Anomalous levels of ?% and 239~240Pu in Florida corals: Evidence of coastal processes

Abstract Strontium-90, a radionuclide whose primary source is fallout from nuclear weapons testing, serves as a tritium-like tracer of ocean circulation. The historical record of 90 Sr activities in the annual bands of island corals have been shown by other investigators to reflect the 90 Sr concentration in surface waters at those site. Strontium-90 activities measured in annual bands in Montastrea annularis from the Florida Keys are 30–120% higher than those in corresponding peak activity years (1960–1965) of a Bermuda coral ( Diploria ). The Bermuda 90 Sr activity record reflects the fallout source only, whereas the additional 90 Sr activity in the Florida Keys is expected to reflect a coastal runoff source as well as the fallout. The coastal circulation patterns off the northern and western edge of the Florida Current further act to concentrate and prolong the exposure of the runoff 90 Sr to the corals. Six measured 239,240 Pu activities in the Florida coral are 30% of 239,240 Pu activities in island coral records previously reported. Since Pu is expected to be scavenged by particles in coastal waters, this decrease in 239,240 Pu substantiates the importance of coastal influences in the Florida 90 Sr record. Strontium-90 activities measured in subannual coral bands from 1973 to 1974 reflect seasonal changes in the 90 Sr concentrations in the surface layer of the coastal waters. This may reflect Loop Current intrusion events. The seasonal and long-term coral 90 Sr data presented in this paper suggests that coastal 90 Sr coral time series may be very useful for documenting coastal circulation patterns.

THE TIME HISTORY of 14C/12C in atmospheric CO2 using annual bands in tree rings was demonstrated by DEVRIES (1958) and many others. Similar records have been obtained from banded corals to track the history of changes in the "C/'2C ratio in the surface ocean (NOZAKI et al., 1978;DRUFFXL and LINICK, 1978; DRUFFEL, 198 1). Annual banding in corals is manifested by regular density changes in the accreted aragonitic skeleton. KNUTSON et al. (1972) was first to show that X-rays of slabs of coral cut along the vertical axis of growth of the scleractinian coral Montastrea annulari~ revealed consecutive dark and light bands. Sectioning the coral into yearly samples retrieves an historical record of the chemical and isotopic composition of the seawater surrounding the corals. The accreted skeleton of the coral has been shown to incorporate minor (i.e.; Sr, Mg, Na;WEBER, 1973WEBER, , 1974G~REALJ, 1977;SMITH et al., 1979;SWART, 1981)  Strontium-90 was introduced to the environment as a product of atmospheric nuclear weapons testing during the period 1952 to 1962. Unlike 3H and 14C, there is no natural source of ?jr. It is an excellent tracer of ocean circulation due to its conservative behavior in the sea and equally important, its incorporation into corals. There is no fractionation of "Sr/Sr in corals relative to its composition in seawater (TOCGWEILER, 1983;NOSHKIN et al., 1975) and the Sr/Ca ratio in corals relative to seawater is affected to only a minor extent by sea surface temperature, coral growth rates and coral species (THOMPSON and LIVINGSTON, 1970;WEBER, 1973;G~REAU, 1977;HOUCK et al., 1977;SMITH et aI., 1979;SWART, 198 1;SCHNEIDER and SMITH, 1982). For example, BENNINGER and DODGE ( 1986) reported the constancy of Sr concentration per gram of coral over a 30 year period in Montastrea annularis from 195 1 -1980, with a Kmsrj = 1.040 f 0.008.
The feasibility to recover 90Sr records from corals and the similarity of these records to the fallout *Sr atmospheric record as measured in New York City was documented by T~GGWEILER (I 983) in corals from the Atlantic (Bermuda) and the Pacific (Oahu) (TOGGWEILER and TRUMBORE, 1985). Measuring ?Sr activity levels in corals has the advantage of providing a complete long-term historical record of the 90Sr activity present in the waters at one site. Previously, only limited time-series measurements were made for locations throughout the Atlantic and Pacific during seawater sampling programs, but these measurements could not represent the dynamic changes occurring constantly throughout the oceans. In addition, annual sampling of coral bands average out seasonal changes which would be reflected in % levels measured in single seawater samples.
