Provenance signature of changing plate boundary conditions along a convergent margin: Detrital record of spreading-ridge and flat-slab subduction processes, Cenozoic forearc basins, Alaska

Along many convergent margins, subduction of oceanic plateaus, aseismic and spreading ridges, and microplates changes the geodynamics of the convergent margin system and can have profound effects on upper plate tectonic processes (e.g., Dickinson and Snyder, 1979; Jordan and Allmendinger, 1986; Hampel, 2002; Fisher et al., 2004; Taylor et al., 2005; Madsen et al., 2006; Pedoja et al., 2006; Pallares et al., 2007; Sak et al., 2009). However, the majority of previous research that contributes to our understanding of forearc basin provenance and depositional systems comes from strata that were deposited during protracted periods of subduction of normal oceanic crust, primarily the Great Valley sequence in the San Joaquin basin of California and the Peninsular Ranges of Baja Mexico (e.g., Ingersoll, 1978a, 1978b; Dickinson and Seely, 1979; Ingersoll, 1979, 1983; Busby-Spera, 1986; Moxon and Graham, 1987; Morris and Busby-Spera, 1988; Linn et al., 1991, 1992; Dickinson, 1995; Busby et al., 1998; DeGraaff-Surpless et al., 2002; Surpless et al., 2006; Surpless and Beverly, 2013). Because the southern convergent margin of Alaska has been subject to subduction of both a spreading ridge and an oceanic plateau during Cenozoic time (Plafker et al., 1994; Haeussler et al., 2000; Bradley et al., 2003; Haeussler et al., 2003; Sisson et al., 2003; Cole et al., 2006; Finzel et al., 2011; Ridgway et al., 2012), the Cenozoic forearc basins of southern Alaska provide an opportunity to evaluate transport and deposition of sediment across a convergent margin during these two different types of shallow subduction.

In order to investigate the provenance record of spreading-ridge and flat-slab subduction, we use U-Pb dating of detrital zircons in both ancient strata and modern river sediment, integrated with clast composition data from conglomerate, from Cenozoic forearc basins in southern Alaska (Fig. 1). Only a few provenance studies have combined U-Pb detrital zircon analyses of both modern fluvial sediment and ancient strata (e.g., Bernet et al., 2006; Cina et al., 2009; van Hoang et al., 2009), but our results indicate that integrating both types of analyses permits better classification of potential source regions and provide useful constraints for establishing provenance and tectonic histories. Integrating data from ancient and modern systems allows us to: (1) reevaluate the existing provenance model for the southern Alaska forearc basins that was proposed before the role of shallow subduction was recognized along this convergent margin, (2) update paleogeographic reconstructions of the region, (3) propose cause-and-effect relationships between subduction-related events and changes in upper plate processes as reflected in provenance, and (4) offer a generalized model for sediment dispersal patterns and detrital zircon signatures for forearc basins affected by spreading-ridge or flat-slab subduction.

The southern margin of Alaska has been the site of continuous convergence and subduction since at least Jurassic time and has experienced several different subduction-related events, including terrane accretions, spreading-ridge subduction, and flat-slab subduction of an oceanic plateau. Terranes that were accreted to the Mesozoic margin by Middle Jurassic time, including the Yukon-Tanana and Farewell terranes, are located inboard of the Alaska Range and are collectively known as the Yukon composite terrane, an amalgamation of Precambrian–Mesozoic microcontinents, continental margin sedimentary assemblages, and continental igneous belts (Fig. 2; Foster and Keith, 1994; Mihalynuk et al., 1994; Monger and Nokleberg, 1996; Gehrels, 2001). During Middle Jurassic to Late Cretaceous time, the Wrangellia composite terrane collided with the outboard margin of the Yukon composite terrane (Pavlis, 1982; McClelland et al., 1992; Trop et al., 2002; Trop et al., 2005; Manuszak et al., 2007). Today, the Yukon and Wrangellia composite terranes are separated by the Alaska Range suture zone, where thousands of meters of Upper Jurassic to Upper Cretaceous marine strata were deposited in the Kahiltna basin and subsequently exhumed (Fig. 2; Ridgway et al., 2002; Hampton et al., 2007, 2010).

Subduction along the outboard margin of the Wrangellia composite terrane continued in Late Cretaceous time (Plafker et al., 1994; Trop and Ridgway, 2007). Between Late Cretaceous and middle Eocene time, the southern Alaskan orocline developed and was accompanied by significant right-lateral motion on major fault systems (Figs. 2 and 3; Coe et al., 1985; Plafker, 1987; Coe et al., 1989; Hillhouse and Coe, 1994; Pavlis and Sisson, 2003; O’Driscoll et al., 2004; Scharman and Pavlis, 2012). Partly coeval with oroclinal bending, during Paleocene and Eocene time (ca. 62–50 Ma), a spreading ridge was subducted from west to east across the entire southern margin prompting a hiatus in arc magmatism, emplacement of slab-window igneous rocks in the forearc and accretionary prism regions, and high-temperature/low-pressure metamorphism of accretionary prism strata (Bradley et al., 2003; Haeussler et al., 2003; Sisson et al., 2003; Cole et al., 2006).

Subduction of normal oceanic crust resumed after spreading-ridge subduction until the Yakutat microplate, which is interpreted as an oceanic plateau (Ferris et al., 2003; Pavlis et al., 2004; Eberhart-Philips et al., 2006; Worthington et al., 2008, 2012; Christeson et al., 2010), was sliced off the continental margin and transported northward (Fig. 1; Plafker et al., 1978; Plafker, 1987; Plafker et al., 1994). The transition from normal- to flat-slab subduction of the microplate is interpreted to have initiated along the outboard margin of south-central Alaska in late Eocene to early Oligocene time (ca. 40–30 Ma) (Plafker et al., 1994; Finzel et al., 2011; Arkle et al., 2013). Flat-slab subduction processes observed in the upper plate of southern Alaska include changes in the style, location, and chemistry of volcanic arc magmatism; variations in sedimentary basin subsidence, inversion, and sediment sources; and accelerated exhumation and surface uplift (e.g., Enkelmann et al., 2008; Finzel et al., 2011; Trop et al., 2012; Arkle et al., 2013). Active flat-slab subduction characterizes the present-day margin of southern Alaska.

The Cook Inlet basin is the active portion of the forearc basin system along the present-day convergent margin (Figs. 1 and 2). The Susitna basin is considered a northern extension of the Cook Inlet basin (Merritt, 1986; Rouse and Houseknecht, 2012). The Matanuska Valley, located along strike to the northeast, represents the Middle Jurassic to Oligocene exhumed remnant part of the forearc basin (Trop et al., 2005; Trop and Ridgway, 2007). The Chugach Mountains, located southeast of the forearc basins, are composed primarily of Permian to Eocene accretionary prism strata (Tysdal and Plafker, 1978; Tysdal and Case, 1979; Winkler and Plafker, 1981; Nilsen and Zuffa, 1982; Nelson et al., 1986; Kusky et al., 1997; Stevens et al., 1997; Bradley et al., 1999; Amato and Pavlis, 2010). The Alaska Range and Talkeetna Mountains, located northwest and north of the forearc basins, are composed of Late Cretaceous to Oligocene granitic arc rocks, minor granitic remnants of an older middle Cretaceous arc, and exposures of the accreted Early to Middle Jurassic Talkeetna arc of the Wrangellia composite terrane (Reed and Lanphere, 1972, 1973, 1974; Berry et al., 1976; Magoon et al., 1976b; Gilbert, 1981; Nye and Turner, 1990; Amato et al., 2007; Rioux et al., 2007; Wilson et al., 2009).

The modern Aleutian-Alaska Peninsula volcanic arc extends northward from the Alaska Peninsula into the western Alaska Range and ends at the interpreted western boundary of the currently shallowly subducting Yakutat slab (Fig. 1). To the east of the subducting slab, the Wrangell volcanic belt consists of northwestward-younging igneous and sedimentary rocks interpreted to record progressive insertion of the Yakutat microplate beneath southern Alaska (Richter et al., 1990; Skulski et al., 1992; Ridgway et al., 1995; Trop et al., 2012). Between these two volcanic regions and lying above the Yakutat slab is a gap in magmatism that is interpreted to be related to flat-slab subduction (e.g., McNamara and Pasyanos, 2002; Qi et al., 2007; Finzel et al., 2011).

