From the methane and the sulfate data we classify the seepage activity in the CMSA in fall 2010 as high. Methane concentrations of up to 3776 nM directly above the seafloor and the observed spatial variations over 3 orders of magnitude are for example comparable to other prominent active margins seep systems such as Hydrate Ridge off Cascadia margin [e.g., Heeschen et al ., 2005 ]. However, in contrast to other active seep areas, which have been visited by multiple measurement campaigns, we are lacking data to quantify temporal changes in seep activity. The only other direct measurement of seep activity from the CMSA are side‐scan sonar data collected in 2008 [ Klaucke et al ., 2012 ], 2 years before the 2010 Maule earthquake. Based on their data, Klaucke et al . [ 2012 ] report an obvious enigma between widespread evidence for fossil cold seeps in the form of high backscatter anomalies on the seafloor that reflect authigenic carbonate precipitates, and little evidence for active seepage. Within their ∼800 km 2 of side‐scan data they found only a couple of gas flares in the water column. Nevertheless, the different techniques used in 2008 (side‐scan sonar) and 2010 (water column and sediment sampling) do not allow to draw a final conclusion if seepage has increased within the 2 years.

The water column overlying the CMSA shows a significant positive methane anomaly (factor 10–1000 above background) indicating active methane release from the seafloor (Figure 3 ). With 8 nM at 55 m water depth (CTD‐6) and 56 nM at 203 m water depth (CTD‐7) it seems plausible that some of the methane in the CMSA even reaches the atmosphere. For the shallow water we can, however, not ultimately distinguish if the high methane concentrations result from a plume that rises from the seafloor or if they originate from a different water mass enriched in methane. In addition to the methane anomalies in the water column, the sulfate penetration depth in surface sediment, which is controlled by microbial sulfate reduction coupled to anaerobic methane oxidation [ Niewöhner et al ., 1998 ; Hensen et al ., 2003 ], can be used as a proxy for the methane flux from subsurface sediment strata [ Borowski et al ., 1996 ]. The comparably shallow sulfate penetration depth in the CMSA (<0.4 m), compared to the cores from farther north and downslope (2–5 m) further indicates a higher seafloor methane flux (Figure 4 A and Table 1 ).

An admixture of methane from greater depth in the CMSA is further consistent with previously published calculations of the stoichiometry of sedimentary sulfate consumption to bicarbonate production across the Chilean margin [ Scholz et al ., 2013 ]. Moreover, the elevated δ 13 C CH4 values in sediments of the CMSA (stars in Figure 4 B) are indicative for a higher proportion of thermogenic hydrocarbon gas compared to cores from outside the CMSA (circles in Figure 4 B). Thermogenic hydrocarbon gas production requires temperatures beyond the 5–60°C [ Claypool and Kvenholden , 1983 ] that prevail within the hemipelagic sediments overlying the paleo‐accretionary prism [ Scholz et al ., 2013 ]. We therefore assume that the methane released in the CMSA is a mixture of shallow biogenic methane with thermogenic methane that is derived from deeper strata. A mixing line calculated by using a microbial hydrocarbon end‐member (PC05 in Figure 4 B and Table 1 ) and a thermogenic end‐member (δ 13 C CH4 ∼ −35‰, C1/(C2+C3) ∼ 100) joins all the CMSA data in the modified Bernard plot (Figure 4 B) [ Faber et al ., 2015 ]. The rough calculation may indicate possible admixing of up to 20% of mature thermogenic methane to shallow microbial methane within the CMSA (i.e., “0.8” in Figure 4 B). A possible thermogenic methane source at greater depth could either be organic‐rich compacted framework rock of the paleo‐accretionary prism or subducted sediments along the plate boundary, or a combination of both.

To investigate the origin of the methane that is released in the CMSA, we measured its stable carbon isotope signature (δ 13 C‐CH 4 ) (Figure 4 B). The methane could be exclusively produced at shallow depth (tens of meters) and transported into the SMTZ by diffusion or it could be derived from greater depth (hundreds to thousands of meters) and transported into the shallow sediments through advective transport along faults or other weak zones [ Schmidt et al ., 2005 ]. If shallow microbial methane production was the exclusive reason for the high methane flux in the CMSA, one would expect sulfate penetration to correlate with the current rate of organic matter delivery which can be roughly estimated from TOC concentrations (Table 1 ). We do not observe such a correlation. In fact, TOC concentrations in slope basins further downslope are substantially higher than in the CMSA, despite a greater sulfate penetration depth (Table 1 ) [cf. Scholz et al ., 2013 ]. This observation suggests that at least some of the methane released in the CMSA must derive from deeper strata.

3.3 Possible Links Between Seafloor Methane Seepage and the Seismic Cycle

Coulomb stress change calculations (Figure 5) suggest that the transfer of stress into the upper plate caused by the coseismic rupture of the 27 February 2010 Mw. 8.8 Maule earthquake could have caused extensional faulting in the marine fore arc in the area of the CMSA. In fact, extensional rupture in the upper plate in the wake of a large plate‐boundary earthquake has been documented in the northern termination of the rupture area of the 2010 Maule earthquake [e.g., Farias et al., 2011; Lange et al., 2012] as well as after the giant 2011 Mw 9.0 Tohoku‐Oki earthquake [Tsuji et al., 2013]. In the case of the 2011 Tohoku‐Oki earthquake comparable open fissures and fractures were visually documented at the seafloor overlying seismically imaged upper plate normal faults [Tsuji et al., 2013]. There, repeated video observations conducted between 2002 and 2012 showed that the fractures were not present in the years before the earthquake suggesting that they formed due to coseismic rupture during the 2011 Tohoku‐Oki earthquake.

