The provenance signal of climate–tectonic interactions in the evolving St. Elias orogen: framework component analysis and pyroxene and epidote single grain geochemistry of sediments from IODP 341 sites U1417 and U1418

The St. Elias orogen and the Surveyor Fan in the adjacent Gulf of Alaska are a coupled source to sink system recording the interplay of tectonics and variable degrees of glaciation during the collision of the Yakutat terrane with the southern Alaska margin since the Miocene. The Miocene to Holocene sediments of the Surveyor Fan were drilled during IODP expedition 341. The recovered material is used to constrain information on changes in erosion centers during the last 10 Ma to study the impact of climatic and tectonic processes on orogen evolution. Point counting of sand- and silt-sized light framework components and geochemical single grain analysis of heavy mineral groups epidote and pyroxene is applied to analyze patterns of sedimentary provenance of two sites on the distal and proximal Surveyor Fan (Site U1417 and U1418, respectively). The studied sands and silts of Miocene to Pleistocene age are slightly enriched in feldspar (plag >> kf) at the proximal site, compositions at both sites do not show systematical changes with time of deposition. Framework component spectra uniformly reflect the expected active margin provenance. Epidote and pyroxene compositions are very consistent and show no change with time of deposition. Associations of epidote and pyroxene with albite, titanite and pumpellyite are in line with near-shore sources in the Chugach Metamorphic Complex and the metabasite belt at its southern border, and in units of recycled detritus exposed in the fold and thrust belt on the western Yakutat Terrane, respectively. Rock fragments indicate input from mainly metamorphic sources during the Miocene and Pliocene and an increase of input from low-grade metamorphic and sedimentary rocks in the Pleistocene, a finding also indicated by the abundance of epidote and pyroxene. This implies increasing erosion of the near-shore areas of the fold and thrust belt with advance of glaciers to the shore since the Miocene, being enhanced by the onset of the Northern Hemisphere glaciation at the beginning of the Pleistocene. Climate changes connected to the mid-Pleistocene transition did not result in appreciable changes in the petrographic compositions. Glaciers seem to have remained nested in their topographically predefined positions, continuously feeding material with uniform characteristics into the fan.


Introduction
The detritus constituting clastic sedimentary rocks is commonly derived from a multitude of source rocks and from several source regions (e.g. Garzanti and Andò 2007). It is therefore often difficult to determine specific provenance from such mixtures, in particular from old mountain belts of fragmentary preservation. Mountain belts may develop pronounced climatic asymmetries if they are oriented perpendicular to prevailing wind systems (Willett et al. 1993;Hay 1996). This results in significant material export being limited to the windward side of the evolving orogen. Sediment transfer paths are relatively short, if the windward side of accretionary, or exterior, coastal mountain belts (Cawood and Buchan 2007;Collins et al. 2011) is facing an ocean basin (Thornburg and Kulm 1987;Potter 1994;Koons et al. 2003;Roe et al. 2006). The stratigraphy and 1 3 provenance of the detritus transported out of coastal accretionary mountain belts and into an adjacent ocean basin is thus much more likely to reflect the climatically and tectonically controlled exhumation history of the orogen without interference from external sources. The long-term sink for this sediment is often found in large submarine fans or trench systems where sediments accumulate at high rates of 35-120 cm/kyr relative to deep-sea pelagic sedimentation (Thornburg et al. 1990;Clift and Vannucchi 2004;Gulick et al. 2015;Bergmann et al. 2020;Garzanti et al. 2020;Pickering et al. 2020a;Reilly et al. 2020).
Mass export from a coastal mountain belt is enhanced if the orogen is affected by varying degrees of glaciation (Hallet et al. 1996;Berger et al. 2008;Koppes and Montgomery 2009;Gulick et al. 2015;Carter et al. 2017). The erosion of mountain ranges by glaciers varies with changes in average temperature resulting in potentially lower rates effected by cold-based glaciers during glacial intervals, and increased rates connected to Alpine glaciers during interglacial periods. These variations also depend on elevation and latitudinal position (Yanites & Ehlers 2012). Examples include the southern Andes and the adjacent overfilled trench (Thornburg et al. 1990;Contreras-Reyes et al. 2010), the Southern Alps and adjacent basins of New Zealand (Griffiths 1979;Jiao et al. 2018,), the Himalaya-Bengal Fan system Pickering et al. 2020b;Reilly et al. 2020), and the St. Elias orogen in southern Alaska and the adjacent Surveyor Fan (Plafker et al. 1994;Reece et al. 2011;Gulick et al. 2015;Huber et al. 2018a,b;Bootes et al. 2019).
The present provenance study enlarges upon the data presented by Huber et al. (2018a,b) which considered U-Pb detrital zircon age and Hf isotope data together with geochemical heavy mineral data on garnet and amphibole recovered by IODP Expedition 341 at sites U1417 and U1418 on the distal and proximal Surveyor Fan, respectively (Fig. 1a). Here, we augment this information by examining the framework component distribution of selected sand and silt layers together with the single grain geochemistry of detrital pyroxene and epidote.
We will use our new data to assess changes in the detrital component compositions of the Gulf of Alaska sediment since the middle Miocene. This will be conducive to further constraining consistencies or variabilities in the source to sink relationship between the main erosion centers in the evolving St. Elias orogen and the Surveyor Fan transport and depositional system. Linking the stratigraphic age and composition of the Gulf of Alaska sediments to the mainland glacial and tectonic history will help to understand the selective impact of glacial and tectonic processes on the rates and foci of exhumation and erosion.

The example of the St. Elias orogen-Surveyor Fan source to sink system
The southern Alaska continental margin hosts the highest coastal mountain range on Earth, the St. Elias Mountains. It formed in the last ca. 8 Ma through accretion and subduction of the Yakutat terrane, and at a time of glaciation of variable intensity (Plafker et al. 1994;Bruhn et al. 2004;Enkelmann et al. 2008Enkelmann et al. , 2015Finzel et al. 2011Finzel et al. , 2015Pavlis et al. 2012;Falkowski et al. 2014). The windward side of the orogen faces the Gulf of Alaska. The interplay of climate and tectonics during orogeny since the Miocene shaped its exhumation and erosion history the information on which is stored in the Gulf of Alaska sediment, and mainly in the Surveyor Fan (Plafker et al. 1994;Gulick et al. 2015). This was originally shown by DSDP Leg 18 (Kulm et al. 1973) and was enlarged upon by IODP Expedition 341 (Expedition 341 Scientists 2014; Jaeger et al. 2014). The St. Elias orogen-Surveyor Fan source to sink system is one of the key natural laboratories to study climate-tectonic interactions during mountain building .

