Subduction erosion is an important process at convergent margins but evidence in the geologic record is scarce because the involved materials are typically lost into the mantle. Detrital zircon dating of accretionary complex sediment allows us to document episodic growth that can be linked to possible triggers. Detrital zircons (n = 2508) from the Mesozoic Chugach accretionary complex, southern Alaska, have U-Pb ages that record progressive subduction accretion punctuated by two periods of tectonic erosion. Lithology and maximum depositional ages permit division of the Chugach accretionary complex into four main units. The oldest, the blueschist-greenschist unit, represents partially subducted sediment associated with a Jurassic oceanic arc, with subduction erosion from 180 to 170 Ma. The end of this erosion period is dated by the oldest maximum depositional age of the Potter Creek assemblage, a <156–169 Ma unit consisting of chert, argillite, and volcanic rocks. A trondhjemite pluton that intrudes the forearc was caused by ridge subduction at 125 Ma and resulted in a second period of subduction erosion lasting until 104 Ma.
The end of Aptian–Albian erosion was marked by deposition of massive sandstone and conglomerate of the 101–91 Ma McHugh Creek assemblage. This influx of clastic sediment is interpreted to have occurred in response to the collision of Wrangellia with North America. This event and the erosional event preceding it steepened the forearc region, allowing mass wasting of forearc crust into the trench, filling it by 89 Ma, the oldest maximum depositional age of the Valdez Group flysch. Clasts in conglomerate include granodiorite from the basement of the Talkeetna arc dated from 199 ± 3 Ma to 179 ± 3 Ma, and sandstone clasts with maximum depositional ages of 100 Ma. From 89 Ma to at least 72 Ma, the Valdez Group flysch was deposited via turbidite fans onto the oceanic crust beyond the trench and accreted as imbricate thrust slices that retained coherent bedding.
The source for detrital zircons in the Potter Creek assemblage is likely a Middle–Late Jurassic oceanic arc, possibly the Talkeetna arc. The abundance of zircons from ash fall tuffs is consistent with easterly winds and suggests the Chugach accretionary complex was south of latitude 25°N in the Late Jurassic. The dominant source for the Albian McHugh Creek assemblage and the Upper Cretaceous Valdez Group flysch was likely the arc associated with the Coast Mountains batholith. Jurassic (ca. 165 Ma) zircons in the McHugh Creek assemblage could have been derived from exhumed plutons, or may be second-cycle zircons derived from the Potter Creek assemblage. The first appearance of Proterozoic and Archean zircons in the Valdez Group records the breakdown of topographic barriers formed by the accreted arc terranes as rivers encroached into continental North America by the middle Late Cretaceous.
Subduction zones are responsible for crustal growth by processes of arc magmatism and forearc accretion, but they are also an important locus for recycling of crust into the mantle via subduction erosion (von Huene and Scholl, 1991; Clift and Vannucchi, 2004; Vannucchi et al., 2004, 2008; von Huene et al., 2004; Clift et al., 2009; Stern, 2011). Subduction erosion has been documented in a large percentage of modern margins and accounts for nearly 30% of the total recycling of continental crust into the mantle (Clift et al., 2009). Identifying subduction erosion in the geologic record is difficult; direct evidence is scarce because the involved materials are lost into the mantle. Furthermore, other mechanisms such as strike-slip faulting can also remove crust from its original position (Karig, 1980; Grove et al., 2008).
At least three areas in North America are inferred to have experienced Mesozoic subduction erosion. These include the Catalina Schist of southern California where subduction erosion ended by 95 Ma (Grove et al., 2008), the Franciscan Complex of northern California where erosion ended by 123 Ma (Dumitru et al., 2010), and the Chugach accretionary complex of southern Alaska. In Alaska, two episodes of subduction erosion have been postulated: one from 180 to 160 Ma, identified mainly on the basis of the migration of the Talkeetna arc (Clift et al., 2005a), and another that occurred from 125 to 100 Ma (Amato and Pavlis, 2010). Subduction erosion has also been identified in ancient accretionary complexes in Japan (Isozaki et al., 2010; Aoki et al., 2012) and Italy (Remitti et al., 2011).
The Mesozoic Chugach accretionary complex (Fig. 1) is exposed in southern Alaska and consists of sedimentary rocks with ages ranging from ca. 200 Ma to 65 Ma (Clark, 1973; Plafker et al., 1989; Amato and Pavlis, 2010). These rocks provide a detailed history of subduction zone processes including arc magmatism, sedimentation in and beyond the trench, metamorphism of subducted sediments, and subduction erosion. The geochronologic and lithologic variations in the accretionary complex record aspects of the tectonic evolution of the southern Alaska margin.
The Chugach accretionary complex is ideal for detailed investigation of its architecture because of the direct path for sediment from the arc to the accretionary complex. Thus, detrital zircon ages from the complex effectively record the age of sedimentation and thus the evolution of the convergent margin. This is in contrast with the Franciscan Complex, where arc-derived sediment and tuffs were mainly transported east, toward the continent and away from the accretionary complex, because of its location in the mid-latitudes with strong westerly winds, resulting in a much lower proportion of contemporaneous zircons reaching the Franciscan trench (Dumitru et al., 2010).
This paper expands on our earlier studies of the detrital zircon ages from the Chugach accretionary complex (Amato and Pavlis, 2010; Kochelek et al., 2011) that focused on a well-exposed transect along Turnagain Arm (Fig. 2). This expanded study includes nearly 2500 detrital U-Pb ages from along and across strike in the accretionary complex.
The magmatic, tectonic, and depositional history of the margin from southern Alaska to southern British Columbia is complicated and controversial (see reviews by Plafker et al., 1994; Trop and Ridgway, 2007; Pavlis and Roeske, 2007). The following summary focuses on the area of southern Alaska between Anchorage and Valdez; there may be significant differences in the timing of similar events along strike, particularly in the Canadian part of the margin because of oblique collisions with the trench. Evidence pertaining to the evolution of the Mesozoic convergent margin of southern Alaska comes from volcanic arc complexes, the accretionary complex, and high-pressure/low-temperature metamorphic rocks.
Lithostratigraphic terranes of southern Alaska (Fig. 1) were invoked to explain the apparent lack of a shared history across major faults (e.g., Jones et al., 1987; Silberling et al., 1994). This terrane nomenclature is useful when assigned to possibly far-traveled island arcs but can be misleading when used to describe accretionary complexes and post-accretionary basins. The Wrangellia composite terrane (i.e., Wrangellia, or the Insular superterrane) includes the Alexander terrane, the Wrangellia terrane, and the Peninsular terrane (Plafker and Berg, 1994). The Peninsular terrane consists mainly of the Talkeetna arc (DeBari and Coleman, 1989; Rioux et al., 2010) and associated sedimentary rocks. The Mesozoic Chugach terrane, now referred to as the Chugach accretionary complex, is inferred to be associated with subduction beneath the Wrangellia composite terrane. The Chugach accretionary complex is separated from the inboard arc terranes by a major structure called the Border Ranges fault system (BRF) that is inferred to have been the plate boundary between the subducting slab and the overriding plate during the Jurassic formation of the Talkeetna arc (Clift et al., 2005b) until it evolved into a major right-lateral strike-slip fault (Pavlis and Roeske, 2007). The Prince William terrane is essentially the Cenozoic continuation of the Chugach accretionary complex (Garver et al., 2012). The Kahiltna terrane represents backarc syn- and post-collisional sedimentation but most workers now refer to it as the Kahiltna Basin or Kahiltna assemblage (Wallace et al., 1989; Kalbas et al., 2007; Hampton et al., 2010). The Yukon-Tanana composite terrane includes both North American basement and cover together with allochthonous Paleozoic–Mesozoic assemblages that were amalgamated during Mesozoic events along the North American margin and overprinted by Early Cretaceous extension (Dusel-Bacon and Williams, 2009).
Mesozoic magmatism coeval with the formation of the Chugach accretionary complex is generally inferred to be related to north- or east-dipping subduction (present coordinates) because of the presence of high-pressure metamorphic rocks and lower-grade subduction-complex sedimentary rocks exposed south of the arcs and because some magmatism north of the arcs was interpreted as occurring in a backarc setting (e.g., Reed and Lanphere, 1973; Roeske et al., 1989; Plafker and Berg, 1994; Clift et al., 2005a; Clift et al., 2005b; Trop and Ridgway, 2007; Hampton et al., 2010). Trondhjemite-tonalite plutons (Fig. 2) intruded the accretionary complex around 130–120 Ma (Pavlis et al., 1988; Barnett et al., 1994; Amato and Pavlis, 2010). These represent magmatism related to reestablishment of subduction following a period of strike-slip faulting (Barnett et al., 1994) or subduction of an oceanic spreading ridge (Pavlis and Roeske, 2007; Amato and Pavlis, 2010).