Cenozoic strata in the Cook Inlet basin crop out in discontinuous belts along the margins of the basin and depositionally overlie or are in fault contact with rocks of the adjacent accretionary prism and volcanic arc (Magoon et al., 1976b). In the center of the basin, the forearc strata unconformably overlie Mesozoic sedimentary and volcanic rocks (Fig. 3; Jones and Silberling, 1979; Magoon and Egbert, 1986).

The West Foreland Formation has been assigned an upper Paleocene or Eocene to Oligocene age based on leaf fossils, palynomorphs, and unpublished industry data (Fig. 3; Kirschner and Lyon, 1973; Magoon et al., 1976b; Swenson, 1997). This formation reaches a maximum thickness of 1259 m along the western margin of the basin but thins abruptly toward the center of the basin (Calderwood and Fackler, 1972; Houston, 1994). It consists of thick, conglomeratic strata deposited in alluvial fan and braided stream systems along the margin of the basin and nonmarine tuffaceous siltstone and claystone with minor coal and conglomerate in the subsurface near the basin axis (Calderwood and Fackler, 1972; Houston, 1994; Wahrhaftig et al., 1994).

The Oligocene age designation for the Hemlock Conglomerate is based on leaf fossils and palynomorphs (Fig. 3; Magoon et al., 1976b; Wolfe and Tanai, 1980). This unit forms a sheet ∼200 m thick across most of the basin, with a maximum thickness of ∼845 m (Flores et al., 2004). The conglomerate and conglomeratic sandstone in this formation were deposited in fluvial and deltaic environments near the basin margins that graded into estuarine depositional systems along the basin center, similar to modern environments in the Cook Inlet basin (Kirschner and Lyon, 1973; Boss et al., 1976; Detterman et al., 1976; Hite, 1976; Magoon and Egbert, 1986).

The Tyonek Formation ranges in age from lower or upper Oligocene to middle Miocene based on leaf fossils and palynomorphs (Fig. 3; Magoon et al., 1976b; Wolfe and Tanai, 1980). Approximately 2400 m of massive sandstone and coal were deposited in alluvial fan and fluvial systems near the basin margins and deltaic and estuarine systems toward the basin center (Detterman et al., 1976; Hite, 1976; Magoon et al., 1976b; Flores et al., 1994; Flores et al., 1997).

The middle to upper Miocene age of the Beluga Formation is based on leaf fossils, palynomorphs, and geochronologic dating of interbedded tuffs (ca.13–5 Ma; Fig. 3; Magoon et al., 1976b; Triplehorn et al., 1977; Turner et al., 1980; Wolfe and Tanai, 1980; Reinink-Smith, 1990; Dallegge and Layer, 2004). This formation consists of relatively thin units of claystone, sandstone, siltstone, and coal deposited in basin margin alluvial fans with possible glacial influence and in shallow braided and meandering streams (Kirschner and Lyon, 1973; Boss et al., 1976; Hayes et al., 1976; Rawlinson, 1984; Flores and Stricker, 1992, 1993b). The maximum thickness is ∼1800 m near the western margin of the basin, but it thins abruptly to the east where it is truncated by the overlying Sterling Formation (Calderwood and Fackler, 1972; Kirschner and Lyon, 1973).

The upper Miocene to Pliocene age of the Sterling Formation is based on leaf fossils, palynomorphs, and geochronologic dating of interbedded tuffs (ca. 9–3 Ma; Fig. 3; Calderwood and Fackler, 1972; Magoon et al., 1976b; Triplehorn et al., 1977; Turner et al., 1980; Wolfe and Tanai, 1980; Reinink-Smith, 1990; Dallegge and Layer, 2004). Massive sandstone, conglomerate, claystone, and minor coal represent deposition in meandering and braided streams (Boss et al., 1976; Hayes et al., 1976; Hite, 1976; Rawlinson, 1984; Flores and Stricker, 1993c). The Sterling Formation is the only Cenozoic formation in the basin that is thickest near the eastern margin, where it is ∼3300 m thick (Calderwood and Fackler, 1972; Kirschner and Lyon, 1973).

The Susitna basin is a broad lowland with few geologic exposures and minimal subsurface data, so its stratigraphy and geologic history are poorly understood. The basin was originally thought to contain at least 3–4 km of siliciclastic strata (Hackett, 1977; Ehm, 1983; Kirschner, 1988; Meyer, 2005) that consisted of Paleocene–Miocene conglomerate, carbonaceous sandstone, mudstone, and coal deposited in fluvial-lacustrine environments (e.g., Hackett, 1977; Merritt, 1986; Meyer et al., 1996; Meyer and Boggess, 2003). Recent work on two exploratory wells, however, has demonstrated the presence of upper Paleocene to lower Eocene interstratified sedimentary and volcanic rocks beneath ∼1400 m of lower to middle Eocene nonmarine sandstone, siltstone, and coal that is unconformably overlain by ∼2500 m of lower or middle Miocene to Quaternary nonmarine conglomerate and sandstone (Fig. 3; Stanley et al., 2013). Strata in the subsurface are not assigned formation names, but strata exposed at the surface are mapped using formation names from the Cook Inlet basin and are interpreted as Miocene or younger (Barnes, 1966; Reed and Nelson, 1980).

Cenozoic strata in the Matanuska remnant forearc basin are Paleocene to Oligocene in age and represent deposition in nonmarine and marginal marine environments. The Paleocene to Eocene Wishbone, Chickaloon, and Arkose Ridge formations consist of ∼4200 m of mudstone, sandstone, conglomerate, coal, and volcanic interbeds deposited in alluvial and fluvial environments along the basin margins and fluvial, lacustrine, and estuarine environments in the basin center (Fig. 3; Clardy, 1974; Winkler, 1992; Flores and Stricker, 1993a; Trop and Ridgway, 2000; Trop et al., 2003; Neff et al., 2011; Trop et al., 2012; Sunderlin et al., 2014). Overlying an angular unconformity with the Wishbone Formation is the ∼200-m-thick Oligocene Tsadaka Formation, which consists of conglomerate and sandstone deposited in stream-dominated alluvial fan environments (Clardy, 1974). There are no known Miocene or Pliocene strata in the Matanuska Valley.

Although significant displacement of potential source terranes along regional strike-slip faults has occurred prior to deposition of the forearc strata, only limited lateral movement has been interpreted for most of these faults since Eocene time. For example, on the northwestern margin of the Cook Inlet basin ∼19–65 km of left-lateral offset has occurred along the Bruin Bay fault prior to the Oligocene (Fig. 2; Detterman and Hartsock, 1966; Detterman and Reed, 1980). Farther northwest, the Lake Clark fault has experienced ∼26 km of right-lateral offset since the late Eocene (Haeussler and Saltus, 2004). North of the Matanuska Valley basin, Late Cretaceous to Cenozoic dextral displacement along the Castle Mountain fault is ∼20–40 km (Grantz, 1966; Clardy, 1974; Detterman et al., 1976; Fuchs, 1980; Trop et al., 2003). On the southeastern margin of the forearc basin system, at least 130 km of Cenozoic strike-slip motion on the western part of the Border Ranges fault system exposed along the Kenai Peninsula is restricted to a brief interval during the early to middle Eocene (ca. 55–45 Ma; Little and Naeser, 1989; Roeske et al., 2003; Pavlis and Roeske, 2007). An additional 140–190 km of strike-slip motion between ∼54–52 Ma is documented by flow within the Chugach Metamorphic Complex (Scharman and Pavlis, 2012).

The only fault that may have experienced significant displacement throughout Cenozoic time is the dextral Denali fault. Previous interpretations have inferred only ∼38 km of offset since ca. 38 Ma along the central Alaska Range portion of the fault (late Eocene; Reed and Lanphere, 1974). More recent work has suggested that the segment of the Denali fault in the eastern Alaska Range has had at least 100 km of right-lateral slip during late Paleocene to late Oligocene time and at least 300 km of right-lateral slip since late Oligocene time (Benowitz et al., 2012b). However, the principal source terrane north of the Denali fault, the Yukon composite terrane, extends >1000 km from northwest of the present-day central Alaska Range southeastward into western Canada. Therefore, even with 400 km of Cenozoic slip on the Denali fault, the modern configuration of the basin relative to its potential sediment sources has remained effectively unchanged since Eocene time.

