THE ORIGIN OF DOLOMITES IN TERTIARY SEDIMENTS FROM THE MARGIN OF GREAT BAHAMA BANK PETER K. SWART AND LESLIE A. MELIM

Transcription

1 THE ORIGIN OF DOLOMITES IN TERTIARY SEDIMENTS FROM THE MARGIN OF GREAT BAHAMA BANK 1 2 PETER K. SWART AND LESLIE A. MELIM 1-Marine Geology and Geophysics, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida Department of Geology, Western Illinois University, Macomb, IL Abstract Based on an integrated geochemical characterized by extremely high Sr and petrographic investigation of dolomites concentrations, which reflect high from two cores drilled on Great Bahama concentrations of Sr in the pore fluids. Bank, we have determined three different The high concentrations of Sr in the pore mechanisms of formation for the dolomites fluids arise through the continued which are common throughout the Pliocene recrystallization of meta-stable aragonite and and Miocene aged portions of these cores. high-mg calcite to dolomite and LMC driven The first mechanism of dolomitization occurs by the oxidation of organic material by in association with development of non- sulfate. Sulfate reduction not only provides depositional surfaces. Dolomite typically the thermodynamic drive for recrystallization, forms below each of these surfaces, the but as the absolute concentration of strontium concentration and extent of which is governed in the pore fluids is governed by the solubility by the length of the period of non-deposition. of celestite, allows the Sr /Ca ratio of the These dolomites are recognized by their interstitial fluid to become much higher than association with the non-depositional normally encountered. The final type of surfaces, characteristic heavy oxygen isotopes dolomite is a massive dolomite which occurs indicative of formation from cold bottom waters, and δ O and Sr profiles with depth which suggest formation in the presence of diffusive temperature and Sr gradients. The second mechanism of dolomitization, occurs in pore fluids where the cation and anion profiles are governed by diffusive processes and forms what we term background dolomite. This is a microsucrosic dolomite and forms both by the recrystallization of the existing sediment and precipitation directly into void space. Dolomitization by this mechanism uses a local source of Mg and consequently the dolomite never comprises more than between 5 and 10% of the sediment. This type of dolomite is c in coarse grained reefal sediments. The pervasive nature of the dolomitization and the relatively normal Sr concentrations, suggest the circulation of normal marine water in a relatively open system. INTRODUCTION It has long been known that a large proportion of the rocks in the subsurface of Bahamas are pervasively dolomitized. The presence of dolomite was established through a series of cores, up to several hundred meters in thickness, taken through Tertiary sediments (Beach and Ginsburg, 1980; Supko, 1977; Gidman, 1978; Pierson, 1982; Williams, 1985). Although numerous modes of 1

2 Swart and Melim Cay Sal Bank Clino Western Line Unda Figure 1: Site location map, showing the position of Clino and Unda near the western margin of Great Bahama Bank. BAHAMAS DRILLING PROJECT SUMMARY Drilling Operations Cores Clino and Unda were obtained using a diamond coring system mounted aboard a jackup barge. The two cores were located approximately 5 and.5 km respectively from the edge of Great Bahama Bank. They were drilled along a seismic profile composed of Western Geophysical lines GBB and 82-03x previously interpreted by Eberli and Ginsburg (1989). Core Clino was drilled m below the mud pit datum (7.3 m above sea level) and recovery averaged 80.8%. Core Unda was drilled m below mud pit (5.2 m above sea level). Recovery in Unda averaged 82.9% In this paper all depths are reported in meters as depths below the mud pit. Facies and Chronostratigraphy Clino, the more distal core, penetrated formation have been suggested for these inclined slope deposits overlain by a reef to dolomites including mixing-zone (Supko, platform sequence. The upper platform to 1977), normal seawater (Swart et al., 1987), reefal interval (197.4 to 21.6 m) consists of 7 reflux (Kaldi and Gidman, 1984), and Kohout sequences, each capped by sub-aerial exposure convection (Simms, 1984), the precise surfaces (Kievman, 1998). The reefal unit mechanism of formation remains uncertain. includes a deeper forereef facies that shallows to reef and eventually backreef facies. The This paper reports on the origin of remainder of the core is a 480 m thick dolomite in Tertiary sediments retrieved from sequence of slope sediments composed of finetwo cores, Clino and Unda, drilled near the sand to silt-sized skeletal and non-skeletal western margin of Great Bahama Bank (GBB) grains interrupted by intervals of coarse- (Fig. 1). These two cores were drilled as part grained skeletal sands. Three hardgrounds are of the Bahamas Drilling Project on a Western present ( , 367, and m.), each of Geophysical seismic line (Eberli et al., 1997) in which represents a break in deposition, the order to date the seismic sequences identified longest of which (2 to 3 Myrs) occurs at by Eberli and Ginsburg (1989) and to m. This surface represents the transition from investigate the nature of the carbonate the late Miocene to the early Pliocene. Based diagenesis in deeper water facies. on a combination of biostratigraphy (Lidz and McNeill, 1995a, 1995b), magnetostratigraphy (McNeill et al., in press) and strontium isotope 2

3 Dolomitization in Great Bahama Bank Figure 2: Summary of the sedimentology, chronostratigraphy, mineralogy, and isotopic composition for Clino and Unda. Data are from Eberli et al. (1997), Kenter et al. (In press), Lidz et al. (1995a, b), Melim et al. (1995; In Press). stratigraphy (Swart et al., in press) the Plio- exposure surfaces overlying a reefal unit based Pleistocene boundary can be placed at on a marine firmground (Kievman, 1998). approximately 110 m (Fig. 2). The middle shallow-water unit (354.7 to Core Unda, the more proximal of the m) is a somewhat deeper water reef two, consists of three successions of shallow- with platy corals and rhodoliths (Budd and water platform sands and reefal deposits, that Kievman, in press) overlain by a sub-aerial alternate with sand and silt-sized deeper exposure surface that also is a phosphatic marginal deposits. The Plio-Pleistocene marine hardground (Melim et al., in press). shallow-water interval (60 to 8.6 m) has 14 The deepest shallow-water unit (454.0 to platform sequences capped by sub-aerial m) consists of shoaling-upward 3

