The Pleistocene volcaniclastic samples from Site 974 are essentially resedimented pyroclastic deposits composed almost entirely of volcanic glass (Comas, Zahn, Klaus, et al., 1996). These ashes are likely the product of individual felsic to intermediate eruptions in that the vitric components are colorless and are homogeneously stained (either red [Ca] or yellow [K]) throughout any given sample. There are three main Pleistocene volcanic provinces in the Tyrrhenian Sea: (1) high-potassic alkaline volcanics of mainland Italy to the east and northeast of Site 974 (Roman and Campanian volcanic areas); (2) isolated alkali-olivine and tholeiitic volcanic centers distributed from northern Sardinia across the axis of the Tyrrhenian Basin to eastern Sicily; and (3) to the southeast, within the Eolian magmatic arc, a calc-alkaline and shoshonitic magmatic province (Savelli, 1988). Given their proximity to Site 974, the first two provinces are the most likely sources of sand-sized ash at this site; however, detailed tephra studies (e.g., McCoy and Cornell, 1990) are needed to better constrain ash provenance through their chemical fingerprints.
The other non-ash sand samples from Unit I contain moderate amounts of plagioclase and potassium feldspar and all but one contain a high percentage of metamorphic lithic fragments; this composition suggests a continental source terrane. Site 974 lies in a north-south-oriented bathymetric low that is defined by basement uplifts and projects toward the northwest-southeast-oriented continental margin of Italy. Carter et al. (1972) and Vanney and Gennesseaux (1985) show a complex series of north-south-oriented submarine channels in the vicinity of Site 974 that extend and head toward the Italian continental shelf near the Tiber River delta. These channel orientations suggest that the Tiber River, or at least that coastal area, may have been a primary source of sediment at Site 974 during the Pleistocene. Outcrops in the Tiber River drainage basin include Pliocene-Pleistocene alkaline-potassic lava flows and pyroclastic rocks, and deformed (and metamorphosed?) Alpine turbidite and pelagic sequences (Bellotti et al., 1986; 1995). Seismic stratigraphy, and paleontological, sedimentological, and geochemical studies suggest that Unit III is a lacustrine ("Lago Mare"), synrift sequence (see Comas, Zahn, Klaus et al. [1996, p. 64] for discussion and additional references). Therefore, it is probable that local basement highs supplied clastic material to the Site 974 area during the Miocene. In fact, dredge hauls along the Monte de Marche, an uptilted basement block just south of Site 974, have yielded a Paleozoic to Neogene suite of phyllite, limestone, and granite fragments (Sartori et al., 1987) that is similar to the suite of lithic fragments present in sand from Unit III at Site 974, as described in this study, and at nearby Site 652, described by Borsetti et al. (1990). According to Selli (1985), these basement rocks are likely part of the Alpine suture zone.
Beach sand samples, particularly those near the mouths of major streams and rivers, were collected to provide information on the compositional characteristics of sand derived from modern Spanish coast terranes that likely served as sources for sand deposited at Sites 976, 978, and 979. These onshore source terranes (Fig. 5) include sedimentary Flysch Trough Units, rocks of the Alboran Domain (metamorphic and nonmetamorphic rocks of the Betic Cordillera), and volcanic rocks (onshore equivalents of the submarine ridges and seamounts that characterize the basin center; Comas et al., 1992). Outcrops of the sedimentary Flysch Trough Units are limited and represented by only the easternmost beach sample taken near Gibraltar (Sample #10; see Table 2 for sample location). This sample has the highest QFL%Q (68%; Table 2) of all the beach sands analyzed, and is thus the most compositionally mature, which could be attributed to the effects of sedimentary recycling. The Alboran Domain, the most extensive terrane that crops out along the coastal region (Fig. 5), contains high-pressure/low-temperature, low-pressure/high-temperature, and low-grade metamorphic rocks, including the Ronda peridotite massif (e.g., Chalouan and Michard, 1990; Goffé et al., 1989). As expected, the series of samples collected from beaches near the mouths of major streams and rivers that cross-cut Alboran Domain rocks are generally quartzolithic and metamorphiclastic with minor plagioclase and sedimentary lithic components. Serpentine fragments within these samples were likely derived from the Ronda peridotite and are limited to samples collected between Malaga and Gibraltar (Fig. 5).