We consider the arc margin of the forearc basin system as the region adjacent to the western Alaska Range in the Cook Inlet basin and all parts of the Susitna basin (Fig. 2). We consider the accretionary prism margin of the forearc basin system as the region adjacent to the western Chugach Mountains in the Cook Inlet basin and all parts of the Matanuska basin. Detailed sample location information is available in Supplemental Table 1¹.

In the Cook Inlet basin, a mudstone sample was collected for palynological analysis at each Cenozoic strata locality. The samples were processed by Russ Harms of Global Geolab Limited in Medicine Hat, Alberta, Canada, and prepared slides of recovered palynomorphs were analyzed by Dr. Pierre Zippi (Biostratigraphy.com, LLC). The results of the analyses are summarized below and published in Zippi and Loveland (2012). Palynology samples were not collected from strata in the Susitna or Matanuska Valley basins as part of this study. In these basins, previously published age control was used in our analysis.

At study sites along the arc margin of the basin, the West Foreland Formation is assigned a middle Eocene age (5 and 6 on Fig. 2), the Tyonek Formation is assigned a late middle Miocene age (3 on Fig. 2), and the Beluga Formation is assigned an early late Miocene age (2 on Fig. 2). At study sites on the accretionary prism side of the basin, the Tyonek Formation is assigned a middle Miocene age (10 on Fig. 2), the Beluga Formation is assigned a late Miocene age (9 on Fig. 2), and the Sterling Formation is assigned an early Pliocene age (8 on Fig. 2).

For samples without palynological data, the maximum depositional age was resolved by determining the youngest peak with three overlapping ages in the detrital zircon signature. For example, the late Miocene age of the Beluga Formation on the accretionary prism margin of the basin is supported by a 6.9 ± 0.6 Ma peak in the detrital zircon population (9 on Fig. 2). In the Matanuska basin, sample 11 yields a peak age of 26.6 ± 0.8 Ma with five overlapping grain ages (11 on Fig. 2). Sample 11 was collected from strata previously interpreted as equivalent to the Paleocene–Eocene Wishbone Formation (Winkler, 1992) or the Oligocene Tsadaka Formation (Magoon et al., 1976b). Given that Tsadaka Formation strata with Oligocene fossils unequivocally overlie the Wishbone Formation at other locations (Clardy, 1974; Winkler, 1992), it is unlikely that the sampled strata represent the Wishbone Formation. Instead, our new data indicate that Sample 11 was collected from strata that represent the Oligocene Tsadaka Formation or the Miocene Tyonek Formation; outcrops of the Tyonek and Tsadaka formations located <1–7 km from the sampled strata yield Angoonian (upper Oligocene to early Miocene) fossils (Wolfe et al., 1966; Clardy, 1974). Lithologically, the strata at the sample 11 locality more closely resemble the Tyonek Formation; so in this study we consider sample 11 to represent upper Oligocene–early Miocene Tyonek Formation.

Clast compositional data were recorded where Eocene and Miocene conglomerate are exposed at sample sites along the arc margin of the Cook Inlet basin. In the West Foreland Formation, four clast counts were completed just south of the Capps Glacier (6 on Fig. 2), and three counts were conducted along the Theodore River (5 on Fig. 2). Two clast counts were conducted in the Tyonek Formation along the Lewis River (3 on Fig. 2). Clast compositional data were collected by counting all the pebble- and cobble-sized clasts in a predefined area on an outcrop face until at least 100 clasts were counted. Fifty-two clast compositions identified in the field were subsequently grouped into eight categories: volcanic, plutonic, volcaniclastic, greenstone, quartz, chert, sedimentary, and metasedimentary types (Fig. 4).

The clast compositions of conglomerate that were observed are non-unique in terms of determining a specific source region, and as a result, they cannot be used to make precise provenance interpretations. The main points that can be taken from the clast compositional data are: (1) the upsection increase in quartz, chert, and sedimentary and metasedimentary lithics from the West Foreland Formation to the Tyonek Formation and (2) the overall difference in clast types between West Foreland Formation at Capps Glacier (mostly volcanic and plutonic) versus at the Theodore River (mostly volcanic, quartz, and chert).

Seven detrital zircon samples were collected from four modern rivers, and nine samples were collected from Cenozoic strata within the Cook Inlet, Susitna, and Matanuska basins (Fig. 2). Modern river samples are from the Susitna and Matanuska rivers, which are the major modern drainages into the Cook Inlet basin along the arc and accretionary prism margins of the basin, respectively (1 and 7 on Fig. 2), and provide a reference for the detrital signal of the modern source regions. The detrital signature of the Susitna River is based on a composite of three samples collected all along the river, from the upper, middle, and lower reaches. The other two modern river samples are from the Tazlina and Tonsina rivers, whose watersheds cover a large part of the northern Chugach Mountains and a small part of the southeast corner of the Talkeetna Mountains (12 on Fig. 2). The Chugach Mountains are primarily composed of accretionary prism strata, whereas the small part of the Tazlina watershed within the Talkeetna Mountains includes some Jurassic, Cretaceous, and Pliocene strata and volcanic rocks of the Talkeetna Formation. The Tazlina and Tonsina samples are used to characterize the bulk detrital zircon signature for the part of the accretionary prism in the northern Chugach Mountains.

In the Cook Inlet basin, samples were collected from the West Foreland, Tyonek, and Beluga formations on the arc margin of the basin and the Tyonek, Beluga, and Sterling formations on the accretionary prism margin of the basin (Fig. 2). In addition, samples were collected from the Tyonek Formation in the Susitna Basin and from the Tyonek Formation in the Matanuska basin.

U-Pb analyses of detrital zircons were conducted by laser ablation–multicollector–inductively coupled plasma mass spectrometry (LA-MC-ICPMS) at the Arizona LaserChron Center following the methods of Gehrels et al. (2006). The analytical data are reported in Supplemental Table 2². Detrital zircon distributions are shown as a composite of all the samples from this study in Figure 5 and as individual distributions for samples in Figures 6 and 7. Figure 8 shows previously published detrital zircon populations of potential sedimentary source rocks as well as new data from the Tazlina and Tonsina rivers (sample 12, Prism strata).

All samples from the arc margin of the basin have Late Cretaceous and Paleocene–Eocene peaks in their age distributions (Fig. 6). The most striking variations present within the arc margin samples are the abundance of Paleozoic and Precambrian grains in the middle Eocene West Foreland Formation (5 on Fig. 6) and the gradual upsection decrease of those detrital zircon age groups.

Jurassic–Early Cretaceous, Late Cretaceous, and Paleocene–Eocene peaks occur in all samples from the accretionary prism margin of the basin (Fig. 7). A transition from dominantly late Early Cretaceous ages in the middle Miocene Tyonek Formation samples (10 and 11 on Fig. 7) to early Early Cretaceous ages in the late Miocene Beluga Formation sample (9 on Fig. 7), to a lack of Early Cretaceous grains in Pliocene Sterling Formation and modern Matanuska River samples (8 and 7, respectively, on Fig. 7) marks the most significant difference among these samples.

For standard detrital zircon analyses of 60–120 grains per sample, the age distribution of zircons analyzed in the laboratory may not reflect the complete age distribution of detrital zircon in the sediment (Dodson et al., 1988; Vermeesch, 2004; Andersen, 2005). Modifying the scheme proposed by Andersen (2005), we employ a ranking system for detrital zircon populations from potential sedimentary sources as well as our new data such that populations comprising <10% of the entire sample are considered accessory groups and are not discussed; populations comprising 10%–20% of the total sample are considered minor populations; and populations comprising >20% of the total sample are considered major populations.

We define six age groups that contain 95% of the ages present in our data based on the age distributions of potential zircon sources in Alaska (Fig. 5). The age ranges are: Paleocene–Eocene (65–37 Ma), Late Cretaceous (104–66 Ma), Early Cretaceous (140–105 Ma), Jurassic to Early Cretaceous (201–141 Ma), Paleozoic (541–252 Ma), and Precambrian (>541 Ma) (ages refer to timescale of Walker et al., 2012).

