The M 5 Earthquake of June 23^rd, 2010 centred 60 km northeast of Ottawa within the Western Québec Seismic Zone (Fig. 1), rattled much of central Canada and is a reminder that intracra-tonic seismicity in heavily urbanized central and eastern North America is still not well understood (Stein and Mazzotti 2007). The Western Québec Seismic Zone poses a direct risk to infrastructure such as mines, lifelines, critical facilities and the large urban areas of Montréal and Ottawa, where soft glaciomarine sediments (Leda Clay) and aged infrastructure compound seismic risk (Motazedian and Hunter 2008; Rosset and Chouinard 2008). There is still inadequate knowledge of earthquake source zones, recurrence intervals, and the effects of postglacial rebound (Ma et al. 2008), so emphasis is placed on the geologic and geomorphic record of seismic activity (e.g. Shilts and Clague 1992; Kelson et al. 1996; Ouellet 1997; Aylsworth et al. 2000; Tuttle 2001; Talwani and Schaeffer 2001). Our purpose here is to highlight the wealth of neotectonic structures that deform the bottom sediments of lakes Kipawa and Timiskaming within the Timiskaming Graben along the border of Ontario and Québec (Fig.2). We also identify a prominent 20 km-long onshore scarp that is inferred to be a product of co-seismic faulting.

The Timiskaming Graben (TG) parallels the Ontario–Québec border and is structurally defined by the Cross Lake Fault and the Quinze Dam Fault (Fig. 2). It is a complex structure that likely first developed as part of the Paleoproterozoic (ca. 2.4 Ga) ‘Cobalt Embayment’; it was much later reactivated during regional extension related to the breakup of Rodinia and Pangea (Barosh 1990; Mazzotti 2007) and during compression related to various Appalachian orogenies (Kumarapeli and Saul 1966; Dix and Molgat 1998). The TG transects the Grenville Front Tectonic Zone, which separates the Grenville and Superior provinces, and merges with the Ottawa-Bonnechere Graben before joining the St. Lawrence Rift (Fig. 2). The entire system is a broad belt of heightened seismic activity known as the Western Québec Seismic Zone (Fig. 1), which encompasses the large urban areas of Ottawa–Hull and Montréal, and extends to Lake Champlain and Vermont, USA (Fig. 2). The history of earthquake activity in this area is therefore of international concern.

Basement to TG consists of Paleoproterozoic and Archean rocks of the Canadian Shield (Lovell and Caine 1970; Lovell and Frey 1976; Russell 1984; Johnson et al. 1992; Dix and Molgat 1998). The graben preserves outliers of Ordovician to Silurian limestones of the Upper Ordovician Liskeard Group and Lower Silurian Wabi Group (Dix et al. 2007) that are downfaulted by as much as 300 m into older Shield strata (Fig. 2). Glacial Lake Barlow was located within the confines of TG between 9000 and 8000 years before present (BP), during Laurentide Ice Sheet deglaciation (Barnett 1992; Dyke 2004; Cronin et al. 2008). The exposed floor of glacial Lake Barlow is referred to as the ‘Little Clay plain’ (informal term; Fig. 2) and is underlain by glaciolacustrine silty clays (varves). Lake Timiskaming (100 km long, 200 m maximum depth) is the postglacial successor to glacial Lake Barlow; Barlow varved clays are present below the floor of Lake Timiskaming as far south as the McConnell Moraine (Fig. 2). The moraine was deposited on a topographic ridge along the Grenville Front Tectonic Zone and forms the southern termination of glacial Lake Barlow (Veillette 1994).

One thousand kilometres of high-resolution seismic data were collected on Lake Timiskaming using an EdgeTech X-STAR digital seismic profiling system (see Lazorek et al. 2006 for details). Multibeam data collected by the Canadian Hydrographic Survey (Canadian Hydrographic Survey 2004) were also used to relate bathymetry to deformation structures seen on seismic profiles. Lake Timiskaming is a large lake ─ 14 km wide in its northern part, and some 100 km in length. To the north, its linear western margin is sharply defined by the Timiskaming West Shore Fault (Figs 2, 3, 4). The central basin is bordered on its Québec side by shallow (<5 m) wave-cut ‘shelves’ and several limestone islands and intervening bays (Fig. 3).

