Links between sedimentary basin development and Pacific Basin plate kinematics recorded in Jurassic to Miocene strata on the western Alaska Peninsula

Convergent plate boundaries are a product of subduction-related processes that include subduction initiation, related growth of new volcanic arc systems, and arc terrane accretion in the upper plate (e.g., Kincaid and Sacks, 1997; Stern, 2004). These processes can lead to the reconfiguration of subducting oceanic plates and commonly result in the creation and expansion of continental crust and, therefore, are important for understanding the evolution of continental crust through time (e.g., Cawood et al., 2009; Draut and Clift, 2013). Sedimentary basins located along convergent margins are considered sensitive recorders of the surface responses to these types of changes in plate dynamics related to the growth of continental margins (Trop et al., 2002; Ingersoll, 2012; Surpless et al., 2014; Surpless, 2015).

Changes in Mesozoic and Cenozoic plate kinematics of the northern Pacific region have significantly modified the upper plate of the northwestern Cordilleran margin (Trop and Ridgway, 2007; Ridgway et al., 2012; Finzel et al., 2015, 2016). This region contains a record of Late Jurassic to Late Cretaceous accretion of the oceanic Wrangellia composite terrane (WCT) and related Talkeetna arc system (Trop et al., 2005; Rioux et al., 2007), the subsequent establishment of a Late Cretaceous subduction-related continental arc system (Trop, 2008), and middle Eocene Pacific plate reorganization and initiation of the Aleutian volcanic arc system (Jicha et al., 2006). Jurassic and Cretaceous plutons related to these first two events occur along the entire northwestern Cordillera, extending from northern Washington to southwestern Alaska (Armstrong, 1988; Brew, 1994; Moll-Stalcup, 1994; Plafker and Berg, 1994). In the best geochronologically constrained part of this magmatic belt, the Coast Mountains batholith in western Canada, large-magmatic-flux events occurred at 160–140 Ma, 120–78 Ma, and 55–48 Ma, separated by magmatic lulls at 140–120 Ma and 78–55 Ma (Gehrels et al., 2009). The Canadian Cordilleran arc systems ceased by the middle Eocene, and subduction and arc formation in the northern Pacific region shifted westward to southern Alaska and the Beringian margin. In western Alaska, this phase of Cenozoic magmatism is defined by the Aleutian–Alaska Peninsula volcanic arc system, which extends over 3000 km from the Kamchatka Peninsula (Russia) to the Cook Inlet of central Alaska (Vallier et al., 1994). The oceanic part of this volcanic arc system in the Aleutian segment is thought to have initiated during middle Eocene time (Jicha et al., 2006).

The shallow-crustal responses to these variations in plate kinematics and related magmatic events are best recorded in the sedimentary basins of the northwestern Cordillera. A >2000-km-long belt of isolated exposures of Jurassic and Cretaceous sedimentary strata located on the inboard side of the WCT has been studied to evaluate the timing of terrane accretion. These studies have focused primarily on southeastern and south-central Alaska (Gehrels, 2001; Hampton et al., 2007, 2010; Hults et al., 2013; Kapp and Gehrels, 1998; Manuszak et al., 2007; Ridgway et al., 2002; Trop and Ridgway, 2007) and to a lesser extent in southwestern Alaska (Kalbas et al., 2007; Wallace et al., 1989). Sedimentary studies focused on the outboard margin of the WCT have been fewer and mainly in the Matanuska and Wrangell basins in south-central Alaska (Trop, 2008; Trop and Ridgway, 2007; Trop et al., 2002, 2005; Finzel et al., 2015, 2016), and the Queen Charlotte Basin in British Columbia (Lewis et al., 1991).

Initiation of Cenozoic subduction along the Aleutian arc has primarily been investigated through the analysis of related igneous rocks on the Aleutian Islands (Fournelle et al., 1994; Jicha et al., 2006; Kay and Kay, 1994). Cenozoic strata are reported from the Aleutian Islands (Vallier et al., 1994; Wilson et al., 2015), but they have not been the focus of basin analysis studies. Rocks located on the western Alaska Peninsula constitute the farthest west exposures of the WCT strata and the Mesozoic plutons of the Cordilleran arc systems, and the easternmost part of the Cenozoic Aleutian volcanic arc system. The pioneering regional mapping and stratigraphic framework studies of Burk (1965) and Detterman et al. (1996) form the foundation for our current understanding of the structure and stratigraphy of the Alaska Peninsula.

In this study, we used the sedimentology and detrital zircon U-Pb geochronology of Late Jurassic to Miocene strata located on the western Alaska Peninsula to investigate the sedimentary record of changes in plate kinematics of the northern Pacific region. The Alaska Peninsula offers some of the best exposures for evaluating long-term sedimentary basin development along a convergent margin, with >14 km in total thickness of sedimentary and volcanic strata well exposed in this remote area. To our knowledge, this is the first analysis of the detrital zircon record of these strata within the context of measured stratigraphic sections. This new data set allows us to evaluate the effects of plate-scale changes on the depositional environments, provenance, and subsidence drivers in sedimentary basins on the western Alaska Peninsula and to provide insight into the changing configuration of sedimentary basins that might be expected over geologic time scales along convergent plate margins.

Convergence along the western margin of North America in the Middle Jurassic (ca. 172 Ma) was associated with the Farallon plate subducting to the northeast, beneath, and relative to, North America (Seton et al., 2012). Terranes of the Intermontane belt that were accreted to the North American margin by the Middle Jurassic are an amalgamation of Precambrian–Mesozoic microcontinents, continental margin sedimentary assemblages, and continental igneous belts (Foster and Keith, 1994; Gehrels, 2001; Mihalynuk et al., 1994; Monger and Nokleberg, 1996). The WCT that sits outboard of the Intermontane belt terranes constitutes the largest addition of juvenile crust to North America since Jurassic time (Wrangellia composite terrane [WCT] on Figs. 1 and 2; Coney et al., 1980; Plafker and Berg, 1994). The WCT consists of three distinct terranes: the Alexander, Wrangellia, and Peninsular terranes (Plafker and Berg, 1994). The Peninsular terrane is partly composed of an oceanic arc, the Early to Late Jurassic Talkeetna arc, which is part of an extensive belt of Jurassic arc rocks (JR on Fig. 1; Rioux et al., 2007). The Talkeetna arc was an oceanic-island arc that was either partly or wholly built upon the already combined Wrangellia-Alexander terrane (Trop et al., 2002, 2005; Rioux et al., 2010) or collided with the combined Wrangellia-Alexander terrane during the Late Jurassic prior to accretion to North America (Clift et al., 2005a, 2005b). By the Early–Late Cretaceous, the WCT had collided with the outboard margin of North America several hundred to >1000 km south of its present-day position (Garver and Davidson, 2015; Manuszak et al., 2007; McClelland et al., 1992; Pavlis, 1982; Stamatakos et al., 2001; Trop et al., 2002, 2005).

During the Early to middle Cretaceous (ca. 140–120 Ma), changes in the configuration of Pacific Basin spreading ridges and increases in the rate of seafloor spreading led to ultrafast spreading rates and a change to SE-directed subduction of the Farallon plate relative to North America (Engebretson et al., 1985; Seton et al., 2012). Widespread Early and middle Cretaceous magmatism is recorded by igneous belts of this age in western Canada and interior Alaska (EK and MK on Fig. 1). At ca. 100 Ma, another significant change in spreading directions reoriented the convergence vector between Farallon and North America back to NE-SW (Seton et al., 2012; Wessel and Kroenke, 2008), and sustained northward translation of the WCT was initiated.

During the Late Cretaceous (ca. 79 Ma), the northern part of the Farallon plate broke into the Kula plate to the west (Atwater, 1989; Lonsdale, 1988; Seton et al., 2012) and the Resurrection plate to the east (Fig. 2; Haeussler et al., 2003). This event was coincident with a shift toward N-directed plate motions relative to North America (Atwater et al., 1993; Seton et al., 2012). Subduction of the Farallon plate prior to ca. 79 Ma and then the Kula-Resurrection plates after ca. 79 Ma continued along the outboard (southern) margin of the WCT, including along the Bering Sea shelf margin, with the establishment of widespread continental magmatism (KT on Fig. 1; Plafker et al., 1994).