Igneous rocks that are partly Paleocene–Eocene in age (light-pink map pattern on Fig. 2) span a large area of south-central Alaska including the Talkeetna Mountains and the western and central Alaska Range (ca. 110–55 Ma; Reed and Lanphere, 1969, 1972, 1973, 1974; Turner and Smith, 1974; Berry et al., 1976; Magoon et al., 1976b; Csejtey et al., 1978; Gilbert, 1981; Reed et al., 1983; Turner and Nye, 1986; Gilbert et al., 1988; Nye and Turner, 1990; Wilson et al., 2009). A small plutonic belt with more restricted middle to late Eocene ages (ca. 43–37 Ma) also occurs in the central Alaska Range and Talkeetna Mountains (Lanphere and Reed, 1985; Moll-Stalcup, 1994). Sparse Paleocene intrusions (ca. 70–50 Ma) occur in interior Alaska in the Yukon-Tanana terrane (Wilson et al., 1985; Foster and Keith, 1994; Moll-Stalcup, 1994). To the south of the forearc basins, small plutons of the Sanak-Baranof belt in the outboard part of the accretionary prism are early Eocene in age (ca. 56–54 Ma; Haeussler et al., 2003). In the Matanuska basin, a potential sedimentary source with a major population of Paleocene–Eocene detrital zircons is the older forearc strata of the Paleocene–Eocene Arkose Ridge Formation (22% of the total distribution; Kortyna et al., 2013; Figs. 2 and 8).

A Late Cretaceous component of igneous rocks (dark-pink map pattern on Fig. 2) is found in the western and central Alaska Range and the Talkeetna Mountains (ca. 110–55 Ma; Reed and Lanphere, 1969, 1972, 1973, 1974; Turner and Smith, 1974; Berry et al., 1976; Magoon et al., 1976b; Csejtey et al., 1978; Gilbert, 1981; Reed et al., 1983; Turner and Nye, 1986; Gilbert et al., 1988; Nye and Turner, 1990; Foster and Keith, 1994; Miller, 1994; Wilson et al., 1998; Wilson et al., 2009). Late Cretaceous intrusions (ca. 110–85 Ma with most ages ca. 95–90 Ma) also occur in interior Alaska in the Yukon-Tanana terrane (Wilson et al., 1985; Foster and Keith, 1994; Moll-Stalcup, 1994). Sedimentary sources with major populations of Late Cretaceous detrital zircons include the Upper Cretaceous Matanuska Formation and the Paleocene–Eocene Arkose Ridge Formation (65%) in the Matanuska basin and accretionary prism strata located in the western (38%) and northern (39%) Chugach Mountains (Figs. 2 and 8).

Known igneous sources for Early Cretaceous populations (red map pattern on Fig. 2) include the Chisana plutonic belt (ca. 105–117 Ma) located along strike from the Cook Inlet basin on the northeastern flank of the Wrangell volcanic belt (Miller, 1994; Short et al., 2005; Snyder and Hart, 2005, 2007), small plutons in the eastern Alaska Range (ca. 110–118; Hart et al., 2004), small plutons in the northern Chugach Mountains (ca. 120–125 Ma; Amato and Pavlis, 2010; Amato et al., 2013), and a more extensive plutonic belt in the Yukon-Tanana area (ca. 110–118; Hart et al., 2004). The only known sedimentary source with a minor Early Cretaceous population is the Kahiltna assemblage located in the Talkeetna Mountains and central and western Alaska Range (Figs. 2 and 8).

Igneous sources of Jurassic to Early Cretaceous detrital zircons include (1) the Talkeetna arc (ca. 201–153 Ma) in the Talkeetna Mountains and the western Alaska Range and (2) the Chitina plutonic belt (ca. 175–135 Ma) located along strike from the Cook Inlet basin between the eastern Chugach Mountains and the Wrangell volcanic belt (purple map pattern on Fig. 2; Csejtey et al., 1978; Winkler and Plafker, 1981; Plafker et al., 1989; Winkler, 1992; Nokleberg et al., 1994; Roeske et al., 2003; Rioux et al., 2007; Hacker et al., 2011). Sedimentary sources with minor or major populations of Jurassic–Early Cretaceous detrital zircons include the older forearc strata in the Matanuska basin (10%), the Kahiltna assemblage (49%), and accretionary prism strata in the western (49%) and northern (23%) Chugach Mountains (Figs. 2 and 8).

A composite population of zircon ages from sedimentary and metamorphic rocks in the Farewell terrane (teal map pattern on Fig. 2) has a major population of Paleozoic grains (23%; Fig. 8). Additional sedimentary sources with minor populations of Paleozoic detrital zircons include the Kahiltna assemblage (14%) and accretionary prism strata in the northern Chugach Mountains (17%).

Composite populations of zircon ages from igneous, sedimentary, and metamorphic rocks in the Yukon-Tanana terrane (tan map pattern on Fig. 2) and sedimentary and metamorphic rocks in the Farewell terrane (teal map pattern on Fig. 2) have major populations of Precambrian grains (95% and 77%, respectively; Fig. 8). Additional sedimentary sources with minor populations of Precambrian grains include the Kahiltna assemblage (19%) and accretionary prism strata in the northern Chugach Mountains (12%; Fig. 8).

The middle Eocene West Foreland Formation south of the Capps Glacier (6 on Figs. 6 and 9A) is inferred to have a local sediment source in the adjacent western Alaska Range. The major Late Cretaceous and Paleocene–Eocene detrital zircon populations may have been sourced from igneous rocks of the western or central Alaska Range or Talkeetna Mountains (quadrangles 8, 10, 11, and 17 on Fig. 2). However, this outcrop is located <3 km from extensive exposures of Paleocene granite (Wilson et al., 2009), is dominated by pebble- and cobble-sized igneous clasts (6 on Fig. 4), and coarsens to the west toward the western Alaska Range (Calderwood and Fackler, 1972; Houston, 1994). These characteristics suggest that the sediment was derived principally from adjacent arc rocks within the western Alaska Range.

In contrast to the West Foreland strata at Capps Glacier, the middle Eocene West Foreland Formation along the Theodore River (5 on Figs. 6 and 9A) contains more diverse detrital zircon ages and more diverse conglomerate clast types. The major Late Cretaceous zircon population (33%) indicates sediment derivation from either the Alaska Range and Talkeetna Mountains (quadrangles 8, 10, 11, and 17 on Fig. 2) or the Yukon composite terrane, which includes both the Yukon-Tanana and Farewell terranes (quadrangles 1–8 on Fig. 2). The other two major populations of Paleozoic (17%) and Precambrian (37%) grains could have two possible sources: either recycled grains from the Kahiltna assemblage exposed within and south of the Alaska Range or from igneous and sedimentary sources in the Yukon composite terrane located north of the Alaska Range. We prefer the Yukon composite terrane as the primary source for these grains because the modern Susitna River drainage encompasses much of the exposed Kahiltna assemblage strata, but the sample from that river lacks these older populations (1 on Fig. 6). We infer that this is the case because derivation of zircons from the Kahiltna assemblage is likely small due to the relatively low abundance of zircons in the sedimentary Kahiltna assemblage strata compared to the Late Cretaceous and Paleocene–Eocene igneous rocks in the watershed. Furthermore, clast compositions of conglomerate in the West Foreland Formation at the Theodore River are more varied than at the Capps Glacier locality and are dominated by quartz and igneous types, which are interpreted as vein quartz and quartz-veined lithologies that are ubiquitous throughout south-central Alaska, including the Alaska Range, Talkeetna Mountains, and Yukon composite terrane (5 on Fig. 4; Nokleberg et al., 2005). Volcanic clasts were likely derived from the active volcanic arc in the western and central Alaska Range. Therefore, the major sediment source areas for these middle Eocene strata are interpreted as the western and central Alaska Range, Talkeetna Mountains, and the Yukon composite terrane north of the Alaska Range.

A comparison between the West Foreland Formation at Theodore River (5 on Figs. 6 and 9A) and lower Miocene Tanana basin strata (Upper Healy Creek–Lower Suntrana formations on Figs. 3 and 9A) from north of the Alaska Range reveals similar major populations of Late Cretaceous, Paleozoic, and Precambrian age zircons (Brennan and Ridgway, 2015). Detrital zircon ages, compositional data, and southwestward paleoflow indicators from the Tanana basin strata are interpreted to represent probable sediment derivation from the Yukon-Tanana terrane (Ridgway et al., 1999; Brennan and Ridgway, 2015). Similarities in the major detrital zircon age populations between the Cook Inlet and Tanana strata, the likely location of sediment sources located in the Yukon composite terrane for both sets of strata, and southwestward paleocurrent flow indicators in the Tanana strata suggest that the forearc may have been depositionally linked to areas north of the present-day Alaska Range. This would require the presence of an extensive axial fluvial system that originated in interior Alaska and flowed southward to the Cook Inlet basin during middle Eocene and early Miocene time.