Provenance framework of the St. Elias orogen
A variety of provenance and thermochronological data derived from the rocks constituting the St. Elias orogen and the sediments of the Gulf of Alaska allowed for tracing the spatial and temporal development of erosion of and mass transfer from the orogen. These data constrain the main source areas to the Chugach Terrane (Huber et al. 2018a,b), a mainly metasedimentary accretionary complex of upper Cretaceous age ( Fig. 1; Pavlis and Sisson 1995;Barefoot et al. 2020). The St. Elias syntax at the transition from the strike-slip Fairweather Fault to the Chugach-St. Elias thrust fault registers the highest rates of erosion, exhumation and uplift (Fig. 1;Falkowski et al. 2014;Dunn et al. 2017;Enkelmann et al. 2017;Huber et al. 2018a,b;Bootes et al. 2019). The Fairweather Fault, in turn, is the fastest-slipping intracontinental strike-slip fault on decadal to millennial time scales (Molnar and Dayem 2010;Lease et al. 2021).

Tectonic setting
The St. Elias orogen at the southern Alaska continental margin formed through the successive accretion of several allochthonous terranes against the North American continent (Coney et al. 1980;Plafker et al. 1994;Colpron et al. 2007). South of the Border Ranges Fault (Fig. 1a, b), accreted terranes, namely the Yakutat, Prince William and Chugach terranes form a broad accretionary complex accreted against the Wrangellia composite terrane (Fig. 1b). Wrangellia had been accreted already in Middle Jurassic to Late Cretaceous time (Trop et al. 2002). The St. Elias orogen developed due to ongoing NW directed subduction of the Yakutat terrane presently taking place at a rate of 50 mm year −1 relative to the North American plate ( Fig. 1a; Elliott et al. 2010;Gulick et al. 2015). In the course of these developments, western Alaska also experienced counterclockwise rotation (Plafker 1987).
The Chugach and Prince William terranes consist mainly of coherent or disrupted turbidite successions and associated melange zones included in the Valdez and Orca groups of Upper Cretaceous-Eocene age (Fig. 1). The deposits originally represent forearc and trench deposits linked to a submarine fan system accumulated by sedimentation rates higher than those of the Bengal-Nicobar fan system (Nilsen and Zuffa 1982;Sample and Reid 2003). The rocks are preserved in greenshist to sub-greenshist metamorphic grade.  Post (1972). The glacier extent during the Last Glacial Maximum after Manley and Kaufman (2002) is marked with a white dashed line. The extent of the Chugach metamorphic complex (CMC) is marked by the gray shaded area. Plate velocity vectors are from Pavlis et al. (2012) Along the present suture to the Yakutat terrane, this complex is more strongly deformed and the metamorphism reaches upper amphibolite metamorphic grade (Pavlis and Sisson 1995). Metamorphism is connected to Eocene ridge subduction and the intrusion of the approximately 2100 km long forearc intrusive zone of the Sanak-Baranof Plutonic Belt (SBPB) between ca. 60 and 50 Ma (Sisson et al. 1989(Sisson et al. , 2003Bradley et al. 1993Bradley et al. , 2003Plafker et al. 1994;Pavlis and Sisson 1995;Haeussler et al. 2003;Farris and Paterson 2009). This higher-grade equivalent of the Valdez and Orca groups is ca. 350 km long and is named the Chugach Metamorphic Complex (Hudson and Plafker 1982;Bruand et al. 2011;Gasser et al. 2011). A metabasite belt of amphibolite facies is juxtaposed against the Chugach Metamorphic Complex along the Contact and Fairweather faults, facing lower-grade metamorphic rocks to the south ( Fig. 1; Bruand et al. 2011;Gasser et al. 2011).
The most outboard terrane, the Yakutat terrane, is tied to the movement of the Pacific plate and is the latest terrane accreting to the southern Alaska margin (Plafker et al. 1994). Northwestward transport of the Yakutat terrane along the dextral Fairweather transform fault system since ca. 30 Ma led to subduction of the Yakutat slab and accretion of the Yakutat terrane along the Chugach-St. Elias and Bering Glacier faults (CSE and BGf, respectively, in Fig. 1b). Since the Miocene, this drove the formation of the St. Elias Mountains (Plafker 1987;Plafker et al. 1994;Bruhn et al. 2004;Enkelmann et al. 2008Enkelmann et al. , 2015Finzel et al. 2011;Arkle et al. 2013;Falkowski et al. 2014). Strong uplift and erosion characterize the orogeny during the last ca. 10 Ma, especially at the Yakutat plate corner, also named the St. Elias Syntaxis, where the transform motion along the Fairweather Fault changes to flat-slab subduction along the Chugach-St. Elias Fault ( Fig. 1; Pavlis et al. 2012;Enkelmann et al. 2009Enkelmann et al. , 2010Enkelmann et al. , 2015Enkelmann et al. , 2017Falkowski et al. 2014;Dunn et al. 2017).
The accreting Yakutat terrane is characterized by a fold and thrust belt located seaward of the Chugach-St. Elias Fault and particularly well developed in its western part (Fig. 1b;Plafker et al. 1994;Enkelmann et al., 2010). It consists of imbricated sediments scraped off the basement and organized into the pre-orogenic Kulthieth (Eocene-Oligocene), Poul Creek (Lower Oligocene to Lower Miocene) and the syn-orogenic Yakataga (Miocene through Holocene) formations (Plafker 1987;Plafker et al. 1994;Worthington et al. 2010;Enkelmann et al. 2015). The eastern part of the terrane is covered by the Upper Cretaceous Yakutat Group comprising fluvial and glaciomarine syn-orogenic strata. It is part of the Chugach flysch and mélange predating the fold and thrust belt development (Plafker 1987;Plafker et al. 1994;Pavlis et al. 2012).
The origin and transport history of the Yakutat Terrane are under debate. The far-travelled terrane hypothesis proposes an origin 1500-2000 km to the south at ca. 44° paleonorth or close to northern California and southern Oregon in the Eocene (Bruns 1983). The short transport hypothesis (ca. 600 km) appears to be more in line with available data and places the origin of the terrane near Prince Rupert, British Columbia (Plafker et al. 1994;Perry et al. 2009).