Sr removal from the atmosphere is much faster than removal of C02, thus the response time of the oceans to fallout ?Sr is on the order of one year versus 10 years for 14C. Although bomb-generated 3H has been used as a tracer for ocean circulation (e.g., OSTLUND et al., 1974;JENKINS, 1980;JENKINS and CLARKE, 1976;FINE et al., 1981), future 'H measurements will be limited by the sensitivity of the analytical instrumentation used for 3H due to its short half-life of 12.4 years. Presently there is no practical method for retrieving the 3H record from the organic matter in corals. Strontium-90's half-life of 28.5 years, combined with the input record in corals, provides a complementary tracer whose potential has not yet been fully realized.
In order to use 90Sr in corals as an effective circulation tracer, the input function of 90Sr to the coral site must be known. For instance, corals growing in coastal areas may be influenced by fluvial transport of WSr from the contitients, in addition to the fallout input. Previous WSr coral work (TOGGWEILER, 1983;TOGGWEILER and TRUMBORE, 1985) has focused on samples from central gyres, far from continents, where the input was dominated by direct deposition from the atmosphere. In contrast, this paper concerns the historical record from the Florida Keys, a near shore site. The results are compared with those from a Bermuda coral representative of a central gyre site. The unexpectedly high ?Sr concentrations in the Florida coral as compared to the Bermuda coral suggest strong coastal influence in the Florida coral record. Extremely low 23y~2?u concentrations in Florida coral relative to a Caribbean coral (BENNINGER and DODGE, 1986) further confirm our suspicion of coastal influence.

COLLECTION OF CORALS AND %r AND =-'% ANALYSIS
Cores from massive colonies of Montastrea annularis were collected in 1983 from "The Rocks" reef (24"57'N, 80"33W) off the Florida Keys from 4 meters water depth (see Fig. 1). This coral reef is 2 km south of Plantation Key. The major water mass to the south of these corals is the Florida Current, which is part of the Gulf Stream system. These corals are influenced by the surface waters ofthe Florida Current and the Florida Bay to the north. "The Rocks" samples were obtained from one core (TR3) collected in February, 1983; it contained growth from A.D. 1939 to 1983. An additional specimen of Montastrea annularis from New Ground (24"4O'N, 82"25W) was collected in 4.0 meters in July, 1983 (see Fig. 1). This site was chosen to represent a second distant Florida Keys reef site that might be influenced by runoff from the land to provide a check of %.r activities recorded in the first primary Florida samples. The New Ground coral was living in green, "murky" water, a sign of coastal influence instead of the dark, purple-blue waters of the Gulf Stream (E. SHINN, pen. commun.). Bands representing four years were selected from this coral, although the 1960 sample was lost.
The   . Each annual sample was sectioned from the core by cutting (with a band saw) through the center of the dense band in Montaskea annularis from Florida; in this way, a given band for year x represented the growth interval from August of year x -1 to July of year x. For the Diploria strigosa samples, the cut was made just below the dense band, representing growth from April to March. Seasonal samples were sectioned from two years (1973 and 1974) in "The Rocks" Florida coral by grinding three-month intervals using a Dremel tool. Annual coral bands were pulverized and dissolved in 8 N HNO, acid. The 1973, 1974 samples and all of the Bermuda samples were dissolved in 4 N HCl to remove the CO2 first for i4C analyses (DRUFFEL, 1989). The Sr extraction and purification chemistry procedures used were reported by BOWEN (1970) and LIVINGSTON et al. (1974). As q (daughter of ?Sr) has a much shorter half-life of 64 hours and a maximum beta energy of 2.27 MeV, the v activity was determined using anticoincidence gas proportional beta counters (NOSHKIN and DEAGAZIO, 1966) and the %r activity derived from this, correcting for any Sr loss by analyzing for stable Sr carrier using Atomic Ab sorption Spectrophotometry. Uncertainties reported here are related to a least square fitting of the data to a 64 hour decay curve assuming a 70% confidence interval. All ?Sr values are decay corrected to the year of accretion and reported as dpm/lOO gm of coral which is essentially equivalent to dpm/ 100 L of Seawater. One hundred grams of coral contains 1 .OO mole of Ca and a Sr/Ca ratio of approximately 0.009. One hundred liters of seawater contains 1.06 moles of Ca, and a Sr/Ca ratio equal to 0.0 1. The Sr/C&,,r&r/C~-, discrimination factor is between 1.03 to 1.08 for a temperature range of25-30°C (SMITH et al., 1979). Thus, this 3-8% enrichment allows for direct comparison of ?Sr activities in corals and seawater. TOGGWEILER (1983, p. 209) concluded that ?lr concentrations in corals were valid recorders of seawater ?Sr concentration by comparing Weathership E ?Ir seawater data near Bermuda with his Bermuda coral record.