The middle Miocene Tyonek Formation from the arc margin of the Cook Inlet (3 on Figs. 6 and 9B) is inferred to have two primary sediment source regions with most of the detritus being derived from the Alaska Range–Talkeetna Mountains region and a minor contribution from far inboard terranes. We interpret the major Late Cretaceous (51%) and minor Paleocene–Eocene (16%) populations to be sourced from igneous rocks of the western or central Alaska Range and/or Talkeetna Mountains (quadrangles 8, 10, 11, and 17 on Fig. 2). Abundant volcanic conglomerate clasts at this locality (3 on Fig. 4) also indicate derivation from a volcanic region such as the Alaska Range or Talkeetna Mountains. Metasedimentary clasts comprise a large percentage of the clast population, and whereas the Kahiltna assemblage is only locally metamorphosed (Nokleberg et al., 1985; Davidson et al., 1992), the Yukon composite terrane, in contrast, consists primarily of widespread metasedimentary and metaigneous rocks. Furthermore, similar to the West Foreland Formation at the Theodore River, minor Paleozoic (13%) and Precambrian (10%) populations are interpreted to be mainly derived from the Yukon composite terrane.

In the Tyonek Formation sample from the Susitna basin (4 on Figs. 6 and 9B), we interpret the Paleocene–Eocene (10%) and Late Cretaceous (39%) populations to be sourced from igneous rocks in the western or central Alaska Range and/or Talkeetna Mountains (quadrangles 8, 10, 11, and 17 on Fig. 2). The minor Early Cretaceous (10%) population is interpreted to represent recycling of Kahiltna assemblage strata also in those regions. The Paleozoic (16%) and Precambrian (16%) populations are again interpreted to be derived from the Yukon composite terrane (quadrangles 1–8 on Fig. 2) due to the lack of these populations in the modern Susitna River sample.

A comparison of detrital zircon populations between middle Miocene forearc strata (3 and 4 on Fig. 9B) and Tanana basin (Upper Suntrana Formation on Figs. 3 and 9B) strata reveals similar abundances of Paleozoic and Precambrian detrital zircon age populations (Brennan and Ridgway, 2015), and both include a major population of Late Cretaceous grains. Similarities between the major age populations in all middle Miocene samples, as well as southwestward paleoflow directions in middle Miocene strata in the Tanana basin (Ridgway et al., 1999), suggest the persistence of the extensive north-to-south fluvial system.

In the upper Miocene Beluga Formation from the Cook Inlet basin (2 on Figs. 6 and 9C), the dominance of Paleocene–Eocene (27%) and Late Cretaceous (52%) detrital zircon ages suggests derivation from igneous rocks of the western or central Alaska Range and/or Talkeetna Mountains (quadrangles 8, 10, 11, and 17 on Fig. 2). The minor Precambrian (10%) population in the Beluga Formation is interpreted to represent recycling from the Kahiltna assemblage located in those regions. In the Susitna River sample (1 on Figs. 6 and 9C), the two youngest peaks in the Paleocene–Eocene population (35 and 45 Ma) and the youngest peak in the Late Cretaceous population (71 Ma) indicate a likely sediment source from igneous rocks in the central Alaska Range and/or Talkeetna Mountains (quadrangles 8, 10, and 11 on Fig. 2) versus the Yukon composite terrane (quadrangles 1–8 on Fig. 2), which is consistent with the extent of the present-day drainage basin.

Late Miocene strata and modern river sediments in the forearc versus the Tanana basin contain different detrital zircon distributions. In the forearc, late Miocene strata (2 on Fig. 9C) and Susitna River sediment (1 on Fig. 9C) are dominated by Paleocene–Eocene and Late Cretaceous populations with a minor population of Precambrian grains in the upper Miocene Beluga Formation. The decrease in abundance of Precambrian and Paleozoic grains from middle Eocene (54%) to late Miocene (15%) time suggests that the Yukon composite terrane was a less significant source of sediment to the forearc by late Miocene time. The small population of Precambrian grains in the upper Miocene Beluga Formation could have been derived from the Kahiltna assemblage south of the Alaska Range. In the upper Miocene Tanana strata (Lignite Creek Formation–Nenana Gravel on Figs. 3 and 9C) and Nenana River sediments (Fig. 9C), profuse Late Cretaceous populations point to the central Alaska Range as a continued source for sediment (Brennan and Ridgway, 2015). In addition, major populations of Paleozoic and Precambrian grains indicate a long-term sediment source from the Yukon composite terrane. Therefore, upper Miocene Tanana basin strata and modern Nenana River sediments contain evidence of significant Yukon composite terrane sources. These patterns suggest that the Tanana basin and forearc fluvial systems had evolved to be independent of each other by late Miocene time and were similar to their modern configurations.

Detrital zircon populations from middle Eocene–early Miocene strata on the accretionary prism margin of the basin have a different provenance signature. The Paleocene–Eocene (36%), Late Cretaceous (30%), and Jurassic–Early Cretaceous (15%) populations in the Tyonek Formation in the Matanuska basin (11 on Figs. 7 and 9B) are interpreted to be sourced primarily from igneous rocks in the Talkeetna Mountains and/or recycled from upper Cretaceous–Eocene strata in the Matanuska basin (quadrangles 7, 11, and 16 on Fig. 2). The lack of significant Paleozoic and Precambrian ages precludes the Yukon composite terrane as a source. In addition, other potential sedimentary sources (i.e., accretionary prism strata and Kahiltna assemblage) do not contain Paleocene–Eocene populations (Fig. 8). Paleocurrent indicators and provenance data from Matanuska basin strata support sediment flux away from the Talkeetna Mountains and toward Cook Inlet during Cenozoic time (Trop et al., 2003; Kortyna et al., 2013; Sunderlin et al., 2014).

The Early Cretaceous (13%) population in the Tyonek Formation could have four possible source regions. Two far-field sources include the middle Cretaceous plutons in the eastern Alaska Range and Yukon-Tanana terrane (quadrangles 3, 4, and 6 on Fig. 2) and the Chisana plutonic belt located on the eastern edge of the Wrangell volcanic belt (quadrangles 13 and 14 on Fig. 2). Two possible proximal sources include the Early Cretaceous plutons in the northern Chugach Mountains (quadrangle 16 on Fig. 2) and recycling from the Kahiltna assemblage in the Talkeetna Mountains (quadrangle 11 on Fig. 2). We prefer a combination of the Chisana plutonic belt and the plutons in the northern Chugach Mountains as the source for the Early Cretaceous population because this sample lacks significant Paleozoic and Precambrian ages, which preclude the Yukon-Tanana terrane and eastern Alaska Range as a source, and Kahiltna assemblage strata are limited to the northern Talkeetna Mountains, which likely created a topographic barrier between exposures of the Kahiltna strata and the Matanuska basin. Early Cretaceous grains in the Tyonek sample are mostly <120 Ma (11 on Fig. 7), which matches the age of the Chisana belt (ca. 105–117); however, the less abundant older Early Cretaceous grains (>120 Ma) in the Tyonek sample require a slightly older source, such as the plutons in the northern Chugach Mountains (ca. 120–125 Ma). If some of the grains were derived from the Chisana plutonic belt, then a regional fluvial system would have flowed more than 350 km along strike from the Chisana belt in eastern Alaska to the Matanuska Valley region during late Oligocene–early Miocene time.