4 Swart and Melim packages of coarse-grained skeletal to non- level caused development of a subaerial skeletal grainstones to rudstones (Kenter et al., exposure surface in Unda (at m) and in press). The two deeper marginal deposits continued hardground development in Clino. sandwiched between the shallow-water units During the early Pliocene a major sea-level rise are fine-sand to silt-sized grainstones to forced eastward backstepping of the shallowpackstones that alternate with coarse-sand water platform and renewed deeper water intervals. Hardgrounds occur at , facies in Unda and formed a condensed , and m and a firmground tops interval in Clino (sequence f). The Unda the deeper water facies at m. The Plio- subaerial exposure surface ( m) was Pleistocene boundary can be placed at overprinted by marine-hardground diagenesis approximately 200 m and the Mio-Pliocene during the transgression. Before progradation boundary at m (Fig. 2). could bring highstand deposits to the margin locations of Clino and Unda, another sea level rise further backstepped the platform renewing Sequence stratigraphy Facies successions document several transgressive deeper margin and slope facies. The subsequent highstand (sequence e) hierarchies of changes in relative sea level in resulted in major progradation of the western cores Clino and Unda (Eberli et al., in press). margin of GBB. The late Pliocene began with These changes resulted in pulses of a relative sea-level fall then rapid rise, resulting progradation of the western margin of Great in a hardground (later partly eroded) in Clino Bahama Bank that are seen on seismic lines as (at 367 m) and a firmground in Unda (at seismic sequences (Eberli et al., in press) and m). In Unda, the following sequence (d) is a in the cores as depositional sequences (Kenter reef, while in Clino a thick package of et al., in press; Kievman and Ginsburg, in proximal slope facies documents rapid press). The sequence boundaries are indicated progradation of the margin. An early by discontinuity horizons (subaerial exposure Pleistocene lowstand resulted in a lowstand on the platform, marine hardgrounds and reef in Clino and platform top facies and firmgrounds on the slope), changes in facies subaerial exposure in Unda (sequence c). and changes in diagenesis (Melim et al., in Sequences b and a were deposited during the press). high frequency, high amplitude sea level Eight seismic sequences (a i) record changes of the Pleistocene. The margin of the relative sea level changes of the middle GBB was located to the west of the cores by Miocene to Recent (Eberli et al., 1997) (Fig. this time resulting in only highstand platform 2). Platform facies of possible middle Miocene facies and numerous subaerial exposure age (sequence i) were deposited during a surfaces in both cores. relative lowstand. The following late Miocene highstand (sequence h) deposited a thick package of deeper margin facies in Unda and METHODS Samples were taken at 1.5 m intervals deeper slope facies in Clino. Sequence g throughout the two cores for X-ray diffraction deposited a late Miocene lowstand reef in (XRD) and stable carbon and oxygen isotopic Unda while a marine hardground was forming analysis. All materials were ground to finer in Clino (at m). A further drop in sea than 63 µm. For XRD analysis, powdered 4

5 Dolomitization in Great Bahama Bank Figure 3: Comparison of 104 peaks for samples with a single dolomite versus samples with two dolomite peaks. samples were smear mounted on glass slides. Peak areas for aragonite, calcite and dolomite were determined using a Scintag XDS-2000 diffraction unit. Mineral concentrations were calculated from peak area ratios assuming that each sample was composed only of calcite, aragonite, and dolomite (the only other minerals present are clays (<<5%) and minor celestite). Peak area ratios were calibrated and the concentrations calculated using calibration curves prepared from results using a series of pure mineral standards (verified by XRD analysis). Duplicate analyses indicate reproducibility of ±3%. In order to isolate the dolomite, sieved samples (> 63 µm) were treated with buffered acetic acid for a period of 2 hours. This procedure selectively leaches the less stable minerals leaving the dolomite behind. This short leaching period was chosen to allow progressive removal of the CaCO 3 components. After each leaching episode, the samples were re-analyzed by XRD and an aliquot was preserved for the determination of δ C and δ O. This process continued until only dolomite remained. The number of leaching episodes varied from 1 to 4, largely controlled by the amount of dolomite initially in the samples (more initial dolomite gave pure dolomite faster). The stoichiometry of the dolomite was determined by XRD analysis of the separates with calcium fluoride as an internal standard. o A step scan was run from 24 to with a o step size of 0.01, count time of 2 seconds per step, source slits of 2 and 4, and receiving slits of 0.2 and 0.1. Scintag XRD software DMSNT version 1.1b was used to identify peaks and determine peak area, but peak position corrections using the calcium fluoride peaks were determined manually. The profile fitting subroutine of the Scintag software fits a Pearson VII profile to a net intensity file with background subtracted, but without the K-alpha-2 peaks subtracted. Eight samples of Paleozoic dolomite were run using the same operating conditions to determine the peak shape for a single stoichiometric dolomite. This provided a measure of the FWHM (the Full Width of the peak at Half of the Maximum intensity) to use during peak fitting of multiple dolomite peaks. Figure 3 shows examples of single, double, and triple dolomite peaks with the Pearson VII profile fits. Mole percent MgCO 3 in the dolomite was calculated using the corrected peak positions and the formula N = d where N is the mole percent Ca and d is the observed d-spacing for the [104] dolomite peak (Lumsden and Chimahusky, 1980). For δ C and δ O analyses, all 5

