionation observed in 10 duplicate samples, prepared at both laboratories : -0.369 for 818() (lcr = 0.08%0) and -0.499 for 813C (lcr = 0.07%0).

Predicted QlB() Water & Hydrographic Data
Because limited sampling for 8 1 80water was preformed in this study, we developed a simple two component mixing model that predicts the yearly variation of 818() of Narragansett Bay water, driven by the measured changes in the bay salinity. The oxygen isotopic values of the two end-members are based upon 818() measurements of Long Island Sound water and measurements of the regional riverine input by . The seawater end-member has a 13 0 18() value of -0.76%0 (SMOW) for a salinity of 32.5%0. The value of the freshwater end-member is taken as -9.5%0, which is an average of the monthly measurements from the rivers in this region, sampled during 1981.
B 1 80water = [S(-0.76%0) + (S-32.5)(-9.5)]/32.5 The salinity measurements from a MERL control tank during the eutrophication experiment show a strong seasonality with a low of 27%0 in the spring and a summer to winter high of approximately 31 %0 (Figure 3a), with little variation 8:1 %0) between the tanks (Frithsen et al, 1985). However, the eutrophication experiment salinity measurements are incomplete, running only from January 1983 to September 1983, less than half the duration of the experiment, with no data available for the header tank (the tank most closely equivalent to conditions in the bay). Figure 3b shows the most recent salinity data (courtesy of J. Frithsen, 1987) collected from 1984to1987, sampled from the MERL header tank and one of the mesocosm control tanks, on a weekly basis.
These salinity data vary between 27.1 %0 and 32.4%0, with more than 95% of the data within the range of 29.5 to 32.4%0. Unlike the salinity data measured during the eutrophication experiment, these data do not exhibit a strong seasonal signal.
The more recent salinity data demonstrate that the MERL seawater inlet is close enough to the surface to reflect short term salinity fluctuations in the surface water, such as fresh water input during high runoff. The MERL mesocosm control tank, also shown in Figure Figure 3a) to calculations made using the average salinity value (29.0%0) of the same data set. The results are shown in Figure 6 and illustrate that the first order variation in the o180aragonite signal is due to seasonal temperature variations.
The greatest consistent difference, 0.5%0, occurs in the late spring when the measured salinities were at a low for the year (Figure 3a). Overall, the observed 3-4%0 salinity variations can affect the 018() of the carbonate shells precipitated in the lower bay. Thus, for detailed isotopic work it is important to measure either the salinity or the 018() of the water in order to accurately use the isotopictemperature equilibrium equations.

_Q18() Water & _Q13C Dissolved Inorganic Carbon
In order to interpret the shell oxygen isotope record in terms of temperature and timing of seasonal growth cycles, it is necessary to document the oxygen isotope variations of the water in Narragansett Bay. The results of the oxygen i~otopic measurements of the water (B180water ) samples collected from the MERL header tank are shown in Figure 7 (and listed in Table 1 values are quite a bit more positive than the values predicted using the 1983 MERL control tank salinity data, which had an average B180water equal to -1.67%0 (SMOW). The majority of this difference is due to the unusually low salinities observed in the spring of 1983, relative to spring conditions measured at other times (Fig. 3a). The use of the measured B180water values will generally result in slightly heavier predicted 618() values of shell carbonate than were estimated using the modeled B180water values.
The results of measurements of the B13C (PDB) of the total dissolved inorganic carbon in Narragansett Bay water collected from the MERL header tank are shown in Figure 8 (listed in Table 2, Appendix I). Samples were taken from August 1986 to September 1987 and the values range from 0.25 to 2.3%o (PDB).
The average of all the data was 1.42 + 0.42%0, with weekly variations of up to 0.5%o. The greatest change took place during late February to mid-March when an approximately 2%o enrichment is observed in the data. The o13C values are increasingly lighter through the fall and into the winter, presumably due to the influx of lighter carbon (carbon-12) from benthic respiration and the low productivity occurring in the bay at this time .

Shell Data
The oxygen and carbon isotope results from the incremental shell samples of Pitar morrhuana and Mytilus edulis are plotted in figures 9 and 10, respectively.
The isotope data are plotted as the delta value (in per mil notation) versus the individual sample number drilled from the shell, roughly equivalent to shell height. The x-axis origin always represents the beginning of shell growth, the last sample on the x-axis is the shell edge sample, i.e., the last shell material formed prior to collection in September 1983. This method of plotting will introduce a bias toward those periods of faster growth, but has the advantage of "flattening" the shell such that the early shell growth is not lost due to the curvature of the The carbon isotope record of the Pitar specimen does not exhibit the regular seasonal cycles observed in the o:Xygen isotope profiles. The seasonal range of values for the 813<: of the shell samples is less than 1 %0 with an overall increase in B13C for the first two and a half years from roughly--0.25 to 0.75%0 and then much more erratic variation for the last two years, from 0.75 to --0.2%0. Within a given year there seems to be a trend toward heavier values early in the (growth) year, followed by a jump of 0.25--0.5%o to lighter values in mid to late summer, followed by a general increase in values through the fall.