Changing degrees of glaciation
Alpine glaciation at the Southern Alaskan continental margin started in the late Miocene (ca. 7 Ma) shortly after the onset of the formation of the St. Elias orogen. It expanded to tidewater glaciation between 6.7 and 5.0 Ma (Lagoe et al. 1993;Fig. 2). After a warm mid-Pliocene period from 4.2 to ca. 3 Ma, the Northern Hemisphere glaciation intensified at ca. 2.7 Ma when global glacial-interglacial climate cycles changed from 40 to 100 kyr at 0.7 to 1 Ma during the mid-Pleistocene transition (Lagoe et al. 1993;Lagoe and Zellers 1996). Beginning at this time, glaciers extended onto the shelf. This coincided with inception of the Surveyor Channel feeding the Surveyor fan (Reece et al. 2011;Gulick et al. 2015). Dropstones recovered at Site U1417 record drifting icebergs by ca. 2.6 Ma (Jaeger et al. 2014;Gulick et al. 2015).
The depositional precursor systems of today's Surveyor fan have been active since ca. 9.7 Ma, moving to the northwest with the Pacific Plate since onset of deposition (Plafker 1987;Stevenson and Embley 1987;Rea and Snoeckx 1995;Reece et al. 2011). How sediment was transported into the precursor of today's Surveyor fan during the early stages of deposition is not resolved so far. The absence of a glacial transport system as well as implications from heavy mineral, zircon U-Pb age and seismic reflection data of the early fan sediments implies the presence of a fluvial tributary system before the onset of glaciation in the late Miocene, funneling material from sources in the North (Stevenson and Embley 1987;Huber et al. 2018a, b).
Today, the Bering, Malaspina and Hubbard glaciers are the largest glaciers of southern Alaska (Pewé 1975;BG, MG and HG in Fig. 1b). The glacial systems directly feed orogenic detritus into today's Surveyor Fan through the Surveyor Channel system which discharges into the Aleutian Trench at its distal end (Fig. 1). The outlets of the glacier systems towards the Gulf of Alaska appear to be tectonically and morphologically controlled as they coincide with major thrust systems. The Bering glacier follows the eastward extension of the Kayak Island Zone, the Malaspina Glacier connects to the continuation on land of the Pamplona Zone (Fig. 1b).

Samples
Samples were taken from fine to medium sands and silts of IODP Expedition 341 sites U1417 and U1418. Site U1417 was drilled ca. 60 km west of today's Surveyor Channel in the distal Surveyor Fan, today being fed by the Alsek, Yakutat and Icy legs through hemipelagic/pelagic rainout and overbank levee deposition from turbidity currents (Expedition 341 Scientists 2014; Dunn et al. 2017;Fig. 1a). Sediments from a maximum depth of ca. 700 m below sea floor of Miocene (ca. 10 Ma) to Holocene age were recovered (Fig. 2). Site U1418 recovered ca. 950 m  Lagoe et al. (1993) and Expedition 341 Scientists (2014) of Early Pleistocene to Holocene sediments from the proximal Surveyor Fan (Expedition 341 Scientists 2014) (Fig. 2). The lowermost lithostratigraphic unit IV formed through an Early Pleistocene massive slope failure (mass transport deposit) (Reece et al. 2011;Expedition 341 Scientists 2014). All younger units were deposited through hemipelagic settling and by overbank turbidity currents from nearby channels (Reece et al. 2011;Expedition 341 Scientists 2014).
Sample names start with the abbreviated name of the site (17 for U1417 or 18 for U1418) followed by the abbreviated name of the stratigraphic interval (Mi = Miocene, Pi = Pliocene, Pe = Pleistocene, EPe = Early Pleistocene, MPe = Middle Pleistocene) and the sampling depth in m core composite depth below seafloor (m CCSF-B) (Fig. 2). The IODP code of each sample can be found in Appendix A. Deposition ages were deduced from bio-and magnetostratigraphic reconstructions and the resulting age model (Expedition 341 Scientists 2014; Jaeger et al. 2014;Gulick et al. 2015;Fig. 2). The sample volume is limited by the amount of material recovered in the sediment core half provided for sampling.
Grains < 0.063 mm provide the largest fraction in all samples, making up over 90% in some samples. Grains > 0.25 mm make up only 0-2% in most samples. The heavy mineral content in all samples is moderately rich (< 5%). The absolute heavy mineral volume for analysis is small. Sediment volumes are limited because of the diameter of the drill core (6.2 cm; sample weights of ca. 10 g to 70 g) and grain size, with many samples providing < 1.5 g of material of 0.063-0.25 mm. Still, this size fraction was the only one providing valuable amounts of heavy minerals for analysis. The small sample volume also made the calculation of sample indices like the Heavy Mineral Concentration index  unfeasible. Restriction to this narrow grain size fraction might lead to some information being lost because of grain size dependence of mineral compositions (e.g., Garzanti et al. 2008;Krippner et al. 2014). Still, intersample comparison is not invalidated (Huber et al. 2018b).

The provenance tool box
Provenance research is most effective when several provenance tools are combined (von Eynatten et al. 2003;Weltje and von Eynatten 2004;von Eynatten and Dunkl 2012;Reimann et al. 2015;Garzanti 2016). This allows the derivation of information on a wide range of source lithologies and their specific origins to approach a comprehensive reconstruction of provenance.
Framework component sandstone petrography is a traditional and well-established method (Krynine 1937;Dickinson 1970;Potter 1994;Weltje and von Eynatten 2004). Evaluation of light minerals and rock fragments in a sediment permits an initial discrimination of provenance (e.g., Dickinson and Suczek 1979;von Eynatten et al. 2003;) and allows for a general insight into the nature of the sediment (Pettijohn et al. 1987). Supported by a statistical assessment, this presents information on source rock lithology and mixing of material supplied from different endmember sources.