Between the Late Cretaceous and the middle Eocene, the modern curvature of Alaska, termed the southern Alaskan orocline, developed (Coe et al., 1985, 1989; Hillhouse and Coe, 1994; Plafker, 1987). The WCT continued to be translated northward by continental margin–parallel strike-slip faults (McClelland et al., 1992; Pavlis, 1982; Stamatakos et al., 2001). Around 55–48 Ma, the Resurrection plate was completely subducted, and the latest significant change in plate motions occurred, involving a counterclockwise shift toward NW-SE convergence between the Kula plate and North America (Fig. 2; Lonsdale, 1988; Madsen et al., 2006; Seton et al., 2012; Wessel and Kroenke, 2008). At about the same time, subduction stepped outboard from the Beringian margin to the present-day Aleutian subduction zone (Jicha et al., 2006). The crust in the Bering Sea north of the Aleutian subduction zone is interpreted as a captured remnant of the Kula plate (Scholl et al., 1986). During the late Eocene, spreading between the Kula and Pacific plates ceased (Seton et al., 2012), followed by minor shifts in Pacific plate motions during the late Oligocene (ca. 28–24 Ma), the early Miocene (17–14 Ma), and the late Miocene (ca. 6 Ma; Fig. 2; Wessel and Kroenke, 2008). Based on outcrop exposures, Cenozoic magmatism along the western end of the Alaska Peninsula has been limited to three short intervals: (1) Paleocene magmatism primarily restricted to the present-day offshore islands located southeast of the study area (Sanak-Baranof belt; KT on Fig. 1), (2) the late Eocene to early Oligocene Meshik volcanics (MT on Fig. 1), and (3) the late Miocene–Holocene Aleutian–Alaska Peninsula volcanic arc (red triangles on Fig. 1; Bradley et al., 2003; Wilson et al., 1999).

Mesozoic and Cenozoic sedimentary and volcanic strata are exposed all along the Alaska Peninsula, with a total cumulative thickness of ∼14,000 m (Detterman et al., 1996). Stratigraphic descriptions, general depositional environments, biostratigraphic ages, and overall provenance interpretations in this section are based on Detterman et al. (1996) unless otherwise noted. The Late Jurassic to Cretaceous stratigraphy of the Alaska Peninsula consists of marine and nonmarine sedimentary strata deposited on the outboard side of the WCT during and after accretion with North America. The Upper Jurassic Naknek Formation is up to 3000 m thick, but averages 1700–2000 m in thickness. This formation unconformably overlies Middle Jurassic strata and was deposited in nonmarine, shallow-marine, and deep-marine depositional environments (Fig. 3). In south-central Alaska, the sediment source region for the Naknek Formation is inferred to have been Talkeetna arc rocks, located to the northwest of the present-day exposures of the Naknek Formation. This interpretation is based on sandstone petrography, heavy minerals, paleocurrent analyses, and grain-size trends (Clift et al., 2005a; Trop et al., 2005). In south-central Alaska, the Naknek Formation is interpreted to represent forearc basin strata located on the outboard side of the WCT.

Near the study area (shown on Figs. 1 and 4), the Lower Cretaceous (Berriasian–Valanginian) Staniukovich Formation is ∼250 m thick and gradationally overlies the Naknek Formation but is unconformable elsewhere, and was deposited in shallow-marine environments (Fig. 3). The overlying Herendeen Formation is also Early Cretaceous in age (Hauterivian–Barremian), and it is conformable with the underlying Staniukovich Formation in some areas and unconformable in others. This unit is ∼270 m thick and also represents deposition in shallow-marine conditions.

The Chignik Formation typically overlies the Naknek Formation, but locally it is found above the Herendeen or Staniukovich Formations. The lower contact of the Late Cretaceous Chignik Formation (Campanian–Maastrichtian) is a major unconformity throughout the Alaska Peninsula that represents ∼45–75 m.y. of Cretaceous time (Fig. 3). The angular discordance with the underlying units is usually minor, with only a few degrees difference between bedding attitudes. This unit consists of 500–600 m of nonmarine coal-bearing strata in the study area that thin rapidly to both the northeast and southwest. The Hoodoo Formation consists of at least 600 m of marine black siltstone and shale that are time equivalent to the nonmarine Chignik Formation. The Chignik and Hoodoo Formations are thought to have been deposited adjacent to a source area northwest of the peninsula based on grain-size distributions.

Cenozoic strata on the Alaska Peninsula consist predominantly of sedimentary strata with a range of marine and nonmarine depositional environments and varying amounts of intercalated volcaniclastic and primary volcanic strata (Fig. 3). The Paleocene to Eocene Tolstoi Formation is ∼660 m thick and consists mainly of fluvial floodplain and shallow-marine deltaic sedimentary strata. The Tolstoi Formation onlaps the underlying Hoodoo, Chignik, Staniukovich, and Naknek Formations along a surface that is characterized by a yellow-orange weathering zone, which is interpreted as an erosional break. Therefore, the lower contact of this formation marks another major unconformity throughout the Alaska Peninsula. Both conglomerate and sandstone compositional data from the Tolstoi Formation are interpreted to indicate recycling of the underlying Mesozoic units (Detterman et al., 1996). Overlying the Tolstoi Formation, the Stepovak Formation consists of ∼2000 m of Upper Eocene to Lower Oligocene volcaniclastic deep-water turbidites and shallow-marine sandstone that intertongue with primary volcanic strata along the southwestern side of the peninsula. The Stepovak Formation was deposited in close proximity to the coeval Meshik arc rocks (Tm on Fig. 4). The lower contact of the Stepovak Formation is considered disconformable based on the abrupt change from nonmarine and shallow-marine deposition in the Tolstoi Formation to the deeper-marine deposition of the Stepovak Formation (Detterman et al., 1996).

Exposures of the overlying Upper Oligocene to lower Middle Miocene Unga Formation are limited to the Pacific side of the peninsula (Tu on Fig. 4); this unit consists predominately of coarse-grained volcaniclastic terrestrial strata. The middle to late Miocene Bear Lake Formation is composed of up to 2360 m of sedimentary strata that locally overlie the Stepovak and Tolstoi Formations (Finzel et al., 2009). The character of the lower contact is typically a disconformity or sometimes an angular unconformity. The Bear Lake Formation is thought to have been deposited during a pause in magmatic activity and during minor thrust faulting in the study area (Decker et al., 2008). The Pliocene Milky River Formation is 465 m thick at its type section in an onshore well. Both in outcrop and in the subsurface, the Milky River Formation is a complex mixture of primary volcanic and volcaniclastic strata that were deposited close to and north of the coeval arc rocks (Detterman et al., 1996).

Stratigraphic sections in the Port Moller area on the southwestern Alaska Peninsula (Fig. 4) were measured as part of a regional basin analysis program conducted in collaboration with the Alaska Division of Geological and Geophysical Surveys. Sections were measured only in the Cenozoic formations because that part of the stratigraphic package, and the Miocene Bear Lake Formation in particular, was the focus of this petroleum exploration–oriented program. Eight stratigraphic sections were measured in the Bear Lake Formation, in addition to one measured section each in the Tolstoi and Stepovak Formations (Decker et al., 2005; Finzel et al., 2005). Dominant lithofacies were defined on the basis of lithology, grain-size relationships, sedimentary structures, bed geometries, stratal stacking relationships, and fossil content of the strata, which were recorded on a bed-by-bed basis. Our study builds on the regional mapping and stratigraphic framework of the Alaska Peninsula defined by Detterman et al. (1996), but with a new emphasis on detailed sedimentology, detrital zircon provenance, and basin dynamics.

Our stratigraphic section through the lower part of the Paleocene–Eocene Tolstoi Formation is 402 m thick and is characterized by thick-bedded, trough cross-stratified sandstone interbedded with carbonaceous shale and coal (Tt on Figs. 4 and 5; Sc on Fig. 6A). Amalgamated sandstone beds average ∼10 m in thickness, but they locally range up to 40 m thick. Sandstone beds have tabular geometries at the scale of the outcrop. Cross-stratification in the sandstone is commonly draped by carbonaceous detritus. Fine-grained intervals in the section are up to 20 m thick (Fm on Fig. 6B) and contain coalified lenses and abundant well-preserved leaf fossils.

The Eocene–Oligocene Stepovak Formation in our 98 m measured section consists of a repetitive succession of coarse-grained sandstone interbedded with thick recessive packages of gray siltstone (Ts on Figs. 4 and 5). The sandstone beds are massive and poorly sorted, have tabular geometries, and average ∼6 m in thickness (Sm on Fig. 6C). The sandstone composition is dominated by angular volcanic lithic fragments, and sandstone beds locally contain large, outsized volcanic clasts of mainly white crystalline tuff (white outline on Fig. 6D). The siltstone-dominated lithofacies is massive, lacks sedimentary structures, and has very rare marine fossils (two mollusk fossils were found during a day of field work; Fs on Fig. 6D). Primary volcanic flows are present throughout the section; the thickest flow is recorded between 16 and 26 m in the measured section.

All eight detailed measured sections for the Bear Lake Formation were published in Finzel et al. (2009), so here we present only a summary of that work focused on the two measured sections from which detrital zircon samples were collected. The Tbl-1 section (Coal Point in Finzel et al., 2009) consists of two main lithofacies associations that are separated by a fault at ∼40 m on the measured section (Figs. 4 and 5). The lowermost 40 m section is interpreted as representing an older part of the Bear Lake Formation and is not discussed further.