The diverse suite of volcanic rocks within the Alboran basin (e.g., calc-alkaline volcanic rocks [7-13 Ma], lamproites and shoshonitic lavas [4.5-9 Ma] and alkali basalts [1.5-6 Ma]; Bellon et al., 1983; Comas et al., 1992; Hernandez et al, 1987) is represented by only one beach sample taken from east of Almería (Sample #2; Table 2). This sample is composed entirely of volcanic lithic fragments, phenocrysts, and zeolite alteration minerals. It is noteworthy that in contrast to this sample, the other nine beach samples are completely devoid of volcanic lithic fragments.
The Site 976 data cluster into three discrete petrographic groups that are stratigraphically distinct. The provenance of each group, from oldest to youngest, is discussed below.
Given the relict pyroclastic textures (e.g., pumice), high plagioclase phenocryst(?) content, presence of embayed biotite phenocrysts, and high percentage of zeolite alteration minerals, Unit IV likely had a significant volcanic component before undergoing burial diagenesis. The uniform composition/alteration of the vitric fragments is consistent with a single eruptive source of pyroclastic material that was likely associated with a middle to upper Miocene pulse of calc-alkaline magmatism that has been documented across the Alboran Basin (Comas et al., 1992). Shipboard biostratigraphic analysis yielded a middle Miocene age (Serravallian: 13.5-10.5 Ma) for Unit IV, but radiometric age dating of biotite from Unit IV may further constrain timing of the source eruption. Unit IV contains interspersed pebbles to sand-sized metamorphic rock fragments similar to underlying basement lithologies at Site 976 (e.g., high-grade pelitic schist, pelitic paragneiss, marble, and calc-silicate rock; Comas, Zahn, Klaus, et al., 1996), as well as onshore outcrops of Alboran Domain rocks. This polymictic lithic assemblage, along with the presence of a mixed deep- and shallow-water fauna (Comas, Zahn, Klaus, et al., 1996), implies that Unit IV is not a primary pyroclastic deposit, but was likely redeposited and is therefore epiclastic. Alonso et al. (Chap. 4, this volume) propose that Unit IV consists of debris-flow deposits.
Because the thin sand layers in the upper part of Unit III (Fig. 3) are very similar in composition to those in Unit II, they likely had the same provenance and are likely part of the same depositional system. The source of the thick turbidite sequence of Units II/III was probably a submarine canyon that headed somewhere along the continental margin to the north; Alonso et al. (Chap. 4, this volume) favor the Fuengirola River as the major source of sediment at Site 976, based on the proximity of the Fuengirola Canyon to Site 976. Onshore beach sands contain components similar to those found in Units II and III at Site 976, but the beach sand is richer in lithic material, with a particularly high proportion of serpentinite fragments likely derived from the Ronda peridotite (Pl. 3). Serpentine is a soft mineral, and it is possible that during increased fluvial transport associated with sea level lowstands, serpentinite fragments were more easily abraded and preferentially lost. Such a scenario would explain the lower content of serpentinite fragments in the Site 976 turbidite samples (maximum framework percent = 5%) as compared with onshore beach samples to the north (maximum framework percent = 10% in Samples #6 and #7; Table 2). If serpentinite (Lmv) is removed from the grain totals and the formulas used for recalculating parameters (Table 1), the mean composition of sand from Units II/III at Site 976 (Q53 F9L38; Qm86K7P7; Lm75Lv0Ls25) is strikingly similar to the average composition of the two beach sand samples north of Site 976 (average of beach samples #7 and #6: Q55F12L34; Qm83K1P16; Lm74Lv0Ls26; Table 2). Feldspar ratios are less correlative, however, with the beach sand relatively depleted in potassium feldspar. This could be a function of a difference in transport history (fluvial/beach vs. fluvial/deltaic/turbidite) and/or climate, both of which are known to affect feldspar ratios (e.g., Kairo et al., 1993; Marsaglia et al., 1996; Sedimentology Seminar, 1988). Note that beach sand to the northwest of Site 976 is even more enriched in serpentinite fragments (maximum content of 33% of the framework grains in Sample #8; Table 2), making a northwestern source less likely.