Similar to the Tyonek Formation sample in the Matanuska Valley, Paleocene–Eocene grains from the Tyonek Formation in the Cook Inlet basin (10 on Figs. 7 and 9B) are also interpreted to be sourced primarily from igneous rocks in the Talkeetna Mountains. Late Cretaceous and Jurassic–Early Cretaceous grains are interpreted to be derived from igneous rocks in the Talkeetna arc or recycled from older forearc strata in the Matanuska Valley or accretionary prism strata located in the western or northern Chugach Mountains. We again interpret the Chisana plutonic belt and the plutons in the northern Chugach Mountains as the source for the Early Cretaceous population in the Tyonek Formation. Most of the Early Cretaceous grains in this sample are <120 Ma (10 on Fig. 7), and the portion of the Wrangell volcanic belt that intruded the Chisana plutonic belt is partly coeval with deposition of the Tyonek Formation and may have triggered exhumation and denudation of the older Chisana belt rocks (e.g., Trop et al., 2012). This interpretation suggests that the inferred Oligocene–early Miocene regional fluvial system along the accretionary prism margin that transported sediment from eastern Alaska into the Cook Inlet basin may have continued during middle Miocene time.

In the Beluga Formation (9 on Figs. 7 and 9C), Paleocene–Eocene (17%), Late Cretaceous (37%), and Jurassic–Early Cretaceous (22%) grains are interpreted to represent sediment derived from igneous rocks in the Talkeetna Mountains or recycled from older strata in the Matanuska basin or accretionary prism strata located in the western or northern Chugach Mountains (quadrangles 7, 11, and 16 on Fig. 2). The Sterling Formation sample (8 on Figs. 7 and 9C) contains comparable populations of Paleocene–Eocene (30%), Late Cretaceous (33%), and Jurassic–Early Cretaceous (20%) grains that suggest a similar source region.

The Early Cretaceous population in the Beluga Formation (10%) was derived from either the Chisana plutonic belt or a combination of the small plutonic belt in the northern Chugach Mountains and recycled from older strata in the Matanuska basin and accretionary prism strata located in the western and northern Chugach Mountains (quadrangles 7, 11, and 16 on Fig. 2). We favor the latter interpretation because the majority of Early Cretaceous grains in both the Beluga and Sterling formations are >120 Ma (8 and 9 on Fig. 7), which better matches the ages of the plutons in the northern Chugach Mountains. Furthermore, the Wrangell volcanic belt has continued to grow since early Miocene time (Richter et al., 1990; Trop et al., 2012) and has created a topographic barrier between a significant portion of the Chisana plutonic rocks and the forearc basin. Therefore, the evolution of Early Cretaceous grains interpreted to be supplied by the Chisana plutonic belt compared to those contributed primarily by plutons in the northern Chugach Mountains strata is supported by the overall decrease in the Early Cretaceous population present in our detrital samples from middle Miocene (24%), to late Miocene (13%), to early Pliocene (6%), to the present-day Matanuska River (7%), as well as the change from Early Cretaceous grains dominantly <120 Ma in late Oligocene and middle Miocene strata, to those mostly >120 Ma in late Miocene and Pliocene strata. Finally, the Beluga and Sterling formations and the Matanuska River sample contain very similar proportions of all the age groups defined in this study (7–9 on Fig. 9C). We infer that this similarity reflects the formation of a less extensive drainage system similar to the modern configuration by late Miocene time.

Previous provenance interpretations for the Cenozoic forearc strata were based on clast composition of conglomerate, sandstone petrography, heavy mineral suites, isopach mapping, and paleocurrent analyses (Table 1; Calderwood and Fackler, 1972; Kirschner and Lyon, 1973; Boss et al., 1976; Hayes et al., 1976; Rawlinson, 1984; Kremer and Stadnicky, 1985). These previous studies interpreted a basin-wide change from local sediment sources, primarily the western Alaska Range, during deposition of the West Foreland Formation (Fig. 10A), to distal sources including eastern Alaska and western Canada during Tyonek Formation time (Fig. 10B), and to sources in the Yukon composite terrane, the Alaska Range, and the Chugach Mountains by late Miocene time (Fig. 10C).

Our new provenance data, based primarily on U-Pb detrital zircon geochronology, suggests a different evolution in the provenance of the Cook Inlet basin. Our interpretation reveals a change from predominantly distal interior Alaska and extensive along-strike sediment sources to a contraction of the drainage systems to the modern-day configuration. From middle Eocene to middle Miocene time, our data indicate that sediment was derived from the region north of the Alaska Range and from the Chisana plutonic belt inboard of the Wrangell belt and transported via regional fluvial systems to the forearc depositional basin (Figs. 10D and 10E). Upper Miocene forearc strata along both the arc and accretionary prism margins contain provenance signals nearly identical to the major contemporary fluvial systems in the Cook Inlet basin, signaling the establishment of these less extensive, modern-day drainage systems by late Miocene time (Fig. 10F).

Because most of the previous provenance studies did not include the original provenance data sets in publication, a direct comparison between our model and most of the older data is not possible. The one exception to this is the study by Rawlinson (1984), who showed that paleoflow in the Beluga and Sterling formations along the accretionary prism margin of the basin was directed toward the west-northwest, and that sandstone compositions were dominated by low-rank metamorphic rock fragments, sedimentary clastic grains, and chert, with an upsection increase in volcanic grains. Rawlinson (1984) interpreted the increase in volcanic grains to represent the reworking of ash-fall deposits from the Aleutian volcanic arc. These data fit well with our interpreted provenance of the accretionary prism strata for the Beluga and Sterling formations on the accretionary prism margin. The accretionary prism strata are located to the east of the basin and are mostly composed of low-rank metamorphic sedimentary strata that would be expected to produce the sandstone compositions observed by Rawlinson (1984).

Arc margin. Our new provenance data support previous sedimentological studies that inferred that transverse alluvial fan–fluvial systems derived dominantly pebble- and cobble-sized sediment from the adjacent western Alaska Range region beginning in middle Eocene time (ca. 48–40 Ma; Fig. 11A; Calderwood and Fackler, 1972; Houston, 1994; Wahrhaftig et al., 1994; this study). Coarse-grained deposition along the arc margin continued through the late Eocene but was replaced by finer-grained alluvial fan–fluvial systems by late Oligocene or early Miocene time (ca. 30–20 Ma; Detterman et al., 1976; Hite, 1976; Magoon and Egbert, 1986; Flores et al., 1994; Flores et al., 1997). The creation of topography required to produce the observed middle Eocene coarse-grained strata is supported by Eocene exhumation in the western Alaska Range documented by K-feldspar thermochronology (ca. 50–40 Ma; Fig. 12A; Benowitz et al., 2012b). We infer that the pulse of Eocene exhumation facilitated deposition of the coarse-grained middle Eocene alluvial fan deposits, but overall slow denudation rates since the Eocene (e.g., Benowitz et al., 2012b) have limited the sediment derived from the western Alaska Range relative to other sources during Oligocene to Recent time.

A regional fluvial system is inferred to have flowed generally southward along the arc margin of the forearc basin during middle Eocene to middle Miocene time (ca. 48–11 Ma; arrows on Figs. 11A and 11B). The presence of this fluvial system is based on middle Eocene, early Miocene, and middle Miocene strata, so it is not well constrained during late Eocene to Oligocene time. The headwaters of this fluvial system were located north of the Alaska Range in interior Alaska. We infer that exhumation of the central Alaska Range during middle Miocene time (ca. 16–11 Ma) created a new source of sediment for this axial system (Fig. 11B). Textural, compositional, and U-Pb detrital zircon data from the Tanana basin to the north of the central Alaska Range suggest exhumation of the range by middle Miocene time (Ridgway et al., 2007; Brennan and Ridgway, 2015). Stratigraphic data from subsurface wells in the Susitna basin, located between the central Alaska Range and the Cook Inlet basin, document a middle Miocene-on-middle Eocene unconformity that also suggests pre-middle Miocene exhumation north of the Cook Inlet basin (Fig. 3; Stanley et al., 2013).

Continued exhumation of the central Alaska Range eventually led to separate drainage systems north and south of the range (Fig. 11C). Thermochronologic data from the central Alaska Range indicate rapid exhumation of that area during late Miocene–Pliocene time (ca. 6–5 Ma; Fig. 12A; Plafker et al., 1992; Fitzgerald et al., 1995). Our new detrital data indicate that by late Miocene time (ca. 11–5 Ma), the previously extensive middle Eocene–middle Miocene fluvial system was partitioned into a configuration similar to the modern distinct forearc and interior Alaskan drainage systems.