Pu was extracted and analyzed in eight of "The Rocks" Florida samples. A 24*Pu spike was added to the acidified coral sample as a yield monitor. Sr was first precipitated with concentrated HNO, and removed from the coral solutions, then Pu was removed by ion exchange and analyzed using the Pu electroplating process reported in LIVINGSTON et al. (1975). The 239,2%r activity was measured in alpha counters previously described in MANN et al. (1975). A Pu/ QXXilDWCaWWW discrimination factor of 1.8 + 1.2 has been previously reported by BENNINGER and DODGE (1986) indicating an enrichment in coral.

RESULTS AND DBXUSSION
Strontium-90 measurements of the Florida Keys and Bermuda corals are listed in Table 1 and shown in Fig. 2. A striking feature of the Florida (The Rocks) '%r curve ( Fig.  2b) is the similarity to the general shape of the New York City (4O'N) WSr atmospher+ depostion (surface air sampling, HASL, 1977;TOONKBL, 1980;Fig. 2a). The highest 90Sr values are found for the 1964 growth band (August 1963-July 1964), slightly later than the maximum observed in the atmospheric deposition record (1963). T~GGWEILER and TRUMBORE (1985) and MADAZAR et al. (1987) have both reported similar shaped curves of WSr in Oahu and Ft. Lauderdale corals, respectively, and BENNINGER and DODGE (1986) in a St. Croix 23g*240Pu coral record. The reduction of "Sr activity in the atmosphere during the '59 to '61 period ( Fig. 2a) is recorded in the coral as a minimum in the '60 to '62 period (Fig. 2b). This OS-l.0 year lag time appears to 30 1 1  (HASL, 1977;TOONKEL, 1980) has been adjusted to represent a Florida coral year. (b) Sr-90 activities in Montastrea annularis from The Rocks, Florida Keys and Diploria strigosa from North Rock, Bermuda. Results from the second Florida Keys site, New Ground (see Fig; 1) are available for three years (see text). The temporal position of each point is fixed at the centroid of a coral year; for Florida, a co& year is August to July, but for Bermuda it is April to March. Vertical bars represent uncertainty in activity. The Sr-90 activity coral record in the Florida Straits is 30-120% higher during the peak fallout years than that for Bermuda. represent the response time for ?Sr in the atmosphere to be deposited in the water and incorporated in the skeletons of shallow ocean corals.

3-90 Annual Deposition 2a New York City
The general decrease of-r levels throu~out the 1970s in the Florida coral is more gradual than the atmospheric deposition curve. This is expected, due to the longer residence time of '?Sr in surface seawater than in the atmosphere. Sr-90 is removed quickly from the atmosphere by precipitation, whereas it remains dissolved in seawater and is removed only by mixing with subsurface waters, horizontal advection and by radioactive decay. Dissolved -r in runoff from continents over a period of several years also adds ?Sr activity to surface seawater, prolonging this isotope's apparent residence time.