The middle and upper parts of the section (40–187 m on Fig. 5) are characterized by trough cross-stratified sandstone (Sc on Fig. 6E) interbedded with coal (C on Fig. 6E) and organic-rich mudstone (Fm on Fig. 6E). Lenticular sandstone beds between 40 and 75 m on the measured section are 3 m in length and 1 m thick. Internally, they contain ripple-lamination with mud drapes on ripple foresets, and they are interbedded with ripple-laminated siltstone. Up section, the sandstone beds become thicker and are dominated by trough cross-stratification. Cross-set thicknesses are up to 70 cm, and individual foresets are draped by organic detritus (Fig. 6F). There is a distinct lack of bioturbation and macrofauna throughout the middle and upper parts of the section. Organic-rich beds include coal seams up to 70 cm thick, but smaller seams are common throughout this part of the section. Interbedded finer-grained siliciclastic facies in the upper section contain fossilized upright tree trunks that are up to 50 cm long and well-preserved leaf fossils.

The Tbl-2 Bear Lake Formation measured section (Left Head in Finzel et al., 2009) consists of three main facies associations (Fig. 5). The lower 70 m section consists of bioturbated, coarse-grained sandstone beds that are 1–2 m in height (Sb on Fig. 6G). These sandstone packages are laterally extensive at the scale of the outcrop and have tabular geometries. Internally, these packages contain inclined heterolithic strata (white arrows on Fig. 6G). In general, most of the stratification is defined by mud drapes along individual foresets. Bioturbation of mud-draped stratification resulted in wispy, discontinuous mud lenses within the sandstone beds. Pebble lags are also common in this facies and locally contain clasts 5–8 cm in diameter. Common clast types are volcanic, quartz, and chert. Coalified wood fragments are also common in this facies. Flaser-bedded siltstone and fine-grained sandstone are interbedded with the coarse-grained, bioturbated sandstone.

The middle part of the measured section (∼70–175 m on Fig. 5) is characterized by bioturbated sandstone (Sb on Fig. 6H) interbedded with thin graded conglomerate beds (Gg on Fig. 6H) that commonly contain disarticulated mussel shells. This facies association contains extensive trace fossils that qualitatively appear to have high abundance and high diversity. Skolithos and Thalassinoides trace fossils are quite common. Distinct horizons of outsized clasts occur in this part of the section, with some clasts having diameters up to 35 cm. The uppermost part of the measured section (∼175–205 m on Fig. 5) contains well-developed trough cross-stratification with thin pebble lags. Disarticulated mussel shell fragments are common. Locally, bioturbation is well developed.

Our measured section in the Paleocene–Eocene Tolstoi Formation is from the well-exposed, lower part of an ∼1400-m-thick reference section presented in Detterman et al. (1996); based on their descriptions, the lithofacies we observed are representative of the entire reference section, and therefore we apply our interpretations to the formation as a whole. The lithofacies in the Tolstoi Formation are interpreted as nonmarine fluvial deposits. The dominance of trough cross-stratification in the thick sandstones and the nonerosive tabular geometries of the sandstone beds are indicative of the medial part of a distributary fluvial system on a delta plain with laterally mobile bed-load streams (e.g., Graham, 1983; Sadler and Kelly, 1993). Thick sections of carbonaceous shale and coals are characteristic of associated extensive overbank environments (e.g., Nichols and Fisher, 2007). The thick finer-grained intervals with coal lenses imply that fluvial channels were stable for relatively long periods of time to allow for peat development.

We interpret the Stepovak Formation at our measured section locality as a submarine fan-delta deposit, based on the poor sorting, angular grains, lack of sedimentary structures, tabular geometries, and scarcity of fossils (e.g., Busby-Spera and Keller, 1985; Busby-Spera and Boles, 1986; Hickey, 1984; Surlyk, 1984; White and BusbySpera, 1987). The lack of tractional sedimentary structures and internal bedding in the sandstones is consistent with mass-flow depositional processes. All of these features are diagnostic of rapid emplacement of sediment in relatively deep-water marine conditions below fair weather wave base. The presence of large, outsized volcanic clasts and the interbedded lava flows suggest proximity to a nearby volcanic source and indicate that these depositional systems were part of the volcaniclastic apron that surrounded subaerial volcanic centers (e.g., White and BusbySpera, 1987). We interpret the strata in this section to have been deposited by sediment gravity flows on submarine fan systems adjacent to an active volcanic arc.

The Tbl-1 section of the Miocene Bear Lake Formation located above the fault at ∼40 m on Figure 5 consists of Facies Association 1 of Finzel et al. (2009), which is interpreted as nonmarine fluvial and swamp deposits. The stacked trough cross-stratified sandstone beds are the product of migrating dunes in shallow fluvial channels (e.g., Friend et al., 1979; Harms et al., 1982; Muñoz et al., 1992). The upward-fining packages represent gradual abandonment of these fluvial channels. The tabular beds of massive and ripple-laminated siltstone represent floodplain environments. In situ tree fossils are consistent with periodic overbank deposition that allowed trees to continue growing after each major flooding event.

The Tbl-2 section of the Bear Lake Formation is composed of Facies Association 3 of Finzel et al. (2009), which is interpreted as intertidal channels and mud flats. The inclined heterolithic strata are inferred to represent lateral accretion deposits produced by point bars in tidal channels (e.g., Gingras et al., 1999; Shanley et al., 1992; Thomas et al., 1987). The flaser- and wavy-bedded sandstone and siltstone indicate that sand- and mud-rich tidal flats were adjacent to the tidal channels (e.g., Shanley et al., 1992). The mussel beds represent areas where mollusks lined the sandy substrate of tidal channels. The monotaxic beds, as well as the types of trace fossils present, are consistent with lower intertidal to very shallow subtidal water depths (Allison and Addicott, 1976). The thin beds of graded conglomerate represent deposition by major storms when coarse detritus was spread across the tidal flat system (e.g., Clifton, 1983).

During Mesozoic time, strata of the Upper Jurassic Naknek, Lower Cretaceous Staniukovich and Herendeen, and Upper Cretaceous Chignik Formations are all interpreted to represent deposition in nonmarine and shallow-marine conditions. Most of these units have coeval deeper-water marine facies that illustrate the persistence of marine deposition along parts of the Alaska Peninsula (Detterman et al., 1996). By the Eocene, however, the Tolstoi Formation in our study area, as well as along the rest of the peninsula, represents dominantly nonmarine depositional environments. An abrupt change to shelfal conditions is recorded in the strata of the Stepovak Formation during the Oligocene, followed by the resumption of nonmarine and shallow-marine environments of the Bear Lake Formation in the Miocene.

One detrital zircon sample was analyzed from the Naknek, Chignik, Tolstoi, and Stepovak Formations, and two detrital zircon samples were analyzed from the Bear Lake Formation. Zircons were separated using standard methods (crushing, water table, sieving, magnetic separation, and heavy liquid separation) and mounting protocols at the University of Iowa. U-Pb analyses of detrital zircons were conducted by laser-ablation–multicollector–inductively coupled plasma–mass spectrometry (LA-MC-ICP-MS) at the Arizona LaserChron Center. The analytical data are reported in Data Repository Table S1.¹ Interpreted ages are based on ²⁰⁶Pb/²³⁸U for grains younger than 900 Ma and on ²⁰⁶Pb/²⁰⁷Pb for grains older than 900 Ma. Analyses for all grains that have >10% uncertainty and for grains that are older than 400 Ma and are >20% discordant or >5% reverse discordant (by comparison of ²⁰⁶Pb/²³⁸U and ²⁰⁶Pb/²⁰⁷Pb ages) are not included in the results and interpretation.

Detrital zircon results for each sample are described here and shown in Figure 7. The detrital zircon signature of the Naknek Formation (n = 302) contains 99% Mesozoic grains and 1% Paleozoic and Precambrian grains. Zircon ages range from 216 to 141 Ma, with the majority between 170 and 150 Ma (62%), as well as one Paleozoic grain with an age of 347 ± 17 Ma (1σ error) and two Precambrian grains with ages of 1643 ± 18 and 1649 ± 25 Ma. The Chignik Formation (n = 312) contains 92% Mesozoic grains and 8% Paleozoic and Precambrian grains. The Mesozoic grains fall into ranges of 103–80 Ma and 231–136 Ma. Significant Precambrian age peaks with three or more grains that have overlapping ages within their 1σ error occur at 1646 Ma, 1430 Ma, and 1143 Ma. Other Precambrian single-grain ages include two at 1782 ± 38 Ma and 1783 ± 14 Ma, and one each at 2054 ± 18 Ma and 2925 ± 14 Ma.