Mytilus edulis
The specimen of Mytilus edulis we examined was collected from the same mesocosm control tank as the Pitar specimen. Mytilus is an epifaunal filter feeder with a bimineralic shell, the outer layer of the shell (beneath the periostracurn) is entirely calcitic and is the layer we sampled for the isotopic analyses. Samples from 52 successive intervals were analyzed isotopically with 9 replicate analyses (Table 4, Appendix I). Sample intervals were more widely spaced than those in Pitar. The shell height of this Mytilus specimen measured 62 mm .
The B18() signal from the Mytilus specimen ( Figure 10) indicates strong seasonal temperature cycles recorded in the outer calcitic shell layer. The measured BIB() values range from 1.5 to -2.3%0 (PDB), an overall range of 3.8%0.
The B18() cycles, interpreted as representing annual temperature cycles, indicate that this Mytilus specimen was less than two years old at the time of collection, suggesting a very high rate of shell growth. The overall shape of the oxygen isotopic profile is somewhat different than that of the Pitar specimen with a predominant summer peak (consisting of 14 samples lighter than -2.0%0), and a greater overall range of isotopic values. Since this specimen is less than two years old, it must have entered the MERL seawater system in the planktonic larval stage.
The carbon isotope data from the Mytilus specimen are also shown in Figure   10. The measured values range between 1.6 and-1.3%0, with an amplitude of 2.9%0. The general trend of the carbon isotopes does not appear to be highly seasonal or to correlate in any way with the oxygen isotope cycles. The B13(: values decrease steadily into the first summer, beginning at 1.6%0 and falling to -0.75%0 by mid-summer (timing inferred from the seasonal cycle of oxygen isotopes). After mid-summer the B13C values are more variable, generally becoming more positive (to approximately-0.5%0) into the fall. The B13C values decrease by 1 %0 during the initial spring growth of the following year, level off at about-1 %0, and then increase to-0.25%0 just prior to collection of this specimen.

Nucula annulata
The isotope data from the specimens of Nucula annulata, a small, sub-surface deposit feeder, are shown in Figure 11 (listed in Table 5 The whole-shell carbon isotopic records from the Nucula specimens have a wide range of variablility ( Figure 11), apparently related to the degree of nutrient loading in the respective mesocosm tanks. The ot3C shell measurements show a trend toward increasing o13C with increasing eutrophication (increasing nutrient addition level) of the mesocosm tanks. The specimens from the 32x tank (32 times the normal bay nutrient loading) have extremely heavy ot3C values, ranging between 4.0 and 5.0%0, with one sample taken from the margin of a specimen measured at 5.8%0. The ot3C values from the shells of specimens recovered from the 8x and 16x tanks overlap somewhat (see discussion below), ranging between 2.0 and 3.0%0. The ot3C values of the control tank specimens are between 0.5 and 1 %0. The 32x tank was observed  to have very low levels of dissolved oxygen (dropping occasionally to zero), high levels of chlorophyll~ and consistently low, but variable, levels of total C02. The heavy o13C values probably reflect the removal of carbon-12 during the sustained high primary productivity of this tank. 2) show that temperatures rise rapidly during April, increasing from 4 to 11°C in 1982, and from 5to13°C in 1983. The isotopic record of the Pitar specimen ( Figure 9) also indicates that shell growth started earlier in the year and/ or ended later in the year with increasing age, with the exception of sample #3, which apparently delineates two later growth (annual) cycles. The amount of shell material formed during the colder water temperatures must be fairly small, as the number of isotopic samples indicating low temperatures is very limited. Slower shell growth results in more time-averaging among incremental samples taken from the shell and subsequent loss of resolution. This sampling effect has been observed by others (e.g., Krantz et al., ,1987 in the isotopic investigations of various species of molluscs using detailed, incremental sampling techniques.  discussed the consequences of unequal sampling of successive growth bands and concluded that the sampling effects were less important than the environmental effects as far as the overall amplitude of the isotopic signal, but that knowledge of life history of a bivalve species can be very important when trying to extract seasonality information using oxygen isotopes.

DISCUSSION
An "average" or bulk shell sample taken from this Pitar shell would probably indicate a temperature of approximately 17-19 °C, skewed toward the higher temperature range due to the greater number of warm temperature samples. This is actually much higher than the average annual bay temperature, which is roughly 10-11°C. Thus, temperature estimates using isotopic analyses of whole shell samples from this species of mollusc would indicate average temperatures much higher than the true value and would not be of much utility.
The Pitar oxygen isotopic profile can also be used to examine the relationship between the annual 818() cycles and the internal growth increments of this specimen. The internal growth increments can be observed as the dark layer, or band, of shell material when viewing a radial cross section of the shell.
These dark bands are generally thought to occur during periods of slow shell growth, the dark color arising from the relatively greater proportion of organic matrix material being incorporated into the shell during periods of slow growth (Jones, 1980, Lutz and. The specimen of Pitar that we have examined isotopically has 5 dark bands, or growth increments that can be observed in the shell cross section. These 5 bands coincide with sample numbers 3, 8, 17, 32, and 45, which are the isotopically heaviest samples measured for each annual cycle ( Figure 9). The internal growth increments are thus an excellent indicator of the age of the specimen.
The annual internal growth increments can be observed in a radial cross section of the shell as the alternation of thin dark layers (bands) with larger white increments of shell material. The dark bands are generally thought to form during periods of slow shell growth. The dark apperance arises from the dense, tightly-packed arrangement of the aragonitic crystallites which results in a translucent shell microstructure, contrasting with the white color of the typical (opaque) microstructure. The specimen of Pitar that we examined isotopically has 5 dark bands, or growth increments, that can be observed in the shell cross section. These 5 bands coincide with sample numbers 3, 8, 17, 32, and 45, which are the heaviest samples (isotopically) measured for each annual cycle ( Figure 9) and formed each winter during the coldest period of shell growth. These annual internal growth increments are thus excellent indicators of the age of the specimen.
The age determination resulting from the B18() record of the Pitar specimen can also be used to construct a standard graph of shell height versus years of growth ( Figure 12). The points on the graph correspond to the observed internal growth increment and to the isotopically heaviest (most positive) sample for each 25 observed annual cycle. The best fit to the von Bertalanffy growth equation is achieved with SHoo = 44.22, k = 0.52, and t 0 = 0.12, such that the shell height (SH) at any time t, is equal to: This specimen of Pitar shows an average growth rate of approximately 9 mm per year, although the growth rate slows dramatically after the second year.
The Mytilus oxygen isotopic profile ( Figure  This supports the validity of modeled BtSOwater estimates. It is known that aragonite is enriched in 18() relative to calcite, although there is some debate about the magnitude of the enrichment. Most of the current research Ku, 1981, 1986)  a week-long interval of time, given 3 months of temperatures above 18°C ( Figure   2). In comparison, the Pitar specimen typically has only 3 to 4 samples with 018 ( The growth rate of this specimen of Mytilus was greater than 3 cm per year, which is quite high relative to most of the reported growth rates (e.g., Seed, 1969, Incze, 1980, although similar values have been observed by Incze et al. (1980) in the Damariscotta River estuary in Maine. This Mytilus specimen had no distinct disturbance bands on its outer shell layer, so we could not compare the disturbance ring method of age determination (e.g., Seed, 1969) to the isotopically determined age.

