The Izu-Honshu collision zone of central Japan has been identified as the detrital source of the Quaternary trench-wedge facies on the basis of sand petrography, paleoflow indicators, facies architecture, clay mineral assemblages, and detailed characterization of the detrital illite component (Taira and Niitsuma, 1986; Marsaglia et al., 1992; Pickering et al., 1992; Taira and Ashi, 1993; Underwood et al., 1993a; Fergusson, this volume). Clay mineral assemblages change, however, in older deposits of the Shikoku Basin. As discussed below, some of those changes probably resulted from diagenetic reactions but diagenesis is superimposed on temporal shifts in the detrital inputs. Dispersal systems for suspended sediment sometimes fluctuate in response to dispersal mechanisms (i.e., surface currents and pathways of sediment gravity flow), uplift of new source areas, climatic controls over the degree of chemical vs. physical weathering, or changes in the amount of exposed volcanic material in a given source. As described by Underwood and Steurer (this volume), we now have enough evidence from ODP sites that have not been affected by diagenesis to reconstruct the temporal evolution of the regional-scale dispersal system for Nankai Trough and Shikoku Basin.
As one moves upsection from the lower to upper Shikoku Basin facies, the increase of wind-blown volcanic ash layers suggests that explosive volcanism gradually intensified from the early Pliocene into the Quaternary. Although the source for each ash layer remains uncertain, contributions could have come from volcanoes on Kyushu, central to northern Honshu, and/or the Izu-Bonin island arc (Cambray et al., 1995; Uto and Tatsumi, 1996; Kamata and Kodama, 1999). Working farther seaward in coeval Shikoku Basin deposits, Chamley (1980) documented a gradual decrease of smectite in progressively younger mudstones. Stratigraphic trends at Site 1177 are more erratic due to the prevalence of turbidites in the lower Shikoku Basin, but the same overall pattern of decreasing smectite through time holds there (Fig. F4). In addition, the content of smectite is relatively high in late Miocene and Pliocene trench-slope sediments at Site 1178 (Underwood and Steurer, this volume).
These data create an interesting paradox: the relative abundance of smectite in mudstone, which forms primarily by weathering of volcanic source rocks, decreases basinwide during the same period of time in which interbedded ash layers become more plentiful. To explain this observation, Underwood and Steurer (this volume) linked the gradual depletion of detrital smectite to intensification of the Kuroshio Current at ~3 Ma; that change in surface water circulation occurred in response to closure of the Pacific-Caribbean Gateway. A stronger and deeper surface current, flowing toward the northeast, would be expected to dampen the transfer of suspended sediment from the Izu-Bonin volcanic source and increase the amount of illite and chlorite carried into the Shikoku Basin and Nankai Trough from the Outer Zone of Japan. It is also interesting to note that simultaneous intensification of the Tsushima Current in the Sea of Japan (i.e., during the Pliocene) evidently caused a similar shift toward higher percentages of illite and chlorite at the expense of smectite (Fagel et al., 1992).
Several lines of evidence show that diagenetic reactions are superimposed on the inferred temporal shifts in the detrital flux into Shikoku Basin. Alteration of volcanic glass to clay minerals is readily apparent in smear slides, for example, and the total clay content in bulk powders increases as the glass degenerates (Shipboard Scientific Party, 2001c, 2001d; Wilson et al., this volume). Masuda et al. (1996) imaged volcanic glass shards from Site 808 with TEM and saw progressive growth of smectite crystals with dioctahedral beidelitic composition. "Spikes" in smectite content of up to 91 wt% occur locally in the Shikoku Basin deposits (Figs. F2, F3); these unusually high values probably come from in situ alteration of volcanic ash layers, although the cryptic bentonites did not stand out from surrounding sediment during visual inspection. Similarly, at Site 1177, sporadic increases in total smectite in the lower Shikoku Basin turbidite facies (Figs. F4, F6) seem too large to attribute exclusively to fluctuating detrital sources. We suggest that replacement of disseminated glass shards in the Miocene mudstone deposits added authigenic smectite to what were already high detrital contributions of both discrete smectite and disordered I/S mixed-layer clay.
Site 1173, the so-called reference site for the Muroto Transect, experienced diagenetic alteration seaward of the deformation front. Illite-smectite reaction progress clearly advances from Site 1173 landward to Site 1174 (Fig. F6). Smectite-illite diagenesis begins at ~390 mbsf at Site 1173 and at ~700 mbsf at Site 1174. At progressively deeper intervals, the percent illite in I/S clays increases (Fig. F6) and the content of smectite remaining in bulk mudstones decreases (Fig. F7). Larger amounts of scatter in I/S proportions at Site 1173 are probably due to mixing between detrital I/S and authigenic I/S. The gradient for Site 1174 is more pronounced because the diagenetic overprint is more extensive. The smectite-illite transition does not reach completion, however, which is consistent with the results from Site 808 (Underwood et al., 1993b). At Site 1174, the maximum illite in I/S is 84%, but there is no evidence of R = 1 ordering. At Site 1173, the maximum illite in I/S is 55%, with no ordering. A transition from R = 0 to R = 1 is supposed to occur when the illite interlayers in I/S reach 60%-70% (Bethke et al., 1986). In older basins this transition occurs at ~100° to 110°C, whereas in younger basins subjected to rapid heating (<3 Ma) this transition is not expected until ~120° to 140°C (Pollastro, 1993). R = 1 ordering was noted at Site 808 at 1220 mbsf and an inferred temperature of 135°C (Underwood et al., 1993b), but newer heat flow data (Shipboard Scientific Party 2001c, 2001d) push that temperature estimate to 144°C.