The Tolstoi Formation (n = 297) contains 7% Cenozoic, 92% Mesozoic, and 1% Paleozoic and Precambrian grains. The age ranges in this sample include 52–47 Ma, 103–61 Ma, and 219–126 Ma. The detrital zircon signature of the Stepovak Formation (n = 148) consists of 44% Cenozoic, 51% Mesozoic, and 5% Paleozoic and Precambrian grains, with age ranges of 40–35 Ma, 100–44 Ma, and 211–142 Ma. One Bear Lake Formation sample from the Tbl-1 locality (n = 312) has 21% Cenozoic, 66% Mesozoic, and 13% Paleozoic and Precambrian grains. The age ranges in this sample are 24–21 Ma, 224–46 Ma, and 369–278 Ma. Significant Paleozoic age peaks with three or more grains that have overlapping ages within their 1σ error occur at 318 Ma and 342 Ma. Precambrian single-grain ages include 1632 ± 17, 1804 ± 22, 1849 ± 25, 1978 ± 18, 2295 ± 16, 2567 ± 16, 2685 ± 19, and 2806 ± 26 Ma. The other Bear Lake Formation sample from the Tbl-2 locality (n = 290) consists of 99% Mesozoic and 1% Paleozoic and Precambrian grains, with age ranges of 118–63 and 225–136 Ma.

In summary, two Mesozoic age populations dominate our new detrital zircon data (Fig. 8). The older Middle–Late Jurassic (ca. 170–150 Ma) population encompasses ∼30% of the zircon ages in the entire data set. The younger middle Cretaceous (ca. 100–80 Ma) population encompasses ∼21% of the total zircon ages. The remaining grains mostly fall into age groups of ca. 210–170 Ma (18%) or ca. 70–50 Ma (8%). Only ∼5% of the total data set are Paleozoic or Precambrian in age, and ∼3% are younger than 50 Ma.

The detrital zircon ages from our study are the first reported from Mesozoic and Cenozoic strata of the western Alaska Peninsula, and they provide an opportunity to better constrain the depositional ages of these formations. In this section, we summarize all available biostratigraphic data integrated with maximum depositional ages (e.g., Dickinson and Gehrels, 2009) from our detrital zircon data for strata at the four measured sections (Tt, Ts, Tbl-1, Tbl-2 on Fig. 4) and the two additional detrital zircon sample locations (Jn and Kc on Fig. 4). The maximum depositional ages were first calculated in Isoplot (Ludwig, 2008) as the mean weighted age of the youngest consecutive age population (n ≥ 3), which consisted of a group of ages that were within their 1σ error of their nearest neighbor in sequence (Table 1). If the resulting mean squares of weighted deviates (MSWD) is close to 1, it suggests that the scatter of ages within the group of grains selected is appropriate based on the precision associated with individual age measurements, and, therefore, the zircons in the selected group are all the same actual age. If the MSWD is much less than or much greater than 1, then either analytical errors have been underestimated, or the scatter is unexplained by the analytical error. For samples where the MSWD is not close to 1, the UNMIX routine in Isoplot was used to model the individual age peaks that comprise the youngest consecutive age population (n ≥ 6; Table 1). The youngest modeled peak age and its 2σ error are reported, along with the relative misfit parameter (as defined by Sambridge and Compston, 1994). In addition, the youngest single-grain age is also reported. The final interpreted depositional age for each sample (Fig. 3) is based on the integration of our preferred maximum depositional age based on zircons (Table 1) with available biostratigraphic information, which is discussed below. The geologic time scale used in this discussion is from Walker et al. (2012).

The detrital zircon sample at the Jn locality is from the Northeast Creek Sandstone Member of the Naknek Formation (Wilson et al., 1999). Based on the scarce presence of the bivalve Buchia concentrica in the upper part of the member elsewhere on the peninsula, the depositional age for these strata was inferred to be middle to late Oxfordian (ca. 162–157 Ma; Fig. 3; Detterman et al., 1996). The youngest peak deconvolved from the youngest consecutive age population (n = 298) present in the detrital zircon signature using UNMIX, however, is 155.7 ± 0.7 Ma, with a misfit of 0.19 (Table 1). Therefore, we interpret the maximum depositional age of the Jn sample to most likely be Kimmeridgian (157–152 Ma).

The Chignik Formation is considered to be late Campanian to early Maastrichtian (ca. 80–70 Ma) based on marine fossils, including the bivalves Inoceramus balticus var. kunimiensis and Inoceramus schmidti and the ammonite Canadoceras newberryanum (Fig. 3; Detterman et al., 1996). The maximum depositional age from our detrital zircon data support this interpretation. The youngest peak deconvolved from the youngest consecutive age population (n = 19) using UNMIX is 82.8 ± 1.6 Ma (early Campanian), with a misfit of 0.41 (Table 1).

The Tolstoi Formation at the reference section, which is also the location of our measured section, is considered to be late Paleocene to middle Eocene (ca. 59–41 Ma), based on an abundant collection of megaflora, including Acerarcticum Heer, Alnus sp., Cocculus flobella (Newberry) Wolfe, Grewiopsis curiculaecrodatus (Hollick) Wolfe, Liquidambar sp., Metasequoia occidentalis (Newberry) Channey, Protophyllum semotum Hickey, and Thuites interruptus (Newberry) Bell (Fig. 3; Detterman et al., 1996; Wolfe, 1981). The youngest peak deconvolved from the youngest consecutive age population (n = 6) using UNMIX is 46.3 ± 2.3 Ma, with a misfit of 0.82 (Table 1). Therefore, we interpret the depositional age of the Tt sample to be middle Eocene (ca. 46–41 Ma).

The detrital zircon sample at the Ts locality is from the upper sandstone member of the Stepovak Formation. At the type locality, which is located near our measured section, megafossils, mainly bivalves and gastropods, from the upper sandstone member are correlated with Eocene or early Oligocene (ca. 41–28 Ma) forms found elsewhere in the North Pacific (Fig. 3; Detterman et al., 1996). The maximum depositional age for the detrital zircon sample is 37.6 ± 0.8 Ma based on the youngest peak deconvolved from the youngest consecutive age population (n = 7) using UNMIX (Table 1). Consequently, the depositional age of the Ts sample is restricted to late Eocene–early Oligocene (ca. 37–28 Ma).

The results of macrofossil paleontological and microfossil palynological analyses for the Bear Lake Formation measured sections were published in Finzel et al. (2009). At the Tbl-1 locality (Coal Point inFinzel et al., 2009), which is in the basal part of the formation, fragments of Metasequoia glyptostroboides and Carpinus cf. C. cappsensis were collected and correlate well with the Seldovian stage (early to middle Miocene; ca. 23–11 Ma) of Wolfe et al. (1966). Five of the six palynology samples from the Tbl-1 section yielded 30%–60% Taxodiaceae-Cupressaceae-Taxaceae (T-C-T) palynomorphs, including some Metasequoia-type grains, as well as probable Pterocarya and Acer grains. This assemblage is indicative of deposition before the temperature decline that began at ca. 15 Ma in the early middle Miocene. There is only one Cenozoic detrital zircon grain in the Tbl-1 sample distribution, and it is much older than the inferred depositional age of the unit; consequently, the depositional age for this sample is based on the biostratigraphy and is inferred to be early Miocene (23–16 Ma; Fig. 3).

At the Tbl-2 locality (Left Head inFinzel et al., 2009), which is in the upper part of the formation, no macrofossils were recovered; however, five palynology samples contain relatively few T-C-T grains, and the Pinaceae dominates this section, especially Tsuga spp., along with the filicales. Angiosperms are represented by Alnus, Betula, Ericales, and Triporopollenites. This palynological assemblage is likely correlative with the early late Miocene Homerian stage of Wolfe et al. (1966) and indicates a late middle to early late Miocene (ca. 14–7 Ma) depositional age for the strata. The detrital zircon signature from the Tbl-2 locality does contain Cenozoic grains, but the mean weighted age of the youngest peak (n = 4) provides a maximum depositional age of 22.8 ± 1.6 Ma, with a MSWD of 1.7 (Table 1), which is older than the inferred depositional age. As a result, the depositional age of the Tbl-2 sample is inferred to be late middle to early late Miocene (ca. 14–7 Ma) based on the biostratigraphy (Fig. 3).

According to previous studies conducted in south-central Alaska, the Naknek Formation was deposited in a forearc basin located on the outboard side of the Jurassic Talkeetna oceanic-island arc during accretion to either the Wrangellia-Alexander terrane (Clift et al., 2005a, 2005b) or the western margin of North America (Trop et al., 2005). The main source of sediment for the Naknek Formation on the Alaska Peninsula and in south-central Alaska is inferred to be the adjacent Talkeetna arc rocks, based on clast composition and age of conglomerate, grain-size patterns, detrital clay composition, and sandstone petrography (Clift et al., 2005a; Detterman et al., 1996; Trop et al., 2005).

