Most of the laminated evaporite intervals at Site 975 consist of what appears to be detrital gypsum, in that the fragments are angular to slightly rounded but generally are rectangular in shape and appear to be broken. Possible sources of gypsum crystals include precipitates that nucleate at the water surface and settle through the water column or nucleate as crusts on the bottom at the sediment-water interface (Schreiber, 1978, 1988). Both may be reworked and transported by water or by eolian processes (Schreiber, 1988). The finer "pure" gypsum laminae at Site 975 could represent direct precipitates at the air-water interface, or the fine gypsum could also be reworked from the basin margins and deposited in deeper water by density currents (Schreiber, 1988). The laminations are episodic, likely reflecting periods of evaporitic drawdown, and meteoric or marine influx; in modern brine ponds, as many as eight laminae may form per year (Schreiber, 1978).
Gypsum laminae exhibiting inverse grading have been ascribed in the literature to a variety of mechanisms including alternating precipitation of fine gypsum and anhydrite where the latter is subsequently replaced by coarse-grained gypsum (Ogniben, 1955), increasing grain size during direct precipitation in the water column (Garrison et al., 1978), mechanical transport and deposition of coarser particles over a finer grained substrate during storms (Hardie and Eugster, 1971), growth of gypsum at the top of laminae as a function of diagenesis (Schreiber et al., 1976), and progressive increase in size of settling crystals as a function of increasing brine concentration (Garrison et al., 1978). Garrison et al. (1978) proposed that irregular relief on planar-wavy laminations may have originated as adhesion or ripple marks and suggested that inverse grading could have been produced by backwash on a beach. Similar features are present in gypsiferous sediments described by Vonder Haar (1976) in the modern Laguna Mormona of Baja California.
We suggest that the most likely origin of the Site 975 planar-wavy and planar-cumulus laminae, and perhaps even the pinch-and-swell structures, may be early-diagenetic overgrowth of detrital gypsum at and just below the sediment-water interface. Schreiber et al. (1976) attributed similar wavy-laminated and nodular clumps of crystals to rapid subaqueous precipitation. The globular masses that characterize the planar-cumulus structures described in this study thus are likely a product of rapid surficial crystal overgrowth. The tendency to subvertical crystal alignment in the coarser grained tops of planar-wavy and planar-cumulus structures also supports such a mechanism in that gypsum crystals that grow as crusts at the sediment-water interface are more likely to have their long axes aligned perpendicular to bedding (Schreiber, 1988). The degree of development of surface topography (wavy vs. cumulus vs. pinch-and-swell) and crystal size in the Site 975 examples may be a function of the time between brine concentration and crystal overgrowth, and burial by matrix sediment. Surficial erosion and dissolution may also be important processes in determining the form ultimately preserved by burial. Such crystal overgrowth would result in interlocking textures, but would also allow for the preservation of larger intercrystalline pores that are present in the pinch-and-swell horizons. The draping and onlapping relationships between laminated matrix and the gypsum "swells" in the pinch-and-swell horizons suggest that the "swells" formed minor topographic highs at the sediment-water interface. Low areas were preferentially infilled with sediment. Growth of the "swells" during matrix deposition is suggested by intertonguing of matrix and coarse gypsum near swell margins. There is no evidence (e.g., inclusions of fine anhydrite or algal filaments) to suggest that the pinch-and-swell gypsum at Site 975 had a supratidal anhydrite precursor.
The sedimentary structures observed in the Site 975 gypsum section appear to represent a continuum of processes associated with the two gypsum cycles outlined in Figure 2. In these cycles there is a gradual progression in the laminate gypsum from planar to planar-wavy to planar-cumulus structures and eventually to pinch-and-swell gypsum. The pinch-and-swell gypsum is abruptly overlain by clay and gypsiferous chalk. This sequence might imply a period of time after clay deposition, when carbonate was preferentially precipitated in the overlying water mass. Much of the character of the gypsiferous chalk appears to be a function of diagenetic displacive growth of gypsum crystals within an unconsolidated sediment. The presence of laminae of densely packed subhorizontal gypsum crystals arranged in a criss-crossing palisade pattern suggests that the chalk was not deposited as a single "event" bed, but was laid down gradually with hiatuses marked by growth of gypsum-crystal crusts on the sediment surface. It is difficult to say whether the clay and gypsiferous chalk represent the beginning or the end of the process that produced the gypsum cycles. The appearance of crenulated gypsum and possible rip-clast conglomerates at the base of the recovered section at Site 975 suggest a down-section change in facies.