The ?Sr results from Bermuda (North Rock) are listed in Table 1 and plotted in Figs. 2b and 3. The Bermuda coral was analyzed to obtain an annual central gyre record (Fig.  2b), as well as to compare our values with previously reported results from Bermuda (~OGGWEILiZR, 1983, Fig. 3). Direct comparisons are difficult, since one year bands were used for this study, while TOGCWEILER (1983) uses two year bands which smooth the overall curve particuh~ly at the peak fallout years (l958-196Oand 1962(l958-196Oand -1966.TheBermudaresultsfrom TOGGWEILER ( 1983) represent a complete record, while our record only covers eight selected years between 195 1 and 1982. TOGGWEILER ( 1983) verified his results by comparing his coral r'?jr activities to surface seawater ?Sr activities at Weathership "E" station (35"N, 48'W) which represents a location in the center of the North Atlantic gyre. We have verified our results by comparing our coral data to surface "Sr seawater data for a 10" by loo area encompassing Bermuda (28-38'N, 60-7O'W). The seawater data presented in   The two Bermuda coral data sets differ by appro~ma~ly 50% following the peak fallout years, But the seawater data, presented to verify the coral results, falls between the coral "Sr curves. Ahhough the seawater data matches our coral data better in 1970, we do not have enough data points throu~out the 1965 to 1970 period to further confirm the closer fit of our coral data to the surface seawater data.
The major difference between the Florida and Bermuda 9oSr coral activities is the absolute ~on~ntration (Table I, Fig. 2). During the years of peak atmospheric delivery (1959)(1960)(1961)(1962)(1963)(1964)(1965) the "Sr activity is 30-120% higher at the Florida coral site than at the Bermuda site. However, based on data of fallout versus latitude, it is expected that the Florida site would receive 10% less '?Sr fallout than Bermuda (JOSEPH et al.,197 1).
Furthermore, ?Sr activities in Florida corals are 1.3 to 2.8 times higher than surface seawater activities in the Caribbean Sea for 1970. SARMIENTO and GWINN ( 1986) predicted an increase of only 1.1 for fallout 90Sr in the surface waters of Florida compared to the Caribbean Sea. Table 2 lists Caribbean 90Sr activities in the surface seawater taken from the location lo*--20"N, 50"-75"W. Based on the above considerations, the Florida coral "Sr record reflects an enrichment of this isotope relative to both the Caribbean surface water and the Bermuda coral. But the enrichment in the Florida coral occurs only during the peak fallout years. During the 197Os, 90Sr fallout was minimal; thus, "Sr levels in the surface ocean during this period were infhrenced primarily by circulation effects. Vpwelling waters derived from the equatorial Atlantic and transported via the Gulf Stream would lower the surface water 90Sr levels. These waters might account for the lowering of the Florida coral "Sr activity relative to Bermuda by 1974.
A confi~ation of the very high activity in the 1964 Florida coral band (100 dpm/ 100 g; Fig. 2b) comes from the duplication of the ?Sr activity in the 1964 band recorded in the New Ground coral (97 dpm/lOO g) which is separated by 100 miles from The Rocks location in the Florida Keys (Fig.  1 1974,1980). But nearshore '37Cs/WSr ratios equal 1 .O. This was accounted for by the combination of enriched "?jr from river runoff and ~oundwater discharge and particle removal of 13'Cs due to high concentrations of particles along coastal regions (BOWEN  et al., 1974).