Outboard, and primarily deposited on Wrangellia composite terrane basement, are a series of sedimentary basins (Fig. 3) formed in the forearc of the Jurassic–Cretaceous arc systems (see summary in Trop and Ridgway, 2007). The Middle Jurassic–Upper Cretaceous basin deposits are marine, >3800 m thick, and include a mid-Cretaceous disconformity that represents uplift and subaerial exposure of the forearc at 130–115 Ma, which was caused by subduction of a spreading ridge (Trop and Ridgway, 2007). The disconformity is coeval with the emplacement of forearc trondhjemite plutons (Pavlis, 1982; Pavlis et al., 1988).
The Chugach accretionary complex (Berg et al., 1972) is the Mesozoic forearc accretionary complex of southern Alaska (Figs. 1, 2) and as traditionally defined includes three major units defined by significant changes in protoliths and/or metamorphic grade (Plafker et al., 1989, 1994). The contacts between these units are generally not exposed. The first is referred to as the blueschist-greenschist unit and is represented by small-volume exposures of blueschist-greenschist–facies rocks that formed during Late Triassic–Early Jurassic subduction (Roeske, 1986; Sisson and Onstott, 1986; Bradley et al., 1999). Protoliths include mafic volcanic rocks, chert, and argillite. The constraints on the timing of blueschist metamorphism comes from metamorphic ages ranging from ca. 205 Ma on Kodiak Island (Roeske et al., 1989) to 192 Ma near Seldovia (Bradley et al., 1999; Lopez-Carmona et al., 2011).
The second unit is the mélange assemblage of Pavlis and Roeske (2007), which includes the McHugh Complex in the study area (Clark, 1973) and the Uyak Complex on Kodiak Island (Connelly, 1978). It is interpreted as the product of continuous subduction from the Jurassic to the latest Cretaceous (e.g., Plafker et al., 1994). Fossil control on the age of sedimentation is sparse, with the youngest radiolarians in the study area yielding ages in the range of Berriasian to Hauterivian (Nelson et al., 1987), or 145–131 Ma (Gradstein et al., 2005). The Uyak Complex has radiolarian ages that are as young as late Valanginian to late Aptian, or ca. 135–113 Ma (Connelly, 1978). These ages fall within the erosional period defined by Amato and Pavlis (2010), but the fossil localities have not been re-examined in the context of modern detrital zircon work. Thus, because the cherts probably represent pelagic assemblages formed far from the paleotrench, their age could pre-date the age of accretion by tens of millions of years (e.g., Pavlis, 1982).
Clark (1973) recognized a metavolcanic unit and a metaclastic unit in the McHugh Complex, but noted that because of extensive mixing and poor age control they were combined into one map unit. Amato and Pavlis (2010) recommended that the McHugh Complex of Clark (1973) be divided into two separate assemblages that have distinctive lithologies and ages (see below). The older assemblage consists of an argillite-chert-volcanic assemblage exposed outboard (south) of the BRF or outboard of the blueschist-greenschist units, where present. Pillow basalts and limestone are rare. Shale is interbedded at a millimeter to decimeter scale with pale green, fine-grained tuffaceous sandstones and siltstones. Aphyric basaltic lava is present in abundances up to 30% locally, and shearing has produced stratal disruption and a steep northwest-dipping cleavage. The dominance of shale and chert indicates deep-water deposition, with tuffaceous sandstones and tuffs indicating the proximity of a volcanic source; thus, the depositional setting was likely a distal turbidite fan receiving ash-fall tuffs from an arc (Clift et al., 2012). The limestone blocks were interpreted as olistoliths from collapsing seamounts with a carbonate carapace (Clift et al., 2012). This assemblage was interpreted as material from a paleo–subduction channel that never reached depths of more than ∼7–10 km (Clift et al., 2012). The younger, outboard part of the mélange assemblage includes abundant greywacke and conglomerate, and minor argillite/shale. The majority of this unit consists of massive medium- to coarse-grained sandstone with a pervasive cleavage. Conglomerate is abundant within the central outcrops along Turnagain Arm (Fig. 2). The conglomerate clast types are dominated by argillite, chert, volcanic rocks, and granitoids, with minor limestone, sandstone, gabbro, and ultramafic rocks (Clift et al., 2012). The paucity of fine-grained material in this unit, the lack of bedding or grading in the sandstones, and the coarse grain sizes of the conglomerate indicate deposition via mass wasting in a setting proximal to the source (Clift et al., 2012).
The youngest unit of the Chugach accretionary complex is the flysch unit known in the study area as the Valdez Group. It consists of Cretaceous clastic turbidite-facies sedimentary rocks extending ∼2000 km along strike and up to 100 km wide (Nilsen and Zuffa, 1982; Dumoulin, 1987; Plafker and Berg, 1994) and includes sandstone, slate, and phyllite flysch deposits with minor conglomerate and mafic igneous rocks. The Valdez Group is interpreted to be correlative with the Sitka Graywacke, Shumagin Formation, Kodiak Formation, and the flysch facies of the Yakutat Group, based on lithology and faunal similarities (Plafker et al., 1994). Fossils (Jones and Clark, 1973) indicate a depositional age of Campanian to Maastrichtian (84–65 Ma using the time scale of Gradstein et al., 2005). Detrital zircons from sandstones yielded maximum depositional ages ranging from 83 Ma to 65 Ma (Kochelek et al., 2011) and a granite clast within the unit was dated at 221 Ma (Bradley et al., 2009). The Valdez Group typically has coherent bedding, but a section of the Valdez Group at the contact with the McHugh Complex near Seldovia (Fig. 1) forms a mélange referred to as the Iceworm mélange (Kusky et al., 1997). The interbedded sandstone and slate is interpreted as a deep-water turbidite facies (Nilsen and Zuffa, 1982; Dumoulin, 1987; Sample and Reid, 2003; Kochelek et al., 2011; Clift et al., 2012). Trenchward of the Valdez Group, Paleocene–Eocene sedimentary and volcanic rocks (Orca Group, Ghost Rocks Formation, and Sitkalidak Formation) have been preserved. These have a similar provenance and depositional setting as the Valdez Group (Dumoulin, 1988; Farmer et al., 1993).
U-Pb GEOCHRONOLOGY RESULTS
For details of the analytical techniques and data reduction procedures, see the Appendix and the GSA Data Repository1. A summary of ages is in Table 1, and the U-Pb data are given in Table 2 and in Table DR1 in the Data Repository.
We dated detrital zircons from the mélange and flysch assemblages. Eight samples of blueschist from Seldovia were crushed for zircon separations, but none yielded zircons. We also dated plutonic clasts from the mélange assemblage and re-dated a trondhjemite pluton intruding the mélange assemblage. Based on the results of Amato and Pavlis (2010) and this study, we now divide the McHugh Complex into two separate units, the Potter Creek assemblage and the McHugh Creek assemblage (Fig. 2).
Potter Creek Assemblage
We dated 16 samples of sandstone from the Potter Creek assemblage in the Anchorage area. The northernmost Potter Creek exposures are found in the Eklutna River area (Fig. 2). Two samples from here yielded maximum depositional ages (MDAs) of 166 Ma and 164 Ma. Approximately 10 km south-southwest at Mount Magnificent, three samples yielded MDA of 168 Ma, 162 Ma, and 160 Ma.
The next section of the study area (Flattop–Turnagain Arm; Fig. 2) is the most densely sampled region. Flattop Mountain is 8 km northeast of Turnagain Arm and one sample west of the peak has a MDA of 156 Ma. A sample from a ridge trending northwest from McHugh Peak, near Rabbit Creek, yielded a MDA of 161 Ma. At Turnagain Arm, seven samples were dated on a cross-section perpendicular to foliation. The westernmost sample (10AnJ-41), located adjacent to the inferred trace of the BRF bounding the east side of Potter’s Marsh (Fig. 4), yielded a MDA of 167 Ma. Re-dating of this sample at the University of California–Santa Barbara Laser Ablation Split Stream facility yielded an MDA of 168 Ma, but with a narrower spread of ages within the youngest peak. A nearby sample (10AnJ-42) yielded a MDA of 170 Ma. There is a 2 km separation from that point to the next two samples that have MDAs of 151 Ma and 149 Ma. The progression of younging ages to the southwest ends with the next samples (09AnJ-17 and 08MS-08) which have MDAs of 160 Ma and 163 Ma.
The southernmost section is at Skilak Lake, ∼70 km southwest of Turnagain Arm (Fig. 1). Three samples of the mélange there have MDAs of 162–158 Ma.
Taken together (Fig. 5), the samples of Potter Creek assemblage have the youngest individual grains ranging from 165 ± 2 Ma to 144 ± 1 Ma, and MDAs ranging from 169 Ma to 156 Ma. The main population of the 910 total ages is from 204 to 144 Ma. These ages form one broad peak centered at 164 Ma but with a long tail toward older ages. There are 18 grains older than 204 Ma. These are mainly Triassic and range from 232 to 209 Ma. Other ages include three grains of Permian age (282 Ma, 290 Ma, and 293 Ma), a single Mississippian age (338 Ma), and a single Devonian age (391 Ma). Only two samples had Proterozoic zircons: sample 10AnJ-42 had one grain at 1854 ± 20 Ma, and sample 10AnJ-49 had one grain at 1644 ± 3 Ma.
