This study presents the long-term exhumation history of the Wrangellia composite terrane of the remote and ice-covered northern St. Elias Mountains in southwest Yukon, northwest British Columbia, and adjacent Alaska. Detrital zircon and apatite fission-track age distributions are presented from 21 glacial catchments. The detrital sampling approach allows for a large spatial coverage (∼30,000 km2) and access to material eroded beneath the ice. An additional five bedrock samples were dated by zircon fission-track analysis for a comparison with detrital results. Our new thermochronology data record the Late Jurassic–mid-Cretaceous accretion of the Wrangellia composite terrane to the former North American margin and magmatism, which reset the older thermal record. The good preservation of the Jurassic–Cretaceous record suggests that Cenozoic erosion must have been limited overall. Nonetheless, Eocene spreading-ridge subduction and Oligocene–Neogene cooling in response to the ongoing Yakutat flat-slab subduction are evident in the study area despite its inboard position from the active plate boundary. The results further indicate an area of rapid exhumation at the northern end of the Fairweather fault ca. 10–5 Ma; this area is bounded by discrete, unmapped structures. The area of rapid exhumation shifted southwest toward the plate boundary and the center of the St. Elias syntaxis after 5 Ma. Integrating the new data with published detrital thermochronology from the southern St. Elias Mountains reveals an evolving concentration of deformation and exhumation, possibly within a large-scale, transpressional structure providing important constraints for geodynamic models of syntaxes.

The Wrangellia composite terrane is the classic example of an accreted and displaced terrane within the terrane mosaic of the western North American margin (Jones et al., 1972; Coney et al., 1980). The composite terrane has been studied to understand continental growth by terrane subduction and accretion processes and its spatial and temporal transitions between compressional, transpressional, and extensional deformation (e.g., Coney et al., 1980; Rusmore and Woodsworth, 1991; Andronicos et al., 1999, 2003; Gehrels et al., 2009; Israel et al., 2013). Due to the long history of accretion, terranes experience continued deformation during their accretion, possible postaccretional displacement, and continued subduction and accretion at their outboard margins.

This study investigates the Wrangellia composite terrane of the northern St. Elias Mountains in southwest Yukon, northwest British Columbia, and adjacent Alaska (Figs. 1 and 2). This area has been influenced by two major accretion events—the Late Jurassic–mid Cretaceous Wrangellia composite terrane accretion to the North American margin and the Late Eocene–Present oblique accretion and flat-slab subduction of the Yakutat microplate (Fig. 1) (e.g., Nokleberg et al., 1994; Plafker et al., 1994). Evidence of the Yakutat flat-slab subduction is found in areas above the subducted slab located in south-central Alaska and the Alaska Range (Fig. 1) (e.g., Enkelmann et al., 2008; Benowitz et al., 2011, 2014; Finzel et al., 2011; Brennan and Ridgway, 2015). Our study area is located east of the Yakutat slab and north and east of the indenting Yakutat plate corner where deformation transitions from transpressional to compressional (Fig. 2B). This area is known as the St. Elias syntaxis and gained recent attention due to its complex structural setting and locally concentrated, high deformation rates (e.g., O’Sullivan et al., 1997; Pavlis et al., 2004; Enkelmann et al., 2009, 2015a, 2015b; Koons et al., 2010). The Wrangellia composite terrane has generally been ascribed the role of the backstop to Cenozoic accretion with limited amounts of exhumation (e.g., Pavlis and Roeske, 2007; Enkelmann et al., 2008, 2010; Spotila and Berger, 2010). However, geophysical data and large-scale geodynamic models suggest that deformation at the current plate boundary is partly transferred inland to the northern margin of the Wrangellia composite terrane and farther to the eastern deformational front of the North American Cordillera (e.g., Mackenzie Mountains in the Northwest Territories; Fig. 1) (e.g., Lahr and Plafker, 1980; Mazzotti and Hyndman, 2002; Koons et al., 2010; Doser, 2014).

We use detrital zircon and apatite fission-track (ZFT and AFT, respectively) analyses on modern sand deposits of glacial and fluvial catchments to investigate the long-term cooling and exhumation history of the remote, rugged, and extensively glaciated area located north and east of the St. Elias syntaxis (Figs. 1 and 2). The detrital thermochronology approach proved to be very powerful to reveal the spatial pattern of rock exhumation occurring above and below the ice fields and glaciers of the southern St. Elias Mountains (Enkelmann et al., 2008, 2009, 2010, 2015a; Grabowski et al., 2013; Falkowski et al., 2014, 2016). The combination of two thermochronometric methods recording cooling through temperatures of 250 ± 40 °C and 110 ± 10 °C, respectively (Gleadow and Duddy, 1981; Brandon et al., 1998), allows identifying changes of cooling through time. We present 3537 new single-grain ages (detrital ZFT and AFT) recording the late Mesozoic–Cenozoic cooling and exhumation in the northern St. Elias Mountains and investigate the temporal and spatial far-field effects of the Yakutat microplate collision.

Key to understanding the cooling record of an orogen is information about the geology in general and particularly the formation age of the rocks, possible sources of thermal overprint, and the tectonic setting. This information allows differentiating between magmatic and/or metamorphic cooling, tectonic and/or erosional denudation, and evaluating climate-tectonic interactions. In the following, we summarize what is known about our 30,142 km2 study area, which includes the northern St. Elias Mountains and the Kluane and Ruby ranges (Fig. 2A). Because most of the area is heavily glaciated, the geology must be inferred from the ice-free ridges and surroundings. Only the two catchments in the northwestern study area and the large Alsek River catchment contain considerable ice-free areas (Fig. 2A).

Tectonic Setting

The North American Cordillera is the classic study area for terrane accretion processes, which started during Mesozoic time and are characterized by major strike-slip faults such as the Tintina and Denali faults (Fig. 1; Jones et al., 1972; Coney et al., 1980). Pacific plate motions have been highly oblique relative to the North American plate since at least 84 Ma, indicating a long-lived transpressional tectonic setting, consistent with suggested large (hundreds to thousands of kilometers) displacements of terranes (e.g., Lowey, 1998; Doubrovine and Tarduno, 2008; Garver and Davidson, 2015). Despite the fact that the paleogeography of some terranes is still under debate (e.g., Garver and Davidson, 2015), it is important for our study that terranes have been situated between major transcurrent faults since the Late Cretaceous, even though plate motion directions changed through time (e.g., Doubrovine and Tarduno, 2008).