Sample preparation and analysis
Samples were wet-sieved to remove grains < 63 µm. Bulk samples > 63 µm were investigated under a binocular microscope. Light and heavy minerals were separated using sodium polytungstate at 2.95 g/cm 3 by density separation. Both light and heavy mineral fractions 250-63 µm were mounted on microscopy slides in epoxy and grinded to ca. 35 µm. On each slide, 300 grains were point counted after the Gazzi-Dickinson point counting method (Ingersoll et al. 1984;Dickinson 1985) using a polarization microscope equipped with a Petrog SteppingStage. Grains > 250 µm were rare and examined under a binocular microscope and exemplary grains of the most common rock fragments were mounted in epoxy and prepared to be examined with a microscope. For pyroxene and epidote analysis, grains were handpicked under a binocular microscope from fractions of heavy mineral separates used in Huber et al. (2018a, b). Pyroxene and epidote major element geochemistry was measured using JEOL JXA-8530F Field Emission Electron Probe Microanalyzer at the Institut für Mineralogie at University of Münster. Minerals intergrown with the pyroxenes and epidotes were measured alongside. Acceleration voltage was at 15 kV, beam current at 10µA and spot size at 5 µm. Pyroxene end-members were calculated using the WinPyrox software (Yavuz 2013).

Framework components
The non-clayey deposits encountered at Sites U1417 and U1418 are fine to medium sands and silts formed as turbidite overbank sediments (Expedition 341 Scientists 2014). Grains larger than medium sand are rare and represent less than 2% in most samples. Most samples consist of subrounded grains, subangular grains occur occasionally. The framework component composition of all samples mainly consists of quartz, feldspar, rock fragments, and accessory heavy minerals (electronic supplement Table 1). The distribution of framework components (QFL) classes the samples of both sites U1417 and U1418 as lithic sands and silts (Dott 1964), or as fLQ or fQL sands (Garzanti 2016). The sand-size fraction is dominated by quartz in all units (Fig. 3). The quartz content consists almost exclusively of monocrystalline quartz. Most of the feldspar is plagioclase, K-feldspar is scarce. Feldspars are relatively fresh with only minor sericitization, plagioclase shows polysynthetic twinning under the microscope (Fig. 4). The samples from the proximal Site U1418 contain more feldspar than those of the distal Site U1417 (Figs. 3, 5), the respective average framework modal compositions are Q 48 F 17 L 35 (U1418) and Q 47 F 12 L 41 (U1417). The main detrital phyllosilicate is biotite, followed by chlorite. The abundance of biotite is highest in the Miocene unit 6 at site U1417 ( Fig. 2) and decreases up section at this site.
The sediments of the oldest Miocene samples recovered from unit VI (Fig. 2) have a very different composition from the younger strata. They contain a high proportion of coalified organic material, biotite and opaque, angular grains (Figs. 6f). Rock fragments, quartz and feldspar make up less than 5% of these sediments. Also, very different are the samples from the mass transport deposit at the base of the core retrieved from unit IV at Site U1418 (Jaeger et al. 2014). They are rich in black angular detrital grains of iron oxide > 0.25 mm.

Principle component analysis
The representation of the samples' framework composition in the QFL diagram (Fig. 5) illustrate that feldspar is slightly more abundant in the samples from Site U1418. This happens mostly at the expense of the lithic grains. Principle component analysis as applied to siliciclastic sedimentary rocks is a means to statistically underpin the assignation of compositional differences obtained by point counting (Vermeesh 2016; Fig. 8). Here, the vector loading of the principle input variables shows that the samples from Site U1418 are in fact characterised by a larger feldspar component whereas the samples from Site U1417 have higher amounts of rock fragments, and of fragments from sedimentary and metamorphic rocks, in particular (Fig. 8).