The location of the large normal fault in the CMSA and its strike direction parallel to the trench (Figure 1B) makes this structure an ideal candidate for the release of some of the stresses transmitted into the upper plate prior to large plate‐boundary earthquakes, and released during the earthquakes. The uneroded edges of the fractures and lack of sedimentary infill (Figure 2) suggest that they represent relatively young (days to couple of years) extensional features, possible related to extensional rupture across the underlying large normal fault during or shortly after the 2010 Maule earthquake. We therefore expect similar seafloor fractures elsewhere in the CMSA. However, we are not able to proof this with the limited spatial coverage (<0.5 km2) of the ROV video observations.

In addition to normal faulting, the extensional stresses that prevail in the upper plate in the area of the CMSA during large plate‐boundary earthquakes (including during the 2010 Maule earthquake) may support hydraulic fracturing and dilation of existing faults. This results in an increase in permeability along both large upper plate faults that crop out in the CMSA; i.e., the normal fault and the eastern splay‐fault. The reopening of faults in an extensional stress regime is one viable mechanism to enhance the permeability of accretionary wedges to considerable depth [Behrmann, 1991]. In the case of indurated lithologies (cohesion of 5 MPa), reopening of weakly cemented hydraulic fractures is possible to depths of over 1 km and an even greater depth of reopening is possible in overpressured environments [see Behrmann, 1991, Figures 3 and 4].

Based on the spatial correlation between the large upper plate faults (normal fault and eastern splay fault), the juvenile extensional fractures, the high water column methane concentrations, and the shallow sulfate penetration depth, we propose a conceptual model (Figure 6) to link seafloor hydrocarbon seepage in the CMSA to the frequently recurring plate‐boundary earthquakes:

Figure 6 Open in figure viewerPowerPoint Conceptual model illustrating how seismic rupture during the 2010 Maule earthquake governed extensional faulting, fluid flow, and hydrocarbon seepage in the Concepción Methane Seepage Area (CMSA).

Large subduction earthquakes induce a regime of extension in the marine fore arc in the area of the CMSA, large enough to reactivate the normal fault and to create extensional fractures at the seafloor. Extension of the upper plate further results in hydraulic fracturing and dilation of the normal‐fault and the eastern splay fault that both underlie the CMSA. This together increases the permeability along the faults, allowing them to become hydraulically communicative with the deep fore arc‐strata and possibly the plate boundary. As a consequence, fluid pulses from the upper plate and/or the plate boundary move upward along the faults resulting in the observed seepage.

The model can well be transferred to other subduction zone settings such as Makran, where the 1945 Mw. 8.1 earthquake seems to have triggered hydrocarbon seepage from the continental slope [Fischer et al., 2013] and where huge upper plate normal faults have been seismically imaged [Grando and McClay, 2007]. Furthermore, the model may also explain basic differences in the spatial distribution of seep sites at the accretionary margin of Central Chile, relative to an erosive plate margin as for example Costa Rica. In the latter case, tectonic erosion at the base of the overriding plate causes widespread subsidence and related normal faulting all over the continental slope. As a consequence, fluid seepage distributes over most of the continental slope, locally extending down to the deformation front [Hensen et al., 2004; Ranero et al., 2008; Sahling et al., 2008]. In central Chile, however, viable conditions for extensional tectonics are only met in the coseismic phase. Here fore‐arc extension, and therewith also seep activity, is spatially limited to the upper continental slope and shelf area landward of the area of peak rupture during large plate‐boundary earthquakes (compare Figure 5B).

In the case of a single earthquake one would expect progressive recementation and healing of the faults and the extensional fractures and concurrent cessation of fluid flow after the earthquake. However, the recurrence rate of large plate‐boundary earthquakes in the study area is only in the range of 100–150 years [Lomnitz, 2004]. Even if not every single earthquake induces extensional stresses large enough to trigger activity of the large upper plate faults, the interseismic period is possibly not long enough to allow for a complete healing of the fluid pathways. It is thus likely that fluid flow along the faults and related seafloor methane seepage continue to some extent over long timescales and do not cease completely during a seismic cycle. This is further supported by the local occurrence of gas flares in the water column in the CMSA, which Klaucke et al. [2012] reported from their thorough investigation of side‐scan sonar data collected in 2008, about 2 years before the 2010 Maule earthquake. Continuously high seep activity over multiple seismic cycles is further in‐line with the widespread evidence for fossil cold seeps in the area reported by Klaucke et al. [2012]. However, repeated measurements in between large plate‐boundary earthquakes are needed to quantify a possible temporal variability of fault‐related methane seepage on the Central Chilean marine fore arc in relation to the seismic cycle.