Today, the geodynamics of southern Alaska are dominated by the collision of the Yakutat microplate, which is a 15–30-km-thick, westward tapering oceanic plateau capped by a remnant Campanian–Maastrichtian accretionary prism (Yakutat Group) in the east and Cenozoic sediments in the west (Fig. 3) (Plafker, 1987; Worthington et al., 2012). The Fairweather–Queen Charlotte transform plate boundary (Fig. 1) formed in the mid-Eocene and resulted in the northwestward translation of the Yakutat microplate (e.g., Plafker et al., 1994; Haeussler et al., 2003). A prominent, westward bend in the plate boundary and associated topography marks the St. Elias syntaxis, which is the transition zone between transform motion and subduction (Fig. 3). During the past decade, various data have been published that illuminate the structural situation and geologic evolution of the St. Elias orogen with a strong emphasis on the southern flanks located in Alaska. Particularly geophysical data, glacial geomorphology, and detrital thermochronology provided insight into the geologic processes underneath the ice cover (e.g., Bruhn et al., 2004, 2012; Berger et al., 2008; Enkelmann et al., 2008, 2009, 2010, 2015a, 2015b; Elliott et al., 2010, 2013; Chapman et al., 2012; Pavlis et al., 2012; Worthington et al., 2012; Van Avendonk et al., 2013), but those data are lacking from the northern side, located in Canada (Fig. 2A). In the following, we summarize what is known about the St. Elias orogeny and highlight some of the major open questions, some of which we are able to address in this study.

The northern Fairweather fault is characterized by transpression with strike-slip motion along the Fairweather fault and thrusting along northeast-dipping faults in the Yakutat foothills (Fig. 3) (Plafker and Thatcher, 2008; Elliott et al., 2010). West of the syntaxis, the Cenozoic cover is being deformed and imbricated by the shallow, northwest-dipping thrusts of the fold-and-thrust belt. It was suggested that the Fairweather fault ends in oblique-extensional splay faulting beneath the Seward Glacier and dextral motion is accommodated along the Bagley fault beneath the Bagley Ice Valley (Figs. 2A and 4; Ford et al., 2003; Bruhn et al., 2004, 2012). The geometry and slip history of the ice-covered Bagley fault is not well known, but it may have accommodated significant vertical motion in the past ∼5 m.y. as a backthrust and dextral motion since ca. 20 Ma consistent with the model of counterclockwise rotation and escape tectonics of southern Alaska (Berger et al., 2008; Bruhn et al., 2012). Our study area is located northeast of the syntaxis and the inferred Connector fault that possibly transfers strain from the Fairweather to the Denali fault (Figs. 2A and 3). Large-scale geodetic studies indicate that strain is transferred from the plate boundary inland to the Denali fault and to the eastern Cordilleran deformation front (e.g., Mazzotti and Hyndman, 2002; Brennan and Ridgway, 2015; Marechal et al., 2015). However, in the southwest corner of Yukon, seismic and GPS station cover is poor (e.g., Doser, 2014; Marechal et al., 2015) and leaves geophysical models ambiguous about specific structures in the study area. It is uncertain whether the Connector fault exists and, if so, when it became active. The role of the Duke River fault and whether it is responsible (rather than the Connector fault) for the bypassing of the eastern Denali fault is also uncertain (see slip rates in Fig. 3) (e.g., Lahr and Plafker, 1980; Doser, 2014; Marechal et al., 2015; Brennan and Ridgway, 2015). The Duke River fault occupies the reactivated suture and thrusts Alexander terrane rocks over Wrangellia terrane rocks (Cobbett et al., 2010). Evidence for its activity and ∼10–13 km of vertical motion comes from muscovite growth (ca. 104–84 Ma) in the Cretaceous, folding of Miocene Wrangell lava in the Pliocene, and current seismicity (Cobbett et al., 2010; Cobbett, 2011; Doser, 2014).

Seismic studies have imaged the Yakutat slab underneath the fold-and-thrust belt, south-central Alaska, and the Alaska Range (Figs. 1 and 3; e.g., Eberhart-Phillips et al., 2006; Bauer et al., 2014), but seismic data have not been able to image the northeastern edge of the slab beneath the St. Elias Mountains better than locating it somewhere east of Mount St. Elias (Bauer et al., 2014). It is certain that the slab is not currently passing beneath the study area (Fig. 3), but it might have in the past.

Geology of the Study Area

The catchments investigated in this study comprise mainly rocks of the Wrangellia composite terrane and, to a smaller extent, rocks of the Yukon composite terrane north of the Denali fault (upper Alsek River catchment) and the Chugach–Prince William terrane south of the Border Ranges fault (catchment KLD29; Figs. 2B and 4). The Yukon composite terrane in the study area comprises mostly early Cenozoic plutonic rocks of the Kluane arc (ca. 85–45 Ma), which are part of the widespread Coast plutonic complex (ca. 175–45 Ma) of the North American Cordillera (e.g., Armstrong, 1988; Erdmer and Mortensen, 1993; Gehrels et al., 2009). Associated with the intrusions, Cretaceous sedimentary rocks have been metamorphosed (Kluane metamorphic assemblage; Fig. 4). Rocks of the Aishihik metamorphic assemblages (late Proterozoic–Mississippian) and Aishihik plutonic complex (ca. 186 Ma) are exposed within the study area as well (Erdmer and Mortensen, 1993; Johnston and Erdmer, 1995; Johnston et al., 1996; Johnston and Canil, 2007; Israel et al., 2011). Along the entire North American Cordillera, the southern margin of the Yukon composite terrane and the northern margin of the Wrangellia composite terrane are characterized by regional, mid-Cretaceous folding, thrusting, and metamorphic overprint, which is interpreted to reflect the final accretion of the Wrangellia composite terrane (e.g., Rubin et al., 1990; Dusel-Bacon et al., 1993; Rusmore et al., 2000; Gehrels, 2001).

The Wrangellia composite terrane comprises in the study area the Paleozoic–early Mesozoic island arc assemblages of the Alexander and Wrangellia terranes (Figs. 1 and 5A). The two terranes were stitched together by Upper Pennsylvanian plutons but share a common tectonic setting since at least the Late Devonian (Gardner et al., 1988; Israel et al., 2014). The history of accretion of the Wrangellia composite terrane to the former North American margin is not entirely resolved, but the two units were likely close in the Late Jurassic, when backarc basin sedimentation occurred along the inboard margin of the Wrangellia composite terrane (e.g., Trop and Ridgway, 2007). In the study area, these backarc basin turbidites were fed by the adjacent Chitina arc (ca. 160–130 Ma) and are represented by the Upper Jurassic–Lower Cretaceous Dezadeash Formation (Figs. 4 and 5; Berg et al., 1972; Eisbacher, 1976; Lowey, 1992). Remnants of the Chitina arc are represented by the widespread St. Elias plutonic suite (Figs. 4 and 5). Approximately 30–50 km north of the Chitina arc, the Chisana arc formed ca. 120–105 Ma (Fig. 5B) due to continued subduction at the outboard margin of the Wrangellia composite terrane. Remnants are the scattered Kluane Ranges plutonic suite in the wider Denali fault zone (Figs. 4 and 5). The Eocene–Oligocene Amphitheatre Formation overlaps the Wrangellia and Alexander terrane rocks in the Denali and Duke River fault zones, where local strike-slip basins formed (Figs. 4 and 5A) (e.g., Ridgway and DeCelles, 1993). During the Early to mid-Miocene, Wrangell lava leaked along transtensional faults and overlies older units in the Duke River fault area (Figs. 4 and 5A) (Skulski et al., 1991, 1992; Israel et al., 2006). Typical intraplate calc-alkaline volcanics and plutonic rocks occur more extensively in the Wrangell Mountains located west of the study area (26–0 Ma; Richter et al., 1990; Trop et al., 2012) and reflect the progressive oblique subduction of the Yakutat microplate and slab edge volcanism (Fig. 1) (Richter et al., 1990).