27
Carbon Isotopes The degree of equilibrium 13C fractionation between biogenic aragonite and seawater bicarbonate is not well established, although Grossman (1984a) has estimated it to be +2.40%o at 25°C (revised from Rubinson and Clayton, 1969).  have demonstrated a temperature dependence of the carbon isotopic fractionation between the dissolved inorganic carbon (DIC) and molluscan aragonite equal to --0.13%0 per°C, that is, less fractionation with increasing temperatures.   The relationship of Grossman and Ku (given above) predicts that at 20°C, the fractionation between the shell aragonite and the o13Cmc should be 0.04%0, or virtually no fractionation at this temperature. The summer o13C values measured from the Pitar specimen vary from year to year, but range from approximately -0.25 to 0.25%0. These measured shell values are much lighter than predicted, using the equation of  and the 1986-1987 water values.
The light shell o13C values may be due to the utilization of metabolic carbon by the mollusc, derived from internal respiration and thought to have an isotopic signature similar to the organic carbon in the tissues.  measured the carbon isotopic composition of the organic carbon in Pitar specimens collected from Narragansett Bay and found the value to be approximately-17.2%0 (PDB  Grossman (1984a), then the contribution from metabolic carbon pool would be roughly 45%.  suggested that the contribution from the metabolic carbon pool ranged from 35-85%, with an average of 56% for several different species of molluscs collected in New Haven Harbor, Connecticut.
The variety of factors affecting the incorporation of carbon isotopes in shell carbonate make it very difficult to ascertain the causes of the observed o13C trends in the Pitar or Mytilus profiles. The first order trend in the o13C profile of the Pitar specimen (an overall increase in o13C for the first several years) may be related to physiological effects such as ontogenetic growth changes and/ or sexual maturity. The increased variability in the record, beginning with sample #13 29 ( Figure 9) could be related to the collection and containment of the specimen in the MERL mesocosm tank, which would have occurred at about this time. The increased variability could also be related to ontogeny, in particular the effects of slower, and probably more intermittent, growth. It is thought that the longer a shell is closed, the greater the build-up of metabolic carbon. This could have occurred at times represented by Pitar samples #3 and #8 (Figure 9)  The second order trend observed in the carbon isotopic profile of the Pitar specimen may be related to the seasonal changes in phosphate in Narragansett Bay. Phosphate concentrations in Narragansett Bay have been observed to fluctuate on a seasonal basis (Figure 13), generally low in the early spring and high in the summer, as discussed by . These same changes in phosphate concentrations are also present in the MERL mesocosm tanks    , Nixon, 1983. In the open ocean the majority of the organic matter never reaches the bottom, but is oxidized. in the water column.
There is no obvious relationship between the measured shell o13C values and the measured 018() values from the same samples for either Pitar or Mytilus.
Such a relationship is suggested by the work of  and to a lesser degree, by Emrich et al. (1970). This indicates that shell o13C values are either insensitive to temperature changes or that the carbon isotopic signal due to the temperature dependence is masked by other isotopic signals or inputs (e.g., changes due to metabolic carbon inputs or changes in the carbon isotopic fractionation due to growth rate variations). For either shell we have examined isotopically, it would be impossible to determine the temperature of shell formation given the o13C composition of the shell carbonate and the isotopic composition of the DIC.
Although shell growth for Mytilus begins approximately one month earlier than for Pitar, it does not begin soon enough to record the anticipated short term changes in the o13C of bay water as a result of the major seasonal (winter-spring) bloom in the bay. This bloom is mainly due to the production of the diatoms Dentonula confervacea and Thalassiosira nordenskioeldii. The effects of this bloom on the nutrients in the bay are not observed in the yearly data tabulation of Pilson (1985b) but may have been averaged out due to the method of reporting the data.
The recent study by  indicates that time lags, on the order of months, are present between the deposition of the organic carbon from the winter-spring bloom and the remineralization of the (deposited) organic material.
The Nucula samples exhibit o13C values generally much more positive than the samples from either Pitar or Mytilus. The Nucula specimens collected from the MERL control tank have o13C values between 1.5 and 2.0%o (PDB) ( Figure 11). As sub-surface deposit feeders, Nucula are immersed in pore waters with very light o13Crnc values, due to the oxidation of isotopically light organic carbon . This apparently has little bearing on the carbon isotopic composition of their shells. Differing levels of primary production (and hence the removal of lighter carbon from the DIC pool) between tanks must account for the observed difference between the Nucula samples. The data presented by  demonstrate that the levels of apparent production and measured chlorophyll-~ are more or less in line with the levels of nutrient addition. During periods of peak productivity (February-April) the Bx, 16x and 32x nutrient addition tanks commonly had chlorophyll a levels ten times higher than the control tanks. The observed high o13C values of Nucula shell carbonate are probably due to the storage of the light carbon isotope in the additional biomass (including accumulated organic matter) present in the nutrient addition tanks.