Seismic data reveal two seismic-stratigraphic successions below the floor of the lake (Fig. 5). Parallel high-frequency reflectors typical of varved glacial Lake Barlow silty clays characterize the older succession (Shilts 1984), which is overlain by a largely transparent, reflector-free Holocene succession. Within Lake Timiskaming, the younger succession is unusually thick (40 m) compared to other glaciated ‘Shield’ lake basins (Shilts and Clague 1992; Lazorek et al. 2006) because of high sediment inputs from several large inflowing muddy rivers that rework glacial Lake Barlow sediment. The seismic survey shows that the late-glacial and postglacial sediment fill of Lake Timiskaming is extensively and severely deformed by horst and graben structures that control the topography of the lake floor (Figs. 4, 5A-E). The remarkably linear, parallel-sided, central deep-water basin of Lake Timiskaming (Fig. 3) is a trench-shaped graben resulting from subsidence between the Timiskaming West Shore Fault (TWSF) and another hitherto undiscovered fault that is named herein the Timiskaming East Shore Fault (Figs. 2, 3, 5B, 5C, 6). The latter trends north–south parallel to the TWSF through Dawson Point and the west side of Mann Island and actually consists of a zone of closely spaced faults (Fig. 5B, C). Shallow-water bays on the eastern margin of the lake, such as Baie Joanne (Fig. 3) and Baie Paulson (Fig. 4A) are the expression of small grabens containing downfaulted late-glacial and postglacial sediments, lying between limestone islands that form horsts (Fig. 6). A detailed 3-D reconstruction of faults is being presented elsewhere, but we can report sidescan records of several examples of open lake-floor fissures above faults, indicating recent or ongoing activity (Fig. 4B, C). Deformation is not restricted to a short phase of heightened glacioisostatic rebound immediately following deglaciation, as reported elsewhere in intracratonic settings (e.g. Lagerback 1992), because such structures would be buried below modern sediment. Instead, deformation has affected the modern lake floor and is ongoing.

Faulting in the northern part of Lake Timiskaming was briefly reported by Shilts (1984) and attributed to late-glacial glaciotectonic processes involving collapse over buried ice blocks, although it was noted that post-glacial sediments had also been displaced. Shilts and Clague (1992) later suggested a link to the M 6.2 Timiskaming Earthquake in 1935. Doig (1991) reported deformation structures in short drill cores and attributed the deformation to seismic shaking. Our substantial seismic coverage demonstrates a pronounced structural control on modern lake-floor topography that is consistent with earthquake-related neotectonic activity. In particular, the faulting of both post-glacial and late-glacial sediments indicates a neotectonic, not glaciotectonic origin.

Tectono-geomorphic features on the surface of the Little Clay plain appear to provide evidence for rapid subsidence during at least one post-glacial earthquake. A remarkably linear, 20 km long, 10 m high, northwest-trending, west-facing scarp between Thornloe and New Liskeard (Fig. 7) coincides with the northern extension of the Timiskaming East Shore Fault interpreted from offshore seismic profiles. Previous workers had suggested that the scarp is the surface expression of a bedrock fault (Lovell and Frey, 1976; Russell 1984) whereas others mapped it as an abandoned glacial Lake Barlow shoreline bluff (Morton et al. 1979). Beach deposits were not observed along the length of the scarp, whose striking linearity is markedly different from other shorelines on the Little Clay plain. The onshore scarp and the Timiskaming East Shore Fault likely record one or more large prehistoric earthquakes and subsidence of the central basin between the Timiskaming West and East Shore faults (Fig. 3). Subsidence of the Little Clay plain has promoted the growth of extensive peat deposits, some of which are covered by ‘run-out debris’ from mudflows that record local landsliding of glaciolacustrine silty clays down the scarp face. It is possible that these mudflows may have neotectonic significance; i.e. they may have been triggered by earthquake activity. Possible ‘sand blows’ or sand volcanoes (Tuttle 2001) where sand has been injected upward through peat, are noted (Fig. 8). Examination and coring of the buried peat stratigraphy below landslide deposits, and in sand blows, may ultimately yield a dateable record of tectono-geomorphic activity.