The Chugach–Prince William terrane represents a vast accretionary complex derived from a long-lived volcanic arc, most likely the Coast plutonic complex (ca. 175–45 Ma) of today’s British Columbia (Dumoulin, 1988; Farmer et al., 1993; Garver and Davidson, 2015). A narrow band of Upper Cretaceous flysch of the Valdez Group is exposed in the southern study area (Fig. 4). This area is part of the Eocene greenschist- to amphibolite-facies Chugach metamorphic complex (Fig. 1) that resulted from near-trench intrusions of the Sanak-Baranof belt associated with a spreading-ridge subduction ca. 62–47 Ma (Hudson et al., 1977a, 1979; Bradley et al., 1993; Pavlis and Sisson, 1995; Gasser et al., 2011, 2012).

Previous Geochronologic and Thermochronologic Data

Higher-temperature chronometric data have been published from the study area, particularly K-Ar and 40Ar/39Ar ages that are largely interpreted as emplacement ages of plutons or lava and some metamorphic overprint (e.g., Hudson et al., 1977a, 1977b; Dodds and Campbell, 1988; Farrar et al., 1988; Richter et al., 1990; Erdmer and Mortensen, 1993; Johnston et al., 1996; Trop et al., 2012). In contrast, only few lower-temperature chronometer data have been reported from the study region including apatite and zircon (U-Th)/He (AHe and ZHe, respectively) and FT ages (Fig. 4) (O’Sullivan et al., 1997; McAleer et al., 2009; Spotila and Berger, 2010). For reasons of clarity, the higher-temperature chronometers are omitted from Figure 4 but are shown in Figure S11. For compilations of thermochronometric ages from the entire St. Elias Mountains, the reader is referred to Enkelmann et al. (2010, 2015a), Spotila and Berger (2010), Gasser et al. (2011), and Falkowski et al. (2016). Overall, the low-temperature chronometric data reflect the ongoing subduction and collision of the Yakutat microplate and efficient glacial erosion starting 6–5 Ma (e.g., Lagoe et al., 1993; Berger et al., 2008; McAleer et al., 2009; Enkelmann et al., 2010, 2015b; Spotila and Berger, 2010). Exhumation is most rapid and deepest in the St. Elias syntaxis area, interpreted from young (≤5 Ma) detrital ZFT and AFT ages (Enkelmann et al., 2009, 2010, 2015a; Falkowski et al., 2014) and bedrock AFT and AHe ages (McAleer et al., 2009; Spotila and Berger, 2010; Enkelmann et al., 2015b). AHe bedrock ages generally increase northward from the plate boundary and within the Wrangellia composite terrane but form a bulge of younger ages northeastward from the indenting Yakutat plate corner suggesting deformation and exhumation occurred also inboard of the plate boundary (Fig. 4; Spotila and Berger, 2010). While the AHe ages record cooling of rocks within the upper ∼2–3 km of the crust (closure temperature of 60 ± 15 °C; Farley, 2000), our new ZFT and AFT data set allows us to study the longer-term cooling history of a broad area and analyze material eroded below the ice.

We sampled 21 glacial and glacio-fluvial catchments in the northern St. Elias Mountains and Kluane and Ruby ranges (Table 1 and Fig. 2). Sampling locations were chosen to include all large glaciers that drain the high Icefield region to the north and east (Fig. 2A). Smaller catchments located farther inland in the Kluane Ranges were sampled to better confine exhumation signals obtained from regions closer to the St. Elias syntaxis (Fig. 2). Three samples are from the Alsek River, which is the only river crossing the St. Elias Mountains and transporting material into the Gulf of Alaska (from north to south KLD20, KLD23, and KLD27; Fig. 2). Note that the Alsek River catchment encompasses most of the catchments in the eastern study area (Fig. 2). Also, catchment KLD110 comprises catchment KLD78, catchment KLD105 is contained within KLD106, and catchment KLD66 lies within the reaches of catchment KLD65 (Fig. 2).

At each sample location, 3–5 kg of medium- to coarse-grained sand were collected for detrital ZFT and AFT analysis. Additionally, five bedrock samples (1–3 kg each) have been collected that allow comparing their ZFT ages to the detrital cooling age record (Fig. 2 and Table 2). Per sample, ∼100 single-grain ages were analyzed using the ζ-calibration method (Hurford, 1990). Analytical details on the fission-track dating procedure can be found in Text S1 (see footnote 1). For extraction of detrital age components, we used the software BINOMFIT (Brandon, 1992, 1996), which employs binomial peak fitting to the single-grain age distributions.

Bedrock Zircon Fission-Track Analysis

Five bedrock ZFT ages from samples located within the analyzed catchments are presented in Table 2. Two bedrock samples from the upper Alsek River catchment yielded ZFT ages of ca. 43 Ma (KLB5) and ca. 110 Ma (KLB91). Sample KLB5 was taken from a granitoid of the Coast plutonic complex (Table 2 and Fig. 4). U-Pb and K-Ar ages from the area (Fig. S1 [see footnote 1]) suggest that KLB5 has Early Eocene crystallization and K-Ar cooling ages. Mudstone sample KLB91 was collected from the Upper Jurassic–Lower Cretaceous Dezadeash Formation and exhibits a mid-Cretaceous cooling age that is probably related to regional deformation during the final stage of Wrangellia composite terrane accretion. This deformational event resulted in low- to medium-grade metamorphism within the flysch deposits in the collision zone (e.g., Dusel-Bacon et al., 1993) and reset the ZFT system of sample KLB91. Three granitoid samples from the east-west–elongated Dusty Glacier catchment (KLD40, Fig. 2) exhibit decreasing ZFT cooling ages of ca. 154 Ma, ca. 101 Ma, and ca. 9.4 Ma from east to west and with increasing elevations (1383–2637 m above sea level [asl]; Table 2). According to the geologic map of southwest Yukon (Israel, 2004) the westernmost sample KLB44 was collected from the Early Permian (290–270 Ma) Icefield Ranges plutonic suite, and KLB41 and KLB42 represent the Upper Jurassic–Lower Cretaceous St. Elias plutonic suite (160–130 Ma; Figs. 4 and 5). This indicates that sample KLB41 cooled below ∼250 °C shortly following its emplacement, while KLB42 either cooled much slower or experienced a heating event and was reset, which left KLB41 unaffected even though the samples were taken only 20 km apart from another (Fig. 2B). The youngest cooling age of ca. 9.4 Ma of KLB44 reflects exhumational cooling and occurs in roughly the same area as a young AHe age of ca. 4.3 Ma reported by Spotila and Berger (2010) and relatively close to ca. 3.6 Ma AHe (McAleer et al., 2009) and 4.5 Ma AFT ages (O’Sullivan et al., 1997) (Fig. 4).