Figure F8 shows how sensitive the numerical simulations of illite-smectite reaction progress are to changes in potassium concentration, burial rate, and heat flow. Variations in heat flow produce the largest changes in the model results. If heat flow at Site 1173 is lowered to 130 mW/m2 or less, the model predicts that smectite-illite diagenesis will not even initiate. At Site 1174, heat flow must be reduced to 70 mW/m2 to prevent initiation of illite-smectite diagenesis, but using the documented heat flow value of 180 mW/m2 overpredicts reaction progress by 15%-20%. Evidently, reaction progress has been retarded either by fluid composition or by insufficient heating time. Present-day potassium concentrations at Sites 1173, 1174, and 808 are more than sufficient to promote the observed reaction progress (Fig. F8). Changing the potassium concentration to minimum values at each site (Site 1173 = 1.8 mM and Site 1174 = 1.5 mM) does not change the model result very much (Fig. F8). Thus, it appears as though potassium has been plentiful enough throughout the basin's burial history. Conversely, increasing the potassium concentration to the maximum value for each site (i.e., normal bottom water) substantially overpredicts the illite-smectite transformation. This scenario is unlikely, however, because it requires continual replenishment of dissolved potassium from an outside source. Changing the age-depth constraints that are listed in Table T3 to more straightforward linear burial gradients results in modest adjustments of the model (Fig. F8). To mimic the effects of faster burial, we reduced the age at each interval by specified percentages; an adjustment of 50% improves the fit between the model prediction and measured values at Site 1174. From this result, we conclude that recent episodes of burial near the toe of the Nankai accretionary prism (i.e., by trench-axis sedimentation, tectonic thickening, and frontal thrusting) may have been too fast for the illite-smectite reaction to keep pace. More sophisticated modeling will be required to explore how reaction progress responded to both gradual and punctuated shifts in heat flow, fluid flow, and burial rates (e.g., Wang et al., 1995; Saffer and Bekins, 1999).
Hyndman et al. (1995) demonstrated that the updip limit of earthquakes along the Nankai subduction boundary occurs within a temperature window of 100°-150ºC. Diagenetic changes in clay minerals are noteworthy in this context because of a hypothesized link between fault zone strength and thermally controlled mineral reactions. Vrolijk (1990), for example, speculated that the updip limit of seismicity matches the depth where 80% of the incoming smectite is transformed to a stronger illite-rich clay assemblage. Smectite has been shown to affect the sliding behavior of faults. In experiments on artificial montmorillonite anhydrite gouges, Shimamoto and Logan (1981) discovered that a bulk content of 15%-20% smectite changes stick-slip behavior to stable sliding. The coefficient of internal friction for natural smectite-bearing gouges from the San Andreas fault zone ranges from 0.21 to 0.41 (Morrow et al., 1982). Logan and Rauenzahn (1987) found that the frictional coefficient for pure montmorillonite ranges from 0.08 to 0.14, whereas the coefficient for quartz gouge ranges from 0.49 to 0.62. In a two-component mixture of quartz + smectite, the weaker mineral must exceed 25% of the bulk total to cause a significant reduction of the coefficient of friction (Logan and Rauenzahn, 1987).
Influenced by such laboratory results, Moore and Saffer (2001) discussed how the transformation of mudstones with initial contents of 30%-50% smectite might lead to changes in frictional properties downdip, but they also pointed out that incoming sediment along the Muroto Transect does not contain enough smectite to create such a scenario by itself. They cited, instead, contributions imparted on frictional properties and effective stress by increases in fluid pressure, hydrocarbon maturation, opal-A to opal-CT transformation, pressure solution, and cementation by zeolites, calcite, phyllosilicates, and silica, all of which occur over the temperature range of 100°-150°C. The logic of Moore and Saffer (2001) certainly holds for strata at Sites 808 and 1174. On the other hand, along the Ashizuri Transect (Site 1177), mudstones in the lower Shikoku Basin turbidite facies and volcaniclastic-rich facies are enriched by much higher percentages of smectite (Fig. F7). The geotechnical response to smectite-illite diagenesis is probably greater landward of the deformation front in the Ashizuri region.
Other geotechnical properties are affected by relatively small amounts of smectite. Robinson and Allam (1998) found that the coefficient of consolidation for montmorillonite decreases with increasing consolidation pressure, whereas other clays show increases in the coefficient with increasing pressure. The coefficient of consolidation also decreases as the proportion of smectite increases (Abeele, 1986). Adding as little as 5% smectite to silty sand increases compressibility, decreases shear strength, and increases secondary compression (Santucci de Magistris et al., 1998). Permeability is likewise affected by small amounts of smectite. In one study, hydraulic conductivity in a mixture of 6% smectite plus sandy silts was nearly equal to that of pure bentonite (Abeele, 1986). Freed and Peacor (1989a) showed that illite packets growing in the smectite matrix in I/S clays reduce the local permeability. In addition, the release of interlayer water during the smectite-illite transition can contribute to excess pore fluid pressures (Bruce, 1984; Colten-Bradley, 1987). Thus, even along corridors where the initial smectite content is not plentiful enough (i.e., <30 wt%) to change the rock's frictional coefficient during illite-smectite diagenesis, expandable clay affects compressibility, fluid migration, and effective stress along the Nankai-Shikoku subduction zone.