6 Swart and Melim Figure 4: X-ray mineralogy on samples at approximately 0.3 m interval from Clino and Unda. Also shown are stable isotopic data for the bulk rock (lines; Melim et al., 1995) and the dolomite separates(symbols, this paper), together with the sedimentology (Kenter et al., in press), and the chronostratigraphy (McNeill et al., in press). Mineralogy: black = aragonite, white = low-mg calcite(lmc), grey = dolomite. samples were dissolved using the common acid transmitted light petrography was o bath method at 90 C and the CO 2 produced supplemented by scanning electron microscopy analyzed using a Finnigan-MAT 251. and cathodoluminescence. Dolomite was Standard isobaric corrections were applied, but no correction has been applied for the differences in the fractionation of oxygen as a result of the dissolution of dolomite and calcite by phosphoric acid (Land, 1980; Vahrenkamp and Swart, 1994). Data are reported relative to V-PDB using the conventional notation. Thin sections were prepared at approximately 3 meter intervals with closer sampling across selected intervals. Standard identified by staining slabs and/or thin sections using Alizarin red-s (after Dickson, 1965) or Titan Yellow (after Miller, 1988). All petrographic descriptions were entered in a computer database to allow rapid retrieval and comparison between different sections in the cores. The strontium concentration of the dolomite separates was determined using atomic absorption (Perkin-Elmer 4500). In 6

7 Dolomitization in Great Bahama Bank % Dolomite this method approximately 100 mg of dolomite separate was dissolved in 10% nitric acid solution, filtered, and the filtrate diluted to 25 3 cm. Corrections were made for the percentage of insoluble residue. Standards were made using specpure CaCO 3 and MgCO3 (Johnson-Matthey) weighed out in approximately the same concentrations as contained in the samples. Standards were then spiked with 1000 ppm Sr standard solution to provide standards with similar intensities to the analyzed samples. Reproducibility of this method is approximately +/-5% % Aragonite Figure 5: Plot of the relationship between the concentration of aragonite and dolomite in Clino and Unda. RESULTS RESULTS X-ray Diffraction Mineralogy Unda bulk mineralogy The mineralogy of core Unda consists primarily of aragonite, LMC, and dolomite Percentage Dolomite Mol % MgCO3 Figure 6: Cross plot of the relationship between the percentage of dolomite and stoichiometry from Clino and Unda. (Fig 4). Minor amounts of dolomite are ubiquitous below 108 m and increase in abundance beneath hardgrounds and firmgrounds (sensu Ekdale et al., 1984) at m, m and m. In addition the sediments comprising the Miocene platform and overlying slope facies are 100% dolomitized ( m). Aragonite is also a common minor component below 108 m. With the exception of the interval beneath the firmground at m, aragonite is <5% (and usually absent) if the dolomite content exceeds 25% (Figs. 4 and 5). Trace amounts of celestite occur in the deeper water sediments. Clino bulk mineralogy Core Clino is composed principally of aragonite, LMC, and dolomite (Fig. 4). Small concentrations of celestite are present below 150 m. Minor dolomite is present everywhere below 150 m and increases in abundance beneath hardgrounds at m, m, m and m. A firmground at 7

8 Swart and Melim m does not contain dolomite. In of different compositions. Peak fit was addition, the increase in dolomite below the surface at m begins 1 m below the surface. In an interval with very little dolomite, a single skeletal grainstone bed at m is over 85% dolomite. Aragonite is much more common in Clino than in Unda with >40% aragonite common between 230 m and 365 m. As in Unda, intervals with >25% dolomite generally have <5% aragonite with the exception of the interval beneath the hardground at m (Figs. 4 and 5). In addition, intervals with >20% aragonite seldom contain more than 5% dolomite (Figs. 4 and 5). However, the concentrations of calculated using a Pearson VII profile without a K-alpha 2 correction to verify the secondary peaks were not artifacts of the correction method (hence the "double" look to the peaks in Figure 3). Samples with a single peak (Fig. 3A; N = 50) show smooth sides. Samples with two dolomites (N = 127) display a shoulder on either the left (Fig. 3B) or right side (Fig. 3C) of the 104 peak. Three samples show a more complex 104 peak that was best described by three separate peaks. In order for two peaks to be resolved, the distance between them must be at least theta, which translates into.1 mole % difference in aragonite and dolomite are not inversely related, rather they are mutually exclusive at higher concentrations. Of the 1231 samples analyzed, <4% contain both >10% aragonite and >10% dolomite, and most of these occur beneath two hardgrounds (Clino m and Unda m; open symbols Fig. 5). Dolomite Stoichiometry Based on the position of 104 peak, all of the dolomite is calcian-rich with values ranging from between 41.8 to 45.8 mole % Mg (Fig. 6). The stoichiometry of the dolomites increase with increasing dolomite the Mg composition. Because of this 2 content (Fig. 6, r = 0.31, statistically significant at the 99% confidence interval). At any given dolomite content, the stoichiometry values have a range of 1 2 mole % Mg (only 3 samples exceed this range), which is 2 reflected in the low r value. Although there are no overall trends of increasing or decreasing dolomite stoichiometry with depth, within the 100% dolomitized interval in Unda, there is tendency for the mole% MgCO to 3 increase with depth. Most of the dolomite samples show a broad 104 peak (Fig. 3) that can be resolved into 2 or more peaks representing dolomites 8 limitation on individual peak resolution, the multiple peaks should be seen as indicating a range of dolomite compositions for each sample rather than two distinct end-member compositions. The average difference between the two calculated 104 peaks is 2.1 mole % Mg (standard deviation, 0.5 mole % Mg; range, 1.0 to 3.7 mole % Mg). The two dolomite peaks are usually not of equal dimensions (Fig.3) and the position of the main peak shifts toward higher stoichiometry with increasing dolomite content. In addition, the position of the secondary peak varies with dolomite content. This is best seen by comparing those samples with <20% dolomite (N = 49) with those sample having >80% dolomite (N = 21). Of the 49 samples with <20% dolomite, 41 have a large peak between mole % Mg with a secondary peak between mole % Mg (Fig. 3C). In contrast, 17 out of 21 samples with >80% dolomite have a large peak around mole % Mg with a secondary peak either at mole % Mg or at mole % Mg (Fig. 3B).