Detrital Fission-Track Analysis

In the following, the results of 21 detrital, glacio-fluvial samples, all of which were dated by ZFT analysis (2187 new ZFT single-grain ages) and 15 of them also by AFT analysis (1350 new AFT single-grain ages), are reported. Results of age component extraction are shown in Figure 6 and summarized in Table 3, which divides the ZFT and AFT age populations into time intervals that combine similar age populations but also follow different regional tectonic settings.

The 15 AFT samples each yielded 2–3 age populations that range between 331.0 ± 70.1 Ma (2.5%, KLD13) and 8.7 ± 2.4 Ma (34%, KLD33), but are mostly Late Cretaceous–Miocene (Table 3). Age clusters can be observed at ca. 90–80 Ma and ca. 30 Ma (Table 3). With the exception of KLD105, samples with a ca. 30 Ma age population (12%–63%; Table 3) occur in the southern part of the study area. Furthermore, some samples contain Late Miocene 13–8 Ma AFT age populations (KLD23, KLD26, KLD27, KLD29, KLD33, KLD40, and KLD85; Table 3), the majority of which are located in the southern part as well.

From the study area, three detrital ZFT samples have been published previously (Enkelmann et al., 2015a). These have been collected from the same glacial catchments as samples KLD40, KLD65, and KLD105. The previous and new data generally agree. For a comparison of previous and new ZFT data for these catchments, see Table S1 (see footnote 1). Each of the new detrital ZFT samples yielded 2–4 different age populations with age peaks that range between 253.8 ± 28.9 Ma (14%, KLD110) and 2.5 ± 1.5 Ma (3%, KLD78). Most age peaks fall into the Cretaceous; some cluster around 130 Ma and 120–115 Ma, while Late Cretaceous age peaks show a wide range (Table 3). Early Cenozoic detrital ZFT age populations show some common age peaks in the range 55–40 Ma; these peaks can be seen in eight of the samples throughout the study area (Table 3 and Fig. 2B). In contrast, three age populations with 25–22 Ma age peaks (KLD17, KLD25, and KLD27; Table 3) occur in the southern study area close to or in the Chugach terrane (cf. Table 3 and Fig. 2B).

The largest young age populations are derived from the two catchments located in the Wrangell volcanic belt (KLD78 and KLD110; Table 3 and Figs. 2B and 4) and are therefore considered to be of volcanic origin and reflect instantaneous cooling at the time of eruption. Looking at the distribution of Wrangell lava in the study area, ZFT and AFT ages from sample KLD85 and the Alsek River samples could reflect volcanic cooling as well (cf. Figs. 2B and 4). Reported ages for the Alsek volcanic field within these catchments range from 13.5 to 10.8 Ma (Stevens et al., 1982; Dodds and Campbell, 1988) to 16.4–15.4 Ma (Trop et al., 2012). If the younger range is valid, the ca. 12.2 Ma AFT age population of KLD85 could be of volcanic origin. Samples KLD20 and KLD23 from the Alsek River contain ca. 10.6 Ma (61%) and ca. 11.3 Ma (37%) AFT age populations, respectively (Table 3), but considering that the mid-Miocene Alsek volcanics make <5% of the catchment area, it is likely that those age populations represent an exhumational rather than a volcanic signal. Mid-Miocene and younger ZFT and AFT age populations in other catchments (other than KLD110, KLD78, and KLD85) are considered as exhumational signals because Wrangell lava is either absent or occurs in only very small areas, and most importantly, are younger than mid-Miocene. No younger than mid-Miocene Wrangell volcanic ages have been reported for the study area (e.g., Skulski et al., 1991; Trop et al., 2012). The youngest (<10 Ma) non-volcanic ZFT cooling age peaks occur in catchments KLD13 (6.5 ± 0.8 Ma, 11%), KLD27 (9.7 ± 1.2 Ma, 4%), and KLD40 (7.6 ± 0.4 Ma, 64%). Catchments KLD13 and KLD40 lie within the ca. 19,200 km2 Alsek River catchment of sample KLD27 (Fig. 2B). Considering the fact that the other two samples collected from the Alsek River upstream of KLD27 do not contain such young ZFT ages (Table 3), the young ages of KLD27 must be derived from the southern part of the Alsek River catchment.

The ZFT age distributions also contain few Jurassic and older age peaks (>145 Ma, mostly >10%; Table 3), mainly 165–150 Ma. It is possible that no older cooling record is preserved, but it should also be noted that with increasing cooling age or uranium content, it becomes increasingly difficult to date zircons by the fission-track method due to overlapping tracks. In many samples, a few grains (<7) could not be dated due to track densities that were too high.

The record of rock cooling in the study area can be divided into phases of terrane deformation that encompass the Mesozoic arc magmatism and accretion of the Wrangellia composite terrane, as well as Cenozoic deformation due to continued accretion of terranes at the new continental margin. While the Mesozoic deformational history of the study area resembles the history of the entire North American Cordillera, the Cenozoic history is more unique due to the inboard location from the St. Elias syntaxis. These aspects of older and younger terrane deformation will be outlined in the following. Older, Paleozoic–early Mesozoic deformation is preserved in the structural and metamorphic record (e.g., Dodds and Campbell, 1992; Dusel-Bacon et al., 1993; Cobbett et al., 2010; Beranek et al., 2014) but not in the detrital FT data.