McHugh Creek Graywacke-Conglomerate Assemblage
We dated 19 samples of this unit (Fig. 6). In the region from the Eklutna River north to Friday Creek, rocks from this assemblage are present in all locations. The samples from Friday Creek, Knik River, Eklutna River, Mount Magnificent, and Eagle River have MDAs ranging from 104 to 91 Ma.
The McHugh Creek greywacke-conglomerate assemblage is exposed on Flattop Mountain, but no samples were dated. Along Turnagain Arm we dated nine sandstones and two sandstone clasts within the Beluga Point conglomerate (Figs. 2, 4). The nine sandstones have an overall younging trend from the northwest (closer to the BRF) to the southeast (away from the arc). The three samples closest to the Potter Creek unit have MDAs of 100 Ma, 101 Ma, and 100 Ma. The next sample, located ∼1 km to the southeast, has a MDA of 93 Ma. The next samples are located another 3 km to the southeast near the conglomerates at Beluga Point. The graywacke sample (08AnMS-05) has a MDA of 98 Ma. Two sandstone clasts were analyzed from near this sample. They yielded MDAs of 99 Ma and 104 Ma. Between Beluga Point and the Eagle River fault, four samples had MDAs of 101 Ma, 101 Ma, 95 Ma, and 97 Ma.
An interesting sample (09AnJ-33) is present just inboard of the Eagle River fault. It has an argillite-chert composition similar to samples within the Potter Creek mesomélange, and the U-Pb detrital zircon ages also match this older unit, with an MDA of 168 Ma, indicating this exposure represents a slice of Potter Creek assemblage intercalated within the McHugh Creek assemblage. One sample east of Skilak Lake (Fig. 1) has zircons as young as 86 Ma indicating that the McHugh Creek assemblage is present there as well.
A summary of all of the ages from the McHugh Creek assemblage (n = 916, excluding the sandstone clasts) shows that the main population has a peak at 95 Ma, with zircon ages ranging from 86 to 113 Ma. Only 19 grains fall into the interval 119–135 Ma. The other significant peak for all of the data is at 164 Ma. Older ages include four Triassic grains from 206 to 220 Ma, and ten grains that have ages between 274 Ma (Permian) and 377 Ma (Devonian), with three of those grains forming a peak at 355 Ma (Early Mississippian). Seven Proterozoic grains range 1245–1676 Ma, with peaks at 1265 Ma and 1654 Ma.
Valdez Group Flysch
The samples southeast of the Eagle River fault zone are taken from meter-scale blocks of Valdez Group flysch that are bounded by faults on either side. Two samples (09AnJ-34 and 09AnJ-35) were collected from east of Falls Creek, where the Eagle River fault was previously mapped (Winkler, 1992). These samples are fine-grained, well-sorted sandstone and yield MDAs of 83 ± 6 Ma (based on the weighted mean of the youngest five grains; the youngest grain was dated at 77 ± 1 Ma) and 89 Ma (Fig. 7). Proceeding eastward, a Valdez Group sample from a faulted block (10AnK-01) yields an MDA of 86 Ma. Outboard of these samples, the Valdez Group has MDAs of 81 Ma, 72 Ma, 68 Ma, and 69 Ma spread over a distance of ∼50 km (this study; Kochelek et al., 2011). The Valdez Group has abundant Jurassic zircon ages (31%), and ∼5% each of Triassic and Paleozoic, with the Paleozoic ages forming a prominent peak on a relative probability distribution diagram in the Mississippian at 341 Ma and a smaller peak in the Devonian at 360 Ma. Precambrian ages are more abundant in the Valdez Group relative to the other units of the Chugach, with Proterozoic ages making up 10% of the total ages. There are nine Archean grains ranging in age from 2.54 to 2.75 Ga.
Plutonic Clasts in the McHugh Creek Assemblage
We dated six granitoid clasts using SHRIMP (sensitive high-resolution ion microprobe; Table 2; Table DR2). Two samples (09AnJ-20 and 09AnJ-25) are hornblende granodiorite. One (09AnJ-21b) is a hornblende-bearing porphyritic intermediate rock intruded at hypabyssal depths. Minor apparent Pb loss resulted in some younger individual analyses, but the weighted mean 206Pb/238U ages of five of the samples are all within error of each other and ranged from 179 ± 3 Ma to 184 ± 3 (Fig. 8). Sample 09AnJ-25 has a weighted mean 206Pb/238U age of 199 ± 3 Ma.
Trondhjemite Pluton in the Potter Creek Assemblage
Using SHRIMP we re-dated sample C80-13 that previously yielded a laser ablation multi-collector inductively coupled plasma mass spectrometry (LA-MC-ICPMS) 206Pb/238U age of 120 ± 3 Ma (Amato and Pavlis, 2010). The goal was to determine if the enhanced spatial resolution of SHRIMP would reveal inheritance. We analyzed nine spots (Fig. 8). The oldest spot (analysis 7, Table DR2) yielded an age of 131 ± 1 Ma that is older and outside of the mean of the other points; this spot location overlaps core and rim zones, as seen on cathodoluminesence (CL) images, and may be an average age of an older and younger part of the zircon. The youngest age is also statistically different from the main population. The remaining seven ages yield a weighted mean 206Pb/238U age of 125 ± 2 Ma, close to the original ICMPS age (Table 2).
Architecture of the Accretionary Complex
It is difficult to map and correlate accretionary-complex sedimentary rocks that have scarce faunal age control and almost no marker beds. With U-Pb detrital zircon geochronology, we can now make some fairly specific age assignments, despite the possibility that the MDAs are overestimating the ages of the rocks by an unknown time span. As stated in Amato and Pavlis (2010), we postulate that the youngest detrital zircon ages from the mélange sedimentary rocks should closely estimate the age of accretion because: (1) the volcanogenic graywackes likely had a rapid, simple depositional path from arc to trench; (2) the majority of the zircons in each sample form the youngest dominant age peaks, consistent with a source within an active arc and/or a rapidly exhuming orogen (e.g., Cawood et al., 2012); and (3) the scarcity of Precambrian grains (in the Potter and McHugh Creek assemblages) is consistent with a local arc source. The coarse material in the McHugh Creek assemblage necessarily had a short transportation distance, further limiting the time lag between deposition and accretion. Thus, we interpret the youngest zircons as within error of, or only slightly older than, the age of accretion.
The detrital zircon ages combined with the lithologic variations observed in this study allow us to define four components of the Chugach accretionary complex which include, from oldest to youngest: (1) the blueschist-greenschist units, most of which are known to be earliest Jurassic or Triassic from cooling ages (Roeske et al., 1989); (2) a unit we define here as the Potter Creek assemblage, which comprises the older part of what was formerly called the McHugh Complex; (3) a younger part of the McHugh Complex that we refer to here as the McHugh Creek assemblage; and (4) the flysch assemblage, which is equivalent to the Valdez Group and the other flysch units of the Chugach accretionary complex.
The blueschist-greenschist unit is exposed as discontinuous fault-bounded bodies scattered along the northern margin of the Chugach accretionary prism, including the Raspberry Schist on Kodiak Island, the Seldovia blueschists on the Kenai Peninsula, and the Iceberg Lake blueschists in the central Chugach Mountains. The Liberty Creek blueschist in the central Chugach Mountains is now known to be a distinctly different assemblage and is described elsewhere (Day et al., 2011). The interpretation of source rocks and depositional setting of the blueschist-greenschist unit is complicated by ductile deformation and recrystallization, but it is clear that the majority of the protoliths were derived from the down-going oceanic plate. The association of cherts, mafic volcanic rocks, gabbroic intrusive rocks, and scattered ultramafic rocks is typical of ophiolitic assemblages characteristic of abyssal oceanic crust (Moores and Vine, 1971). The protolith of the schistose rocks in the blueschist-greenschist unit was likely fine-grained pelitic sediment, but it is unclear whether this mud was generated in the oceans as siliceous ooze, or had a terrigenous source and was deposited in the abyssal plain, or had a terrigenous source and was deposited closer to the trench.