Evidence of earthquake-triggered sediment slumping that can be related to a specific earthquake is found on the floor of Lake Kipawa, east of Lake Timiskaming (Doughty et al. 2010; Figs 1, 9). Like Lake Timiskaming, its sediment infill consists of Barlow silty clays (maximum thickness 30 m) overlain by a thin (<5 m) succession of organic-rich Holocene sediment. High-resolution seismic profiling on Lake Kipawa (not shown) reveals some 35 distinct locations where the sediments of both successions have experienced downslope failure; these large slumps are characterized by hummocky mounds of thickened and contorted deposits. Shilts (1984) reported their presence in Lake Kipawa and also recognized them in nearby Lake Tee (Fig. 9). Detailed mapping of these landslides reveals that they occur within a broad 600 km² zone around the epicentre of the 1935 Timiskaming Earthquake (Fig. 9).

To our knowledge, the scale and type of active neotectonic deformation in the Timiskaming Graben briefly reported herein is, so far, unique in eastern North America (Kelson et al. 1996; Tuttle 2001; Talwani and Schaeffer 2001; Stein and Mazzotti 2007). The deformation is interpreted to reflect frequent seismic activity and associated crustal deformation in the Western Québec Seismic Zone. This finding is consistent with Mazzotti’s (2007) identification of the Timiskaming Graben as a persistent ‘weak zone’ within the otherwise rigid North American craton. Kumarapeli and Saul (1966) compare the Proterozoic–Phanerozoic development of the St. Lawrence Rift to the modern East African Rift, and, indeed, the graben infill of Lake Timiskaming is similar to that in other rift basins worldwide (e.g. Scholz et al. 1998; Gawthorpe and Leeder 2000; Withjack et al. 2002).

The Western Québec Seismic Zone is one of the most active seismic zones in eastern North America; large, damaging earthquakes occurred in 1732 (Montreal; M 5.8), in 1935 (Temiscaming; M 6.2) and in 1944 (Cornwall; M 5.6). The epicentre of the 1935 earthquake was located 10 km east of Lake Timiskaming near Lake Kipawa and is recorded by extensive lake floor slumping (Hodgson 1936; Bent 1996; see previous section). The 20-km-long scarp between Thornloe and New Liskeard appears to mark subsidence of the surface of the Little Clay plain along the Timiskaming East Shore Fault (Fig. 7), and points to a heightened seismic risk for the communities of New Liskeard and Hailey-bury, which lie on the same downfaulted block but just above lake level (Fig. 6). Earthquakes (M>3) occur every two years in the Timiskaming district (Adams et al. 2000). Geomorphic evidence for recurring, large (M>6) Holocene earthquakes along the Ottawa-Bonnechere Graben (Fig. 2) was identified by Aylsworth et al. (2000) and is consistent with the data reported here from the Timiskaming Graben. The most recent earthquake (M 5.2) in the Timiskaming area occurred on January 1, 2000 under Lake Kipawa, within 15 km of the 1935 epicentre. This scale of activity is not consistent with a model of seismicity driven by postglacial rebound alone (e.g. Chung 2002) but is more consistent with ongoing deformation within the North American Plate as a consequence of the interaction of plate movement and old structures (Mazzotti 2007).

The flat floor of glacial Lake Barlow (i.e. the Little Clay plain) is an ideal substrate for the rapid identification of neotectonic structures and fault offsets (e.g. Hume 1925) in a way not possible on the rough exposed bedrock of the surrounding Shield surface. It is apparent that a wide range of structurally controlled geomorphic features occur in the area and await detailed investigation (Figs. 4, 5). Given the implications of the findings presented herein for hazard assessment of the cities of Ottawa and Montréal (Motazedian and Hunter 2008; Rosset and Chouinard 2008), a program of tectono-geomorphic mapping is being conducted across the Timiskaming Graben and Little Clay plain. Sub-bottom seismic reflection profiling will also be carried out on the many other lakes in the region, to constrain the record of postglacial seismicity in the Western Québec Seismic Zone. The Timiskaming Graben is an ideal candidate for deep continental drilling. It is possible that a long, late Cenozoic sediment and possible paleoclimate record is preserved below the central graben basin of Lake Timiskaming. Deep drilling would also constrain the long-term tectonic history of the graben.