Mesozoic Cooling

Carboniferous–Early Cretaceous ZFT and AFT age populations are dominated by magmatic phases and associated metamorphism that reset almost the entire previous thermal record of the Cambrian–Upper Triassic basement rocks of the Wrangellia composite terrane. The Late Jurassic cooling age populations (around 155 Ma) of the Alexander terrane are assignable to magmatic activity and subsequent cooling of the Chitina arc–St. Elias plutonic suite (Fig. 5) (e.g., Dodds and Campbell, 1988; Plafker et al., 1989, 1994). After cessation of Chitina arc activity (ca. 130 Ma), rocks cooled probably due to thermal relaxation, as recorded by the large group of ≤130 Ma ZFT and AFT age peaks (Table 3). Magmatic intrusions and heating of the crust might have uplifted the crust and resulted in erosion and exhumation. Renewed magmatic activity of the Chisana arc began ca. 120 Ma (Fig. 5), and remnants can be found only in the Wrangellia terrane part of the study area (KLD110 and KLD78, Alsek River catchment), but a characteristic cooling signal like that of the St. Elias plutonic suite is not detected from these catchments (Table 3). However, some catchments in the Alexander terrane indicate late Early Cretaceous cooling. The area has been in a forearc position at that time, but no sedimentary strata are preserved. Therefore, cooling age populations of this time might reflect an uplifted area experiencing erosion.

This kind of setting must have continued throughout the mid- and Late Cretaceous but with enhanced exhumation indicated by the large cooling age populations dating this time interval that are found throughout the study area (Table 3). Final accretion of the Wrangellia composite terrane to North America caused deformation that is recorded along the entire length of the composite terrane during the mid-Cretaceous (e.g., McClelland and Gehrels, 1990; Dodds and Campbell, 1992; Dusel-Bacon et al., 1993; Gehrels, 2001; Ridgway et al., 2002), while the southern margin deformed due to continued subduction and accretion of sediments (e.g., Plafker et al., 1989). Furthermore, the Denali and Border Ranges faults became active as major dextral strike-slip faults in the Late Cretaceous due to very rapid oblique subduction, placing the study area in a transpressional setting (Engebretson et al., 1985; Little and Naeser, 1989; Smart et al., 1996; Doubrovine and Tarduno, 2008). At the northern margin, the Duke River thrust fault also became active or reactivated, placing greenschist-facies Alexander terrane rocks over prehnite-pumpellyite–facies Wrangellia terrane rocks (Cobbett et al., 2010). Thus, our cooling age populations represent a mixture of tectonic and erosional denudation and possibly metamorphism associated with this mid- to Late Cretaceous orogeny. Similarly, the Coast plutonic complex in British Columbia records very high exhumation and erosion rates during the Late Cretaceous–Paleocene resulting in the vast Chugach–Prince William accretionary complex (e.g., Hollister, 1982; Farmer et al., 1993; Garver and Davidson, 2015).

Even though the detrital thermochronologic data also record Cenozoic cooling, the fact that the Late Jurassic–Cretaceous cooling record is so dominant and well preserved suggests a limited amount of erosion in the study area since that time. In contrast, the detrital FT record from the Chugach–Prince William terrane and Yakutat microplate contains barely any older than Cenozoic cooling (Enkelmann et al., 2008, 2009, 2010, 2015a; Grabowski et al., 2013; Falkowski et al., 2014, 2016). Figure 7 illustrates the detrital ZFT and AFT cooling signals of the study area in comparison to the northern Fairweather fault area and the southern and western St. Elias Mountains in form of single-grain pie charts. It is important to note that single-grain ages have large uncertainties and are not used for the interpretation of cooling phases. Nevertheless, the visualization with pie charts gives a good overview of the spatial occurrence of cooling ages. The abrupt younging of cooling age populations south of the Border Ranges fault (except the St. Elias syntaxis area, which is discussed below) is well observed in KLD29 with mainly Oligocene and younger cooling ages of both ZFT and AFT systems (Table 3 and Fig. 7), and in samples CH46, CH21, and WR23 (Fig. 7; Enkelmann et al., 2008). AHe bedrock samples from the western St. Elias Mountains show the same trend (Spotila et al., 2004; Enkelmann et al., 2010; Spotila and Berger, 2010).

The upper Alsek River catchment is located in the Yukon composite terrane and mostly comprises rocks of the Coast plutonic complex. Based on the 43 Ma ZFT age of bedrock sample KLB5 (Table 2) and other K-Ar cooling ages (Fig. S1 [see footnote 1]) we expected Eocene ZFT cooling ages from the plutons and mid-Cretaceous ages from the Wrangellia suture zone, e.g., from the Dezadeash Formation (ca. 110 Ma ZFT age of KLB91; Table 2 and Fig. 7). However, those cooling ages are barely present in the Alsek River samples (KLD20, KLD23, and KLD27; Table 3). This is most likely because of the morphology of the large Alsek River catchment and long transport distance (up to 270 km). The catchment is characterized by low-relief topography north of the Denali fault, and sediment is stored in the wide valleys before the river enters the more rugged topography of the St. Elias Mountains (Figs. 2 and 7). The sand-sized material we collected along the Alsek River thus originates likely from the large glaciers that drain the high Icefield region but not from the upper Alsek River catchment.

Cenozoic Cooling

We discuss the Cenozoic cooling recorded in our study area in terms of three different phases: (1) Eocene spreading-ridge subduction and slab-window passage followed by plate reorganization; (2) Oligocene–Present Yakutat flat-slab subduction; and (3) rapid, deep St. Elias syntaxis exhumation.

Eocene Spreading-Ridge Subduction and Plate Reorganization

Paleocene–mid Eocene ZFT and AFT age populations (ca. 60–45 Ma; Table 3) occur throughout the entire study area and are similar to the ZFT age populations found in the fold-and-thrust belt and in the Yakutat foothills, south of the active plate boundary (Fig. 7; Enkelmann et al., 2009; Falkowski et al., 2014). However, the provenance of the Eocene–Oligocene fold-and-thrust belt sediments and the Yakutat Group is much farther south along the western North American margin (Plafker et al., 1994; Perry et al., 2009; Garver and Davidson, 2015). Eocene cooling has also been reported from farther west, based on ca. 45 Ma detrital ZFT and AFT age populations found in the eastern and central Alaska Range (Lease et al., 2016) and thermal history modeling of K-feldspar 40Ar/39Ar data from the western Alaska Range (Fig. 1; Benowitz et al., 2012). This widespread Eocene cooling phase is well recognized in the geologic record of southern Alaska and documents oblique subduction of a spreading ridge that resulted in diachronous (ca. 62–47 Ma), east-to-west–migrating, near-trench plutonism within the Chugach–Prince William accretionary prism (Hudson et al., 1977a, 1979; Bradley et al., 1993; Pavlis and Sisson, 1995; Gasser et al., 2011, 2012). The inboard effects are not well understood, but the sedimentary record shows Eocene episodes of uplift and subsidence inferred from the change from marine to coarse-grained, non-marine sedimentation and magmatism farther west on the Wrangellia composite terrane (Matanuska Valley and Talkeetna Mountains; Fig. 1). This was interpreted as an effect of the migrating slab window beneath southern Alaska (Trop et al., 2003; Trop and Ridgway, 2007).