Accretionary prism margin. Along the accretionary prism margin of the basin, our new provenance interpretation suggests a late Oligocene to middle Miocene (ca. 26–11 Ma) regional fluvial system that stretched from the Chisana plutonic belt east of the present-day Wrangell volcanic belt to the Cook Inlet forearc basin (Fig. 13A). We infer that during middle to late Miocene time, surface exhumation along strike to the east of the Cook Inlet basin isolated sediment sources located east of the Matanuska Valley from the active forearc depositional basin.

The Wrangell volcanic belt, situated east of the Matanuska Valley along the Alaska-Yukon Territory border, has produced geochronologic and palynologic data that illustrate northwestward younging from late Oligocene to Pleistocene time (Richter et al., 1990; Skulski et al., 1992; Ridgway et al., 1995; Trop et al., 2012). The Alaskan part of the Wrangell belt is mainly late Miocene and younger and is interpreted to have created a northwestward-migrating topographic barrier that sequestered sediment sources to the east from the Cook Inlet depositional system, including the Chisana belt. The Copper River basin, between the Matanuska Valley and the Wrangell volcanic belt, has accumulated a relatively insignificant amount of Paleogene and Neogene strata (<1.2 km maximum; Trop and Ridgway, 2007). This suggests that this basin has either served primarily as a Cenozoic sediment-bypass region or that it has been exhumed and much of the sedimentary package has been eroded and transported to other basins. North and south of the Matanuska Valley, thermochronologic data from the Talkeetna Mountains and northern Chugach Mountains, respectively, indicate that exhumation began during late Eocene to early Oligocene time (ca. 35–30 Ma) and continued into early-middle Miocene time (ca. 16–11 Ma; Fig. 12A; Little and Naeser, 1989; Hoffman and Armstrong, 2006; Arkle et al., 2013). We propose that this exhumation resulted in the Matanuska Valley region being an area of sediment erosion and bypass by middle Miocene time.

In summary, we infer that by late Miocene time (ca. 11 Ma), the topographic expression of the Wrangell volcanic belt, along with surface uplift of the Copper River basin and Matanuska Valley region, exceeded the erosional power of the late Oligocene to middle Miocene (ca. 26–11 Ma) fluvial depositional systems that had previously transported sediment from eastern Alaska to the Cook Inlet basin. Consequently, the headwater regions of the middle Miocene fluvial system shifted westward from eastern Alaska to the modern Matanuska Valley (Fig. 13B).

Previous studies suggest that relatively continuous magmatic activity is typically recorded in detrital zircon signatures from forearc basin strata (e.g., DeGraaff-Surpless et al., 2002; Clements and Hall, 2011; Jacobson et al., 2011). This record in forearc basins is represented in the detrital zircon populations by the youngest peak, which is typically: (1) one of the largest peaks in the population, (2) overlapping with or slightly older than the depositional age of the sediment from which the zircons are extracted, and (3) systematically younger up through the stratigraphic section.

In contrast, the largest peaks in our detrital zircon data from the southern Alaska Cenozoic forearc basin system are Late Cretaceous or Paleocene–early Eocene regardless of the depositional age of the host rock (Figs. 6 and 7). We infer this trend to indicate that magmatic activity adjacent to the forearc basin sharply decreased after early Eocene time and was at a minimum during middle Eocene–Pliocene time. A histogram of Cenozoic detrital zircon ages from the forearc strata and two modern river systems (Susitna and Matanuska rivers) from our study combined with data from a recent compilation of the current bedrock geochronologic database from the western and central Alaska Range (Finzel et al., 2011) reveals that the most abundant detrital zircon ages are Paleocene–early Eocene (65–45 Ma; Fig. 12B). A slight increase in ages during middle-late Eocene time (40–34 Ma; Fig. 12B) is interpreted to be related to slab-window magmatism in the Caribou Creek volcanic field associated with subduction of a spreading ridge (e.g., Lytwyn et al., 2000; Bradley et al., 2003; Sisson et al., 2003; Cole et al., 2006) and a brief pulse of magmatic activity (<10 m.y.) in the Aleutian-Alaska Range volcanic arc (e.g., Moll-Stalcup, 1994). Therefore, it appears that late Eocene–Pliocene magmatic arc activity in south-central Alaska was minimal. This relative magmatic gap, during which ∼8 km of Oligocene–Pliocene strata were deposited in the forearc basin, indicates that the typical forearc basin model with coeval volcanic arc sources of sediment is not appropriate for the Cenozoic southern Alaska system.

New U-Pb detrital zircon geochronology and compositional data from conglomerate of the southern Alaskan Cenozoic forearc basin system demonstrate profound long-term changes in sediment sources. We suggest that these changes were linked to two major Cenozoic tectonic events along the southern Alaska convergent margin: Paleocene to Eocene spreading-ridge subduction and Oligocene to Recent subduction of an oceanic plateau.

In Late Cretaceous time, a series of marine forearc basins extended east from the Cook Inlet to the Wrangell–Saint Elias Mountains in eastern Alaska and southwestward along the Alaska Peninsula (CIB, MB, and WB on Fig. 14A). Within these basins, the Cretaceous Matanuska and Kaguyak formations and other unnamed units record dominantly marine depositional systems with the majority of sediment input from the adjacent Jurassic–Cretaceous volcano-plutonic belts that represent arc magmatism (Fig. 14A; Magoon et al., 1976a; Magoon et al., 1976b; Fisher and Magoon, 1978; Detterman and Miller, 1985; Magoon and Egbert, 1986; Trop, 2008). During this time period, the southern Alaska forearc basin system was similar to the models based on the Great Valley sequence (Dickinson and Seely, 1979; Dickinson, 1995; DeGraaff-Surpless et al., 2002) with the bulk of the sediment derived from an adjacent coeval magmatic arc system and deposition in marine settings (Trop, 2008).

During Paleocene and Eocene time (ca. 62–50 Ma), a spreading ridge was subducted from west to east across the entire southern Alaska convergent margin, encountering the Cook Inlet–Matanuska Valley region at ca. 57–54 Ma (Fig. 3; Bradley et al., 2003; Haeussler et al., 2003; Cole et al., 2006; Kortyna et al., 2013). Spreading-ridge subduction causes progressively younger oceanic crust to be consumed, producing a shallow slab that can extend both several hundreds of kilometers or more away from the spreading ridge and inboard from the subduction zone beneath the overriding plate (Cloos, 1993). This process often results in significant lithospheric thinning through physical erosion of the mantle lithosphere from beneath the overriding plate and thermal erosion from upwelling asthenosphere (Cole and Stewart, 2009; Jacobson et al., 2011; Ling et al., 2013). The net effect of this event upon the forearc region is typically short-term isostatic uplift of as much as several kilometers, followed by rapid subsidence to near original elevations after the ridge passes (Cloos, 1993; Madsen et al., 2006; Breitsprecher and Thorkelson, 2009; Groome and Thorkelson, 2009; Benowitz et al., 2012a). Furthermore, development of a slab window due to ridge subduction can lead to temporary cessation of the volcanic arc (e.g., Dickinson and Snyder, 1979; Thorkelson, 1996; Gorring and Kay, 2001).

The detrital record from the forearc in southern Alaska reveals the effect of these processes on the forearc basin dynamics and sediment dispersal. During ridge subduction, an unconformity developed in the forearc basin that records either erosion of the Late Cretaceous forearc strata or nondeposition during early Paleocene time as the ridge passed from west to east along the margin (Fig. 3; Trop et al., 2003; White and Bradley, 2006; Ridgway et al., 2012). This Paleocene unconformity is interpreted to represent a brief period of isostatic uplift of the forearc region. Following ridge subduction, nonmarine depositional systems were quickly reestablished, reflecting rapid subsidence that frequently occurs after the ridge passes. Provenance studies of forearc strata in the Matanuska Valley that were deposited contemporaneously with ridge subduction have sand-sized and conglomeratic modes dominated by igneous detritus and record exhumation of the adjacent arc rocks, as well as erosion of local volcanic centers attributed to slab-window magmatism (Kortyna et al., 2013). Strata adjacent to the accretionary prism locally contain detrital modes and ages that indicate erosion of the accretionary prism as well (Little, 1988; Trop et al., 2012).