32
The observed increase in o13C with increasing nutrient addition is not a smooth trend. The ot3C results of the Nucula collected from the 8x and 16x nutrient addition tanks are approximately the same (2 to 3%o), although distinct from either the control, or the 32x results. The ot3C results from the control and 32x tanks indicate heavier isotopic values for the Nucula shells from the 16x tank than the 8x tank. Whether or not the similar 8x and 16x ot3C values are due to factors operating between the tanks, or within the carbonate deposition mechanism of the mollusc, is very difficult to determine. There are a great number of factors affecting the isotopic composition of dissolved inorganic carbon and the incorporation of this carbon into shell carbonate. ff the total biologic production was the same in both tanks, and the benthic organic carbon storage equal, then the draw-down of the lighter carbon isotope would have to be approximately the same, leading to the observed results. ff we use the chlorophyll-~ measurements as an indicator of production, we must conclude that the production in the 16x tank was almost always higher than the 8x tank, leading to higher ot3Cn1c values in the 16x tank. The observed oxygen concentrations in the 8x tank had peak (spring-time) values that exceeded those of the 16x tank for brief periods of time, although the net seasonal production was greater in the 16x tank. On the other hand, the 8x tank had the greatest benthic biomass, including the greatest biomass of bivalves, of any of the mesocosm tanks at the end of the eutrophication experiment in 1983. This bentic biomass would 'lock-up' more of the lighter carbon, making the ot3Cn1c in the 8x tank more positive than that in the 16x tank. The 8x tank was also the only tank to show a potential for a large net accumulation of organic matter ) and thus, the storage of the lighter carbon isotope relative to the 16x tank.
ff the benthic organic carbon regeneration in the 16x tank operated at a higher rate than that of the 8x tank, the isotopic effects of the production draw-down on 33 the 813Co1c would have been balanced more quickly in the 16x tank. Further isotopic investigations are needed to resolve the relative contributions from these various sources.

CONCLUSIONS
The isotopic analysis of incremental shell samples from Pitar morrhuana, Mytilus edulis and Nucula annulata demonstrates the utility of this method for determining shell growth patterns and reconstructing environmental conditions as recorded in the shell carbonate. The isotopic analyses, along with the detailed data available from the MERL mesocosms, have enabled us to accurately interpret the 818() and 813C signals in the shells of these bivalves in terms of known variations in the chemical and physical properties of the water in which they grew. The data indicate that the aragonite shell of Pitar appears to be forming at oxygen isotopic equilibrium with the ambient water (as defined by the aragonite-water-temperature equilibrium equation of  and that the 818() profile from this specimen is primarily controlled by temperature changes.

Fig.
Concentrations of phosphate in Narragansett Bay and in MERL tanks. The stippled area represents the field in which were found approximately 95% of the monthly means of weekly observations in l ower Narragansett Bay, during a six-year period of observation. The two dashed lines connect the monthly means of weekly observations in three MERL control tanks during a two-year period. The solid line connects the monthly means of weekly observations in three tank s run in batch mode, with no sea water added during eight months of observation. There are no seasonal trends in 5 13 C of either the shell edge samples or the incremental. The 5 13 C values are more depleted than predicted, indicating that the carbon isotopic signal is influenced by the uptake of isotopically light metabolic carbon.

INTRODUCTION
Detailed isotopic analyses of the shell material of molluscs have revealed information about growth patterns of the molluscs as well as environmental conditions present during growth. When used in conjunction with growth increment analysis , a better understanding of factors such as growth rate and population size versus environmental influences, including temperature, salinity, pollution, and food availability, may be revealed. This study uses such techniques to examine the growth history and ecology of the hard clam Mercenaria mercenaria (Linne), collected from various locations in Narragansett Bay, Rhode Island (Fig. 1).
The hard clam is common along the east coast of the United States, ranging from the Gulf of Saint Lawrence to Florida and into the northern Gulf of Mexico . M. mercenaria is a infauna! filter feeder that inhabits the intertidal and shallow subtidal zones of bays and estuaries, tolerating a wide variety of temperature and salinity regimes. Although M. mercenaria is geographically widespread and well studied, virtually no isotopic studies have been undertaken on the shells of this species. We believe that stable isotope studies can significantly improve the understanding of rate and season of shell growth. Extensive work has been done on the shell microstructure of this species (see Grizzle and Lutz, in press), and in this paper we compare the isotope results from the shells to the observed shell growth patterns. In addition, to develop a better understanding of ecological influences on growth rate, we set out to assess the suitability of M. mercenaria as a monitor of environmental conditions in the bay. We hoped that these results, for example, could also be used to interpret the shell record of specimens collected from archeological sites in terms of environmental change. Many of the more recent isotopic studies of molluscs have lacked the detailed sampling of the local growth environment that has such a strong influence on the isotopic record, relying instead on inferred relationships, often from differing environments. This study combines the detailed stable isotopic analyses of M. mercenaria specimens collected from Narragansett Bay with sampling of the bay water in order to determine the S 1 3c of the total dissolved inorganic carbon and the 8 1 80 of the water.