Pu-239,240 results in the Florida coral are listed in Table  3 and shown in Fig. 4. The Florida coral 23q*2'@Pu activity levels, like '*Sr, reflect a direct response to the atmospheric fallout changes. Similar to the "Sr activity record, the peak 23q,240Pu activity year occurs in 1964, about one year aRer the atmospheric maxima. The 1966 Pu sample in Table 3 appears to be anomalously low. It is not clear whether this resulted from unknown analytical error or is a real effect due to coastal circulation effects and the position of the Loop Current during that year (see Oceanographic Considerations section below.) Our z39~2*Pu data for The Rocks Florida coral are strikingly different from those reported by BENNINGER and DODGE ( 1986) for St. Croix corals (Fig. 4, Table 3). Though the shapes of the two records are similar, the Florida coral contains approximately 20-35% of the *3g,240Pu levels in the St. Croix corals, comparing six individual year results. (The 1966 result has not been induded in this comparison.) The *39,*?u,@Sr ratios for The Rocks Florida coral are reported in Table 3. This ratio has two maxima of 0.0039 in 1961 and 0.0037 in 1964 which reflect the fresh fallout increases seen in the atmospheric record before the nuclear weapons test ban in 1962. This ratio in fresh fallout is 0.0 18 (HARLEY, 1975;SHOLICOVITZ, 1983). Following the 1964 maximum, this ratio decreases significantly with time. The decrease in the ratio illustrates that less fallout 23y*24'?u is available to the corals relative to fallout '?Sr as time progresses, because less of the 239*24~u remains in the water column relative to 90Sr. Sr is a conservative element and therefore remains in the dissolved form, whereas Pu is more particle reactive and is scavenged from the water column. Pu removal is particularly pronounced in coastal areas where high concentrations of particles exist in the water column (SANTSCHI et al., 1980, SHOLKOVITZ andMANN, 1987). Relative to Pu delivered to the surface land or sea, only 3% of Pu is removed from the surface water column in open ocean, whereas 90% is removed along coastal regions and 97% in lakes (SHOL-KOVITZ, 1983). Consequently, the 239*2@Pu in the surface ocean nearer the continents is removed quicker due to the combined effects of dilution and scavenging from the water column as compared to the 90Sr radionuclide which must be decreased by dilution and decay only.

OCEANOGRAPHIC CONSIDERATIONS
To interpret these *Sr activity levels in relation to ocean circulation and mixing processes, we consider the differences in vertical mixing for the Caribbean Current vs. the Atlantic Ocean. Changes in the Loop Current position (see Fig I) within the Gulf of Mexico may explain differences in the 90Sr seasonal record in the Florida Keys coral.
Ocean water entering the Caribbean Sea/Yucatan Straits region is stratified due to subtropical temperature and salinity effects (WUST, 1964) Table 3.) This is due to the geochemistry of Pu in coastal waters (adsorption on particulates) as compared to the oceanic waters influencing the Caribbean coral.
the Florida Straits region. DRumI. (1989) studied upper ocean ventilation in the Sargasso Sea gyre using h&h precision bomb A14C measurements. An increase in A14C occurred earlier in Florida corals than in Bermuda corals. This is not due to coastal influences. Radio~rbon is introduced to the surface ocean via isotopic gas exchange, and with a relatively long turnover time (ten yeam, DRUFFEL and LIMCK, 1978).
Modelling the coral results indicates that the bomb 14C signal was damped at Bermuda during 1960-1970 due to increased mixing with "C-poor subsurface waters (200-400 m) during 18" or mode water formation that occurs during late winter (WORTHINGTON, 1976). Conversely, the thermocline in the Florida Straits during the winter months deepens to only 50-75 m, thus allowing a much smaller volume into which the bomb 14C can be mixed, resulting in a higher concentration of this isotope in surface waters overall. Seasonal ?!Zr activity levels in Florida corals are shown in Fig. 5b and listed in Table 4. Both 1973 and1974 show the lowest activity during the winter months of January to March. Since it has been shown that temperature affects the Sr/Ca ratio in corals to only a minor extent (less than 10% change over 12°C range reported in SMITH et al., 1979) or not at all (GOREAU, 1977;SWART, 198 l), the seasonal variation in 90Sr activity in corals is likely due to a change in the source of water to the coral site. The 30% lower coral %Sr values during the winter could not be due to increased depth of the mixed  (STURGES and EVANS, 1983). No data points are available for months 11, 12/73 and 7,8,9, 12/74. A complete time series (1965 to 1978) assigning monthly points for the Loop Current position based on a cubic spline fit through the available ~ydr~phic data appears in STUIZGES and EVANS (1983). (b) Seasonal "sr activities for 1973 and 1974 in Man-&s&en a~nula~s from The Rocks reef, Florida Keys. Three months of coral growth were sectioned from the core for %r a&y&s for two consecutive years. Vertical bars represent uncertainty in activity value. See Table 4 for comparison of actual values. layer and entrainment of lower '?$r waters, since 90Sr levels in the subsurface waters are the same or higher than those in surface waters (BOWEN et al., , 1974LIVINGSTON et al., 1985). There is prelimin~ evidence that this seasonal variation in 9oSr activity in corals is related to the position of the Loop Current.