9 Dolomitization in Great Bahama Bank Petrography At least trace amounts of dolomite are present throughout most of both cores except Figure 7: Photomicrograph of completely dolomitized grainstone with fabric-preserving dolomite. Note the well preserved Halimeda and red algae grains. Sample Unda m. Field of view is 3 mm. in the upper m (Fig. 4). Two principal textural types (fabric-preserving and microsucrosic) have been identified, but intermediate fabrics are also present. It is unlikely that all the dolomite formed at the same time and some of the variation represents individual stages in a process that begins with nucleation of dolomite and ends with 100% dolomite. The following discussion, therefore, focuses on describing each fabric and its distribution, leaving interpretation of dolomite timing to a later section. Fabric-preserving dolomite: Fabricpreserving dolomite is found in scattered locations in Clino and in the middle reef to platform section ( m) of Unda. This form of dolomite occurs exclusively in blocky spar-cemented skeletal grainstone to packstone lithologies (Fig. 7). Samples with fabric-preserving dolomite are always >80% dolomite, even when found in intervals with minor dolomite (e.g. Clino m). Carbonate needle mud has recrystallized to micrite or fine microspar but this could have occurred prior to, rather than during, dolomitization. Grains with originally very fine fabric (e.g., red algae) also have recrystallized to a coarser fabric (Fig. 7). Some dolomitized Halimeda grains retain some of the brown pleochroism typical of neomorphic Halimeda elsewhere in these cores (Melim et al., in press). The pleochroism in the dolomite is not as dark as in the calcitic neomorphic spar, but it is distinct from the clear dolomitized blocky spar cement infilling primary pores in the Halimeda. In addition to the dolomitized blocky spar, most of the fabric-preserved rocks also contain euhedral rhombs of dolomite spar partially filling primary and secondary pores. The outer rims of this dolomite spar occasionally shows dull luminescence, while other dolomite is nonluminescent. A variety of the fabric-preserving Figure 8: Photomicrograph of completely dolomitized hardground with micritic dolomite. Light colored oval is a boring coated with iron oxides and phosphate. Sample Unda m. Field of view is 3 mm. dolomite is a micritic type which occurs both as a replacement of micritic grains and 9

10 Swart and Melim apparently as a primary cement associated with dolomite is non-luminescent under marine hardgrounds (Fig. 8). The most cathodoluminescence. common type of grains replaced are red algae, In Clino, microsucrosic dolomite but trace amounts of dolomitized micrite rims forms up to 50% of the lower slope facies and micritized skeletal grains are also present in the deeper water facies. Micritic dolomite o δ O /oo cement is apparently the primary lithification element in several hardgrounds. Micritic 500 dolomite formed during hardground formation as it is sometimes directly overlies phosphate crusts. This dolomite is nonluminescent under cathodoluminescence Microsucrosic dolomite: The most pervasive type of dolomite corresponds to the microsucrosic variety as described by Dawans and Swart (1988). This dolomite consists of small euhedral rhombs (1 to 40 Fm in size). It 600 is similar to dolomites commonly found in deep sea cores (Swart and Guzikowski, 1988; Dix and Mullins, 1988). In the low permeability, aragonite-rich interval in Clino around 300 m, dolomite is <5% of the sediment and occurs only as very fine (<1µm) crystals. More commonly, 10-20% dolomite is present (Fig. 4) and has a variety of dolomite crystal sizes (1-30 µm). This Depth (m) Carbonate Dolomite o 36 C/1000m Figure 10: Oxygen isotopic composition of dolomites and co-existing calcites below the m hardground in Clino. Note the steady increase in the oxygen isotopic composition towards the non-depositional surface. Figure 9: Photomicrograph of completely dolomitized slope sediment with microsucrosic dolomite. Large pores are probably molds of bioclasts. Sample Unda m. Field of view is 1.3 mm. between 536 and 600 m. In Unda, % microsucrosic dolomite occurs in the upper slope facies (at m) and throughout the middle reef to platform section ( m). Samples with fabrics intermediate between fabric-preserving and microsucrosic are also present in the middle reef to platform section, usually in packstones. The microsucrosic dolomite forms as 10