The detrital zircon signature of the Naknek Formation at the Jn locality on the western Alaska Peninsula contains Mesozoic ages that range from 216 to 141 Ma (Fig. 7). Exposures of the Talkeetna arc in south-central Alaska have U-Pb ages that range from 202 to 181 Ma in the Chugach Mountains and 183–153 Ma in the Talkeetna Mountains and easternmost Alaska Peninsula (Fig. 1; Rioux et al., 2010). The area to the northwest of the Jn locality today is the marine continental shelf submerged beneath the Bering Sea. Dredge and subsurface data in the St. George basin (SGB on Fig. 1) and elsewhere on the shelf provide evidence that igneous rocks of the accreted Talkeetna arc exist beneath the currently submerged shelf (Worrall, 1991). Furthermore, grain-size trends in the Naknek Formation near the study area suggest both close proximity to the sediment source and that the source region was located generally to the northwest of the outcrop belt (Detterman et al., 1996). Therefore, we infer that the sole sediment source for the Jn sample was the Talkeetna oceanic-island arc rocks located offshore beneath the Bering Sea shelf today. This implies that the Naknek Formation on the Alaska Peninsula, similar to south-central Alaska, also represents deposition in a forearc basin positioned on the outboard side of the WCT (Fig. 9A).

The Late Cretaceous Chignik Formation is thickest in the study area and is interpreted as nonmarine, but the formation thins rapidly and quickly transitions to marine facies of the Hoodoo Formation both to the northeast and southwest (Detterman et al., 1996). Based on grain-size, facies, and thickness trends, the primary sediment source in the study area is inferred to have been located to the northwest (Detterman et al., 1996).

The detrital zircon signature from the Kc locality contains a dominant age peak of 231–136 Ma, which suggests that Talkeetna arc rocks were still a primary source of sediment during the Late Cretaceous. A younger 103–80 Ma peak, however, indicates the presence of a middle Cretaceous volcanic arc also supplying sediment to the basin during deposition. These ages match well with a sparse but widespread suite of Cretaceous plutons (ca. 110–80 Ma) that extends from the southern Brooks Range to the Alaska Range in interior Alaska and continues south into western Canada (MK on Fig. 1; Plafker and Berg, 1994). This magmatic belt is inferred to be mainly related to shallow subduction beneath a continental-margin volcanic arc, although crustal melting may have also played a role (Crawford et al., 1987; Pavlis et al., 1993; Hudson, 1994; Miller and Richter, 1994; Plafker and Berg, 1994; Hart et al., 2004).

Plutonic rocks of this middle Cretaceous age, however, have not been reported within ∼500 km of the study area. Furthermore, based on grain-size distributions, the sediment source for the Chignik Formation is inferred to have been adjacent to and northwest of the peninsula (Detterman et al., 1996). Therefore, we infer one of two source options for this population. Middle Cretaceous plutonic rocks could be situated beneath the Bering Sea shelf. Subduction along the shelf margin is inferred to have occurred until the Paleocene (Plafker and Berg, 1994), which means related arc rocks could be submerged beneath the shelf today. Alternatively, given the Late Cretaceous to Eocene translational history of the WCT (Butler et al., 2001; Cowan et al., 1997), another possible source could be the middle Cretaceous igneous belts in interior Alaska and western Canada. Tectonic models suggest that the terrane may have been transported ∼1000 km to up to ∼3000 km along the western margin of North America since the middle Cretaceous. From paleomagnetic studies, volcanic tuffs in the Upper Cretaceous MacColl Ridge Formation in the Wrangell Mountains forearc basin of south-central Alaska are interpreted as having undergone ∼1000 km of tectonic transport along major margin-parallel strike-slip faults (Stamatakos et al., 2001). These strata are age equivalent to the Chignik Formation, and we infer a similar amount of tectonic displacement for both parts of the Mesozoic forearc basin. The inferred translation aligns the study area with south-central Alaska, in a position that could easily receive detritus from the inboard middle Cretaceous igneous belts located there. Furthermore, Cretaceous strata in basins inboard of the WCT and exposed in south-central and southwestern Alaska today also have middle Cretaceous detrital zircon populations (Fig. 8; Amato et al., 2007; Bradley et al., 2009; Hampton et al., 2010; Hults et al., 2013; Kalbas et al., 2007). Either way, the presence of an inboard continental-margin arc implies that the Chignik Formation represents continued deposition in a forearc basin located on the outboard margin of the accreted WCT (Fig. 9B). The middle Cretaceous population in the Chignik Formation is relatively small, which may reflect the geographically sparse and widespread nature of the Cretaceous magmatic belt. The minor increase in the amount of Paleozoic and Precambrian grains between the Naknek Formation (1%) and the Chignik Formation (∼8%) suggests that a small amount of detritus was being derived from either the older inboard terranes or was being recycled from Jurassic–Cretaceous retroarc strata caught in the collision zone between the WCT and North America (e.g., Trop et al., 2002; Manuszak et al., 2007).

Deposition of our Tolstoi Formation sample is interpreted as occurring in the middle Eocene (ca. 46–41 Ma). This unit is dominantly nonmarine in character, both in the study area and along most of the Alaska Peninsula, but a limited extent of the formation located to the southwest is interpreted as shallow marine (Detterman et al., 1996; Turner et al., 1988). Microfossils from the Tolstoi Formation in the North Aleutian COST well, located northwest of the study area, suggest a nonmarine depositional environment that extended at least from our Tt locality to the offshore well location (Fig. 4; Turner et al., 1988). Farther north, Campanian to middle Eocene strata form a thin layer across the entire Bering Sea shelf (Worrall, 1991). Based on subsurface and seismic data, the Bering Sea shelf is interpreted as being an exposed highland during the Campanian to middle Eocene (Worrall, 1991). Furthermore, the overall sand:shale ratio in the Tolstoi Formation on the Alaska Peninsula increases from west to east (Detterman et al., 1996). All of these patterns suggest a sediment source region located to the northeast of the study area during deposition of the Tolstoi Formation (Fig. 9C).

Framework grains in sandstone of the Tolstoi Formation are dominated by lithic detritus, with a subordinate component of volcanic detritus, which has been interpreted to represent dominantly recycled sediment input from the underlying Mesozoic units (Detterman et al., 1996). The detrital zircon signature of the Tolstoi Formation at the Tt locality is compatible with reworking of the underlying Mesozoic units (Fig. 7). The prominent Talkeetna arc peak found in the underlying Jn and Kc samples is subdued in the Tt age distribution, but it is still present. In addition, an Early Cretaceous population of grains suggests that magmatism to the northeast of the study area may have extended into Early Cretaceous time.

The large population of middle Cretaceous grains (103–78 Ma) in the Tt sample matches well with the middle Cretaceous population found in the underlying Chignik Formation and could represent recycling from that unit. In addition, as previously discussed, middle Cretaceous plutonic rocks may be present beneath the Bering Sea shelf today. The youngest detrital zircon population in the Tt sample ranges from 52 to 47 Ma. These ages overlap with the young end of widespread Late Cretaceous to Tertiary igneous belts in south-central and interior Alaska (KT on Fig. 1). Based on aeromagnetic anomalies on the Bering Sea shelf and contemporaneous rocks located on St. Matthew Island (Fig. 1), this belt may extend further to the south along the northern side of the Alaska Peninsula and continue northwest along the continental shelf margin (Moll-Stalcup, 1994). In addition, Eocene volcanics encountered in subsurface drilling in the St. George basin and radiometric ages of basement rocks on St. George Island suggest the presence of early Cenozoic volcanic centers on the Bering Sea shelf (Fig. 1; Hopkins and Silberman, 1978; Worrall, 1991). If a volcanic arc was active inboard of the study area during the middle Eocene, then the Tolstoi Formation represents deposition in a forearc basin. The modern arc that is situated outboard of the study area, however, is inferred to have initiated during the early Eocene in the Aleutians (Jicha et al., 2006), and the oldest arc-related igneous rocks on the Alaska Peninsula near the study area are the middle Eocene to Oligocene Meshik arc (Wilson et al., 1991). Therefore, the Tolstoi Formation was potentially deposited in a basin undergoing a transition from a forearc to a backarc position (Fig. 9C).

The Stepovak Formation is a volcaniclastic unit that is interbedded with coeval andesitic and basaltic volcanic rocks of the Meshik (ca. 43–25 Ma) volcanic arc. The lower part of the Stepovak Formation is the siltstone member, which is interpreted to represent deep-water turbidite deposits (Detterman et al., 1996). The upper part of the formation, where our measured section is located, is the sandstone member, which is interpreted as being deposited by submarine fan processes on the volcaniclastic apron that surrounded subaerial volcanic centers (e.g., White and BusbySpera, 1987). The sandstones in the upper member are almost entirely composed of fresh-appearing and unaltered volcanic grains; this has been interpreted to signify a change in the composition of the sediment source from older Mesozoic igneous suites and recycled strata inferred for the Tolstoi Formation to input from the coeval volcanic arc located to the south during Stepovak time (Detterman et al., 1996; Helmold et al., 2008). In the subsurface, the Stepovak Formation is thickest to the south near the Alaska Peninsula and thins to the north (Finzel et al., 2005; Worrall, 1991). This pattern suggests that the sediment source was located to the south (Fig. 9D). The position of a volcanic arc in an outboard position during deposition of the Stepovak Formation means that the study area had fully transitioned to a backarc basin by the late Eocene.