The thermochronologic record of the spreading-ridge subduction has been more ambiguous because it is difficult to distinguish between cooling due to thermal relaxation and rock exhumation (e.g., Benowitz et al., 2012; Lease et al., 2016). Because the slab window passed beneath the central and eastern Alaska Range ca. 60–56 Ma, Lease et al. (2016) suggested that the Eocene detrital ZFT and AFT age populations (ca. 45 Ma) record cooling due to the 47–43 Ma reorganization of the Pacific plates that followed the spreading-ridge subduction. This change in relative plate motions resulted in a more orthogonal convergence, westward directed subduction in south-central Alaska, the establishment of the Aleutian arc, and initiation of the Fairweather–Queen Charlotte transform plate boundary (Fig. 1; Engebretson et al., 1985; Atwater and Stock, 1998; Haeussler et al., 2003; Doubrovine and Tarduno, 2008). Even after plate reorganization, relative plate motion was primarily oblique rather than orthogonal in our study area, and ZFT and AFT cooling age populations are closer in time to the suggested spreading-ridge subduction at that portion of the margin (ca. 53–50 Ma; Bradley et al., 1993; Gasser et al., 2011, 2012). However, it is not possible to determine the dominant driver of cooling, and it might have been a combination of different influences during the Eocene. Prior to spreading-ridge subduction, the downgoing oceanic crust becomes increasingly younger, thicker, and buoyant, which results in a stronger coupling with the upper plate and possibly uplift. Then, the subduction of the spreading-ridge and slab-window passage allows for an upwelling of hot mantle material, which together can result in the uplift and erosion of the upper plate as well as a significant increase in the geothermal gradient (e.g., Thorkelson and Taylor, 1989; Cloos, 1993; Sakaguchi, 1996; Bradley et al., 2003). The change in relative plate motions might have had an effect on rock exhumation as well, suggested by the 43 Ma increase in exhumation rates found at Mount Logan (O’Sullivan and Currie, 1996). However, our Eocene age populations do not cluster around the time of plate motion change (47–43 Ma) and are generally older, suggesting that spreading-ridge subduction and slab-window development had a strong impact on the rock heating and subsequent cooling in our study area. Benowitz et al. (2012) also described a scenario of spreading-ridge subduction, flat-slab subduction, and slab-window evolution to be responsible for uplift and Eocene cooling ages in the western Alaska Range and entire southern Alaska rather than plate reorganization.

Oligocene–Present Yakutat Flat-Slab Subduction

Oligocene–Early Miocene cooling age populations are dominant in the detrital AFT samples from the southern catchments of the study area, but they are generally sparse in the detrital ZFT record (Table 3; orange bins of the pie charts in Fig. 7). The fact that only apatite records this cooling phase indicates that the total amount of exhumation since the Oligocene was limited to ∼3.0–4.5 km (assuming a paleogeothermal gradient of 25–35 °C/km). In contrast, Oligocene and younger cooling ages are far more abundant in the detrital ZFT and AFT data farther south and west in the St. Elias Mountains (Fig. 7; Enkelmann et al., 2009, 2010, 2015a; Grabowski et al., 2013; Falkowski et al., 2014). Oligocene and younger detrital and bedrock cooling ages have also been found farther west in the Chugach and Talkeetna Mountains and in the Alaska Range (Fig. 1), and they have been associated with plate coupling due to the flat-slab subduction of the Yakutat microplate (Benowitz et al., 2011, 2012, 2014; Arkle et al., 2013; Lease et al., 2016). The effect of the flat-slab subduction on the upper plate has also been shown by sedimentologic and provenance studies (e.g., Trop and Ridgway, 2007; Brennan and Ridgway, 2015; Finzel et al., 2015) and numerical modeling studies (Jadamec et al., 2013). During the same time, deformation farther southeast is dominated by transpression along the Fairweather fault (Chapman et al., 2012; Falkowski et al., 2014, 2016).

We suggest that the Oligocene cooling signal in our study area is related to exhumation due to subduction of the Yakutat microplate. Even though the Yakutat slab is not presently beneath the study area (Fig. 3), it probably was in the past. The Late Eocene–Oligocene Amphitheatre Formation was deposited in a strike-slip basin that developed in the Denali and Duke River fault zones in transpressional and transtensional settings, and sediment was sourced from local, high-relief topography (Ridgway and DeCelles, 1993). This indicates oblique convergence and subduction of the thick Yakutat microplate during that time and is consistent with erosion and exhumation in the study area. Furthermore, leaking of Wrangell lava at transtensional faults in the Miocene and Wrangell arc volcanism beginning 26 Ma are consistent with the passage of the Yakutat microplate beneath the study area and the location of the slab edge closer to the Duke River fault zone at that time (Richter et al., 1990; Skulski et al., 1991, 1992; Trop et al., 2012). Compared to regions farther west, the amount of exhumation is smaller in the study area possibly due to the shorter time that the slab was present beneath the study area.

Coupling between the Yakutat microplate and the upper plate probably increased over time as the crustal thickness of the subducting plate increased (Worthington et al., 2012). A mid-Miocene cooling signal (ca. 15–12 Ma) has been observed in bedrock and detrital samples from the St. Elias syntaxis area and the northern Fairweather fault and was interpreted to reflect increasing resistance to subduction of the Yakutat microplate (O’Sullivan and Currie, 1996; Meigs et al., 2008; Grabowski et al., 2013; Falkowski et al., 2014). A few ca. 18–10 Ma AFT and ZFT age peaks are also found in the southern study area, especially in the southernmost KLD29 catchment, where a large (54%) ca. 13 Ma AFT age peak and a smaller ca. 14 Ma ZFT age peak are found (Table 3). The large ca. 12 Ma (64%) AFT age peak of catchment KLD85 (Table 3) farther north in the study area is left out here because it might be a volcanic signal as discussed earlier. Likewise, KLD110 and KLD78 are excluded here. All other mid-Miocene cooling age populations, however, likely reflect distal response to Yakutat–North American plate coupling and collision.