ENVIRONMENTAL SETTING
Narragansett Bay is a weakly stratified, temperate estuary with strong tidal mixing and low fresh water input as described by   Salinity values were obtained from the literature , Kremer and Nixon, 1978, the MERL group (Oviatt, pers. comm., 1987) for the lower bay, and from Scott Nixon and Steve Granger at GSO (pers. comm., 1987) for the upper bay and Providence River. The majority ( > 90%) of the (weekly) salinity measurements at MERL, which are essentially surface water values, fall within the range of 29.5 to 32.0%0, and are representative of the surface salinities in the middle and lower bay. Short-term, low-salinity excursions observed at 77 MERL ( Fig. 3A) are probably due to periods of high runoff and related to the height (near sea-surface) of the MERL seawater intake. Figure 3a shows the weekly salinity measurements at the MERL header tank from October 1984 to January 1987 and from one of the MERL control tanks (tank-12) during part of that time period. The lower-bay surface salinity data presented by  has a range of values similar to those measured at the MERL tanks (all years). Samples for isotopic analyses were roasted under vacuum at 390°C for one hour and then reacted in "100%" orthophosphoric acid at a constant temperature of 50°C (±0.5°). The resultant C02 gas was subsequently analyzed on a V.G.
Micromass 602-D dual collection isotope ratio mass spectrometer. The isotopic data are given in the conventional delta (o) notation as the per mil (%0) enrichment or depletion in 1 8a and 13 C relative to the Pee Dee Belemnite (PDB) standard (Craig, 1957). NBS-20 was used at the start and end of each session on the mass spectrometer as a standard. The analytical precision for the shell samples was +0.1 %0 for both o 1 8a and o 13 C (standard deviation for the NBS-20 duplicates was the same).
We have found that roasting pure aragonite samples, such as theMercenaria shell samples, under vacuum at 390°C results in an oxygen isotopic fractionation of--0.35%0 (la= 0.06%0), relative to the unroasted sample; a very slight carbon isotopic fractionation equal to --0.06%0 (la= 0.06%0) was also observed (see Appendix II). The values given in this paper have not been corrected for this observed fractionation.
We estimated the o 1 8a of Narragansett Bay water using a two component linear mixing model, controlled by salinity changes (Fig. 3a)     The more commonly used aragonite-water equilibrium temperature equation of  yields slightly different results than their more recent equation . Most of the more recent work involving biologically formed aragonite (e.g.,  has used the earlier Grossman (1982) equation. The earlier aragonite equation was given in the same form as the calcite equation of Epstein et al. (1951,1953)    equation is slightly less than 7%o.
The majority of the difference between the two equations occurs at temperatures below approximately 5°C, corresponding to o 1 8aaragonite values greater than about 2.5%o (PDB), using the conditions given above.