Implications of the framework petrography data
Petrographic framework component compositions are likely but not unequivocal guides to the tectonic setting of source areas. Original detrital signals may be obscured by mixing from different sources (Dickinson 1985;Garzanti 2016). Therefore, the QFL diagram does not unambiguously constrain the tectonic provenance of a sedimentary deposit but may serve as an initial graphical classification tool guiding Fig. 3 Light mineral spectra of sites U1417 and U1418. Sedimentary, volcanic and metamorphic rock fragments correspond to the lithics categories Ls, Lv and Lm, respectively, of the Gazzi-Dickinson point counting method (Ingersoll et al. 1984) further study which may combine the framework data with the more conducive single grain geochemical analysis of heavy minerals (von Eynatten et al. 2003;Garzanti 2016). Application of the classical QFL Dickinson diagram is most unerring when applied to detritus derived from active margins as it does not really allow for the consideration of anorogenic provenances including continental rifts (Garzanti 2016).
The classical Dickinson diagram does not include a specific active margin provenance. This provenance encompasses the magmatic arc and the recycled orogenic provenance types. As evidenced by the Andean cordillera orogen, climatic asymmetry leads to a magmatic arc provenance in sands derived from the western rain shadow pro-side of the mountain belt (Potter 1994;). Contrastingly, erosion on the windward eastern and retro-side of the orogen is focused in fold and thrust belts exposing mainly Paleozoic siliciclastic sedimentary rocks results in sands with a recycled orogen provenance (Fig. 5;Potter 1994;). In southern Alaska, the pro-wedge of the St. Elias orogen is located on the windward side, which also includes the fold and thrust belt on the Yakutat Terrane ( Fig. 1; Trop and Ridgway 2007), thus combining both provenance types of  on the southern, windward pro-side of the orogen.
We also note that sands derived from subduction complexes representing another leading edge provenance indicate a recycled orogen provenance ( Fig. 1; Dickinson & Suczek 1979). On the windward side of the St. Elias orogen, the Chugach and Prince William terranes represent Mesozoic and Cenozoic subduction complexes, respectively ( Fig. 1; Trop and Ridgway 2007). The provenance of the Surveyor Fan sediments can therefore be expected to potentially encompass both the classical magmatic arc and recycled orogen provenance, including subduction complexes. Concluding from this discussion, we will henceforth in this contribution consider the combined fields of magmatic arc and recycled orogen provenance as the realm of active margin provenance (Fig. 5).
The considered modification gives us confidence to use the QFL diagram in the analysis of the detrital compositions presented here as the active margin tectonic setting of sources in the St. Elias Orogen and the sink in the Surveyor Fan are well established since the seminal work by Plafker (1987). Compositional variations will rather reflect input from different sources in the exposed subduction complexes within the active margin St. Elias strike-slip orogen than changes of tectonic setting and overall provenance.
The studied samples from both sites have relatively similar average framework component compositions of Q 48 F 17 L 35 (U1418) and Q 47 F 12 L 41 (U1417) with the samples from site U1418 having slightly more feldspar (Figs. 1,2,3,5,8). In the QFL diagram of Dickinson et al. (1983;Fig. 5), the samples plot around the subduction complex provenance in the "recycled orogen" provenance field also including the clastic wedge provenance of , all representing detritus derived through recycling of older orogenic detritus encompassing foreland basin systems, remnant ocean basins and collisional orogens (Fig. 1).
The analyzed samples from Sites U1417 and U1418 are predominantly fine sands and silts. They are poor in polycrystalline quartz. Considering relatively short transport distances from source to sink on the one hand, and predominantly robust subglacial to glaciofluvial transport processes on the other, it is likely that mechanically less stable polycrystalline quartz grains disintegrated to finer grain sizes along crystallographic domain boundaries. These finer quartz grains may have been additionally concentrated by sorting during transport. Original framework compositions may thus have been richer in fragments of polycrystalline quartz and rock fragments like chert and volcanic rocks (Figs. 5,6), examples of which are contained in the Fig. 4 Representative microphotographs of detrital grains from site U1417 and U1418. a Plane-polarized light, b cross-polarized light rare coarser fractions (Fig. 6). This would result in shifting the framework compositions of Sites U1417 and U1418 samples towards the magmatic arc provenance fields of the QFL diagram (Fig. 5). Conversely, the decrease of feldspar and increase of quartz from the older Yakutat Terrane units incorporated and exposed in the fold and thrust belt to the younger Surveyor Fan sediments (Fig. 1) may reflect the recycling process and loss of more labile components from the thrust belt into the ocean. However, both scenarios would not influence the general interpretation of an active margin provenance of the detritus.
The St. Elias orogen exposing Mesozoic and Cenozoic subduction complexes is presently a strike-slip orogen accommodating the northwestward movement of the Yakutat Terrane ( Fig. 1; Elliott et al. 2010;Gulick et al. 2015). Available provenance data on strike-slip orogens were obtained from the deeper marine turbidite and contourite record (Hayes 1973;Valloni and Maynard, 1981;Bender-Whitaker et al. 2018) and plot in the dissected arc field of  Dickinson et al. (1983) with apices Q = quartz, F = feldspar and L = lithics showing the samples from site U1417 and U1418 plotting in the recycled orogenic provenance field, and averages of onshore reference samples (Fig. 1) from the literature. The sample 17Mi704 from unit VI (Figs. 2, 5f, 6) could not be plotted in this diagram because of the lack of grains of the respective groups. Black stars, average reference compositions; strike-slip regimes: Alaska Abyssal Plain sandstones, DSDP Leg 18, Site 178, middle Miocene (Hayes 1973); Leading edge strike-slip (Valloni and Maynard 1981); Canterbury Basin, and Canterbury rivers, South Island, New Zealand (Bender-Whitaker et al. 2018), the black arrow links the onshore river sand to the offshore turbidite sand compositions; subduction zones: Central America trench (Yerino and Maynard 1984); Leading edge subduction, Peru-Chile trench (Yerino and Maynard 1984;Thornburg and Kulm 1987); Remnant ocean basin, Bengal and Nicobar fans (Ingersoll and Suczek 1979); all others from Dickinson and Suczek (1979). Clastic wedge provenance and pro-side and retro-side indication according to  1 3 the Dickinson diagram (Fig. 5). River sands from the Canterbury province east of the Southern Alps strike-slip orogen of New Zealand (Molnar and Dayem 2010) indicate a recycled orogen − clastic wedge provenance like the studied samples form the Gulf of Alaska (Fig. 5;Bender-Whitaker et al. 2018). In the New Zealand case, the significantly increased immaturity of the deep-sea deposits relative to the river sands is considered to be due to a mixture of the Canterbury River detritus with other material transported along shore and offshore by marine currents (Fig. 5; Bender-Whitaker et al. 2018).
The deep-marine sands recovered from the Gulf of Alaska abyssal plain by IODP Expedition 341 have a framework component composition very similar to the Canterbury river sands. An admixture of more immature detritus like in the New Zealand case is not observed for our samples but may have played a role in the genesis of the deposits recovered at DSDP Site 178 ( Fig. 5; Hayes 1973). Glacial transport of detritus across the Yakutat shelf and onward transport to the Fig. 6 Microphotographs of the silicate framework components in the grain fractions > 0.25 mm from site U1417 and U1418. ag, argillite; gn, gneiss; int, intrusive rock fragment; kf, K-feldspar; mf, mafic volcanic rock fragment; pg, plagioclase; sd, sandstone; sh, shist; qtz, quartz; sl, slate; st, shist; F, coalified organic fragments. Red scale bar represents 0.5 mm deep sea via trough mouth fans allowed for a direct transfer of material without admixture from other sources (Laberg and Vorren 2000;Gulick et al. 2015) and without significant intermediate storage (Clift et al. 2014). This also permits the interpretation of the studied samples as being derived from a strike-slip source exposing older subduction complexes. As the sand petrography does not allow for distinguishing between a subduction complex and a potential strike-slip Onshore, the mature, quartzose part of the Kodiak Formation as part of the upper Cretacous Valdez Group turbidites has a QFL composition of Q 49 F 15 L 36 (Sample and Reid 2003;Fig. 5) similar to the studied samples. Rocks belonging to the younger part of the Kodiak Formation are richer in volcanic detritus with Q 10 F 22 L 68 and plot in the undissected arc field of the QFL diagram (Fig. 5) reflecting a change of their sources from the erosion of older, partly Proterozoic sources to increasing input of the contemporaneous Eocene arc (Sample and Reid 2003;Barefoot et al. 2020). Large parts of the Valdez Group are of a compositional maturity between the end-members of the Kodiak Formation, with an average framework composition of Q 20 F 46 L 34 ( Fig. 5; Dumoulin 1988). The average composition of the Valdez Group resembles that of Modern trench sediments (Valloni and Maynard 1981;Thornburg and Kulm 1987) and falls in the transitional arc provenance field of the QFL diagram (Fig. 5).
Coeval and younger lithostratigraphic units exposed onshore in the St. Elias orogen fold and thrust belt, from the Upper Cretaceous Yakutat Group to the Miocene − Present Yakataga Formation, are richer in feldspar than rock fragments, and comparatively slightly enriched in quartz (Figs. 1, 5). They uniformly reflect a dissected magmatic arc provenance in the QFL diagram (Fig. 5). Their composition is also similar to Modern deep-sea sands derived from leading edge strike-slip orogens (Valloni and Maynard 1981), a finding that correlates directly with the tectonic setting of the St Elias orogen (Plafker et al. 1994;Elliott et al. 2010;Pavlis et al. 2012). Sands of Miocene age obtained by DSDP Leg 18 at Site 178 on the Alaska Abyssal Plain ( Fig. 1; Hayes  1973) have a similar composition with Q 33 F 49 L 18 , and thus have more feldspar and less rock fragments than the coeval and younger, and more quartz-rich, sands at Site U1417 and U1418 (Fig. 5).
Evaluation of the main rock fragments shows a difference between the samples from sites U1417 and U1418. The Miocene to Pliocene samples from site U1417 give higher numbers of metamorphic grains compared to the Pleistocene samples from site U1417 and site U1418. This corroborates the shipboard results of IODP Expedition 341 where macroscopic inspection of clast compositions > 2 mm in diameter gave a similar result (Figs. 6, 9). Furthermore, this is confirmed by principal component analysis (Vermeesch et al. 2016) showing all Pleistocene samples besides sample 17Pe181 to cluster in the topmost right corner of the plot while the Miocene to Pliocene samples plot towards the bottom right corner (Fig. 8). The rock fragment composition obtained in this study is consistent with a subduction complex − clastic wedge, active margin provenance and corroborates the results of Zuffa et al. (1980).