Waters from the North E@atoriat Current and the Guiana Current primarily pass through the Lesser Antilles into the Caribbean Sea, with eventual passage through the Yucatan Channel (WUST, 1964) into the Gulf of Mexico. During some periods of the year this water travels nearly directly north toward the Florida Keys, then turns east into the Florida Straits. However, the degree of penetration of this cument into the Gulf of Mexico typically increases with time, forming a progressively larger anticyclonic Loop Current into the Gulf (Fig. 1) before exiting into the Florida Straits (MAUL, 1977;LEIPPER, 1970;HUH et aI., 1981;VUKOVICH et ai., 1979;MOLINARI and MAYER, 1982;Now-LIN and MCLELLAN, 1967;BEHRINGER et al., 1977). An annual cycle of Loop Current growth and eventual decay due to eddy separation has been documented over various periods STURGES and EVANS, 1983;see Fig. 5a). It has been shown that this Loop Current gyre in the Gulf has extended almost to the Mississippi Delta in August 1973(MAUL, 1977 and onto the West Florida Continental shelf passing within 8 km of shore in February 1977(HUH et al., 1981. These extensive intrusions mix with and entrain north and east coastal Gulf waters significantly changing the properties (i.e. salinity, tern~mt~, etc.: MAUL, 1977;MOL.INARJ and MAYER, 1982) of the water mass exiting cycloni~ly into the Florida Straits. For example, very low salinity water (24k) observed all along the edge of the Florida Current off the Florida Keys coincided with the August 1973 maximum intrusion event (MAUL, 1977). Figure 5a represents the Loop Current position time series for 1973 and 1974 taken from STURGES and EVANS (1983). When the maximum intrusion occurred in these years, the Florida coral 90Sr activities coincide with seasonally high values (Fig. 5b). We suspect that the coral registers more 90Sr during the summer months, possibly due to increased en-t~nment of coastal water as a result of northward Loop Current in~sion into the GuIf of Mexico. It is difficult to resolve the time lag between the ?$r activity increase in the coral as a response to the Loop Current penetration into the Gulf since crucial position data is not available for November and December, 1973. Additionally, the resolution in the coral data is in three month intervals, whereas it is in one month intervals for the Loop Current data.
An extensive and prolonged Loop Current intrusion occurred in 1966 (STURGES andEVANS, 1983). This major intrusion event might account for our very low 1966 coral Pu value due to particle laden coastal water influence in the Gulf Stream which would remove Pu from the water column.

COASTAL CIRCULATION NEAR THE FLORIDA KEYS
To investigate the possibility of coastal influence on the '%r coral record at The Rocks reef in greater detail, it is necessary to examine the circulation patterns in the Florida Straits and around the Florida Keys. Extensive meanders of the Gulf Stream through the Florida Straits create localized circulation patterns off the Florida Keys and southeastern Florida coast important to this study. Currents in the axis of the Gulf Stream reach speeds up to 200 cm/set. However, cyclonic spin-off eddies, caused by offshore meanders of the stream, can occur ah along the north and western edge of the Florida Current causing current reversals along the adjacent coastal regions (LEE and MAYER, 1977;LEE and MOOERS, 1977;LEE, 1975;BROOKS and NIILER, 1975;see Fig. 6). These cyclonic eddies continually exchange Florida Current water with coastal water on the shelf. The Pourtales Terrace is the extensive 30-50 km platform (<200 m depth) off the lower and middle Florida Keys (Fig. 6a). It is characterized by persistent coastal and bottom countercurrents and longer duration cyclonic gyres than the weekly produced inshore cyclonic gyres created downstream off Miami (Fig. 6b;LEE, 1975;BROOKS and NIILER, 1975). These persistent cyclonic flows provide a mechanism for fish larvae transport from remote upstream spawning grounds (i.e. north of Key West and inside Florida Bay) to downstream coastal and nearshore nursery areas such as inside the upper Keys (T. N. LEE, pers. commun., REHRER et al., 1967;MURPHY et al., 1975). Additionally, the Gulf Stream is in a curving mode off the Florida Keys and is therefore located more offshore compared to its location along the coast off the narrow Miami Terrace. The Gulf Stream runs straight all along the lower eastern Florida coast. Therefore, corals along the eastern coast of Florida are expected to be more directly influenced by Gulf Stream water than those along the southern coast off the Keys. MYLE et al. (1984) showed that copper enrichments in the surface waters of the Florida Current could be accounted for from river sources to the Gulf Stream. However, we have concluded that the "Sr in the Mississippi River runoff could not account alone for our 90Sr Florida coral enrichments if integrated throughout the upper 50 m of Gulf transport. Gulf Stream waters carrying a surface fallout 90Sr activity of approximately 50 dpm/ 100 L for 1964. However, as the axis of the Gulf Stream is held offshore due to meanders and the Pour-tales Terrace, coastal eddies carrying constantly exchanged waters from the coastal shelf and the edge of the Gulf Stream might concentrate coastal waters over the Florida Keys corals.