Zircons were absent from all eight samples separated for this study, as they were from eight samples of blueschist from near Iceberg Lake (Day et al., 2011). The apparent lack of zircons in these blueschists is consistent with an abyssal-plain protolith, as zircons would be less likely to be suspended and transported to the abyssal plain from an arc source. Alternatively, the lack of zircons could be related to sedimentation prior to the onset of arc magmatism. The initiation of the Talkeetna arc is dated at ca. 212–207 Ma (Pálfy et al., 1999; Amato et al., 2007; Rioux et al., 2010). Blueschist metamorphism occurred at 205–192 Ma (Sisson and Onstott, 1986; Roeske et al., 1989; Lopez-Carmona et al., 2011). Thus, the sedimentary protoliths could be close in age to the arc initiation and thus either pre-date or only slightly post-date the initiation of arc magmatism. If the sedimentation post-dated the onset of arc magmatism, the lack of zircons could be related to a newly formed arc that may have been erupting mafic or mafic-intermediate compositions (basaltic andesite) that would be less likely to contain zircon (e.g., Kelemen et al., 2003).
Potter Creek Assemblage
The Potter Creek assemblage is equivalent to the “mesomélange” of Amato and Pavlis (2010) and is named after Potter Creek located south of Anchorage, where the unit is well exposed. Structurally and lithologically, the Potter Creek assemblage is a type II mélange in the classification of Cowan (1985). Assemblages of this type are widespread in the Cordillera of western North America (Cowan, 1985) and from our observations are also typical of the regionally mapped exposures of the McHugh Complex. The Potter Creek assemblage is probably a combination of deep-water sediment deposited in the trench and abyssal plain together with arc-derived hemipelagic clastic sediment. The relative paucity of volcanogenic litharenite in this assemblage indicates that erosion of the arc edifice provided only a minor component of the sediment volume. In any case, most volcaniclastic material is generated as a result of explosive volcanic eruptions rather than direct erosion of the volcanic basement (Draut and Clift, 2006). The abundance of chert, shale, and volcanic rocks (both pillow lavas and tuff) is diagnostic of deep-water sediment closely associated with oceanic crust. We agree with Clift et al. (2012) that the presence of rare limestone blocks juxtaposed with volcanic rocks is consistent with a carbonate-topped seamount colliding with the trench, or with shallow-water reefs developed on an outer forearc high and mass wasted into the trench. The presence of zircons in the majority of separated samples, however, points to a source relatively close to the arc. We suggest that the abundance of zircons in the fine-grained, distal sediment may largely be related to water-lain ash-fall tuffs that mixed with the mud in the trench. The similarity in age between the youngest zircons in the Potter Creek assemblage rocks (ca. 146 Ma, sample 09AnC-08) and the youngest radiolarian ages in adjacent chert blocks (Berriasian to Hauterivian, 146–130 Ma; Nelson et al., 1987; Gradstein et al., 2005) indicates that there was mixing of arc-source sediment and pelagic biochemical sediment carried into the trench.
The maximum depositional ages in the Potter Creek assemblage range from 169 Ma to 156 Ma (Amato and Pavlis, 2010; this study). This unit, however, might have a slightly longer time gap between deposition and accretion relative to the other units because the fine-grained muds and chert were likely deposited in a deeper ocean-basin setting away from the arc, rather than directly into the trench or slightly past the trench as in the case of the McHugh Creek assemblage and the Valdez Group flysch. The dominant zircon population in the Potter Creek assemblage has a peak on a relative probability diagram at 162 Ma, and 65% of the analyses fall into the age range of 155–180 Ma. Because most of the ages are between 150 and 209 Ma, we suggest that the Talkeetna arc was the dominant source, consistent with the interpretation of Amato and Pavlis (2010). However, the abundance of zircons with ages overlapping those of the Chitina arc (153–135 Ma) is consistent with either that arc also being a source, or the two arcs being essentially part of the same subduction system (Plafker et al., 1989, 1994). In addition, other coeval oceanic arcs along strike could be equally likely as Middle to Late Jurassic sources given regional evidence for post-accretionary strike-slip offsets (Pavlis and Roeske, 2007).
Zircons with Triassic ages (232–209 Ma) may have been derived from one of the following sources: (1) oceanic sedimentary rocks and supra–subduction zone volcanic rocks that pre-dated the Talkeetna arc (Detterman and Reed, 1980; Plafker et al., 1989; Amato et al., 2007); (2) Triassic igneous rocks of the Wrangellia arc (e.g., Greene et al., 2008), although these are more commonly mafic and also zircon poor; (3) the Alexander terrane (Fig. 1), which has a Permo-Triassic plutonic suite with ages from 280 to 220 Ma (Gehrels and Saleeby, 1987) that could be a source for the Triassic grains and the three Permian grains within this suite; or (4) plutonic rocks in the Yukon-Tanana region (Aleinikoff et al., 1981), although this source seems unlikely given the scarcity of older zircons. The remaining older zircons include Mississippian (one grain), Devonian (one grain), and Proterozoic (two grains), but these are too scarce to be statistically significant.
McHugh Creek Assemblage
The McHugh Creek assemblage is the graywacke-conglomerate unit of Amato and Pavlis (2010). We retain the name “McHugh” because McHugh Creek enters Turnagain Arm in an area that has abundant exposures of these rocks. The MDAs are 101–91 Ma (Amato and Pavlis, 2010; this study), although no fossils have been reported. The contact between this and the Potter Creek assemblage is marked by both the first appearance of massive sandstones and a topographic expression produced by strong cementation of the sandstone that results in higher, rugged modern topography relative to the Potter Creek argillites that are generally more weathered and result in relatively gentle slopes.
The McHugh Creek assemblage has a dramatically different depositional setting and structural style relative to the Potter Creek assemblage; aside from the difference in depositional age, this is one of the compelling reasons to divide the McHugh Complex into these two assemblages. The abundance of massive volcanogenic litharenite and coarse conglomerate suggests that mass-wasting processes were dominant. The sediment source is interpreted to have been dominantly from an active arc with significant topography, based on the age of the youngest zircons (ca. 100–90 Ma) and the proximal high-energy facies. The source for these ages is likely the volcanic arc that was associated with the Coast Mountains batholith and the associated mid-Cretaceous orogen because this arc was experiencing voluminous magmatism and significant shortening at the time (Rubin et al., 1990; Gehrels et al., 2009). Minor magmatism occurred inboard of the Talkeetna arc around 100 Ma (Amato et al., 2007), but an open ocean basin then still occupied what is now southwest Alaska (Trop and Ridgway, 2007; Hampton et al., 2010). Whether the zircons were derived from tuffs or from erosion of the arc edifice has some implications for the lag between eruption and deposition. In the former there would be no time lag, and in the latter there is potentially a lag of a few million years. Nonetheless, it is likely that any lag would have been somewhat less than a few million years unless a significant forearc basin first stored sediment and then inverted to release large volumes of material to the forearc (Draut and Clift, 2013).
The six plutonic clasts with ages of 199–179 Ma provide unambiguous evidence that a Jurassic arc, most likely the Talkeetna arc (Rioux et al., 2010), was being exhumed and shedding coarse (clasts up to 1 m) detritus into the trench. It is interesting that despite the presence of ca. 200 Ma and ca. 180 Ma boulders, this Early Jurassic age population is not as significant in the litharenites within the same unit as the Middle–Late Jurassic ages among the detrital zircons (Fig. 9A). Of the ages between 220 and 150 Ma (n = 247), only 69 (28%) are dated at 200–175 Ma, whereas 171 (69%) are between 175 Ma and 150 Ma. This points to another source besides a Late Jurassic arc, which provided detritus in the form of both the granitoid boulders and the Early Jurassic detrital zircons. Rioux et al. (2010) noted that the youngest Talkeetna arc plutons were ca. 153 Ma and that the younger plutons were located inboard (north) of the older part of the arc. Perhaps the longer transport path of the exposed Middle Jurassic plutons resulted in sand-sized particles being deposited in the trench, whereas the older arc, which would have been more proximal to the trench, was shedding cobbles and boulders with fewer individual zircons. A less probable alternative is that a younger arc such as the Chitina arc and mid-Jurassic plutons of the Coast Plutonic Complex were being exhumed and the two sources combined to produce 180 Ma boulders and 165 Ma sand. Evidence from U-Pb dating in the modern Nankai Trough indicates that the source rocks were likely not supplying sediment to the accretionary prism hundreds of kilometers along strike but more likely directly across strike (Clift et al., 2013).
Another possible source for the 165 Ma zircon population in the McHugh Creek assemblage is the Potter Creek assemblage. Recycling of these zircons could have occurred if the older part of the accretionary complex was exposed above sea level and then eroded. Evidence supporting a recycling origin includes: (1) the nearly identical matching of the age populations (Fig. 9A); (2) the abundance of argillite and chert cobbles in the Beluga Point conglomerate; and (3) the presence of chert grains in the litharenite, which requires a chert source formed by subaerial erosion and not simply ground up by tectonic processes in the subduction zone. Evidence against this being a significant source comes from the limited size of the Potter Creek assemblage, because a shorter wedge length would indicate a lower topographic profile unlikely to have emerged above sea level. However, an unknown amount of this unit, as well as the crystalline backstop to the assemblage above (north) of the BRF, was removed by erosion and faulting. We place the younger limit of deposition of the Potter Creek assemblage at the time of ridge subduction and trondhjemite magmatism at 125 Ma. Resolution of this question may come from lower-temperature thermochronologic cooling ages from the zircons with older U-Pb ages in the McHugh Creek assemblage. If they were exhumed in the Late Cretaceous they will have young cooling ages, whereas if they were initially deposited in the Potter Creek assemblage in the Early Cretaceous and recycled, they should have older cooling ages.