Heavy minerals, pyroxene and epidote
The heavy mineral spectra at both Sites U1417 and U1418 are characterized mainly by amphibole, pyroxene, epidote and garnet, in decreasing order of abundance. Biotite and chlorite represent scarce components (Huber et al. 2018b; Fig. 10). There are minor variations between sites. Epidote and amphibole are most abundant at Site U1417, and amphibole and pyroxene dominate at Site U1418. Garnet represents a minor but appreciable component (Fig. 10). The details of the geochemistry and provenance of amphibole and garnet have been presented by Huber et al. (2018b).

Pyroxene
Pyroxenes reflect the composition of the source rock and they form predominantly in mafic to intermediate magmatic lithologies and high-grade metamorphic rocks (Rösler 1984). Clinopyroxene is a common component in volcaniclastic sandstones derived from magmatic arcs (Cawood 1983;Nechaev and Isphording 1993). Clinopyroxene is, however, one of the most unstable detrital heavy minerals in sandstones (Morton and Hallsworth 2007). The geochemical composition of pyroxenes from Sites U1417 and U1418 allows for the classification based on major elements Mg, Ca and Fe (Morimoto et al. 1988).
The strongly predominant variety of pyroxene encountered in the studied samples is brown clinopyroxene  (Fig. 11a). Most pyroxene occurs intergrown with feldspar, mostly albite, pumpellyite and titanite (Fig. 11a). The most abundant pyroxene endmember at both sites is the clinopyroxene augite, with minor occurrences of diopside and hedenbergite ( Fig. 12a; electronic supplement Table 2). All augite is relatively rich in aluminum (Appendix B). A few grains approaching the composition of the pyroxenoid wollastonite occur ( Fig. 12a; Morimoto et al. 1988). Wollastonite is mostly derived from metamorphic calc-silicate lithologies and rarely from nephelenitic volcanic rocks (Pichler and Schmitt-Riegraf 1987). Nisbet and Pearce (1977) established that the major element concentrations in pyroxenes reflect the plate tectonic setting of the mafic volcanic source rocks. In the triangular space of MnO-TiO 2 -Na 2 O, the pyroxenes of Sites U1417 and U1418 plot in a highly distributed pattern (Fig. 12b). There is no systematic variability in pyroxene composition between the two sites and between samples from different stratigraphic intervals (Fig. 12a, b). One can, however, state that the majority of sample compositions plot in the fields A, F and G, indicating a preferred provenance from volcanic arc and within plate alkalic basalts. The joint consideration of the predominance of green magnesiohornblende in the samples (Huber et al. 2018b) and the absence of olivine assigns the predominant brown clinopyroxenes to a convergent margin provenance according to Nechaev and Isphording (1993).