In contrast to the strong clastic pulse of Unit II, during the accumulation of Unit I sediments, Site 976 only received sporadic sand input. The composition of this sand is distinct from that found in the underlying units in that it is very quartzose and lithic poor, and resembles modern quartz-rich sands described by Huang and Stanley (1972) from shallow piston cores in the deeper part of the Western Alboran Basin. Huang and Stanley (1972) propose that the major source of the piston-cored sand is the Strait of Gibraltar area and adjacent Spanish slope, where coastal outcrops consist of sedimentary Flysch Trough Units (e.g., Aljibe quartz arenites with >98% quartz grains; Stromberg and Bluck, 1998) that would likely produce second-cycle, quartzose sand (Fig. 5). The more quartzose composition of beach sand from the Gibraltar area (68%; Sample #10; Table 2) supports this hypothesis. Therefore, we propose that the compositional shift from Unit II/III to Unit I sand likely results from a westward shift in source area. The strong east-flowing bottom currents needed to erode and transport this sediment from the Gibraltar area may have been generated by estuarine circulation during glacial minima when surface waters were freshened (Huang and Stanley, 1972). This scenario seems more plausible than that of the Shipboard Scientific Party (1973) who proposed that the quartzose sand in the correlative Quaternary section at DSDP Site 121 (Fig. 5) may have been deposited by melting icebergs carried into the Mediterranean from the Atlantic Ocean during glacial maxima.
The Almería channel is one of the major submarine canyons/channels in the Alboran Sea and the only channel that extends into the Eastern Alboran Basin (Alonso and Maldonado, 1992; Carter et al., 1972; Vanney and Gennesseaux, 1985); as such, it likely provides the main conduit for terrigenous sediment carried into the Eastern Alboran Basin and Sites 977 and 978. This long, well-developed channel initiates in the coastal embayment between Almería and the Chella Bank, and extends southward feeding into the Alboran Channel, a major Pleistocene-Pliocene sediment depocenter that extends from the middle of the Alboran Basin eastward toward Sites 977 and 978 (Figs. 3, 5; Alonso and Maldonado, 1992). The main drainage of the Almería coastal embayment is the Andarax River, a seasonal ephemeral stream that empties into the Alboran Sea at Almería. Modern beach sand collected near the mouth of the Andarax River shows it to be predominantly composed of quartz-mica tectonite fragments, carbonate lithic fragments, and quartz (Pl. 3; Table 2).
Although detrital modes of sand samples from Unit I at Site 977 and Unit I at 978 are highly variable, they exhibit similar mean compositions. For example QFL%Q varies from 8% to 100% at Site 977, and from 49% to 73% at Site 978, but the mean QFL%Q for each is 57%. In contrast, beach sand from the mouth of the Andarax River has a relatively low QFL%Q of 28%. The metamorphic and sedimentary lithic populations within Site 977 and 978 sands are similar to those found in onshore beach sand samples, but the proportions of these components vary. This lack of direct correlation could be a product of one or more of the following: (1) the beach sampled is not representative of modern Andarax River sediment; (2) the composition of Andarax River sediment has changed through time, perhaps because of changes in drainage basin size; (3) there is an inherent dependence of composition on grain size in sand derived from Alboran Domain rocks, with lithic components enriched in coarser beach sand and quartz enriched in the finer grained turbidites at Site 977 and 978; and(4) metamorphic and sedimentary lithic fragments are preferentially abraded/removed during sand transport down the Almería Channel and through the Alboran Trough. Although the precise source of Unit I sand at Site 977 and 978 is equivocal, the lack of volcaniclastic debris effectively rules out derivation from onshore or offshore volcanic provinces and the submarine physiography strongly supports an Andarax River source via the Almería Channel.
In contrast, the single sand sample analyzed from Unit II at Site 977, as well as the underlying gravel described by shipboard scientists has a significant volcanic lithic component. As discussed in Comas et al. (1996), Unit II gravel not only contains rhyodacite pebbles similar to samples from the Al-Mansour Seamount, but also basalt and andesite pebbles similar to the volcanic rocks found to the west on the Alboran Ridge and to the north on Cabo de Gata. Thus Unit II sand and gravel may have been locally derived, or, alternatively, far-traveled.