The detrital zircon signature from the Ts locality contains ∼95% pre–Meshik-arc-aged grains and is therefore in apparent disagreement with previous published provenance interpretations. Populations of Late Triassic to Early Cretaceous (211–142 Ma) and middle Cretaceous to middle Eocene (100–44 Ma) grains are interpreted here as indicating a significant proportion of recycled sediment derived from the underlying Naknek, Chignik, and Tolstoi Formations (Fig. 7). Sandstone petrographic studies demonstrate the minor presence of sedimentary and metamorphic rock fragments in Stepovak Formation sandstones (Helmold et al., 2008), lending support to a partially recycled provenance signature. A younger population of grains (ca. 40–35 Ma) that contains ages coeval with Meshik volcanism comprises only ∼5% of the sample. The conflicting sandstone compositions and detrital zircon signature can be reconciled by considering zircon abundance. Abundance may be low in volcanic rocks of the Meshik arc because they are dominantly intermediate to mafic in composition (e.g., Detterman et al., 1996; Wilson, 1985). Therefore, even though the arc rocks may have supplied much of the sediment to the Stepovak Formation, that signature would not be strongly expressed in the detrital zircon ages.

The detrital zircon signature from the Miocene Bear Lake Formation suggests continued principal input from older sedimentary units and a small component of sediment derived directly from Cenozoic igneous suites. At the Tbl-1 locality, the Late Triassic to Early Cretaceous (225–136 Ma) and Late Cretaceous to Paleocene (118–63 Ma) populations match well with those found in the older Naknek, Chignik, Tolstoi, and Stepovak Formations (Fig. 7). We infer that sediment derived from these units was the dominant source for the Bear Lake Formation at the Tbl-1 locality (Fig. 9E).

The detrital zircon signature at the Tbl-2 locality also contains similar populations, which we infer to reflect sediment recycling from older stratigraphic units, including Late Triassic to Eocene ages (224–46 Ma). In addition, this sample also contains a Late Devonian to Early Permian population. While not a significant presence in any other sample, all of our detrital zircon samples from the older stratigraphic units do contain at least a few grains that fall into this range. Therefore, this population in the Tbl-2 sample also likely represents a recycled component from the pre-Miocene strata. The youngest age group in the Tbl-2 sample comprises only ∼1% of the detrital zircon signature, is late Oligocene to early Miocene (24–21 Ma), and is interpreted to represent sediment input directly from Meshik arc rocks exposed nearby (Tm on Fig. 4).

In the study area, a broad, doubly plunging anticline that is associated with dextral transpression along the David River zone exposes the older Mesozoic units at the surface and is interpreted to have formed coeval with deposition of the Bear Lake strata (Fig. 4; Decker et al., 2008; Sralla and Blodgett, 2007). Furthermore, the uppermost part of the Bear Lake Formation is characterized onshore by a complex series of angular unconformities between the locally steeply dipping and tightly folded Bear Lake Formation and the overlying gently dipping Milky River Formation (Decker et al., 2005; Finzel et al., 2009). Publicly available seismic data suggest that this unconformity may be a regional feature, but its characteristics change offshore. There, gently dipping reflectors of the Bear Lake Formation are locally truncated with subtle discordance by an interpreted erosional surface onto which strong reflectors, representing clinoforms in the Milky River Formation, prograde northwestward. The position of the study area inboard of a volcanic arc system and the presence of the David River zone suggest that the Bear Lake Formation represents deposition in a backarc basin within a transpressional stress regime (Fig. 9E; e.g., Ingersoll, 2012).

During the Late Jurassic and the Late Cretaceous, the Naknek and Chignik Formations were deposited in a forearc basin on the outboard side of the WCT. The dominant sediment sources for these units were located to the north of the basin and included Jurassic Talkeetna arc and middle Cretaceous igneous rocks. During the late Paleocene–middle Eocene, the basin transitioned from a forearc to a backarc position, with recycled sediment derived from the underlying Mesozoic units coming primarily from the northeast. By the late Eocene, the study area had fully transitioned to a backarc basin, with sediment of the Stepovak Formation being derived from the south. In addition to the coeval Meshik arc, structures associated with transpressional deformation in the study area during late Miocene time provided an additional sediment source in the form of uplifted and exhumed Mesozoic and early Cenozoic strata.

Collision of the WCT with North America has been inferred to have initiated between the Middle Jurassic and the Late Jurassic along different parts of the Cordilleran margin (McClelland and Gehrels, 1990; McClelland et al., 1992; Lewis et al., 1991; Thompson et al., 1991; Trop et al., 2002, 2005). Based on the integration of stratigraphic studies in the Queen Charlotte Basin in British Columbia (Lewis et al., 1991; Thompson et al., 1991) with the Wrangell basin and Matanuska Valley in south-central Alaska, as well as structural relationships along the margin, previous studies have inferred that the collision occurred in a progressive east-to-west (present coordinates) closing pattern beginning with present-day British Columbia (Wallace et al., 1989; Ridgway et al., 2002; Trop et al., 2002, 2005). Based on stratigraphic relationships in the study area, we suggest that this pattern continues along strike to the western Alaska Peninsula.

Near the easternmost exposures of the WCT in the Queen Charlotte Basin in British Columbia, terrane accretion is inferred to have initiated at the close of the Middle Jurassic (Lewis et al., 1991; Thompson et al., 1991). Accretion in this region is manifested by the development of a regional unconformity that marks the end of prolonged stable shelf sedimentation represented by the Maude Group (Fig. 10) and that extends for most of the Late Jurassic. Coeval shortening in the region is inferred to be the most significant deformational event for all of Mesozoic or Cenozoic time (Lewis et al., 1991). Stable marine depositional conditions did not return until the latest Jurassic with deposition of the Longarm Formation, resulting in an ∼25 m.y. period during the Middle and Late Jurassic that was characterized by nondeposition or erosion, with limited sediment accumulation occurring only locally in the Yakoun and Moresby Groups.

To the west in south-central Alaska, the stratigraphic records contained in the Wrangell basin, Matanuska Valley, and Cook Inlet basin reveal a slightly younger accretionary history (Trop et al., 2002, 2005; LePain et al., 2013). Each region has Middle Jurassic strata that record relatively stable shallow- to deep-marine depositional conditions, including the Nizina Mountain and lower Root Glacier Formations in the Wrangell basin, and the Tuxedni and Chinitna Formations in the Matanuska Valley and Cook Inlet basin. During the Late Jurassic, however, regional shortening and synorogenic sedimentation resulted in deposition of the coarse-grained upper Root Glacier and Naknek Formations (Trop et al., 2002, 2005). These units are inferred to record initiation of WCT accretion to North America or collision of the Peninsular terrane with the combined Wrangellia-Alexander terrane. Regional unconformity development along with relatively thin, mostly localized sedimentation ensued in the Late Jurassic and continued for ∼30 m.y. until the late Early Cretaceous, when prolonged shallow- and deep-marine depositional systems were once again established in the Kennicott and Matanuska Formations and unnamed equivalent strata in the south-central Alaskan basins (Fig. 10).

On the western Alaska Peninsula, the Mesozoic stratigraphy suggests a more prolonged accretionary history. Following Middle Jurassic shallow-marine sedimentation represented by the Kialagvik and Shelikof Formations, development of a brief erosional unconformity was succeeded by deposition of the shallow-marine Naknek Formation. Unlike other regions in Alaska, however, there is no evidence for regional Late Jurassic shortening in the study area, and widespread shallow-marine sedimentation continued into the Early Cretaceous with deposition of the Staniukovich and Herendeen Formations. Subsequently, a regional unconformity developed in the late Early Cretaceous that extends until deposition of the Chignik Formation began in the latest Cretaceous. This unconformity spans ∼45–65 m.y. and is characterized by a slight angular discordance between the Upper Cretaceous Chignik Formation and the underlying Upper Jurassic and Lower Cretaceous units.

To summarize, the stratigraphic records contained in basins from all along the WCT suggest that the timing of initiation, duration of unconformity development, and magnitude of deformation associated with terrane accretion are time transgressive. In general, accretion initiated earlier in the east during the Middle Jurassic and later in the west during the Late Jurassic. The duration of unconformity development increased from east to west, spanning ∼25–30 m.y. in British Columbia and south-central Alaska, but ∼45–65 m.y. on the western Alaska Peninsula. The magnitude of deformation associated with terrane accretion as recorded in the forearc region, however, decreased in general from east to west. Basins positioned between the inboard margin of the WCT and the outboard margin of North America also contain evidence for a general westward-younging and westward-softening pattern of final suturing from Aptian to Campanian time (Nokleberg et al., 1992; Ridgway et al., 1997, 2002; Eastham and Ridgway, 2002; Trop et al., 2004; Davidson and McPhillips, 2007; Hampton et al., 2007; Kalbas et al., 2007; Manuszak et al., 2007).