11 Dolomitization in Great Bahama Bank Dolomite o O /oo δ O δ o /oo Coexisiting Carbonate Figure 11: Relationship between the oxygen isotopic composition of the precursor and the δ O of the dolomite. There are two trends in the data. Above the m hardground there appears to be no trend, but below (circled data points) there is a positive correlation suggesting that the precursor was already diagenetically altered along the geothermal gradient prior to dolomitization. both a primary void filling cement and by replacing fine micritic sediments, red algae and echinoderm grains. Using partially dolomitized samples, the following sequence has been identified. The first dolomite forms as very fine (1-10µm) euhedral rhombs within the matrix of packstones to wackestones. This early stage also includes replacement of HMC grains such as echinoderms (usually by a single dolomite crystal) and red algae (as micritic dolomite). Aragonitic skeletal grains are dissolved to produce molds before 10% dolomite forms; aragonitic peloids last longer but are dissolved by the time dolomite reaches about 20%. As the percentage of dolomite increases, the size of the rhombs becomes larger and they impinge on surrounding micritic grains. At a composition of 50-70% dolomite, dolomite rhombs 1 to 50µm in size occupy nearly all of the matrix and grow into pores. As dolomitization approaches 100%, most remaining LMC skeletal grains (mainly Foraminifera and molluscs) are dissolved leaving an open network of 10 to 50µm, subhedral to euhedral rhombic crystals (Fig. 9). This dissolution appears to coincide with the final stage of dolomitization as partially dissolved LMC grains are common when dolomite content is 90 to 95%. Classic Dolomite o /oo C δ C δ o /oo Coexisiting Carbonate Figure 12: Relationship between the carbon isotopic composition of the precursor and the carbon isotopic composition of the dolomite. The intercept of approximately +1 corresponds to the expected equilibrium difference in the δ C between dolomite and calcite (Sheppard and Schwarcz 1970). sucrosic dolomite is not found as the dolomite retains a range of crystal sizes rather than the uniform crystal size of true sucrosic dolomite. This dolomite is nonluminescent under cathodoluminescence. Stable Oxygen and Carbon Isotopes Oxygen: The δ O of the bulk sediments varies between -6.8 and +5.2 (Fig. 4). The lowest δ O values occur in the upper section of both cores which have been shown to be diagenetically altered by meteoric waters 11

12 Swart and Melim 0 A B Depth (m) Strontium (ppm) Figure : Concentration of strontium in the dolomites from Clino and Unda. Note that near the non-depositional surfaces at m in Clino and at and m in Unda, the concentration of strontium approaches that which we would expect in dolomites formed from normal seawater. The massively dolomitized interval in Unda ( m) also has values indicative of normal seawater. The eroded hardground at 367 m in Clino has higher values as does the hardground at 263 m Clino. (Melim et al., 1995). Zones rich in dolomite have more positive δ O values, corresponding to the fact that the dolomites from Clino and Unda are approximately +3 enriched in δ O relative to the limestone (Fig. 10). This corresponds to the known oxygen isotopic difference between dolomite and calcite (Land, 1980). The mean δ O values of dolomite separates in Clino and Unda are (+/ ) and

13 % Dolomite Strontium (ppm) Fluid Sr/Ca Strontium Dolomitization in Great Bahama Bank 1000 statistically significant difference between the δ O values of the microsucrosic and 500 fabric-preserving dolomites. Stoichiometry Carbon: The δ C of the bulk carbonate lies 0 between and +3.8 (Fig. 4). The mean δ C of the dolomite from Clino is Dolomite Mol% MgCO (+/-0.49 ) and Unda Figure 14: a) Relationship between Sr (+/-0.61 ). As in the case of the δ O (determined by AA) and MgCO 3. The lower line values, the lowest δ C values are associated represents the relationship between Sr and MgCO identified by Vahrenkamp (1988) and 3 Vahrenkamp and Swart (1990). The vertical extent of the data represent changes in the Sr/Ca ratio of the pore fluids. Figure 14b) The relationship between the percentage of dolomite in the sediment and the Sr concentration of the dolomite. (+/-0.62 ) respectively. The highest δ O values of dolomites are those associated with non-depositional surfaces. Above the m hardground in Clino there is no relationship between the δ O of the dolomite and the co-existing dolomite (Fig. 11). However, below m there is a positive relationship between the limestone and the dolomite. There appears to be no with the upper portion of the core affected by meteoric diagenesis (Fig. 10). Dolomites at the non-depositional surfaces tend to have lower δ C values relative to sediments above and below the surface. The δ C of dolomite is positively correlated with the δ C of the co-mingled carbonate with an intercept of +1, approximately equivalent to the estimated equilibrium difference between calcite and dolomite (Sheppard and Schwarcz, 1970) (Fig. 12). Strontium The strontium concentration of the dolomites ranges from 70 to over 2000 ppm (Fig. ) with the concentration of strontium being inversely related to both the stoichiometry and concentration of dolomite in the rock (Fig. 14a and b). The lowest concentrations of Sr occur in dolomites near non-depositional surfaces or associated with the massively dolomitized interval in Unda. The eroded hardgrounds at 367 m in Clino and hardground at 263 m in Unda have slighter elevated values compared to the non-eroded

14 Non-depositional surface Depth (mbsf) Celestite Solubility Swart and Melim Strontium (um) Figure 15: Changes in the estimated strontium concentration of the pore fluids below the hardground at 586.3m. Concentrations are based on an estimated increase of 5 mm in the concentration of Ca over the same interval. surfaces. Trending away from nondepositional surfaces there is a tendency for the concentration of strontium to increase with depth (Fig. ). DISCUSSION Although the interpretation of the geochemistry of dolomite is still equivocal (Land, 1980 and others), the concentration of strontium, δ C, and δ O of the dolomites from Clino and Unda can be used to place constraints on the mechanism, source of magnesium, and location of dolomite formation in both Clino and Unda. Strontium: The observed variation in the concentration of Sr measured in these dolomites ranges between 70 and 2378 ppm. High concentrations of Sr are particularly unusual in dolomites (Guzikowski, 1987) and the inverse relationship between the Sr concentration in the dolomite and the original dolomite concentration of the sediment might suggest the presence of a residual contamination of aragonite or HMC derived from the accompanying sediments. However, contamination can be discounted for several reasons. First, the nature of the treatment to which these samples were subjected (See methods section) meant that the sieved crushed samples were leached with acetic acid for continually until X-ray diffraction showed them to consist entirely of dolomite. Second, there was no correlation between the amount of dolomite and aragonite (Fig. 5). In fact most of the samples with dolomite contained no aragonite, suggesting that either these samples never contained aragonite, or more likely that the aragonite was largely dissolved before dolomitization occurred. The concentration of strontium in dolomites formed from normal marine waters lies between 70 to 250 ppm and has been suggested to be related to the calcian nature of the dolomite, (Vahrenkamp and Swart, 1990; Malone et al, 1996). Vahrenkamp and Swart (1990) suggested that the distribution coefficient (D ) for the incorporation of Sr Sr into dolomite varied with the MgCO content 3 of the dolomite according to equation 1. 14