Provenance data from Late Cretaceous strata on the arc margin of the basin indicate dominantly local arc sources prior to ridge subduction, whereas data from middle Eocene strata on the arc margin that postdate ridge subduction suggest sediment sources located north of the present-day Alaska Range inboard of the mostly dormant Late Cretaceous to Paleocene volcanic arc (Figs.14A and 14B; Magoon et al., 1976a, 1976b; Fisher and Magoon, 1978; Detterman and Miller, 1985; Magoon and Egbert, 1986). We infer that after passage of the spreading ridge, widespread subsidence of the upper plate occurred due to crustal thinning that resulted from a combination of physical and thermal thinning of the lithosphere (e.g., Cole and Stewart, 2009) as well as denudation of the Late Cretaceous to Paleocene arc associated with isostatic uplift of the upper plate. Crustal thinning in combination with early Eocene cessation of previously prolific arc activity (>50 Ma; Fig. 12B) allowed for the expansion of the forearc drainage networks into the Late Cretaceous–Paleocene retroarc area.

Along the accretionary prism margin of the basin, comparison of previous (Trop, 2008) and our new provenance data reveals a shift from dominantly local arc sources prior to ridge subduction (Fig. 14A) to interpreted source areas located along strike to the east in the exhumed part of the forearc basin (Fig. 14B). Due to passage of the spreading ridge from west to east, the areas to the west, including the Cook Inlet basin, subsided earlier than areas to the east, including the Matanuska Valley and areas farther east, which were still undergoing isostatic uplift. We infer that this progression caused an overall west-dipping tilt to the forearc region that resulted in erosion of the eastern part and deposition in the western part of the forearc basin system.

Large-scale axial fluvial systems in modern or ancient forearc basins are not well documented. However, uplift and erosion of forearc basin strata followed by axial transport of the detritus in marine settings have been interpreted in areas exposed to spreading-ridge or flat-slab subduction. Examples include subduction of a spreading ridge during late Cenozoic time in southern California (Fletcher et al., 2007), an aseismic ridge along the Kronotsky Peninsula in the Kamchatka forearc after Miocene time (Marsaglia et al., 1999), and the Nazca aseismic ridge during Cenozoic time between 13.5°S and 17°S along the South American Andean margin (Hsu, 1992; Machare and Ortlieb, 1992); in all these areas, forearc uplift resulted in erosional denudation of older forearc strata and recycling of the sediment into younger forearc basinal strata. Conversely, low-angle Laramide subduction beneath the Late Cretaceous to Eocene forearc in California resulted in the introduction of detritus derived from areas inboard of the arc but did not trigger the development of large-scale axial transport of sediment (Sharman et al., 2014).

Flat-slab subduction of the Yakutat oceanic plateau is interpreted to have initiated along the outboard margin of southern Alaska in late Eocene to early Oligocene time (ca. 40–30 Ma; Benowitz et al., 2011; Finzel et al., 2011; Ridgway et al., 2012; Arkle et al., 2013). Yakutat subduction resulted in a zone of broad, diffuse deformation as it increased the transfer of compressive stresses inboard and upward to the upper plate, similar to other regions affected by flat-slab subduction (e.g., Dickinson and Snyder, 1979; Jordan and Allmendinger, 1986; Gutscher et al., 2000; Hampel, 2002; Lallemand et al., 2005; Espurt et al., 2008).

Extensive geochronologic and thermochronologic data from the area above the present-day flat-slab region have been inferred to demonstrate the links between flat-slab subduction of the Yakutat microplate and magmatism and exhumation in the upper plate (Finzel et al., 2011). Bedrock and detrital thermochronologic data from above the flat-slab region indicate a significant increase in exhumation beginning ca. 40–35 Ma with generally sustained exhumation until another increase starting at ca. 10–5 Ma (Fig. 12A). Bedrock and detrital geochronologic data show that magmatism briefly resumed in the western and central Alaska Range after the passage of the spreading ridge (ca. 40–30 Ma on Fig. 12B), but very limited arc-related magmatic activity is recorded after 30 Ma until the initiation of the modern arc outside the present-day flat-slab region in the western Alaska Range in Quaternary time (2–0 Ma on Fig. 12B).

The detrital record from the Cenozoic forearc basin system in southern Alaska records the effect of this temporally sustained exhumation across a widespread region in the upper plate. Provenance signatures from both the arc and accretionary prism margins of the forearc basin suggest a shift in sediment source regions from interior and eastern Alaska, respectively, to south-central Alaska by late Miocene time (Fig. 14C). The similarity between detrital zircon populations in late Miocene and early Pliocene strata and those in the major modern fluvial systems in the modern forearc basin indicates that the late Miocene–Pliocene watershed was comparable to the modern-day configuration (1 and 2 on Fig. 6 and 7–9 on Fig. 7). We infer that progressively increasing exhumation above the flat-slab region since Oligocene time has resulted in a change from a middle Eocene to middle Miocene extensive, regional drainage system with its headwaters in interior and eastern Alaska to a more localized late Miocene to Recent drainage system with its headwaters in the central Alaska Range and Matanuska Valley.

U-Pb dating of detrital zircons in both ancient strata and modern river sediment, integrated with clast compositional data from conglomerate, has resulted in new provenance and paleogeographic interpretations that incorporate the effects of spreading-ridge and flat-slab subduction for the southern Alaska Cenozoic forearc basins. We place these results within the context of a general provenance and depositional model for forearc basins that form under changing geodynamic plate boundary conditions.

Subduction of a spreading ridge or an oceanic plateau can significantly modify the sediment transport networks and provenance of forearc basin systems. The existing provenance and depositional models generated by studies of “more typical” forearc basins that were subject to long-term subduction of primarily normal oceanic crust demonstrate that (1) the depositional environments in the basin shallow gradually over time from marine to nonmarine systems; (2) the primary sources of sediment in the early stages of forearc basins are the proximal magmatic arc and subaerially exposed parts of the accretionary prism; and (3) provenance potentially evolves into the more distal retroarc region as the arc becomes more dissected and drainage networks are expanded (e.g., Ingersoll, 1978a, 1978b; Dickinson and Seely, 1979; Ingersoll, 1979, 1983; Busby-Spera, 1986; Moxon and Graham, 1987; Morris and Busby-Spera, 1988; Linn et al., 1991, 1992; Dickinson, 1995; Busby et al., 1998; DeGraaff-Surpless et al., 2002; Surpless et al., 2006; Surpless and Beverly, 2013). These processes result in detrital zircon populations from forearc strata that become more heterogeneous upsection in response to more varied source terranes with time (DeGraaff-Surpless et al., 2002; Cawood et al., 2012).

Forearc basins that have experienced shallow subduction episodes, in contrast, will contain a very different provenance record. Subduction of a spreading ridge may result in an abrupt change from local arc and prism sediment sources to sediment derived from the former retroarc region and exhumed portions of the adjacent forearc. The detrital zircon signature should reflect these processes by recording an abrupt increase in the diversity of detrital zircon ages that reflects new sediment sources from outside the forearc system after passage of the ridge has resulted in reconfiguration of the upper plate. Flat-slab subduction of an oceanic plateau or microplate may lead to (1) sustained exhumation, erosion, and sediment bypass of the part of the forearc basin above the flat slab, (2) a contraction of the forearc system with dominant sediment sources located above the flat-slab region, and (3) long-term cessation of arc activity. The detrital zircon signature in this case should reveal exhumation of the region above the flat slab, including older forearc strata and exhumed parts of the extinguished arc. In addition, the youngest peak in the signature should not get younger with younger depositional age of the host strata because the maximum depositional age based on detrital zircons is pinned by the last period of major arc activity prior to insertion of a flat slab. This study provides a foundation for new tectonic and provenance models of forearc basins that have been modified by shallow subduction processes. Our results may also help to facilitate the use of U-Pb dating of detrital zircons to understand basins with composite tectonic histories that form during changing plate boundary conditions.

We would like to thank the Alaska Division of Geological & Geophysical Surveys for providing logistical support for this project. Additional funding for E.F. was received from ExxonMobil. K.R. was supported by funding from the National Science Foundation (NSF). J.T. was supported by funding from the NSF and Alaska Division of Geological & Geophysical Surveys. We also appreciate useful discussions about Cook Inlet geology with David LePain, Bob Gillis, and Dave Doherty. The Arizona LaserChron Lab was used to acquire the data presented here; the NSF Instrumentation and Facilities Program provided support for the lab. Kathy Surpless and Associate Editor Terry Pavlis provided useful reviews that helped us to greatly improve the manuscript.