In the case of the western Alaska Peninsula and south-central Alaska forearc region, the end of collision of the WCT was followed by the establishment of a middle to Late Cretaceous continental arc system. The sedimentary record of this new arc system on the Alaska Peninsula is marked by the introduction of 80–103 Ma detrital zircons as documented in the Chignik Formation (Fig. 7).

Evidence for Paleocene–Eocene spreading ridge subduction in the northwestern Cordillera was first documented by near-trench, slab-window igneous rocks in the accretionary prism and high-temperature/low-pressure metamorphism of accretionary prism strata (Bradley et al., 2003; Haeussler et al., 2003; Sisson et al., 2003). Haeussler et al. (2003) proposed the presence of two subducting spreading ridges in the northern Pacific during Paleocene–Eocene time: one that began in western Alaska and swept eastward, and one that remained relatively stationary at ∼48°N (Fig. 2). The study area on the western Alaska Peninsula sits directly inboard from the westernmost extent of the near-trench plutons found in the Sanak-Baranof belt, which are interpreted to record migration of a slab window associated with spreading ridge subduction (Fig. 1; Bradley et al., 2003). Eastward migration of the northern ridge was inferred to have produced the eastward-younging pattern documented in the Sanak-Baranof belt (ca. 62–50 Ma). Igneous rocks in the forearc region of south-central Alaska have also been inferred to be related to slab-window magmatism associated with spreading ridge subduction (Cole et al., 2006).

More recently, a stratigraphic and provenance record of spreading ridge subduction has been inferred from the forearc basins of south-central Alaska in the Matanuska Valley (Kortyna et al., 2014) and Cook Inlet basin (Finzel et al., 2015, 2016). In those regions, the stratigraphic record of spreading ridge subduction is represented by regional unconformity development that coincides with a transition from dominantly marine to nonmarine depositional systems. For example, in both the Matanuska Valley and Cook Inlet basin, widespread latest Early Cretaceous to latest Late Cretaceous sedimentation is epitomized by deposition of the Matanuska Formation and unnamed equivalents (Fig. 10). These units are inferred to represent predominantly shelf and slope depositional environments (Grantz, 1964; Trop, 2008). These strata are overlain by a regional unconformity that underlies dominantly nonmarine early Cenozoic strata.

On the western Alaska Peninsula, the Late Cretaceous Chignik Formation is nonmarine in the study area, but it thins and becomes marine in character to both the northeast and southwest. The coeval Hoodoo Formation is inferred to represent only marine deposition. These units are separated from the overlying Tolstoi Formation by a regional unconformity (Fig. 10). The Tolstoi Formation onlaps the underlying Hoodoo, Chignik, Staniukovich, and Naknek Formations along a surface that is interpreted as a distinct erosional break. Previous clast compositional analyses of conglomerate and sandstone petrographic data (Detterman et al., 1996), as well as our new detrital zircon data from the Tolstoi Formation, are interpreted to indicate recycling of the underlying Mesozoic units. The Tolstoi Formation is inferred to represent deposition in fluvial systems that were widespread across the entire peninsula.

In summary, characteristics that are consistent with the same type of depositional and deformational history inferred to be associated with subduction of a spreading ridge in south-central Alaska, including an abrupt transition from dominantly marine to nonmarine depositional systems and development of an angular unconformity (Kortyna et al., 2014; Ridgway et al., 2012; Finzel et al., 2015, 2016), are also present in early Cenozoic strata on the western Alaska Peninsula. We suggest that this similarity could imply an analogous tectonic history that involved spreading ridge subduction along the western Alaska Peninsula.

Subduction along the Aleutian Island portion of the present-day southern Alaska subduction zone is interpreted to have initiated ca. 50–55 Ma (Jicha et al., 2006). Bond et al. (1988) suggested that crustal thickening along the edge of the North American plate along the Alaska Peninsula–Aleutian arc due to relative plate motions might have created a tectonic load that contributed to progressive basin subsidence beginning in the Eocene. They speculated that the magnitude of the load created by the emplacement of the volcanic arc was not large enough to drive the observed subsidence. However, that study did not consider a dynamic contribution for basin subsidence in the study area, and evidence for crustal thickening is generally lacking in the region. Walker et al. (2003) called upon both fault-controlled subsidence related to a right-lateral strike-slip fault in the approximate position of the David River zone (Fig. 4) from the early or middle Eocene to the Miocene, and then flexural subsidence from the volcanic load beginning in the late Eocene. They acknowledged, however, that the responsible fault zone is relatively narrow (∼10 km wide), that there is no evidence that it is a lithosphere-scale structure that could cause the magnitude of subsidence that is observed (>3 km), and that only ∼20% extension occurred in the basin as suggested by crustal thinning. Furthermore, based on stratal geometries observed in seismic data, any fault-controlled subsidence that may have occurred is interpreted to have ceased by the middle Miocene (Walker et al., 2003). Therefore, previous subsidence studies in the region of the study area have been unable to conclusively explain the magnitude of Oligocene–Holocene subsidence observed in the basin.

The establishment of a new subduction zone along the northern Pacific margin provides an opportunity to evaluate subsidence and basin geometry as predicted by geodynamic models. The dynamic topography behind modern and modeled subduction zones has been documented by many previous studies (e.g., Bond, 1976; Burgess and Moresi, 1999; Davies, 1981; Gurnis, 1990, 1992, 1993; Mitrovica et al., 1989; Stern and Holt, 1994). In general, mantle convection propelled by the dense, sinking slab is inferred to drive subsidence in the backarc region and beyond. Compared to flexural subsidence, the wavelength (500–2000 km) is generally much larger, and the magnitude of subsidence can be somewhat smaller (∼500–2000 m).

Gurnis (1992) proposed a three-stage model for dynamic subsidence that consists of (1) ∼10 m.y. of rapid subsidence associated with subduction initiation, (2) ∼50 m.y. of continued but decreasing subsidence associated with slab shallowing, and (3) ∼150 m.y. of decreasing dynamic subsidence as younger oceanic crust enters the subduction zone and the ocean basin closes. In the Taranaki retro-arc foreland basin in New Zealand, Stern and Holt (1994) demonstrated that initial rapid subsidence was driven by both dynamic and flexural mechanisms. Marsaglia (2012) suggested that the stratigraphic record of subduction stepping outboard of a collided terrane along a convergent margin is an angular unconformity overlain by shallow-marine and nonmarine strata, which are in turn overlain by deeper-marine volcaniclastic deposits in the upper plate.

The post–middle Eocene sedimentary record on the western Alaska Peninsula can be interpreted to represent a similar progression of events. The rapid transition from nonmarine deposition of the Tolstoi Formation to submarine shelf deposition of the Stepovak Formation occurred in the late Eocene. Geochronologic evidence indicates that sparse Aleutian arc volcanism began ca. 46 Ma (Jicha et al., 2006), and that by the latest Eocene (ca. 37 Ma), the Aleutian arc to the west and Meshik arc in the study area appear to have been well established (Fig. 1; Jicha et al., 2006; Wilson, 1985). Many workers postulate that initiation of the Aleutian arc may have been related to a change in plate motions at ca. 47 Ma (Duncan and Keller, 2004; Wessel and Kroenke, 2008; Seton et al., 2012), although some controversy exists about Paleogene Pacific Basin plate motions (e.g., Doubrovine and Tarduno, 2008; Wright et al., 2015). This change is interpreted to have caused Pacific plate motion to shift from northeast- to northwest-directed and would have resulted in more orthogonal subduction in the study area (Lonsdale, 1988). Therefore, we infer that the relatively abrupt transition from nonmarine to fully marine depositional environments during the late Eocene was related to a combination of dynamic and flexural subsidence mechanisms. Dynamic subsidence during this time would have been associated with initiation of the adjacent Aleutian arc subduction zone, as well as a potential change to more orthogonal plate motions along the Alaska Peninsula. Flexural subsidence was driven by emplacement of a tectonic load via the creation of the volcanic arcs.

Sedimentation in the basin continued in the Miocene and Pliocene with a maximum of ∼3000 m of strata deposited in the southern part of the basin. This clastic load in combination with the modern volcanic load contributes to the overall shape of the flexed basement. We infer that this pattern is overprinted by <1 km of additional subsidence produced by dynamic topography (e.g., Gurnis, 1993). According to numerical models, the subsidence produced by dynamic topography should be fairly uniform over large distances and should have a magnitude between 400 and 2000 m (Burgess and Moresi, 1999; Stern and Holt, 1994). The flooded continental shelf beneath the Bering Sea is ∼200 m below the global average, which is shallower than what is expected but may be the result of the thick sedimentary package present there (Gurnis, 1993). Therefore, we suggest that the modern basin profile is a product of both flexural subsidence associated with the volcanic and sedimentary loads, and dynamic subsidence produced by the subducting slab.