15 Dolomitization in Great Bahama Bank D Sr ' Sr s %20(x Ca dolomite Sr Ca fluid (1) In this equation, Sr s = the strontium concentration of dolomite with an ideal 50:50 stoichiometry and x= number of moles of excess CaCO. Using a seawater Sr/Ca ratio 3-3 of 8.67 x 10 then a value for D Sr of can be calculated assuming a Sr s of 70 ppm (Vahrenkamp and Swart, 1990). Applying this equation to the Sr concentration of dolomites measured in this study we are able to estimated the Sr/Ca ratios of the pore fluids from which the dolomites formed and hence constrain the environment of dolomitization. The massive dolomites in Unda have a mean Sr concentration of 230 ppm. Using equation 1 and the mean stoichiometry of dolomites in this interval an equilibrium Sr concentration of 190 ppm would be expected in these dolomites assuming formation from normal seawater. Although it is possible that the fluids which dolomitized Unda had slightly elevated Sr concentrations, we consider that these data suggest that Unda was dolomitized by seawater with near normal composition. The hard ground dolomites in both Clino and Unda typically show lower concentrations of Sr near the hardground surface, but increasing Sr concentrations with depth (Fig. ). Using the hardground at m in Clino as an example, the Sr concentration increases from between 200 and 250 ppm in the 30 meters below the hardground to over 1500 ppm at a depth of 650 m. The low concentrations of Sr near the hardground surface are consistent with formation from fluids with marine Sr /Ca ratios, while with increasing depth the higher Sr concentrations suggest formation from fluids with increasing Sr /Ca ratios. An estimate of the Sr concentration in the fluids below the m hardground are shown in figure 15. In this calculation we have used a D Sr based on equation 1 (Vahrenkamp and Swart, 1990) and assumed an increase in the concentration of Ca over the thickness of the sequence of 5 mm. The increase of Ca is only an approximation and it is possible that the change may be greater than 5 mm as increases of over 10 mm were noted in porewater retrieved from these sites (Swart et al. In Press). The nature of the estimated profile is very similar to the types of Sr profiles observed in deep-sea and peri-platform sediments (Baker et al., 1982; Baker et al., 1982; Swart and Guzikowski, 1988; Swart and Burns, 1990; and others ) and arise from the recrystallization of meta-stable forms of calcium carbonate such as high-mg calcite and aragonite which contain relatively high Sr concentrations compared to low-mg calcite and dolomite. The limit of the maximum amount of Sr in the pore-fluids has been shown to be dictated by the solubility product of celestite (SrSO 4) (Baker and Bloomer, 15

16 Swart and Melim Strontium (ppm) Sr/Ca (x 1000) Mol% MgCO 3 Figure 16: Three dimensional plot showing the relationship between mole% MgCO 3, fluid Sr /Ca ratio, and dolomite Sr concentration. 1988). This relationship can be readily seen by the pore waters at this location. In the the presence of celestite in many cores and in absence of sulfate reduction there is an upper 2- the ion molar product of Sr and SO 4 in many limit on the estimated Sr concentration of ODP sites (Swart and Burns, 1990). Celestite approximately 600 µm. In the absence of was detected in both Clino and Unda and therefore it is likely that similar relationships changes in the concentration of Ca, such as concentration of Sr in the pore fluids could also control the concentration of strontium in

17 Dolomitization in Great Bahama Bank form a dolomite with approximately 1500 ppm downward and the local-dissolution of HMC Sr. There are four non-depositional surfaces in Clino ( m, m, m and 544 m) and three Unda ( m, m and m) where the concentration of dolomite is at a maximum at or slightly below the surfaces and tends to decrease downward away from the surface (Fig.4). For the three surfaces associated with the highest concentration of dolomite (Clino 367 and m and Unda m), the Sr concentration of the dolomites below the surfaces increases downward away from the non-depositional surfaces (Fig. ). The best developed of these trends is at m in Clino which represents the longest time of non-deposition (2-3 My). We suggest that the captures this gradient, thereby constraining the timing of dolomite formation. The slightly higher Sr concentrations found in dolomite from eroded non-depositional surfaces are consistent with this hypothesis as erosion would have removed the dolomites formed near the interface between the seawater and the underlying sediment with seawater Sr concentrations. Dolomites which are situated well away from the non-depositional surfaces in Clino and Unda, have the highest Sr concentrations, sometimes in excess of 2000 ppm. Based on the previous discussions these dolomites must have formed from porewaters with elevated Sr /Ca ratios and in which there was a significant depletion in sulfate increase in the Sr concentration of the and/or an enrichment in Ca. Using the dolomite with depth represents formation of dolomite along a gradient in which Sr is diffusing upward out of the sediments and Mg is diffusing downwards from the overlying seawater. Several studies have highest concentration of Sr measured in the dolomites from Clino (2397 ppm), a seawater Ca concentration of 15 mm, then the estimated concentration of Sr in the pore fluids at the time of formation in the interval shown that concentrations of dolomite similar where the dolomite was found was to those measured in this study can be formed approximately 1600 µm. A model of the by Mg diffusing in to the sediments from overlying seawater (Baker and Burns, 1986; Compton and Siever, 1986). Increases in Sr, such as seen in the dolomites closely resemble relationship between stoichiometry, fluid Sr /Ca ratio and the Sr concentration in dolomites is shown in figure 16. In order for the pore fluids to have such a high Sr the Sr profiles seen in pore water from deep sea and peri-platform sediments (Baker et al., 1982; Swart and Guzikowski, 1988; Swart and Burns, 1990; Swart et al., 1994 and others). Deep within the sediments, Sr is being added to the pore waters through recrystallization of aragonite and HMC and precipitation of LMC. The Sr subsequently diffuses upwards towards the relatively low concentration of Sr in the overlying seawater. Dolomite forming along this gradient with Mg being supplied both by the diffusion of Mg concentration, the concentration of SO 4 in the pore fluids needs to be below 10 mm compared to normal seawater concentrations of 28 mm. These data suggest that the pore fluids, from which these dolomites formed, have experienced significant amounts of sulfate reduction. Notwithstanding any potential influence that the removal of sulfate may have on the kinetics of the dolomitization reaction (Baker and Kastner, 1981), the fact that sulfate reduction is taking place provides an