Closer range influences must address Florida Bay runoff effects. LANDSAT images and ship track records in a Loop Current study (MAUL, 1977) and digital thermal infrared data (ROBERTS et al., 1982) record evidence of Florida Bay runoff advecting through the Keys. In fact, winter runoff water passing from the Bay through the Keys producing lethally cold (< 12OC) shallow-water accounts for the absence of a living reef tract across tidal passes within the Keys (ROBERTS et al., 1982). The Florida reef tract is generally protected by the buffering effect of warm oceanic water.
HUDSON (198 1) transplanted Montastrea annularis along an inshore-offshore transect from Snake Creek to Cracker Reef to study the effects of environmental stress on reefbuilding corals. He attributed the death or reduced growth of corals closest to the tidal pass at Snake Creek to water coming from the Florida Bay which can have temperatures outside the tolerance range of 13.9-32°C for M~n~astre~ annularis. Since our coral (The Rocks) is located 4 km from Snake Creek in 4 m ofwater, it is possible that it is influenced by runoff from the Florida Bay. We believe the combination of close range runoff and coastal eddies could concentrate the ?Sr activities in the shallow waters off the Florida Keys, much like a blender that recirculates fluid within a finite volume.

CONCLUSION
The time histories of atmospheric introduction and fallout of-r and 239*240Pu are clearly reflected in a coral from the Florida Keys site (The Rocks). The coral activity curve for these isotopes reflects a 0.5 to 1.0 year lag relative to the atmospheric activity curve. The 90Sr activity reflects a more gradual reduction in the 1970s in coral relative to the atmosphere, due to longer residence time of Sr in the surface ocean than in the atmosphere.
The Florida coral record documents enrichment of 90Sr and depletion of 239~2*Pu compared to fallout predictions and previous coral records of these isotopes from central gyre localities. These two and three fold differences in isotope activities in the coral cannot be accounted for by changes in temperature and chemical composition of the seawater or differences in physical parameters of the coral such as growth or coral species. We conclude that the enrichment of %Sr in the Florida Keys coral record is likely the result of continental runoff which has extended exposure to the coral due to coastal circulation patterns in this region. The con~mmi~nt depletion of 239P2?u is due to removal from the water column by particles in coastal waters.
There appears to be evidence that the 90Sr and 23g,24~u activities in corals along coastal regions may be strongly influenced not only by geochemistry, but by circulation effects. Even two Florida Keys corals, although they had similar 90Sr levels in 1964, differed by 30% in 1974. The oceanic inputs of ?jr and 239*2% from fallout and runoff ceased by the early 1970s. Ocean and coastal circulation patterns became the dominant factor influencing these activities in corals. This suggests that site specific considerations must be addressed before coral %Gr records can be interpreted. These considerations might be viewed as limitations to the use of coral "sr records for large scale circulation information. However, coastal coral ?Sr records may become a very useful tool for interpreting local circulation patterns.