The flysch assemblage throughout southern Alaska is the most voluminous part of the accretionary complex. In addition, the depositional style is different from the older units. We concur with previous workers that the coherent bedding of this assemblage is indicative of deposition not in the trench, but on the oceanic plate beyond the trench (Sample and Reid, 2003; Kochelek et al., 2011). We suggest that the filling of the trench occurred during rapid deposition of the McHugh Creek assemblage. The source for the flysch was also most likely the arc associated with the Coast Mountains batholith exposed in southeastern Alaska and western British Columbia (Fig. 9B) based on petrographic comparisons (Nilsen and Zuffa, 1982; Dumoulin, 1987), paleocurrent measurements (Nilsen and Zuffa, 1982), Nd, Sr, and Pb isotopic data (Farmer et al., 1993; Sample and Reid, 2003), and detrital zircon ages (Haeussler et al., 2005; Kochelek et al., 2011). The Kodiak Formation flysch was interpreted as a combination of recycled orogenic and magmatic arc sources that was deposited in an extensive fan that filled the trench, thus allowing deposition onto the oceanic plate beyond the trench, and that rapid, voluminous deposition occurred following a collision between offshore terranes and continental North America (Sample and Reid, 2003; Kochelek et al., 2011).
The sources for the pre–Late Cretaceous zircons in the Valdez Group are similar to those for the McHugh Creek assemblage (Fig. 9A). The main difference is the proportion of Proterozoic and Paleozoic zircons in the Valdez Group (Fig. 4A). In the older units of the Chugach accretionary complex, Proterozoic and Paleozoic zircons compose <2% of the total population. In the Valdez flysch, 19% of the zircons are pre-Mesozoic and 12% of the total zircons are Precambrian (Fig. 9C). It is worth noting that the Orca Group exposed near Prince William Sound contained a similar percentage of Precambrian zircons (Davidson et al., 2011) as found in the Valdez Group. Paleozoic zircons could be sourced from Wrangellia or continental North America. The sources for the Precambrian zircons must be from continental North America, inboard of the accreted Wrangellia composite terrane. For the Precambrian zircons, the most prominent peak on a relative probability distribution diagram is at 1855 Ma (Fig. 9C). These ages are found in zircon cores in gneisses in the parautochthonous Yukon-Tanana assemblage (Dusel-Bacon and Williams, 2009), in rocks of the Trans-Hudson–Penokean orogeny (Corrigan et al., 2009), in the Wathaman batholith in central Canada (Meyer et al., 1992), and in the Great Falls tectonic zone at the northwest margin of the Wyoming craton (Mueller et al., 2002). All of these sources are north of 45°N latitude, raising the possibility that the accretionary complex was near that latitude at 85–75 Ma, but it is unknown whether the Precambrian zircons in the Valdez Group are recycled. Regardless of the sources for Precambrian zircons, it is clear that beginning at ca. 85 Ma, transportation pathways from the continental interior were formed along which sediment was delivered to the trench. This sediment mixed with young arc material and either newly exhumed Early Cretaceous plutonic rocks or recycled Early Cretaceous zircons from the Potter Creek assemblage.
Detrital Zircon Age Trends within the Accretionary Complex
Relative probability distribution diagrams for the majority of the samples from this study show the youngest peak as the most prominent peak (Figs. 5–7). This is a hallmark of convergent margins in general and of trench and forearc basin settings specifically (Cawood et al., 2012). However, not all accretionary complexes have this pattern. For example, the Franciscan Complex of California has very few zircons that were deposited contemporaneously with their volcanic source, and it has been suggested that in the Franciscan this is the result of the trench being upwind of the arc, preventing zircons in volcanic ash from reaching the accretionary complex (Dumitru et al., 2010). In these cases the youngest zircons may be sourced from the exhumed plutonic roots of the arc, so that the depositional ages would be much younger than the youngest zircons in accretionary complex sediment.
In the Chugach accretionary complex, a typical sample has a youngest peak that consists of 60%–90% of the total zircons analyzed (Figs. 5–7). In addition, there are no significant gaps between the maximum depositional ages of the samples within each of the three main units. These characteristics are strong evidence for nearly coeval eruptive and depositional ages. The scarcity of pre-Mesozoic zircons in the mélange assemblages is consistent with the widely held hypothesis that the Wrangellia composite terrane was an oceanic arc throughout much of its history, isolated from continental sources. Thus, the forearc structure may have allowed relatively open, simple sediment pathways from the arc to the trench, as in many modern western Pacific arcs (Draut and Clift, 2006). Regardless of the cause, the abundance of nearly coeval ages allows the sediment in the accretionary complex to provide a detailed record of the history of the convergent margin.
Another important aspect of the detrital zircon data is the overall trend of decreasing ages away from the arc and toward the trench (Fig. 10). This trend is best developed in the Potter Creek assemblage along Turnagain Arm, where our high sampling density reveals a trend from inboard samples dated at 167 Ma toward outboard samples at 156 Ma. The range of ages in the McHugh Creek assemblage along Turnagain Arm reveals a moderate trend of decreasing ages from inboard samples at 101 Ma and outboard samples at 97–95 Ma. This limited range of ages is consistent with the apparently rapid deposition of large volumes of sand and cobbles in this unit. The five Valdez Group samples have maximum depositional ages from inboard to outboard of 87 Ma, 81 Ma, 72 Ma, 68 Ma, and 69 Ma.
The pattern of decreasing ages with distance away from the arc is a hallmark of models of progressive outboard/trenchward accretion and is a consequence of the normal stacking sequence of thrust systems, but in most previously documented cases these were documented with biostratigraphy (e.g., Nishi, 1988). The ages of the fossils would not correlate with the accretionary ages in a straightforward way if the oceanic sediment was deposited on oceanic crust far from the trench, in which case the depositional age would be much older than the accretionary age. Amato and Pavlis (2010) showed this general pattern for Turnagain Arm but with a limited data set (n = 5). Likewise, Dumitru et al. (2010) demonstrated this pattern for the Franciscan Complex (n = 7) but the MDAs were inferred to be much older than the actual depositional ages. Isozaki et al. (2010) showed a general younging trend of MDAs, but the spatial distribution of the samples is unclear. The age pattern in the Catalina Schist had older ages in the structurally highest units and younger ages in the structurally lower units (Grove et al., 2008). The age pattern that we document in this study demonstrates that within each package there are generally younger ages in the more outboard/structurally lower rocks, consistent with the progressive outboard accretion model. We also document significant gaps in maximum depositional ages between the major rock packages, consistent with a tectonic erosion model in which growth of the accretionary complex is episodically truncated.
Syn- and Post-Accretionary Faulting of the Chugach Accretionary Complex
The current geographic distribution of the four units of the Chugach accretionary complex can be used to evaluate post-accretionary faulting. This problem is significant because the BRF is known to carry a complex history of reactivation, including large-magnitude dextral slip in the latest Mesozoic and Paleogene, as well as unknown amounts of older strike slip (Pavlis and Roeske, 2007). Thus, post-accretionary dextral slip has transported the Chugach accretionary complex laterally relative to coeval arcs, but the magnitude of that strike slip is only resolved as greater than ∼130 km, but potentially 600 km or more (Pavlis and Roeske, 2007).
The oldest rocks of the accretionary complex, those of the blueschist-greenschist unit, have the poorest paleogeographic control as a result of both their relative antiquity and their scattered positions. Presently these rocks are restricted to either fault-bounded lenses within the Potter Creek assemblage (Iceberg Lake blueschist of Winkler et al., 1981; Sisson and Onstott, 1986), or fault-bounded bodies adjacent to the BRF, i.e., the Raspberry Schist (Connelly and Moore, 1979; Roeske et al., 1989), the Seldovia blueschist (Bradley et al., 1999), and the Liberty Creek blueschist (Sisson and Onstott, 1986). The scattering of these blueschist slices along the BRF could originate from strike-slip dispersal or from limited preservation of blueschists in the interval of subduction erosion prior to the accretion of the Potter Creek assemblage (Clift et al., 2005a).
The Potter Creek assemblage is entirely absent north of the Eklutna River (Fig. 2), but a broad band of metamorphic rocks within the BRF system (JPzm of Winkler, 1992), also referred to as the metamorphic subterrane of the Knik River terrane (Pavlis, 1983), has detrital zircon ages that closely match those of the Potter Creek assemblage (Labrado et al., 2012). A type II mélange with Late Jurassic MDAs also was found in the central Chugach Mountains near Iceberg Lake (Day et al., 2011) and lithologically similar rocks compose the bulk of the mélange assemblage in the central Chugach Mountains (Winkler et al., 1981; Plafker et al. 1989). Nonetheless, the central Chugach Mountains carry a known, complex record of Cretaceous–Eocene dextral strike-slip faulting that clearly produced significant dispersal of the mélange assemblage in this region (Pavlis and Roeske, 2007).