The physiography of the Southern Alboran Basin where Site 979 is located (Fig. 3) suggests that possible sand sources are limited to the North African coast between Melilla and Al-Hoseima and the submarine volcanic ridges that surround the basin (Fig. 4). The Unit I sand at Site 979 is generally quartzolithic, but with variable lithic proportions. Volcanic lithic fragments (glassy shards) are most abundant in the uppermost sample (<0.085 Ma) and the lowermost part of the section (Pliocene; > 2.63 Ma), below a possible hiatus at 476 mbsf (Comas, Zahn, Klaus, et al., 1996). Given that the basin is surrounded by volcanic terranes, the source(s) of this vitric component is equivocal. In addition, no discrete source canyon is present along the North African coast south of Site 979, where the coastal zone is characterized by a broad shallow (<100 m) shelf (Fig. 3). The high percentage of glauconite and bioclastic debris (the percentage of bioclastic debris was so high it was not tallied in most samples) in Site 979 samples is consistent with redistribution from the shallow shelf environment into the basin via turbidity currents. In contrast to the southern Spanish margin, outcrops along the North African coast, due south of Site 979, are dominantly sedimentary, with lesser volcanic and metamorphic rocks (Fig. 4); this change in source terrane is reflected in the relatively high sedimentary lithic proportions of many samples from Site 979 (Fig. 10). However, there is a crude tendency for the older (deeper) samples to contain more sedimentary lithic fragments and the younger samples to contain more metamorphic lithic fragments (Table 2).
Quantitative detrital modes of sandstones have been closely tied to the tectonic settings of their provenance terranes. Based on compilations of detrital modes, Dickinson et al. (1983) outlined three main provenance fields on a QFL ternary diagram: Continental Block, Magmatic Arc, and Recycled Orogenic. Quartzose sandstones derived from subduction complexes or fold-thrust belts constitute the Recycled Orogenic provenance group, which Dickinson et al. (1983) further divide into Quartzose-, Transitional- and Lithic-Recycled subfields on a QmFLt ternary plot. Mean values for most of the subgroups outlined at the Alboran and Tyrrhenian Sea sites plot in the Recycled Orogenic field and near the boundary of the Quartzose- and Transitional-Recycled subfields (Fig. 11). Exceptions include Unit I Pleistocene sand at Site 974 and Unit IV Miocene sand at Site 976, which plot in Magmatic Arc fields, and the Unit I Pleistocene sand at Site 976, which plots near the join between Recycled Orogenic and Transitional Continental fields (Fig. 11). Note that the likely diagenetic alteration and dissolution of volcanic debris within Unit IV sand at Site 976 suggest that the sand's true mean composition should be shifted towards the lithic end of the Magmatic Arc field (arrow in Fig. 11), and that the composition of the lower Pliocene to Miocene sand from Unit II at Site 977 falls in the Dissected Arc field. Sand within the Alboran and Tyrrhenian Basins has a mixed metamorphic, volcanic, and sedimentary provenance reflecting the unique "Mediterranean-style" tectonic histories of these basins: collisional orogen followed by extension and related volcanism. Crude temporal trends, such as that from more volcaniclastic (Miocene) to metamorphiclastic (Pliocene-Pleistocene) compositions in the Alboran Basin, and from metamorphiclastic (Miocene) to volcaniclastic (Pleistocene) compositions in the Tyrrhenian Sea, can be linked to changes in tectonic regime, with the volcaniclastic intervals representing periods of active extension and volcanism. It is important to note that Pleistocene sand from Site 974 in the Tyrrhenian backarc basin (Q19F22L59; Lm46Lv50Ls5) is very similar in composition to Pleistocene sand from the Japan Sea continental-backarc basin (e.g., Site 299; Q13F25L62; Lm39Lv38Ls23; Marsaglia et al., 1992; Marsaglia and Ingersoll, 1992), although Site 974 is distinguished by a higher metamorphic lithic fraction. The latter could be considered as characteristic of sand deposited in backarc basins that develop in collisional settings.
Petrographic data from the Alboran and Tyrrhenian Seas form the basis of a preliminary actualistic model for interpreting sand provenance within remnant ocean basins caught up in complex suture zones. For example, the range of volcaniclastic, metamorphiclastic, and quartzose sand compositions that we find across these basins and within individual stratigraphic columns (e.g., Sites 976 and 974; Table 2) is strikingly similar to the suite of petrofacies documented by Critelli (1993) within the Liguride Complex, an accretionary wedge of the Southern Apennines. His Upper Cretaceous to middle Eocene quartzose petrofacies (Q90F9L1) corresponds to Unit I at Site 976 (Q76F17L6), his upper Oligocene volcanolithic petrofacies (Q15F25L61), corresponds to Unit I at Site 974 (Q19F22L59), and his upper Oligocene to lower Miocene quartzolithic petrofacies (Q54F10L36) corresponds to Unit I at Sites 977, 978, and 979 and Units II and III at Site 976 (Q(51-57)F(2-9)L(38-41)).