Most previous workers have favored a shallow-water to supratidal (sabkha) depositional environment for the Messinian sedimentary section (e.g., Garrison et al., 1978). There are, however, conflicting opinions as to whether the basin was shallow or deep (see Sonnenfeld, 1985). Evidence cited by Nesteroff (1973) in favor of a deep depositional basin include the deep-water fauna in the earliest Pliocene sedimentary fill and the presence of deeply cut submarine canyon systems and terrigenous coarse clastic sand within the basins. Criteria called upon by Nesteroff (1973), Nesteroff et al. (1973), and Friedman (1973) to support shallow-water to supratidal (sabkha) deposition include: laminated stromatolitic structures, nodular to chicken-wire anhydrite, littoral or brackish benthic microfaunas (foraminifers, ostracodes, diatoms), isotopic studies (e.g., Fontes et al., 1973) and the presence of halite and dolomite.
In our shipboard analysis of the Site 975 sequence, we tentatively compared the gypsum cycles at Site 975 to evaporite cycles defined by Garrison et al. (1978) for Site 374 in the Ionian Sea (Fig. 1) and suggested that they formed in a supratidal to shallow, subaqueous environment. Closer inspection shows significant differences between the Site 374 and Site 975 facies. For example, in the cycles at Site 374, the basal member consists of dolomitic to diatomaceous mudstone, in some cases displaying stromatolitic-like laminations. Displacive crystals and pinch-and-swell structures of gypsum, interpreted to have replaced anhydrite, disrupt the fabric of the mudstones. In contrast, the clays and micritic silty clays interbedded with the gypsum at Site 975 are calcitic and do not have textures or displacive fabrics similar to those of the mudstones at Site 374. The gypsiferous chalks of Site 975 are somewhat similar in appearance to some of the more disrupted mudstones at Site 374, but again, the chalks at Site 975 are calcitic, not dolomitic. Furthermore, the pinch-and-swell evaporites at Site 374 consist of anhydrite or gypsum after anhydrite that formed displacively in a carbonate mud. As stated above, we see no evidence for anhydrite precursor mineralogy in the pinch-and-swell gypsum at Site 975. The pinch-and-swell gypsum at Site 975 is also relatively pure gypsum, with only low concentrations of calcite, quartz, or dolomite (Table 3).
The Messinian evaporite sequence recovered at Site 372 (Menorca Rise, Fig. 1), contains evaporite lithofacies similar to those at Site 975. Although there are differences between the two sites in the carbonate facies and in the cyclic sequence of evaporite lithofacies, the clastic and crystalline textures in the laminated to "micro-laminated" gypsum intervals in Core 8 at Site 372 are nearly identical to those found in the planar, planar-wavy, and planar-cumulus laminated intervals at Site 975. The evaporite sequence at Site 372 contains cross-laminated intervals and better developed crenulated and intraclastic (rip-up) beds; noticeably lacking, however, is the more massive, pinch-and-swell gypsum present at Site 975. This lithofacies may be present at Site 372, but not recovered. Garrison et al. (1978) interpreted the Site 372 evaporites as having been deposited on a shallow-water evaporite flat or lagoon above storm wave base and within the photic zone.
We believe that our textural and sedimentological evidence, coupled with the sequential change from planar to planar-wavy to planar-cumulus to pinch-and-swell gypsum in the cycles at Site 975, suggests a subaqueous origin for the evaporative facies. Current data provide no evidence for a supratidal or sabkha-type environment. Laminated evaporites can be the product of intertidal, shallow-water, or deep water-processes. Lateral continuity of gypsum laminae is often used as an indication of deep-water basinal deposition (see Warren, 1983, for discussion). Our view at Site 975 is limited to a few centimeters within the core. However, we are fortunate that the ship's position was offset 20 m to the north after drilling Hole 975B and before drilling Hole 975C, and a unique sequence of evaporite beds can be directly correlated between the two holes. As Figure 4A, B indicates, there is a direct correlation of pinch-and-swell gypsum, clay and gypsiferous chalk across this interval and the finest details in overlying and underlying gypsum laminae are also directly correlative on a millimeter-scale.
Additionally, we observed no cross stratification, desiccation features, brine shrimp fecal pellets, algal filaments, or cerithid gastropods that would indicate, as suggested by Schreiber et al. (1976) and Schreiber (1988), a shallow-water origin. Thus, our observations are consistent with a deeper water depositional setting for the accumulation of these sediments, where "deep" is defined only as below wave base. The question of whether or not these sediments accumulated within the photic zone as discussed in Schreiber et al. (1976), could only be answered by using a scanning electron microscope to establish the presence or absence of fine algal structures. In fact, within a laminated interval (Section 161-975B-33X-CC, 22-27 cm), we observed circular pores in fine gypsum crystals that could either be dissolved nannofossils or possibly coccoid algae (see Awramik, 1978).
The submillimeter to fine millimeter-scale laminations observed in the laminated gypsum at Site 975 could represent various basin-wide, climatic phenomena such as seasonal changes in temperature, precipitation, and humidity, or perhaps even singular rainfall events or pulses of seawater input into the basin that decrease salinity and result in carbonate precipitation (Warren, 1983).