Our new detrital zircon data reveal a record of magmatism for the western Alaska Peninsula that can be compared to magmatic records for other regions of the northern Cordillera. For example, Gehrels et al. (2009) produced an average magmatic flux curve based on the volume of exposed igneous bedrock for the Coast Mountains batholith in northern British Columbia that demonstrates periods of high-flux magmatism in the Late Jurassic (ca. 160–140 Ma), middle Cretaceous (ca. 120–78 Ma), and Eocene (ca. 55–48 Ma; Fig. 8D). A probability density plot of detrital zircon U-Pb data from Cenozoic strata of the Cook Inlet basin shows peaks with ages at ca. 110–90 Ma and ca. 80–50 Ma (Fig. 8C; Finzel et al., 2015). A probability density plot of geochronologic ages from bedrock in south-central Alaska (Wilson et al., 2015), however, has peaks around ca. 70–40 Ma and younger than ca. 10 Ma (Fig. 8C). A composite probability density plot of detrital zircon U-Pb data from this study has peaks with ages at ca. 170–140 Ma and ca. 100–80 Ma (Fig. 8A). A probability density plot of geochronologic ages from bedrock on the Alaska Peninsula (Wilson et al., 2015), however, is dominated by younger peaks of ca. 40–25 Ma and younger than ca. 10 Ma (Fig. 8A). We infer that some of the differences between these curves can be explained in terms of the Mesozoic–Cenozoic plate kinematics in the northern Pacific region.

The Coast Mountains and Alaska Peninsula data sets suggest widespread Middle–Late Jurassic magmatism along the northern Cordillera. These magmatic events were most likely related to an extensive subduction zone that facilitated subduction of the Farallon plate beneath large expanses of western North America and outboard terranes during the Middle and Late Jurassic (Fig. 2; e.g., Gehrels et al., 2009). The Jurassic population may be subdued in the Cook Inlet basin because that data set only contains detrital ages from Cenozoic strata, but Jurassic detrital zircon populations have been reported in preliminary studies of Mesozoic strata from the adjacent Matanuska Valley (Stevens et al., 2012; Reid and Finzel, 2016). The Alaska Peninsula, however, includes data from Jurassic and Cretaceous strata, and the Coast Mountains curve is based on geologic mapping of plutonic belts where Jurassic igneous rocks are well exposed.

The Coast Mountains and detrital zircon curves from the Cook Inlet and Alaska Peninsula indicate elevated magmatic activity during the middle Cretaceous (ca. 120–80 Ma), consistent with the inference of development of an extensive continental arc system and continued long-term subduction along the outboard margin of western North America (Seton et al., 2012; Wessel and Kroenke, 2008). Only the Cook Inlet detrital and south-central Alaska bedrock records, though, suggest significant magmatic activity during latest Cretaceous–Paleocene time (ca. 80–50 Ma). Gehrels et al. (2009) inferred a decrease in the magmatic flux in the Coast Mountains during this time due to a shift toward N-directed plate motions relative to North America that resulted in a higher obliquity between the subducting and upper plates along the British Columbia region. This may also be the reason why the magmatic activity was sustained in the Cook Inlet basin. If the WCT were proximal to its present-day position and the margin had a similar geometry as today, then a northward shift in plate motions would not introduce more obliquity to the subduction system along the south-central Alaska margin, and magmatism would have continued, in contrast to cessation of magmatism in the Coast Mountains of British Columbia (Fig. 2). There is uncertainty, however, in the position of the WCT along the North American margin during the Late Cretaceous (e.g., Cowan et al., 1997; Butler et al., 2001). Another possibility is that subduction of a spreading ridge beneath south-central Alaska increased non-arc-related magmatic activity at least during the Paleocene. For example, the Coast Mountains curve displays another increase in magmatic flux during the Eocene (ca. 55–48 Ma) that has been interpreted to represent the creation of an extensional regime with high-flux magmatism associated with the subduction of the same spreading ridge (Haeussler et al., 2003; Gehrels et al., 2009).

The Coast Mountains and detrital zircon curves from the Cook Inlet and Alaska Peninsula lack evidence for abundant magmatic activity after the middle Eocene (ca. 45 Ma). The bedrock records, however, suggest persistent but decreasing magmatic activity in south-central Alaska into early Oligocene time (ca. 30 Ma), a magmatic pulse at ca. 40–30 Ma on the Alaska Peninsula, and increased magmatic activity for both regions after ca. 10 Ma. This trend in the Coast Mountains has been interpreted to reflect the transition from convergence to a transform boundary along the western margin of North America as the Pacific plate shifted toward more NW-directed motion (Gehrels et al., 2009). In the Cook Inlet basin and south-central Alaska, magmatism may have been hampered by subsequent flat-slab subduction events. Beginning in the late Eocene–Oligocene and continuing to the present day, flat-slab subduction of the Yakutat microplate has been interpreted to have hindered magmatic activity in that region (McNamara and Pasyanos, 2002; Qi et al., 2007; Finzel et al., 2011, 2015, 2016).

Flat-slab subduction events have not been recognized on the Alaska Peninsula, and geologic mapping and radiometric dating document abundant Eocene–Quaternary volcanic bedrock all along the western Alaska Peninsula (Fig. 8A; Wilson et al., 1999, 2015; Jicha et al., 2009). Based on these observations, magmatism was especially productive during the late Eocene to early Oligocene and late Miocene–Holocene; however, these ages are virtually absent in our data set. This difference may again be reconciled by considering zircon abundance. Abundance is likely low in volcanic rocks of the Meshik and modern arcs because they are dominantly intermediate to mafic in composition, and the associated plutonic rocks have for the most part not yet been exhumed (e.g., Detterman et al., 1996; Wilson, 1985). Therefore, even though the presence of these mainly extrusive arc rocks suggests widespread magmatism during certain periods of Cenozoic time, that signature would not be strongly expressed in the detrital zircon ages of the Cenozoic strata.

New stratigraphic and detrital zircon geochronologic data from Jurassic to Miocene strata located on the western Alaska Peninsula provide additional constraints on the tectonic and paleogeographic history of far western Alaska. The main regional implications of our study are summarized in the following points.

• (1) Sedimentary basins on the western Alaska Peninsula record major changes in plate dynamics related to Late Jurassic to Late Cretaceous accretion of the oceanic WCT, subsequent establishment of a Late Cretaceous continental volcanic arc, and middle Eocene Pacific plate reorganization and subduction initiation marked by the Alaska Peninsula–Aleutian volcanic arc.

• (2) Regional stratigraphic variations between the study area on the western Alaska Peninsula and forearc basins located along strike in south-central Alaska and British Columbia reflect variations in the dynamics and timing of Late Jurassic–Late Cretaceous collision and accretion of the WCT and Paleocene–Eocene spreading ridge subduction.

• (3) Initiation of the Aleutian and Meshik volcanic arcs followed final subduction of the Resurrection plate during the early Eocene. The stratigraphic response to this event was a shift from nonmarine to marine depositional systems accompanied by a transition to a backarc basin setting. Dynamic subsidence associated with the change in subduction parameters and flexural subsidence associated with loading from both the volcanic arc and backarc basin strata produced the basin geometry observed today.

In a broader sense, our results show that along convergent margins with changing plate kinematics, major depocenters may shift between forearc and backarc locations. In addition, long-term recycling of sediment should be expected within and between forearc and backarc basins. Finally, a direct link between primary arc sources of sediment and depocenters in adjacent basins may not be as common as suggested by classic models of convergent margin basins.

This article is dedicated to the memory of our colleague, the late Rocky Reifenstuhl. Rocky was an amazing field geologist and impacted greatly both the understanding of Alaskan geology and the development of new talent in that arena. Field operations for the 2004–2005 field seasons were funded from multiple sources: a U.S. Department of Energy grant from the Arctic Energy Technology Development Laboratory at the University of Alaska–Fairbanks to R. Reifenstuhl (Alaska Division of Geological & Geophysical Surveys) and P. McCarthy (University of Alaska–Fairbanks), contributions from the Bristol Bay Native Corporation, and the Alaska Division of Geological & Geophysical Surveys and Division of Oil and Gas operating budgets. Analytical work was funded by start-up funds from the University of Iowa to E. Finzel. We thank David LePain, Marwan Wartes, and Robert Gillis for facilitating access to samples that were collected during the 2004–2005 field seasons and analyzed for this study, Phil Finzel for his help in collecting samples, and the staff of the Bear Lake Lodge for assistance with logistics. Tony Doré, Jamey Jones, and Marwan Wartes provided constructive reviews that improved the manuscript.