18 additional mechanism whereby pore waters undersaturated with respect to HMC and aragonite can be produced, causing dissolution and precipitation of LMC and dolomite. As there are two calcium carbonate minerals (aragonite and LMC) present with different solubilities, once the pore waters become undersaturated with respect to aragonite, continued sulfate reduction by the oxidation of organic material is not necessary to drive the system as the dissolution and precipitation is controlled by the solubility difference between the two minerals. Oxygen Isotopes: The δ O of carbonates normally responds to the temperature of formation and the δ O of the water. The Swart and Melim as the dolomite, that is it is isotopically positive near the hard ground and then decreases with increasing depth. This covariance of the dolomite and the precursor suggests that the sediment was altered to LMC prior to dolomitization and supports the conclusion that aragonite alteration was largely complete prior to dolomitization (Fig. 5). δ O of the dolomites confirm the methanogenesis (Irwin et al., 1977). The δ C interpretation of dolomite formation beneath of the sediments and dolomites from Clino hardgrounds based on the strontium and Unda show lower δ C values associated concentrations discussed previously. Dolomites near the non-depositional surfaces are all enriched in O suggesting formation from relatively cold bottom waters (Fig. 17). With increasing depth from the nondepositional surface, the δ O of the dolomites becomes isotopically more negative reflecting the normal increase in temperature with depth. Eventually at some distance beneath the non- depositional surface the δ O of the dolomites approaches the δ O found in the dolomites which contain high concentrations of Sr. In fact the approximate magnitude of the geothermal gradient can be estimated by using the change in δ O with increasing depth (Fig. 17). This calculation estimates an increase in temperature of approximately 5 C over a depth of 100m, equivalent to a typical geothermal gradient over continental crust. It is interesting to note that the calculated δ O of the sediment without the dolomite, follows the approximate same trend Carbon Isotopes: The δ C of diagenetic carbonates principally changes in relationship to the amount of organic carbon being oxidized. Lower δ C values therefore are usually interpreted as reflecting the input of oxidized organic carbon, while higher δ C values might reflect CO 2 associated with with hardground surfaces (Fig. 16). These lower values do not extend very far below the hardground surface and are probably caused by the oxidation of organic material near the surface of the hardground. The occurrence of a depletion in the δ C at hardground surfaces is contrary to the current dogma which suggests that only subaerial exposure surfaces and not hardgrounds exhibit depletions in the δ C (Allan and Matthews, 1982). A more pronounced change in carbon isotopic composition occurs above the hardgrounds. This change typically manifests itself as an enrichment (Fig. 17) and is probably a result of a change in the carbon isotopic composition of the sediment as sea level rises and carbonate sediment production on the adjacent carbonate platform is turned on bringing material which is higher in δ C compared to the pelagic material which

19 Dolomitization in Great Bahama Bank dominated the sediment below the non-depositional surface (Shinn et al., 1989). Stoichiometry: Most dolomites isolated from Clino and Unda contain dolomites with several different Mg/Ca ratios (Fig. 3). The exception to this are the dolomites found immediately below the hardgrounds at 367 and m in Clino and m in Unda which are more uniform in their composition. The dolomite in the reefal section, although containing dolomites with more than one composition, contain slightly more Mg than those in Clino. Although, it has been suggested that calcian dolomites form in association with lower salinity fluids (Lumsden and Chimahusky, 1980), the origin of the Depth (m) differences in dolomite stoichiometry is still a matter of speculation (Morrow, 1982). It is generally believed that when first formed, most dolomites are calcian in composition and approach an ideal composition with increasing age and depth. In the dolomites investigated in this study, there is no trend with increasing age or burial. However, the trend between the stoichiometry and the percentage of dolomite (Fig. 6) suggests that early formed dolomite O C A Figure 17: Changes in the carbon and oxygen isotopic composition of the bulk sediment relative to the hardground surfaces in Clino and Unda. Note the depletions in carbon isotopic composition close to the non-depositional surfaces B are calcian in composition and that as more dolomite forms, the bulk composition becomes more Mg rich. As there is also an increase in crystal size as the amount of dolomite increases, there may be an element of Oswald rippening similar to that documented by Gregg et al. (1992). Although near-surface samples in the study of Gregg et al (1992) did not show a change in 19