Despite dispersal along the margin, the Potter Creek assemblage generally is located inboard (northwest) of the McHugh Creek assemblage, consistent with a general trenchward-migrating accretionary sequence. We identified one section of the Potter Creek assemblage near the Eagle River fault along Turnagain Arm based on lithology, structural style, and maximum depositional age (sample 09AnJ-33; 162 Ma), but this ∼400-m-wide block is either a local fault slice that was transported along the Eagle River fault or an olistolith of the Potter Creek assemblage that was mixed into the McHugh Creek assemblage. Because of its presence along the Eagle River fault we prefer the former interpretation.
Collectively these regional relationships are consistent with previous studies that indicate that strike-slip systems have produced significant dispersal of the Mesozoic accretionary complex along the BRF. Modern analogs preclude long-distance transport of sediment parallel to the trench axis (Clift et al., 2013). To the north and east of the Knik River (Fig. 2) the strike-slip overprint is intense. In this area, cataclastic zones up to 2 km wide form anastomosing fault networks that complexly disrupt the assemblage (Pavlis et al., 1988; Pavlis and Roeske, 2007). Pavlis and Roeske (2007) interpreted these as a system of dextral faults that linked up to the dextral Castle Mountain fault system through the Matanuska Valley. Thus, although major advances have been made in understanding this dextral overprint, it is difficult to draw firm conclusions about subduction zone processes from this segment because of the intensity of the overprint. In contrast, following the rationale of Pavlis and Roeske (2007), we suggest that to the south of the Knik River the strike-slip overprint is less intense and largely restricted to a few discrete structures. Particularly prominent is the widespread preservation of what we interpret as a primary thrust-contact relationship along the Eagle River fault through much of this region. Along Turnagain Arm, the McHugh Creek assemblage is located structurally between the Potter Creek assemblage and the Valdez flysch assemblage and it has a progression of MDAs that, in general and within the 2σ uncertainty of the ages, decreases with distance away from the BRF (Fig. 10). This regular progression suggests that there is little internal mixing of this unit, at least where it was sampled in detail along Turnagain Arm, and the deformed zone of the Eagle River fault records a primary, subduction-related contact relationship. In this same area, the Potter Creek–McHugh Creek primary thrust-contact relationships are also probably preserved, but more work is needed to fully assess which parts of the boundary carry other overprints.
Collectively these regional observations suggest that for the segment of the margin south of the Knik River (Fig. 2) we can make an assumption that the history of the accretionary complex is directly tied to subduction history recorded by arc systems carried on the Wrangellia composite terrane. We recognize that the true, middle-Mesozoic across-strike record may now lie in southeast Alaska or even British Columbia, but because this record is broadly similar throughout the region, this distinction may not be particularly significant to interpretations that tie the forearc to the arc record. Thus, in the absence of other information, we assume the sedimentary records of the southern Alaskan forearc basin and the forearc accretion are closely tied and that this tie is important in regional syntheses.
Evidence for Subduction Erosion
Roeske et al. (1989) first proposed that subduction erosion may have affected the Chugach accretionary complex. They suggested that if the blueschists on Kodiak Island were correlated to the arc, a hypothesis consistent with their ages of the high-pressure–low-temperature rocks, then the coeval forearc must have been completely removed, possibly via subduction erosion. Because the dense blueschist-facies rocks were preserved along the BRF, Roeske et al. (1989) preferred a model in which oblique convergence resulted in strike-slip faults that juxtaposed the accretionary complex against the arc system.
We suggest that there were two episodes of subduction erosion in the Chugach accretionary complex (Fig. 11): one during the Middle Jurassic and another in the Early Cretaceous. Clift et al. (2005a) first interpreted the removal of the forearc region and migration of magmatism inboard between 180 and 160 Ma as a period of subduction erosion following the blueschist-forming event. They noted that the limited amount of sediment subduction below the Talkeetna arc was consistent with modern systems in which low sediment thickness in the trench typically leads to subduction erosion of the forearc region (e.g., Clift and Vannucchi, 2004). Based on our detrital zircon data from Turnagain Arm, subduction erosion must have ended by 169 Ma. This subduction erosion event was responsible for forming the backstop against which the Chugach accretionary complex was accreted and for the inboard migration of arc magmatism prior to the onset of rapid accretion (Fig. 11; Clift et al., 2005a; Rioux et al., 2010).
The second episode of subduction erosion, originally proposed by Amato and Pavlis (2010), is supported by our expanded data set (Fig. 11). The youngest preserved samples from the Potter Creek assemblage that we have dated have MDAs of 156 Ma, but the oldest part of the McHugh Creek assemblage is dated at 101 Ma, a gap of over 50 m.y. Following the earlier hypothesis of Amato and Pavlis (2010), we propose that the subduction erosion interval occurred during the Aptian–Albian from ca. 125 Ma to 101 Ma, but cannot rule out initiation of subduction erosion earlier than 125 Ma, although it must have been after 156 Ma. This conclusion is based on the inference that the trigger for erosion was a ridge subduction event at 125 Ma. As shown by von Huene and Scholl (1991), the most efficient process driving subduction erosion would be the subduction of large submarine topographic features, and that study emphasized the role of seamounts, plateaus, and ridges. Critical to this inference is the evidence for a ridge subduction event coeval with the subduction erosion event.
Pavlis (1982) recognized that Early Cretaceous plutons were emplaced within the Chugach mélange assemblage (i.e., Potter Creek assemblage), and that their forearc position was consistent with a ridge subduction hypothesis. However, Pavlis (1982), Pavlis et al. (1988), and Barnett et al. (1994) favored the alternative hypothesis that this event recorded reestablishment of subduction following a strike-slip event or temporary cessation of subduction because a significant magmatic gap occurred on the Wrangellia composite terrane just prior to the forearc plutonism. Pavlis and Roeske (2007) reviewed this problem and concluded that new data favor the ridge subduction hypothesis, and that other regional data also provide further support for the ridge subduction hypothesis. The interval 130–110 Ma was a time of major changes in sedimentation within the adjacent forearc basin including a major unconformity below an Albian–Aptian clastic succession that records rapid forearc subsidence following a basement uplift (Trop, 2008). In addition, the period ca. 135–120 Ma has long been known as being a magmatic lull in the northern Cordillera (e.g., Armstrong, 1988). Subsidence and a magmatic lull are typical of ridge subduction events. Uplift of a forearc basin occurred where the Nazca Ridge is colliding with the central Andean forearc (Clift et al., 2003), and also occurred throughout the southern Alaskan margin during latest Cretaceous–Paleogene ridge subduction (Trop et al., 2003). Uplift is occurring in southern Chile, as well as the Woodlark Basin of Papua New Guinea where active spreading centers are subducting (e.g., Taylor and Exon, 1987). Similarly, there is a magmatic gap associated with the slab window in Chile (Kay et al., 1993) and other areas where ridge subduction has occurred (e.g., Sisson et al., 2003).
Based on these relationships we suggest that regional data all broadly support Early Cretaceous ridge subduction, and we infer that this event produced an episode of subduction erosion that is recorded in the detrital zircon ages (Fig. 11). This conclusion carries a caveat that post-accretionary strike slip may have disrupted the primary accretionary architecture, or even be the source of the juxtapositions we observe. Nonetheless, several observations suggest this process is minor within the area of this study, the Anchorage–Kenai Peninsula area:
(1) Post-accretionary strike-slip deformation is most prominent in the central and eastern Chugach Mountains, and Pavlis and Roeske (2007) used this observation to conclude much of this slip was transferred to other structures like the Castle Mountain fault. Thus, although younger complications are clearly present, the most extreme strike-slip effects are not present in the Anchorage area where the bulk of our data were collected.
(2) Although disruptions by strike-slip faults are conspicuous throughout the region, there are clear vestiges of primary, subduction-related structural contacts throughout the region. In the western Chugach Mountains (Fig. 2) strike-slip faults disrupt the geology, but the system can be restored to realistic paleogeography consistent with a subduction accretion system: that is, a band of ductile deformation representing an Early Cretaceous sinistral-oblique thrust zone, passing structurally downward into a complex transition with the middle Cretaceous McHugh Creek assemblage through a transitional structural contact with the Valdez Group (Chugach flysch) (Pavlis, 1982; Pavlis et al., 1988; Pavlis and Roeske, 2007). Similarly, the exposure along Turnagain Arm (Fig. 2) appears to be a relatively intact assemblage from the Potter Creek assemblage through the McHugh Creek assemblage to the flysch assemblage with no obvious strike-slip disruptions. Finally, the Kenai Peninsula preserves large vestiges of the Potter Creek assemblage with lesser amounts of the McHugh Creek assemblage in what appear to be primary thrust relationships onto the flysch assemblage, and similar large-scale geology continues to Kodiak Island.