St. Elias Syntaxis Exhumation

Detrital ZFT data revealed that exhumation with extremely high rates and from greater depth occurred below the ice in a ca. 4800 km2 area encompassing the Seward and Hubbard-Valerie glacial catchments and the northern Fairweather fault (Fig. 7; Enkelmann et al., 2008, 2009, 2010, 2015a; Grabowski et al., 2013; Falkowski et al., 2014, 2016). Detrital ZFT cooling age populations of <5 Ma are much younger than expected from surrounding bedrock data, even when considering the lower elevation sources of detrital samples (below the ice compared to bedrock samples from ice-free ridges) and heat advection in response to rapid exhumation (Spotila and Berger, 2010). The regional extent of rapid and deep exhumation was defined based on the occurrence of young detrital ZFT age populations in glacial catchments (cf. composite pie charts in Fig. 7 and reddish-colored catchments in Fig. 8A), whereby the northern and eastern boundaries were uncertain due to the lack of data from catchments draining to the north. Our new data show that no <5 Ma detrital ZFT age populations occur on the northeast side of the speculated Connector fault (Table 3; excluding volcanic cooling signals). Whether the fault trace is the limit of the rapidly exhuming area is unknown due to the large catchments, but the location of the inferred Connector fault is an obvious choice. Seismic activity of the Art Lewis Glacier fault, the southern part of the Connector fault, indicates dextral shearing (Bruhn et al., 2012), but it is possible that the fault accommodated significant vertical motion in the past as it is suggested based on geodynamic models that imply a south-dipping reverse fault or shear zone north of the Fairweather fault in the syntaxis area (Koons et al., 2010). While geologic and seismic studies suggest that inland strain transfer along the Connector fault might be a young feature (<1 m.y.; Lahr and Plafker, 1980; Doser, 2014), the Art Lewis Glacier fault has probably been active for several million years (written communication between G. Plafker and Kalbas et al., 2008). However, the ice cover hampers geologic mapping, and glacial erosion potentially removes marker features.

Regardless of the structural accommodation of exhumation, further evidence for rapid and deep exhumation starting ca. 5 Ma in the syntaxial region comes from 6 to 5 Ma detrital ZFT age populations found in locally derived sediments deposited 5–4 Ma and exposed in the fold-and-thrust belt (Enkelmann et al., 2008, 2009). This short lag time between cooling below ZFT closure temperature and deposition requires very rapid exhumation (Enkelmann et al., 2010). The occurrence of rapid exhumation is also supported by the finding that detrital AFT cooling age peaks from the syntaxis region are essentially the same as the detrital ZFT age peaks (Enkelmann et al., 2015a), suggesting rapid cooling below both the ZFT and AFT closure temperatures (i.e., ca. 250 °C and ca. 110 °C, respectively) in 1–2 m.y. The amount of exhumation was, however, limited (ca. 10 km), which is implied by the facts that higher-temperature 40Ar/39Ar systems of detrital samples from the Seward and Hubbard glaciers (Falkowski et al., 2016) and bedrock ages at the surrounding ridges are much older (e.g., at Mount Logan; O’Sullivan and Currie, 1996). Enkelmann et al. (2015a) showed that the focus of deformation and the area of most rapid exhumation shifted south of the plate boundary (Fig. 8B) after ca. 2 Ma based on thermochronology data integrated with geophysical, structural, and surface processes data. This southward shift was explained by the rheologic modification of the incoming Yakutat microplate, which resulted from increasing erosion rates due to the onset of glaciation at 5.6 Ma and glacial intensification with the global climate shift at 2.6 Ma and onset of 100 k.y. glacial cycles in the mid-Pleistocene (Enkelmann et al., 2015a). The resulting high sedimentation rates on the Yakutat microplate (>2 mm/yr up to 14 mm/yr; Hallet et al., 1996; Lagoe and Zellers, 1996; Jaeger et al., 1998; Gulick et al., 2015) allowed the development of a décollement and a shift of deformation to shallow thrusts at the coast. With the emergence of the southern St. Elias Mountains that intercept with atmospheric circulation, the modern precipitation pattern was established and caused a focusing of erosion in the southern syntaxis region (Enkelmann et al., 2015a).

Our new data from the northern region of the syntaxis suggest that rapid exhumation has already been focused in the wider syntaxial region prior to 5 Ma. The youngest ZFT age peaks of our study area (ca. 7.6–6.5 Ma; Table 3) in KLD40 and KLD13 occur adjacent to the eastern and southeastern Hubbard Glacier catchment, which is also reflected in a small ca. 10 Ma ZFT age peak in the southernmost Alsek River sample (KLD27) (Figs. 7 and 8A). Furthermore, the southern parts of catchments KLD27 and KLD23 and catchment KLD26 record major ca. 10 Ma AFT age populations suggesting Late Miocene exhumation (Table 3). Bedrock data from the Dusty Glacier catchment (KLD40) allow to further confine the ca. 7.6 Ma age peak (64%; Table 3) to the upper (western) part of the catchment. There is a stark contrast between young bedrock ages at higher elevations (ca. 9.4 Ma ZFT age; this study; and ca. 4.3 Ma AHe age; Spotila and Berger, 2010) occurring in the western catchment and much older ZFT (>100 Ma; this study) and AHe (ca. 16.6 Ma; Spotila and Berger, 2010) bedrock ages in the eastern catchment (Fig. 8A). This contrast suggests a vertical offset across an unrecognized structure with significant exhumation on its western side after ca. 10 Ma. This is consistent with the KLD40 detrital ZFT and AFT age populations with peak ages at ca. 7.6 Ma and ca. 8.8 Ma, respectively, which are the same within 1σ errors (Table 3 and Fig. 8A). Considering the different elevations of bedrock and detrital samples, as well as heat advection and the dependence on fault geometry and slip history, we do not attempt to estimate rates or exact timing of exhumation here, other than a window of 10–5 Ma.

Taken together, the spatial exhumation pattern confirms the suggested “bull’s-eye” pattern for the St. Elias syntaxis, similar to other colliding plate corners where efficient erosion occurs, e.g., in the Himalaya (e.g., Enkelmann et al., 2011; Koons et al., 2013; Bendick and Ehlers, 2014) but adds a temporal dimension to it. We suggest a progressive focusing and increasing rates of post–ca. 10 Ma exhumation in the syntaxial area, accommodated by discrete structures such as the proposed fault in the Dusty Glacier catchment and the inferred Connector fault (Figs. 8B and 8C). The southwestern limit of suggested ca. 10–5 Ma rapid exhumation is not clear because the area overlaps with the area where <5 Ma exhumation signals prevail and potentially conceal 10–5 Ma exhumation. The focus of exhumation shifted south over time and into the area between Connector and Fairweather faults where we find clear evidence in the detrital ZFT age population record (Syntaxis/Nunatak Fjord composite chart in Fig. 7 and reddish polygon in Figs. 8A and 8B; Falkowski et al., 2014). Few, small ≤5 Ma detrital ZFT age populations were also found in the hanging wall of the Yakutat fault south of the Fairweather fault, suggesting that exhumation increased there as well, along shallower paths, however (Falkowski et al., 2014). The exhumation pattern in combination with inferred structures north of the Fairweather fault and known structures (thrust faults) south of the Fairweather fault in the Yakutat foothills indicate that exhumation might have been accommodated within a large-scale, two-sided flower structure with a migrating focus of deformation (Fig. 8C). Partial flower structures, i.e., linked strike-slip and thrust fault systems that develop to accommodate transpressional deformation, have been suggested previously to occur in the area, e.g., the Yakutat foothills thrusts and Fairweather fault or the Art Lewis Glacier fault and thrust faults at Mount Logan (Bruhn et al., 2004, 2012). However, our finding of a much larger region of rapid exhumation since ca. 10 Ma requires geodynamic models to take into account the larger amount of deformation accommodated within the syntaxis area in the past.