Water Oxygen & Carbon Isotopic Results
The results of the oxygen isotopic measurements of the water (o 1 8awater> samples collected from the MERL header tank are shown in Figure 4b (listed in Appendix I, Table 1 A summer salinity value this low is highly implausible given the various sets of (bottom-water) salinity measurements made in Narragansett Bay over the years.
It appears that the oxygen isotopic composition of the aragonitic shells of M. mercenaria are displaced from isotopic equilibrium, as defined by the equilibrium oxygen isotope fractionation equation of  for aragonite and water.
The shell edge data from the Providence River M. mercenaria specimens collected on July 23, 1984 are also plotted in Figure 7a. These molluscs were forming shell material that was generally isotopically lighter (ave = -4.35%0) than that of the specimens in the bay, just to the south. The difference between the two sets of values (0.75%0) is probably due to either the decrease in salinity prevalent in the Providence River, or the slightly elevated temperatures that commonly occur in the river in the summer . The literature data suggest that the salinity gradient is stronger than the temperature gradient at the head of the bay (e.g.,  and it is likely that most of the observed 0.75%0 o 1 8a shift in the shell carbonate is due to salinity variations rather than temperature variations. A -0.75%o change in the o 1 SO of the water corresponds to a reasonable salinity change of approximately 3%o, using the mixing model described previously. When the shell edge o 1 SO values are plotted against the lengths of the specimens (Figure 8) we see that in general, the smaller specimens have heavier shell edge material than the larger specimens. This indicates that the smaller specimens, which are usually the younger specimens, are forming shell material in colder water than the larger (older) specimens. The smaller specimens apparently have a longer growth season than the larger specimens, judging by this data. The M. mercenaria specimen collected from station 12 has an oxygen isotope profile ( Figure 11) that is quite different from the specimens previously described. to-1.0%0 over most of the shell; but ,with increasing age, the shell o 13 C becomes increasingly lighter, ranging between-1.75 and-3.0%0 for the outermost shell.
This mimics the trend observed in the oxygen isotope record during later growth in this specimen.
The equilibrium fractionation between aragonite and dissolved inorganic carbon (DIC) is somewhat uncertain. Rubinson and Clayton (1969) give a value of 2.7%o (for a mixture of aragonite and calcite) at 25°C whereas Grossman (1984a) suggests a value of 2.40o/oo between dissolved inorganic carbon and aragonite at 20°C.  indicate that the 13 C enrichment Much of the recent work involving detailed isotopic analyses of mollusc shells (e.g., Romanek, et al., 1987) has focused on the correlation between the isotopic record and the growth patterns that are observed on the exterior of the shell and/ or the growth increments observed in a cross sectional view of the shell. The term growth increment can cover a wide variety of scales. The internal increment patterns that can be resolved with isotopes are usually seasonal or annual in nature whereas the growth increments that can be observed with the aid of a microscope have been ascribed to daily, bidaily, fortnightly and lunarmonth growth periods (see  for a review) and are much too fine to be individually sampled for isotopic analysis.
Most bivalves are known to have internal growth increments of one kind or another  and in many cases, such as observed with Spisula solidissima , the more obvious internal growth increments have been shown to be annual in nature, allowing for quick and unambiguous age determinations. The exterior growth bands, or growth checks, common to many bivalves and often very distinct, are not always related to the annual (internal) growth increments . Age estimates that make use of these exterior growth bands may not reflect the true age of the specimen. For the remainder of this paper the term growth increment will refer to the easily observable (with the naked eye), dark and light internal growth increment that is observed in a shell cross section.
The detailed isotopic analyses of the M. mercenaria shells allow for the comparison of the isotopic record (or profile) to the major growth increments 94 observed in these shells. The relative positions of the dark internal growth increments are marked with a small black arrow along the x axis in each o 1 8a profile (Figures 9, 10, & 11). In the outer shell layer the dark growth increments may appear somewhat diffuse and will often split into thinner, multiple bands or a "packet" of bands, as they curve toward the external shell surface. The sharp, narrow, dark growth increments observed in the shells of the specimens from stations 2A and 9 generally appear to represent annual increments based on the oxygen isotope results. Because the dark increments usually split into multiple bands and intersect the outer shell layer at several different sampled intervals, it is difficult to pin down the actual timing of the initiation of these bands. For example, 4 individual dark bands, apparent as a single dark increment near the umbo, intersect the outer shell layer of the specimen from station 9 between samples #63 and #71. These samples have isotopic signatures that indicate summer through fall growth with the last, and sharpest, of the bands coinciding with the heaviest value recorded between two successive growth years. Counted separately, these 4 bands do not represent annual growth. When grouped together these packets occur in different growth years (as defined by the isotopic profiles). The first few dark increments (when the molluscs are less than 3 years old) are fairly distinct and are not always subdivided into multiple bands. These increments are also lighter and thinner than the dark increments that occur later.
After the third or fourth year, the dark increments are always split into multiple bands and are darker than previous bands. This is probably indicative of more erratic growth in such years, a notion supported by the isotopic evidence and the number of samples taken per growth year. The isotopes indicate that after the third year, the multiple splits of a dark increment (a packet) begin appearing in the summer and occur through the fall. There is usually a thin (much thinner than any sample), sharp band that coincides with the heaviest value recorded.
Thus, the most obvious dark increments, or groups of finer increments, are not representative of slow, winter growth, but correspond to periods of erratic growth occurring throughout the summer and fall. These data indicate that the counting of the dark growth increments (counting each increment packet as one) in M. mercenaria will give approximately the correct age of the mollusc, although care must be exercised in correctly identifying all the increments and in using the dark increments to estimate the season of death of the specimen.
The specimen collected from station 12 exhibits a great number of dark growth increments (shown on Figure 11). Most of these increments are too thin to resolve with the sampling methods used here. The growth increments toward the ventral margin (those formed most recently) are too numerous to differentiate by eye, making the outer edge fairly dark in color. The inability to correlate the oxygen isotope record with any seasonal temperature fluctuations in this specimen precludes any determination of the timing of the dark increments.
The shell microgrowth patterns of M. mercenaria have been documented by a number of workers (e.g., Pannella and MacClintock, 1968; as indicating nearly continuous shell growth during a year. These researchers have reported a hierarchy of growth patterns in the shell microstructure, ranging from subdaily to seasonal annual growth increments, as well as abrupt breaks in shell growth related to spawning, thermal shock (including freezing), and storm effects.  reported that the daily growth increments of M. mercenaria range from 1 to 50 mm in winter, and from 15 to 150 mm during summer in specimens collected from a temperate New Jersey estuary (seasonal temperature range of -1.4 to 28.0°C).
Ansell (1968) concluded thatM. mercenaria stops growing at temperatures below approximately 9°C, although temperature is apparently not the only factor limiting growth. The work of  indicates that shell growth in M. mercenaria is continuous for specimens collected from a temperate estuary in New Jersey but, the growth rate decreases rapidly below approximately 10°C.
Although the isotopic evidence indicates that the aragonitic shells of M. This arguement is strengthened if comparisons are made relative to the fine detail of the shell microstructure when examined in thin section. If we use the growth rates given by  then three months of continuous shell growth at lµm per day would result in 0.12mm of total shell growth during the winter months, too small to be sampled with the sampling techniques used here. A shell in its fastest growth years, growing continuously for three months at lOmm per day, would grow approximately 1.2µm in height over that period of time.
Similarly, shell growth of 30µm per day would result in 3.6mm of shell growth recording winter temperatures values (when the o 1 8a of the carbonate is examined). Both of these winter growth intervals would be nearly impossible to miss using the sequential shell sampling methods described above. It is possible that samples filed from the shell margin of smaller specimens (1 to 3 cm) could be used for a more accurate assessment of the total range of isotopic values. The shell edge samples have the advantage of being more detailed, due to their narrower (sampling) size.
The majority of the measured oxygen isotopic ratios are intermediate in value, indicating that the majority of shell growth in any well sampled year occurs in the spring and early summer. Previous studies (e.g.,   is merely the steadiest growth, occurring at a slower but more consistent rate than the growth rates observed in the summer in the studies cited above. H the rate of growth in the summer was faster but infrequent, or erratic, perhaps due to 98 the energy required. for spawning or the availability of food, the result could be an isotopic profile similar to that observed.