(3) The erosional period coincided with forearc trondhjemite magmatism, uplift, and erosion in the forearc basin. These age relations provide a direct tie to a ridge subduction event, whereas omission by strike-flip faulting would not result in the occurrence of these two linked events.
The end of the second subduction erosional episode is marked by the influx of massive volumes of clastic sediment that constitutes the McHugh Creek assemblage (Fig. 11). Higher sedimentation rates are characteristic of convergent margins associated with long-term accretion (Clift and Vannucchi, 2004). Mass-wasting events indicated by coarse conglomerate in the McHugh Creek assemblage are inferred to have resulted from the oversteepening of the forearc region during the preceding period of tectonic erosion. Similar gravitational instability of the forearc and subsequent mass wasting was noted in the Nankai accretionary prism of Japan (Park et al., 2002) and in the Ligurian subduction complex of Italy (Remitti et al., 2011) in the aftermath of seamount collisions.
More importantly, the onset of accretion of the McHugh Creek assemblage is tied closely in time to major regional orogenic events, and the influx of clastics from a collision was undoubtedly critical to this transition. This sedimentary influx culminated in the transition to a different style of accretion, the flysch assemblage that was accreted in a classic fold-thrust style consistent with large-volumes of sediment entering the trench (e.g., Byrne and Fisher, 1987; Sample and Reid, 2003). In our initial analysis of the detrital zircon data along Turnagain Arm (Amato and Pavlis, 2010), it appeared that the structural contact between the McHugh Creek assemblage and flysch assemblage (Eagle River fault) represented an age break. With more data (Kochelek et al., 2011; this study), however, it is clear this transition is primarily a change in structural style coincident with changes in sedimentation. That is, the Eagle River fault along Turnagain Arm is marked by a mélange zone composed of rocks that are lithologically akin to the flysch assemblage, and these rocks are a significant part of the age transition along this break. Thus, although there is a continuous age progression across the transition from the McHugh Creek assemblage to the Valdez Group flysch assemblage, the structural break between them (the Eagle River fault) is a significant structural contact that represents a broad zone of deformation, but the traditional contact shown on maps is a lithologic change reflecting a change in depositional style from a sandstone-dominant assemblage to a sand-mud assemblage. The Eagle River fault thus represents both a structural and lithologic contact, and this is important for understanding the regional geology because it implies that primary thrust contacts can be distinguished in this system through careful detrital zircon sampling across the transition in conjunction with structural mapping.
Implications for Collisions along the Cordilleran Margin
The period between the latest accretion of the Potter Creek assemblage and the beginning of the accretion of the flysch assemblage (ca. 156–89 Ma) was the period when most of the northern Cordilleran orogen was assembled from various crustal fragments. It is widely accepted that by the end of this period, all ocean basins were closed, but the size and extent of pre-existing oceans remains controversial. Our detrital zircon data span this time interval and record a history on the trailing edge of this orogenic system that is distinct from other regional data sets. Thus, these data have broader implications for regional tectonic models of the northern Cordillera.
There is widespread agreement that a major middle Cretaceous contractional orogen developed in the northern Cordillera (e.g., Rubin et al., 1990), and recent work in Alaska provides strong evidence that this event was associated with south-to-north closure of an ocean basin (Trop and Ridgway, 2007; Hampton et al., 2007, 2010). What is not clear, however, is the nature of this ocean basin. In the model of Monger et al. (1982), the Wrangellia composite terrane was exotic to North America and mid-Cretaceous events recorded the closure of an ocean basin. Later models (Pavlis, 1982, 1989; Pavlis et al., 1988; Rubin et al., 1990; Trop and Ridgway, 2007) built on this hypothesis, attributing mid-Jurassic deformation to intra-oceanic collisions that formed the Wrangellia composite terrane followed by a diachronous history in which the Gravina-Nutzotin-Kahiltna ocean basin system was formed during a north-to-south closure of the ocean basin between 110 Ma and 85 Ma. In contrast, work in southern British Columbia led others to suggest that the mid-Jurassic event represents the collision of the Wrangellia superterrane with North America followed quickly by a Late Jurassic backarc opening to form the Gravina-Nutzotin ocean basin (McClelland et al., 1992; van der Heyden, 1992). In this model, mid-Cretaceous shortening represents closure of a backarc basin (e.g., see review by Gehrels et al., 2009). In both models, a major ocean basin separated Wrangellia from North America between Late Jurassic and mid–Early Cretaceous time, isolating North American sedimentary sources from the forearc except in southern British Columbia where the two-phase collision model implies a direct tie to North America (Gehrels et al., 2009).
Distinguishing backarc closure versus collision is a persistent problem in tectonic studies because overprinting within the suture zone typically obscures the history and because it is generally impossible to establish the scale of a consumed ocean basin other than in cases where paleomagnetism can resolve the issue. Modern detrital zircon techniques provide new hope for resolution of these problems, however, because they potentially can resolve evolving sources to distinguish models. In the case of the northern Cordilleran system, our detrital zircon data together with other regional data sets are consistent with a long-held view that the Wrangellia composite terrane was isolated from North American sources until Late Cretaceous time. Thus, a Jurassic collision–backarc opening model is difficult to reconcile with the new evidence.
Within the suture zone, detrital zircon ages from sedimentary rocks in the Kahiltna Basin indicate a continental source starting around 135–125 Ma, suggesting that the collision of the arc system with North America could have begun just prior to that time (Hampton et al., 2010). Kalbas et al. (2007) noted that the Kahiltna Basin probably represents sedimentation in response to an oblique arc-continent collision and that the timing of shortening was likely diachronous along the North American margin, proceeding from south to north (Trop and Ridgway, 2007). Diachronous collisions or backarc closures have long been known to characterize modern and ancient systems, but the evidence for a south-to-north closure has important implications for our data because axial drainages developed in this type of system (e.g., Kalbas et al., 2007) could persist over long intervals of time, influencing sediment dispersal during and after the collision.
In our detrital zircon data from the trailing edge of the orogen, the oldest accretionary units (Potter Creek and McHugh Creek assemblages) have no direct evidence of sources outside of the Wrangellia composite terrane. The first appearance of a clear North American source in this assemblage is at ca. 89–85 Ma in the oldest phases of accretion of the flysch assemblage, consistent with the view that this flysch assemblage was derived from erosion of the northern Cordilleran mid-Cretaceous orogen (e.g., Kochelek et al., 2011, and references therein). To us, the absence of a clear North American source in Jurassic Potter Creek assemblage is difficult to reconcile with a Jurassic collision model because we would expect that detritus shed from that collisional orogen should have been carried to the coeval trench.
Perhaps the most surprising observation from our detrital zircon data is the prominence of local sources in the McHugh Creek assemblage during the time interval ca. 100–90 Ma. By this time there is no doubt that a major contractional event had produced a large orogen in southeast Alaska extending into British Columbia (Chardon, 2003; Gehrels et al., 2009). This suggests that either the Chugach accretionary complex was far from the mid-Cretaceous Coast orogen or that this orogen maintained a drainage divide that continued to isolate North American sources well into Late Cretaceous time. Given that by ca. 90–80 Ma the Chugach (Valdez) flysch appears to be dominated by a Coast orogen source, the former appears unlikely and thus, our data imply intriguing details for the mid-Cretaceous northern Cordilleran orogen.
We suggest that the oblique collision model provides a simple explanation for this observation (Fig. 11). A characteristic feature of oblique (and some orthogonal) collisions is the development of axial drainages that disperse sediment laterally (e.g., Kalbas et al., 2007, and reference therein), and these drainages can persist over long intervals of time. Thus, the simplest interpretation of our detrital zircon data is that through-going drainages to North America could not develop until late in the contractional history, after full closure of the intervening ocean basin. This relationship presumably holds regardless of what type of basin was closed, and this general conclusion needs to be considered in future paleogeographic analyses of this system.