The new thermochronology data from the northern St. Elias Mountains reveal the long-term cooling history of the Wrangellia composite terrane that comprises two major phases of terrane collision and accretion to the western margin of North America. The first phase was the accretion of the Wrangellia composite terrane during Late Jurassic–mid-Cretaceous time. This phase was characterized by widespread magmatism that almost fully reset the thermal record of the study area. The fact that the Jurassic–Cretaceous record is so well preserved in the study area indicates the strong thermal effect of magmatism to the entire upper crust and hinders the record of coeval cooling due to erosion or tectonic rock exhumation by means of thermochronology. The subsequent, Late Cretaceous phase of exhumation is, however, recorded. It was related to a transpressional tectonic setting with rapid, oblique plate convergence that resulted in crustal shortening and thickening, and translation of terranes, similar to observation from the entire North American Cordillera (e.g., Farmer et al., 1993; Hollister and Andronicos, 2006).

The second phase of terrane accretion is related to the ongoing oblique flat-slab subduction and collision of the Yakutat microplate. After passage of the spreading ridge during the Eocene, the North American margin was affected by the subduction of the Yakutat microplate that initiated in the Late Eocene. Both events deformed particularly the outboard accretionary wedge (e.g., Hudson et al., 1977a, 1979; Bradley et al., 1993; Pavlis and Sisson, 1995) but are also evident in the thermochronometric record of the regions that are located farther inboard. The collision of the Yakutat microplate is inherently different from the Wrangellia composite terrane accretion due to the smaller size and its nature as an oceanic plateau with increasing crustal thickness (Christeson et al., 2010; Worthington et al., 2012). While accretion of the Wrangellia composite terrane was associated with arc magmatism, flat-slab subduction suppresses magmatism above the slab and confines it along the slab edges. However, the stronger coupling between downgoing and upper plate results in surface uplift above the slab (e.g., Chugach and Talkeetna Mountains), focused deformation in weak crustal sections and reactivation of preexisting structures (Alaska Range and Denali fault; e.g., Fitzgerald et al., 2014; Lease et al., 2016). The lack of magmatism and the inboard transfer of strain results in rock cooling due to exhumation that is recorded in the thermochronology data in the areas above the flat slab. Similar to southern Alaska, flat-slab subduction of the Juan Fernandez Ridge of the Nazca plate transfers strain much farther inboard and east of the Andes than in the regions north and south of the subducted ridge in central Argentina (e.g., Alvarado et al., 2009; Richardson et al., 2013). This process reactivated old basement structures and formed the Sierras Pampeanas. However, attempts to quantify the associated rock exhumation in the Sierras Pampeanas by means of thermochronology failed due to limited erosion in the semi-arid climate that results in rock uplift but limited exhumation (e.g., Jordan et al., 1989; Richardson et al., 2013; Enkelmann et al., 2014). Thus, climate conditions that favor efficient erosion are a precondition for rock exhumation in convergent settings and will result in significant cooling that is recorded by thermochronology data. In south-central Alaska, the exhumation associated with flat-slab subduction reaches several hundreds of kilometers inboard from the plate margin.

In this study, we investigated the thermochronologic record in the area located northeast of the indenting Yakutat plate corner and thus provide insight on how the upper plate is deforming as a response to the oblique convergence. We found that upper-plate deformation clearly occurs northeast of the plate boundary but reaches inboard only several tens of kilometers. In this region north of the syntaxis, significant exhumation might be accommodated along south-southwest–dipping faults such as the southern Connector fault, i.e., the Art Lewis Glacier fault, but also along a previously unrecognized structure farther inboard. We clearly show that the stresses from the Yakutat corner do not only result in dextral strike-slip motion to connect with inboard structures such as the Denali fault and Totschunda fault, but result also in shortening perpendicular to the plate motion. Our data suggest a southward migration of the focus of exhumation at the syntaxis (Fig. 8). We show that the previously recognized area of rapidly exhuming rocks extended farther inboard in the Late Miocene (ca. 10–5 Ma) with deformation and exhumation probably accommodated along the Fairweather fault and southwest-dipping faults north of it. The driver or drivers for the focusing of exhumation in the syntaxial region are uncertain. However, a 20° clockwise rotation of Pacific–North American relative plate motion has been attributed to increased transpression and changes in the structural setting facilitating vertical motion and impeding lateral slip. Even though 5 Ma (Engebretson et al., 1985) is often cited as timing of plate motion change (e.g., Fitzgerald et al., 1995; O’Sullivan and Currie, 1996; Bruhn et al., 2012; Falkowski et al., 2016), other estimates that do not rely on fixed hot spots might be more appropriate at ca. 8 Ma (Atwater and Stock, 1998), which is more consistent with a ca. 10–5 Ma onset of exhumation processes in the syntaxis. With glaciation of the area, and increasing thickness of the subducting slab, the focus of deformation shifted southward ca. 5 Ma and caused rapid and deep-seated exhumation in the center of the structure between Connector fault and thrust faults in the Yakutat foothills. The focusing of deformation in the North American plate might have been due to its weakness relative to the indenting thick oceanic basement rocks of the Yakutat microplate. The southward shift of deformation continued and since <2 Ma deformation and the most rapid exhumation is concentrated along northeast-dipping faults that parallel the northern Fairweather fault and imbricate the metasedimentary rocks of the Yakutat Group, and along the northwest-dipping, shallow thrusts of the fold-and-thrust belt comprising thick Cenozoic cover strata (Enkelmann et al., 2015a). This migration of deformation and exhumation over time presents an important observational data set for future geodynamic models that aim to understand mechanisms at plate corners, which must be treated as four-dimensional systems.

This study was funded by the Deutsche Forschungsgemeinschaft (DFG grants EN-941/1 and EN-941/1-2). We are grateful to our very skilled helicopter pilot, Doug Makkonen, for his efforts during our sampling campaign and Philipp Widmann for his assistance in the field. We thank editor Kurt Stüwe, Sarah M. Roeske, and an anonymous reviewer, whose comments and suggestions helped us to improve this manuscript.

1GSA Data Repository Item 2016150, Table S1: Comparison of previous and new detrital ZFT data from catchments KLD40, KLD65, and KLD110; Text S1: Analytical procedures; Figure S1: Geologic map of the St. Elias Mountains with higher-temperature chronometer data from the sampled catchments and their immediate surroundings; Datasets S1 and S2: ZFT and AFT single-grain ages, respectively, is available at, or on request from