Carbon Isotopic Disequilibrium
The B 13 C of the shell carbonate most likely reflects, or is regulated by, a wide variety of processes operating concurrently in the bay, including : phytoplankton production in the water column, regeneration of organic matter by the benthos, seasonal variations in 13 C input from fresh and/or oceanic water, the rate of growth, and the utilization of metabolic carbon by the molluscs, commonly refered to as the "vital effect". The seasonal phytoplankton production in Narragansett Bay is highly variable both in timing and in the quantity of biomass production (Karentz and Smayda, 1984;. This seasonal primary production can alter the B 13 C of seawater bicarbonate by depleting the water in 12C during periods of high productivity and some researchers  have attributed changes in mollusc shell l> 13 C to this process. The degradation of this same organic biomass will eventually return much of the lighter carbon to the water column.  have shown that in Narragansett Bay this remineralization process lags the initial phytoplankton production by a few months. The starting point for their study was the Narragansett Bay winter-spring bloom which usually occurs in late January (plus or minus a few weeks) and is the strongest of the seasonal plankton blooms occurring in the bay. The water samples measured for B 13 C DIC do not show an increase in B 13 C values early in the year that might accompany this bloom. There are sharp changes in the B 13 Co1c of the water in Narragansett Bay and there is a trend toward lighter values through the spring and summer.
Since M. mercenaria does not precipitate sufficient shell material to measure isotopically during the winter months, it is impossible to resolve the B 13 C signal that might occur in the water column as a result of the winter-spring bloom. An examination of the isotopic record from any of the M. mercenaria shells sampled here (Figures 9, 10 The summer o 13 C values may indicate a fairly rapid rate of remineralization of the organic matter from the winter and spring blooms, as measurements in the bay  reveal higher levels of most nutrients throughout the summer.
The same data also show high nutrient levels continuing from summer through the fall, which makes it difficult to postulate a mechanism that would draw down the lighter carbon-12, thus enriching the shell carbonate in carbon-13 as observed.
There is an established negative correlation between surface water phosphate levels and o 1 3C values of total dissolved carbon (TOC) in seawater (Broecker, 1982) involving the uptake of phosphate and carbon-12 during photosynthetic production which depletes the surrounding water in 12(: relative to 13C. It has been suggested that a source of the isotopically light carbon observed in molluscan shell carbonate could be derived from the pore waters surrounding the (infauna!) organisms. It is well documented (e.g., Grossman, 1984a, Mc.Corkle et al., 1985 that the carbon isotopic composition of pore waters decreases with increasing depth in the sediment, due to the bacterial remineralization (oxidation) of the isotopically light organic matter within the sediments. Organic-carbon rich sediments, such as those found in Narragansett Bay , would be expected to have a more pronounced pattern of o 13 C depletion with depth than sediments from the open ocean and one might expect a higher advection rate of light o 13 C out of the sediment pore waters in this environment. Although o 13 C values from Narragansett Bay pore waters are lacking, it is doubtful that this is an important source of light carbon for M. mercenaria. As a large infauna! filter feeder M. mercenaria is probably capable of pumping water right through the benthic boundary layer. Considering the strong tidal circulation of Narragansett Bay, coupled with the additional (and constant) estuarine circulation, it seems that any advecting (isotopically light) pore waters would be rapidly mixed before uptake by M. mercenaria could occur.
The incorporation of metabolically derived carbon, depleted in 13C, into the M. mercenaria shell carbonate is the most likely reason for the observed o 13 C disequilibrium. The idea that this process might be occurring is not new.
McCrea (1950) proposed that molluscs' internal environment could cause the deposition of shell carbonate that was out of isotopic equilibrium with the dissolved bicarbonate of the water (e.g., Craig, 1953, Hammen andWilbur, 1959;I<eith et al., 1964;Tanaka et al., 1984;. Shell formation in molluscs is thought to occur entirely within the environment of the extrapallial fluid layer which is in contact with the shell surface and is a part of the mantle (see , for a recent review). The fluxes of ca2+ and HC03= are thought to be bidirectional (Greenway, 1981a) through the mantle membranes and it has been suggested ) that the ttco 3 = is provided by direct ionic transfer from both the dissolved bicarbonate and from metabolic COi (which includes that C(h, produced (respired) by the body tissues and that produced under the pH control of the extrapallial fluid).
The addition of a small amount of C(h, derived from the molluscan tissue could alter the isotopic signature of the shell carbonate quite easily as the tissue carbon is much lighter (isotopically) than the dissolved bicarbonate in marine waters.    This conclusion is similar to the results of  and it is important to note that the use of different fractionation factors will yield different results. Values for the fractionation factors between the various components of the COi, -CaCOJ system can be found in Friedman and O'Neil (1977), and more recently in Grossman (1984a), Sugimura (1985), Wannindhof, (1985), Herczeg andFairbanks (1987). It is not clear how molluscan metabolic carbon would vary from day to day, or over longer periods of time, as many factors can contribute to the quantity of metabolic C{)i, produced. There is apparently a change in the utilization of the metabolic carbon with increasing age (see below) and it has been demonstrated ) that the organic carbon in molluscan tissue varies with the available food sources. The values for carbon isotopic fractionation between aragonite and bicarbonate and carbonate ion is poorly constrained. Considering the effects of metabolic-derived C02, on shell carbonate, and the problems involved in the separation of metabolic carbon from DIC, it is likely that fractionation factors derived from the study of biologic organisms will probably be in error.
Another possible source of light carbon results from the necessity of molluscs to maintain, or at least moderate, the pH of the extrapallial fluid during shell deposition. The precipitation of calcium carbonate from Ca2+ and HC03= produces hydrogen ion, or protons, which decreases the pH of the extrapallial fluid. It has been proposed ) that carbonic anhydrase, which is present in the extrapallial fluid and is thought to act as a catalyst, catalyzes the reaction: HC03=+ H+-> C02, + H20. The evolved CQi, is expected to diffuse from the extrapallial fluid to the mantle and out to the body fluid (blood) and eventually to the surrounding medium.    Romanek et al., 1985;Krantz et al., 1987). The trend toward lighter o 13 C values is usually concurrent with a change in the recorded otB() values. The oxygen isotopic values continue to change with seasonal temperature variations, but shell growth is apparently restricted to warmer temperatures as evidenced by the lower o 1 8a amplitudes. The o 13 c trend has been attributed by Romanek et al. (1985 and to physiological changes occurring during the transition from a primarily juvenile stage to the adult stage and the resultant shift in energy from shell growth to gametogenesis.