Across the Mediterranean, the marly interbeds of the upper Miocene are characterized by dwarfed marine foraminiferal and oligotypical faunas that probably are the result of ecologic stress (Cita, 1973; Hsü, Montadert, et al., 1978). In addition to foraminifers within gypsiferous chalk beds at Site 975, we have found whole and/or fragmented foraminifers in almost every matrix (gypsum) lamina examined in this study. This seems to imply that the marine incursions thought to be responsible for the interbedded marls may have in fact been stronger signals of continuous seawater influx into the basin during the latest Messinian. The alternative explanation, that these foraminifers could be eolian contamination from exposed marine facies on the margin of the basin, is probably best evaluated by detailed paleontological studies.
The calcareous and mixed calcareous/terrigenous sand interval (Core 161-975B-33X-CC, 3-11 cm) recovered from Hole 975B may grade laterally into a pebbly sand. At Hole 975C (offset of 20 m), a pebble 4.5 cm in diameter, but no sand, was recovered in this interval. In thin section, the pebble consists of tectonized granule conglomerate, with twinned carbonate cement and clasts of quartz, quartzite, metaquartzite, and chert. The composition of terrigenous material in this sandy unit, particularly the common presence of siliciclastic and metamorphic lithic grains, is consistent with derivation from Paleozoic sedimentary, metasedimentary, and schistose basement rocks that crop out on Menorca and have been dredged from off Menorca (Bourrouilh and Gorsline, 1979; Jenkyns et al., 1990).
In contrast, the carbonate- and volcanic-lithic suite found in the sand at Section 161-975B-33X-3, 141-144 cm, could easily be attributed to the erosional products of Mesozoic rocks exposed on the island of Menorca. The Triassic section on Menorca includes red-bed siliciclastic rocks, dolomites, intertidal and subtidal carbonates, evaporites and alkaline mafic volcanic and pyroclastic rocks (Jenkyns et al., 1990; Rodríguez-Perea et al., 1987). This sequence is in turn overlain by Jurassic platform carbonates that include stromatolitic, pelletal, and intraclastic facies, followed by pelagic carbonates that locally include red nodular limestone (ammonitico rosso), and Fe-Mn hardgrounds (Jenkyns et al., 1990; Rodríguez-Perea et al., 1987). At Site 122 to the north of the Balearic Islands in the Valencia Trough, gravel recovered just above Messinian evaporites (Ryan, Hsü et al., 1973) is amazingly similar in composition to the volcaniclastic sandy interval described at Site 975 (Section 975B-33X-3, 141-144 cm). Thus, it is possible that Site 975 and Site 122 both received detritus from an uplifted Mesozoic section on Menorca during the latest Miocene.
Messinian evaporite sequences throughout the region are commonly overlain by foraminifer- and quartz-bearing sandy intervals (Fig. 1; e.g., Sites 371 and 372 [Hsü, Montadert, et al., 1978]; Sites 124 and 132 [Ryan, Hsü, et al., 1973]; Sicily [Hsü et al., 1973]). At Site 975, the basal post-evaporite sand facies is likely a product of increased input of marine water into the South Balearic region, as suggested by the abundance of foraminifers. The sandstone bed is laminated and moderately well sorted, and could be the result of a catastrophic flooding event (e.g., like that envisioned by Cita et al., 1978) or alternatively of eolian or beach processes acting along the margin of a rising proto-Balearic Sea. The abrupt up-section shift from sand composed of foraminifers and terrigenous debris to sand composed of carbonate- and volcanic-lithic debris remains problematic.
The sediments of Unit II at Site 975 have a number of distinctive characteristics that distinguish them from Messinian post-evaporite deposits at other drilling sites:
The micritic sediments of Unit II appear to have formed in an environment characterized by inorganic calcite precipitation, low planktonic productivity, and episodic, variable influx of terrigenous material. Only during times of isolation from the terrigenous sediment sources could pure micrite accumulate. The lack of bioturbation throughout Unit II indicates an absence of benthic burrowing fauna. Ostracodes and the benthic foraminifer Ammonia tepida found in Section 161-975B-33X-CC (Shipboard Scientific Party, 1996b) indicate a brackish-water environment for the sediments at the base of Unit II. In addition, foraminifers from Sample 975B-33X-3, 80 cm, to the bottom of the hole are dwarfed forms, indicative of a stressed environment.
In the Site 975 chapter (Shipboard Scientific Party, 1996b), it was suggested that the micritic sediments were deposited in a shallow-water, low-energy, environment, and that the fine laminations that characterize all but the most clay-rich intervals could be due to either (1) seasonal fluctuations of productivity, salinity, or terrigenous input, or (2) layering of algal/microbial mats. An alternative interpretation of origin is suggested by the similarity of the pure micrites to chemically precipitated lacustrine chalks such as those described by Hsü and Kelts (1978) from the Black Sea. These chalks are thought to have formed in a deep, largely stagnant, freshwater lake. Additional paleontological analyses, as well as scanning electron microscopic investigations and isotopic analyses (both currently in progress) are necessary to refine further an interpretation of the depositional environment of the Unit II micritic sediment.