Epidote
Minerals of the epidote-zoisite group make up a large fraction of the heavy mineral spectra of site U1417 and U1418. All grains show a very similar geochemistry and consist predominantly of Fe bearing epidote (Table 1; electronic Supplement Table 3), zoisite is rare. Epidote is a heavy mineral which commonly forms during low-grade metamorphism of marls and other clay and carbonate bearing rocks, and of mafic rocks (Pichler and Schmitt-Riegraf 1987;Spiegel et al. 2002). Epidote also forms under medium pressure metamorphic conditions in mafic and intermediate magmatic rocks (Grapes and Hoskin 2004), and in high and ultrahigh pressure regimes in mafic magmatic and pelitic lithologies Fig. 9 Diagram showing the lithology of the rock fragments in samples from sites U1417 and U1418. The red and blue stars represent the average abundances of the main types of clasts larger than 2 mm in diameter at sites U1417 and U1418, respectively, as determined during on-board core description by IODP Expedition 341 Scientists (2014).  (Enami et al. 2004). It may also be the product of the metasomatic alteration of pyroxene, amphibole and anorthite in magmatic rocks (Pichler and Schmitt-Riegraf 1987). The geochemistry of epidote minerals is highly variable. It is controlled by temperature, pressure, oxygen fugacity (ƒO 2 ), and bulk rock composition. In metamafic rocks, epidote  shows a tendency to higher Fe (Enami et al. 2004). Epidote is unstable under burial conditions, but less so than clinopyroxene (Morton and Hallsworth 2007;Garzanti et al. 2020).
Most grains of the epidote group at Sites U1417 and U1418 are intergrown with albite or quartz (Fig. 12b). There is no systematic variability in epidote composition with time of deposition or between sites. The amount of epidote, however, varies between the two sites. At the distal site, the epidote content of the heavy mineral fraction rises from ca. 16% in the Miocene to 58% in the Pleistocene sediments (Huber et al. 2018b). Within the Pleistocene samples from the proximal site (U1418), epidote content is relatively constant at ca. 22% to ca. 18% (Huber et al. 2018b).
The X Fe value of epidote is a measure of metamorphic grade and increases towards lower grades (Grapes and Hoskin 2004;Enami et al. 2004). X Fe is calculated as [Fe 3+ / (Al + Cr + Fe 3+ + Mn 3+ )] (Enami et al. 2004). The X Fe values from both sites are rather similar and range between 0 and 0.9 (Fig. 13). The modes of the individual and combined distributions are at ca. 0.5 (Fig. 13a). The sample averages cluster between X Fe values of 0.4 and 0.5 (Fig. 13b). These are relatively high and indicative of metamorphic conditions of relatively low temperatures and low to intermediate pressures (Enami et al. 2004). The less abundant low X Fe values indicate that some epidote had been derived from highpressure/high-temperature metamorphic terrains (Fig. 13).

Fig. 12
Geochemical composition of pyroxene from site U1417 and U1418. a Major element composition of pyroxene and the pyroxenoid wollastonite according to Morimoto et al. (1988). b Composition of pyroxene according to Nisbet and Pearce (1977) showing no trend in correlation of sampling depth and stratigraphic position to tectonic setting. a volcanic arc basalts (VAB); b ocean floor basalts (OFB); c within plate alkalic basalts (WPA); d, all; e, volcanic arc basalts + within plate tholeiitic basalts + within plate alkalic basalts (VAB + WPT + WPA); f, volcanic arc basalts + within plate alkalic basalts (VAB + WPA); G, within plate alkalic basalts (WPA) Fig. 13 X Fe values of detrital epidote, combined from sites U1417 and U1418. X Fe calculated according to Enami et al. (2004). a Histogram of X Fe values from both sites; b X Fe vs SiO 2 diagram

Implications of the pyroxene and epidote data
Augite is the predominant pyroxene in the studied samples (Fig. 12a). It is the most abundant mafic mineral in calcalkaline and tholeiitic magmatic rocks (Pichler and Schmitt-Riegraf 1987). The geochemical characteristics of the analyzed mineral grains assign their original formation predominantly to a convergent (active) margin provenance (Fig. 12b). A potential source lithology may have been the metabasite belt at the southern margin of the Chugach Metamorphic Complex Gasser et al. 2011). The studied epidote minerals have geochemical characteristics indicating an origin in a high temperature-low-pressure metamorphic belt like the Chugach Metamorphic Complex. 40 Ar/ 39 Ar ages on hornblende obtained from several samples from Sites U1417 and U1418 group mainly at ca. 145-133 Ma (very early Cretaceous), and between 55 and 37 Ma (Eocene; Huber et al. 2018b). Detrital zircon U-Pb age distributions from Site U1417 and U1418 sediments have a pronounced maximum at 60 to 50 Ma (Paleocene-Eocene; Huber et al. 2018a). In the evolving St. Elias orogen, Paleocene-Eocene time saw the intrusion of the Sanak-Baranof Plutonic Belt and the metamorphic overprint of the Chugach Metamorphic Complex caused by ridge subduction (Fig. 1; Sisson et al. 2003;Farris and Paterson 2009). Both lithotectonic units are very likely sources of pyroxene and epidote even though no age information on both minerals is available. The Sanak-Baranof Plutonic Belt and the Chugach Metamorphic Complex with the metabasite belt at its southern margin probably sourced the bulk of the pyroxenes and epidotes. The Chugach Metamorphic Complex and the metabasite belt have already been determined as the major source of hornblende and garnet in the sediments of Sites U1417 and U1418 (Huber et al. 2018b). Rocks of the Early Cretaceous Chisana arc on the Wrangellia Terrane may have been additional sources (Trop and Ridgway 2007).
The combination of our new pyroxene and epidote data with the already available heavy mineral data of Huber et al. (2018b) allows for the consideration of overall heavy mineral compositions in terms of likely tectonic setting of the sources (Fig. 14; ). The abundances as displayed in A − &tHM − O + P + S (amphibole-all other heavy minerals − olivine + pyroxene + spinel) space emphasize the derivation of the minerals from metamorphic terrains of greenshist to amphibolite facies grade (Fig. 14) corresponding to the axial-belt orogenic provenance of .
As particularly pyroxene is a very unstable mineral during transport and burial significant recycling of mineral grains seems unlikely. The mentioned sources must have been contributing detritus more or less directly to the evolving Surveyor Fan.   showing the heavy mineral composition of site U1417 and U1418 sediments. A = amphibole, &tHM = all other transparent heavy minerals, O + P + S = olivine + pyroxene + spinel. Heavy mineral data from this study and Huber et al. (2018b)