Oxygen Isotopic Disequilibrium
The previous discussion on the sources and magnitude of the carbon isotopic fractionation and causes of disequilibrium may shed some light on the apparent oxygen isotope disequilibrium observed in M. mercenaria. The depleted shell o18Q values can also be attributed to the incorporation of isotopically light oxygen, derived from metabolic sources that may have affected the carbon isotopic compositions . The carbonate formed by molluscs must be in isotopic equilibrium with metabolically derived sources of carbon dioxide and water, as well as the ambient, non-biogenic sources. The main sources of isotopically light oxygen that could be incorporated into molluscan shell carbonate are respired (metabolic) carbon dioxide and metabolic water, via the (highly simplified) reaction: . The assumption is made that the respired water and carbon dioxide are in isotopic equilibrium with each other.  has shown that the isotopic ratio of the respired oxygen is depleted relative to dissolved oxygen not used in the metabolic processes (dissolved oxygen is approx. +24%o SMOW, or -7%o PDB, respired oxygen is approx. +3%0 SMOW, or -28%0 PDB). Grossman's model (1987) for the incorporation of metabolic oxygen into biogenic calcite predicts that a pure "metabolic" calcite, formed near the surface, will be depleted by 17%0 relative to an inorganic calcite precipitated in the same water. The values estimated for aragonite would presumably be enriched by approximately 1 %0. The 1 to 1.5%o 8 18 0 depletion that we observe in warmest M. mercenaria samples would be accounted for by the incorporation of 6-9% of the depleted metabolic oxygen source, using this model. This percentage range is lower than that predicted using metabolic carbon (36%), which may indicate that 1) the metabolic oxygen isotope model developed by Grossman for calcite formation is incorrect; 2) the metabolic carbon model is incorrect; 3) the equilibrium fractionation factors between aragonite (both biogenic and inorganic) and the important chemical species (C{)i, Hc03=, CHi{)) are not known well enough for accurate modeling; 4) other processes are involved that moderate the contribution from metabolic oxygen (i.e., kinetic fractionation, 1 8a reequilibration with water); or 5) all of the above.
The kinetic fractionation of isotopes that may affect the oxygen isotopic composition of shell carbonate. The rate of carbonate formation will affect the isotopic composition of the shell, if kinetic fractionation is an important process, with faster precipitation rates resulting in depleted shell material.  has shown that precipitation rates affect the carbon isotopic composition of inorganic calcite--aragonite mixtures with slow rates of precipitation producing enriched carbonates (+3.4%o PDB) relative to carbonates formed by fast precipitation (+0.4%0).   The isotopic exchange of metabolically derived oxygen with the oxygen of the ambient water is a process that could account for carbonates with near 106 equilibrium oxygen isotopic compositions, and disequilibrium carbon isotopic compositions . This is an intriguing possibility, as a great number of biogenic carbonates are formed at, or near, oxygen isotopic equilibrium, but out of carbon isotopic equilibrium with respect to dissolved inorganic carbon. This process could occur within the extrapallial fluid or at the site of calcification.
Mytilus edulis sample number 0 is from the (ventral) shell edge, and represents a sample of the last shell material formed before collection. Only the outer calcitic layer of the shell was sampled.         A section of M. mercenaria shell was ground and seived (through a 163 micron seive) for this experiment. The samples labelled unroasted (untreated) were run after the grinding; the samples labelled roasted were heated in vacu for 1 hour at 390°C (our standard carbonate preparation technique); the samples labelled hypochlorite were soaked in a 53 hypochlorite solution for 24 hours, air dried at 40°C then run on the mass spectrometer. The 8 18 0 values of the samples that were roasted were typically 0.43%0 lighter (more depleted) than the samples that were not roasted. The samples treated with hypochlorite solution were 0.31 %o lighter than the samples that were not roasted, but 0.12%0 heavier than the samples that were roasted.