20 Swart and Melim stoichiometry with crystal size, these workers were looking at a change from 0.4 Fm to 1Fm, while in this study there is a change from <1Fm to.50fm. Recrystallization during dolomitization is also supported by the shift in the dominant dolomite peak toward more stoichiometric values in samples with greater amounts of dolomite (Fig. 3). The dolomites associated with the hardgrounds tend to have more uniform compositions compared with the rest of the dolomite and also tend to have a more uniform crystal size (although not necessarily a larger crystal size) perhaps indicating less recrystallization. SUMMARY The dolomites which are found in Clino and Unda formed by three different mechanisms which we will term (i) hardground dolomitization, (ii) background dolomitization, and (iii) massive dolomitization. Hardground Dolomites These dolomites formed in response to the presence of a non-depositional surface. The time represented by the period of nondeposition allows Mg from the overlying seawater to diffuse into the sediments and therefore the concentration of dolomite is greatest nearest the non-depositional surface and decreases downward. The dolomite closest to the surface has the heaviest oxygen isotopic composition, reflecting formation at low bottom water temperatures, and the lowest concentration of strontium, indicating fluids with normal seawater Sr /Ca ratios. The longer the period of non-deposition the greater the concentration and thickness of the dolomite rich zone. A mature hardground would therefore contain a thick zone of dolomite, with a concentration of dolomite decreasing downward and one in which the Sr and O concentration of the dolomite mimics that in the steady state pore water profile which developed over this time period. Background Dolomites The background dolomites comprise less than 10% of the sediment and possess very high Sr concentrations, typically in excess of 1000 ppm. In such locations the Mg necessary for dolomite formation is supplied by that present in the pore fluids and by local diffusion. The high Sr content of the dolomite identifies the region of formation as being an area characterized by high a Sr /Ca ratio 2- in the pore fluids and depletion in the SO 4 concentration of the pore fluids, probably as a result of the oxidation of organic material. Massive Dolomites The massive dolomites found in the middle reefal and overlying deeper margin sections of Unda clearly formed by a different mechanism and from a different fluid than the dolomites found in the deeper water facies of Clino and Unda. The two principal clues in constraining their formation are the pervasive dolomitization, the sediments are 100% dolomitized in this interval compared to Clino, and the relatively low Sr concentrations compared to the hardground and background dolomites. The low Sr concentrations may be explained by the fact that most of the dolomite followed extensive diagenesis of aragonite to LMC in an open system that actually removed substantial carbonate forming secondary porosity (Melim et al., 1995). Any elevated Sr concentrations formed during this earlier aragonite diagenesis had apparently been flushed prior to 20

21 Dolomitization in Great Bahama Bank dolomitization. Clearly, a mechanism had to p exist to allow circulation of large quantities of seawater with normal Sr /Ca ratios to supply the needed Mg for the extensive dolomitization. Since this massive dolomite extends up to.260 m, the underlying reef (354.7 to 292.8) was dolomitized at a minimum of 50 to 100 m burial depth. Our data do not allow distinguishing between the various models for circulating seawater in carbonate platforms. ACKNOWLEDGEMENTS The authors would like to thank Dr. R.N. Ginsburg whose ideas provided the inspiration for many of the ideas developed in this project. Drilling of the BDP holes was supported by NSF grants OCE and We are also indebted to G. Eberli for discussion and friendship in this project. This project was supported by a grant from DOE grant DE-FG05-92ER14253 to G. Eberli and P.K. Swart and the Industrial Associates of the Comparative Sedimentology Laboratory. REFERENCES Allan, J.R. and Matthews, R.K Isotope signatures associated with early meteoric diagenesis. Sedimentology, v. 29, p Baker, P., and Kastner, M., 1981, Constraints on the formation of sedimentary dolomite: Science, v. 2, p Baker, P.A., Gieskes, J.M., and Elderfield, H., 1982, Diagenesis of carbonates in deepsea sediments; evidence from Sr/Ca ratios and interstitial dissolved Sr data: Journal of Sedimentary Petrology, v. 52, 21 Baker, P.A. and Burns, S.A Occurrence and formation of dolomite in organic-rich continental margin sediments, Bull. Am. Assoc. Petrol. Geol., v. 69, p Baker, P.A. and Bloomer, 1988 The origin of celestite in deep-sea carbonates, Geochim. Cosmochim. Acta, v. 52, p Beach, D. K., and Ginsburg, R. N., 1980, Facies succession, Plio-Pleistocene carbonates, Northwestern Great Bahama Bank. Amer. Ass. Petr. Geol. Bull., v. 64, p Beach, D.K., 1993, Submarine cementation of subsurface Pliocene carbonates from the interior of Great Bahama Bank. J. Sed. Pet., v. 63, p Budd, A.F. and Kievman, C.M., In Press, Coral assemblages and reef environments in the Bahamas Drilling Project Cores. In:SEPM Contributions in Sedimentology (ed Ginsburg, R. N.), SEPM. Compton, J., and Siever, R., 1986, Diffusion and mass balance of Mg during early dolomite formation Monterey Formation: Geochimica Cosmochimica Acta, v. 50, p Dawans, J. M. & Swart, P. K., Textural and geochemical alternations in late Cenozoic Bahamian dolomites, Sedimentology, v. 35, p Dix, G., and Mullins, H., 1992, Shallow-burial diagenesis of deep-water carbonates, northern Bahamas:Results from deepocean drilling transects: Geol. Soc. America Bull., v. 104, p Eberli, G.P. and Ginsburg, R.N., Segmentation and coalescence of platforms, Tertiary, NW Great Bahama