Implications for the Baja British Columbia Hypothesis
The Border Ranges fault likely accommodated northward translation of the Chugach accretionary complex, with post-Paleocene displacement estimates of ∼2000 km based on paleomagnetic data (Coe et al., 1985). Geologic evidence suggests that ∼1100 km of translation (Cowan, 1982) occurred after 50 Ma (Cowan, 2003), with other workers suggesting that the amount of offset was closer to 700 km (Pavlis and Roeske, 2007). Cretaceous (90–70 Ma) displacements of Wrangellia may have been >1000 km based on paleomagnetic data used to support the Baja British Columbia hypothesis in which Wrangellia originated at a latitude near the modern location of Baja California and was subsequently translated >3000 km to the north (Cowan et al., 1997; Cowan, 2003; Pavlis and Roeske, 2007). We agree with Cowan et al. (1997) and others who state that Wrangellia and its associated Jurassic–Cretaceous arcs were not likely at the same latitude as the Franciscan Complex in California at this time. We base this on the comparison of the detrital zircon populations in the Chugach accretionary complex from this study with those from the Franciscan Complex (Dumitru et al., 2010). The Franciscan Complex has a small percentage of coeval volcanic arc–derived zircons in the sediment deposited in the accretionary complex, a situation ascribed to westerly winds blowing arc-derived ashes to the east, away from the trench. If the abundance of arc-derived zircons in the Chugach accretionary complex is a consequence of easterly winds depositing ash-fall tuffs in the Potter Creek assemblage, then we argue that the Chugach accretionary complex would have been located far enough south at ca. 160 Ma to encounter the trade winds that are generally developed south of 30°N latitude. Although the dominance of 1855 Ma zircons in the Valdez Group suggests that the Chugach accretionary complex was north of latitude 45°N by the end of the Cretaceous, more data are required to evaluate this aspect of the Baja British Columbia model.
The Mesozoic Chugach accretionary complex consists of at least four distinct parts that reveal episodic accretion punctuated by two periods of subduction erosion. The U-Pb detrital zircon ages provide a high-resolution record of tectonic activity, magmatic activity, changing source regions, and the assembly of the accretionary complex over time. The oldest of these four units is a blueschist-greenschist unit that does not contain detrital zircons. Previously published earliest Jurassic metamorphic ages provide a link to the Talkeetna arc, but it is possible that deposition of the sedimentary protoliths could pre-date the arc. Following subduction erosion from ca. 180 Ma to 170 Ma that removed the forearc and caused magmatism to migrate inboard (Clift et al., 2005a), collision of the Talkeetna arc with the Wrangellia terrane resulted in an influx of sediment that ended the erosional period, forming the Potter Creek assemblage. These rocks were deposited from 169 until 156 Ma and possibly until 125 Ma, but no record of 156–125 Ma sediment remains because ridge subduction peaking at around 125 Ma resulted in a period of subduction erosion that removed previously accreted material and lasted until 100 Ma.
Evidence for ridge subduction culminating at ca. 125 Ma comes from dating of trondhjemitic plutons intruding the older part of the Chugach accretionary complex and from a coeval disconformity in the forearc basin. This is analogous to the ridge subduction, trondhjemitic plutonism, and subduction erosion in southern Chile since 4 Ma (Bourgois et al., 1996). Massive sandstone and coarse conglomerate of the McHugh Creek assemblage were deposited in response to the collision of the Wrangellia superterrane with North America at ca. 101 Ma. This event steepened the forearc region allowing mass wasting of forearc and arc crust into the trench, filling it by 89 Ma. From 89 Ma to at least 72 Ma, the Valdez Group flysch was deposited via turbidite fans onto the oceanic crust beyond the trench, after which it was accreted in large blocks that retained their coherent bedding.
Detrital zircon ages from a cross-section of the Chugach accretionary complex show progressive outboard accretion within each of the main units. This has been observed in other accretionary complexes but it is particularly well documented in this study because of the inferred limited lag time between arc magmatism and the supply of volcanic zircons to the accretionary complex. Sources include the Talkeetna arc and possibly the Chitina arc or another coeval Middle–Late Jurassic arc for the Potter Creek assemblage, and the arc associated with the Coast Mountains batholith for the McHugh Creek assemblage and the Valdez Group flysch. Jurassic (165 Ma) zircons in the Upper Cretaceous McHugh Creek assemblage could have been derived from erosion of exhumed Jurassic plutons or they could be second-cycle zircons derived from the Potter Creek assemblage.
Precambrian zircons first appear in significant quantities in the Upper Cretaceous Valdez Group flysch. Their appearance probably reflects the breakdown of topographic barriers formed by the accreted arc terranes by rivers eroding into the continental interior of North America, consistent with changes in the heavy mineral assemblage at that time (Clift et al., 2012). This transition from outboard (arc) to inboard (continent) sources with time was also observed in the Cretaceous schists of southern California (Jacobson et al., 2011). The abundance of zircons from ash-fall tuffs in the Potter Creek assemblage is consistent with easterly winds and suggests that the Chugach accretionary complex may have been south of 25°N latitude in the Late Jurassic–Early Cretaceous. The main population of Precambrian zircons in the Upper Cretaceous Valdez Group is centered at 1855 Ma, a time of magmatism in areas generally found north of latitude 45°N.
Subduction erosion has thus been identified in at least four accretionary complexes: (1) the Chugach accretionary complex (Amato and Pavlis, 2010; this study); (2) the Franciscan Complex (Dumitru et al., 2010) and the coeval Catalina-Orocopia-Pelona Schist (Grove et al., 2008; Jacobson et al., 2011); (3) in Japan (Isozaki et al., 2010); and (4) in Liguria, Italy (Remitti et al., 2011). We suggest that accretion punctuated by periods of episodic subduction erosion is a hallmark of most long-lived accretionary complexes and that detrital zircon dating is an ideal method to identify these episodes, particularly when specific triggers for erosion can be identified.
APPENDIX: GEOCHRONOLOGY METHODS
U-Pb geochronology was carried out using two methods. SHRIMP-RG (sensitive high-resolution ion microprobe reverse geometry) dating of igneous rocks was conducted at the Stanford–U.S. Geological Survey Ion Probe Facility. The primary beam excavated an area of about 25–30 µm across to a depth of about 1 µm. The analytical routine followed Williams (1998) and Strickland et al. (2011). Data reduction utilized the SQUID program (Ludwig, 2005). Isotopic compositions were calibrated by replicate analyses of zircon standard R33 that has an age of 419 Ma (Black et al., 2004). 206Pb/238U ages were corrected for common Pb using the 207Pb method. Common Pb compositions were estimated from Stacey and Kramers (1975).
Laser ablation multi-collector inductively coupled plasma mass spectrometry (LA-MC-ICPMS) dating was conducted on sedimentary rocks at the Arizona Laserchron Laboratory at the University of Arizona. Beam diameter was generally 30–40 µm, though a 20 µm beam diameter was used on sample 10AnT-09 because of its smaller average zircon size. Errors on spot ages of individual zircons grains are reported in the text and tables at 1σ, and we report weighted mean ages in the text and figures at the 2σ level. Data are presented on concordia diagrams and relative probability distribution diagrams (Ludwig, 2003). “Peak” ages refer to the peaks on a relative probability distribution diagram. Maximum depositional ages (MDAs) were calculated using the weighted mean of the grains that make up the youngest peak consisting of n > 2 ages, using the Zircon Age Extractor of Ludwig (2003). This is different from the technique used in Amato and Pavlis (2010).
We used several filters for the data. We used a discordance filter of ±15% for grains <250 Ma [(206Pb/238U age)/(207Pb/235U age)] but note that the 207Pb/235U age is much less precise than the 206Pb/238U age for grains <250 Ma. We do not report the 207Pb/206Pb ages unless the 206Pb/238U age is >250 Ma. For these older ages we use a 10% discordance filter and a 5% reverse discordance filter. We also apply a fairly strict uncertainty filter where ages with a 1σ uncertainty of >5% are discarded. We made an exception to this for sample 10AnT-09, which had few total ages and few ages with <5% 1σ uncertainties, because this is the only sample in its region. For the Valdez Group, which has zircons with generally low U concentrations and younger ages, we used a 10% 1σ uncertainty filter. One sample was re-analyzed at the University of California–Santa Barbara at the Laser-Ablation Split-Stream Dual ICPMS Facility (LASS) using the techniques described by Kylander-Clark et al. (2013).
This work was funded by National Science Foundation (NSF) grants EAR-0809608 to Amato and EAR-0809609 to Pavlis. George Gehrels and the staff at the Arizona LaserChron Center helped acquire detrital zircon data, and the laboratory was supported by NSF grant EAR-1032156. Amato acknowledges support from the Cooperative Institute for Research in Environmental Sciences at the University of Colorado at Boulder. Brad Hacker and Andrew Kylander-Clark provided some additional zircon analyses. We appreciate helpful discussions with Dwight Bradley, Lang Farmer, Rebecca Flowers, and Craig Jones. This paper benefited from detailed reviews by Nancy Riggs, Becky Dorsey, Sarah Roeske, and Brian Hampton.
↵§ Present address: Chesapeake Energy, 6100 N. Western Ave., Oklahoma City, Oklahoma 73118, USA.
↵# Present address: Chevron, 1550 Coraopolis Heights Rd., Moon Township, Pennsylvania 15108, USA.
Science Editor: Nancy Riggs
Associate Editor: Rebecca Dorsey
- Received 6 November 2012.
- Revision received 7 June 2013.
- Accepted 12 August 2013.
- © 2013 Geological Society of America