Large-scale provenance
The sands and silts recovered at Sites U1417 and U1418 are relatively well-sorted quartz-intermediate deposits of turbidity currents building up the Surveyor deep-sea fan on the Gulf of Alaska sea floor (Plafker et al. 1994;Gulick et al. 2015). They range in age from the middle Miocene to the Pleistocene (Fig. 2; Jaeger et al. 2014). The sediments from both sites have similar overall compositions. In terms of light mineral framework components, they are rich in feldspar and rock fragments and have an active margin orogen provenance directly connected to the adjacent accretionary St Elias orogen (Figs. 2, 5).
Joint consideration of petrographic characteristics, U-Pb age data on detrital zircons (Huber et al. 2018a) and detailed geochemical data on the heavy minerals amphibole and garnet (Huber et al. 2018b), and of pyroxene and epidote (this contribution) allows to identify the ultimate sources of the detritus. These are the Sanak-Baranof Plutonic Belt and the Chugach Metamorphic Complex ( Fig. 2; Huber et al. 2018a,b) now exposed in subduction complexes in an active margin orogen, with significant contributions from the axial metamorphic belt (Figs. 5, 14).
Partial cycling of original detritus through intermediate storages represented by the Kulthieth, Poul Creek and Yakataga formations exposed in the St. Elias fold and thrust belt is likely given the syn-orogenic control on all Cenozoic deposits on the mainland and the Gulf of Alaska abyssal plain (Plafker et al. 1994;Enkelmann et al. 2008;Perry et al. 2009). The ubiquitous and abundant presence of unstable clinopyroxene indicates, however, that destruction of labile minerals during recycling and transport has been a minor factor.

Framework components
Considering the temporal evolution of provenance, a major change occurred in the late Miocene, from mainly coalified organic material and mica to abundant quartz, feldspar and rock fragments (Site U1417, Fig. 2). This implies a modification of sediment transport mechanism on-and/or offshore, and/or a change in erosion patterns in the potential source areas. As it predates the onset of glaciation (Reece et al. 2011), tectonic processes are most likely to have triggered the change in sediment composition by changes in the loci of erosion and avenues of sediment dispersal.
The abundance of coalified organic material in Unit VI (Fig. 2, 6f) implies coal bearing strata to have fed most of these deposits. Jaeger et al. (2014) suggest the coaliferous alluvial and shallow marine Kulthieth Formation (56-33.9 Ma) on the Yakutat terrane as main source. Today, this formation is exposed in the fold and thrust belt on the Yakutat Terrane which belongs to the Cenozoic subduction complex at this margin (Trop and Ridgway 2007). Due to the northwestward movement of the Yakutat terrane since 10 Ma, these rocks must have been located further southeast during deposition of the site U1417 sediments. Geochemical and biomarker analyses of the coals from the Kulthieth Formation and the site U1417 sediments confirm the origin of the material (Childress et al. 2013). The Kulthieth, Poul Creek and Yakataga formations have been found to differ in the biomarkers of their organic carbon . For now, the source region of these sediments cannot be further specified but a Yakutat terrane source is likely.
The Yakataga Formation, in turn, has been deposited coevally with the Surveyor Fan sediments. It has been suggested to be sourced by two or more sources, most likely the Chugach-Prince William terrane and the Poul Creek and Kulthieth formations on the Yakutat Terrane (Plafker et al. 1994;Enkelmann et al. 2008;Perry et al. 2009). This is in line with erosion and transport of material also from the Kulthieth Formation via the Yakataga Formation or directly into the precursor of today's Surveyor Fan at ca. 11 Ma.
Sediment accumulation rates increased in the Gulf of Alaska between 9.87 and 9.47 Ma with the initiation of Surveyor Fan turbidite deposition (Rea and Snoeckx 1995). At the same time, unit VG ( Fig. 2; sample 17Mi642, ca. 8.5-9.5 Ma) has been deposited at Site U1417. This unit has been found to be very likely sourced by the Chugach Metamorphic Complex (Huber et al. 2018a,b). The high amount of metamorphic rock fragments (Fig. 9, sample 17Mi642) supports this finding.
Because of the Miocene position of site U1417 further to the southeast, an efficient transport system is required to feed material into the Site U1417 region. Seismic data suggest transport of material from sources to the north into the precursor of the Surveyor Fan (Stevenson and Embley 1987). Exhumation rates of the northeastern side of the St. Elias Syntaxis were extremely high at ca. 10 Ma and a fluvial system is considered to have existed, effectively eroding and exporting the material out of the orogen . A link between these two inferred fluvial systems might explain the early input from the St. Elias orogen into the Surveyor Fan precursor. The change in sediment composition from mica and organic rich material to higher amounts of heavier minerals and rock fragments also implies a change to a higher energy transport regime.
While detailed analysis of heavy minerals from the Miocene to Pleistocene sediments from sites U1417 and U1418 shows changes with depositional age (Huber et al. 2018b), the light mineral composition is uniform from late Miocene onwards for Site U1417 and for the Pleistocene at the proximal Site U1418. This is not surprising as light mineral analysis has already been found to be less discriminative provenance indicator than heavy mineral analysis (von Eynatten et al. 2003). In fact, many lithologies at the southern Alaskan continental margin provide high amounts of quartz and feldspar to the erosional systems (Bruand et al. 2014;Hudson and Plafker 1982). Biotite, quartz and plagioclase have been found to make up 90% of the central part of the schist zone of the Chugach Metamorphic Complex (Hudson and Plafker 1982).
Considering framework component compositions together with transport and sorting effects, all samples from site U1417 and U1418 essentially imply an active margin orogen source. The slightly higher feldspar content at Site U1418 might result from the more proximal position of this site relative to the continental sediment routing systems to the Surveyor Fan. Plutons along the Southern Alaskan continental margin in the eastern and central Chugach Metamorphic Complex show granitoid to tonalitic compositions, while plutons further east tend to have less quartz and more plagioclase, showing a more gabbroic composition (Bradley et al. 1993).
Site U1418 sediments have been found to be sourced through the Bering-Bagley glacial system (Fig. 1) whereas site U1417 was sourced from a much wider part of the orogen with longer transport paths (Dunn et al. 2017;Huber et al. 2018a,b), being in line with the differences in feldspar abundance. K-feldspar is a minor component present of all samples (0-3%). In the potential source region, K-feldspar has only been found in appreciable amounts in the sedimentary units of the Cretaceous units in the Yakutat area, on the Chugach terrane (Zuffa et al. 1980). On average they show 5% K-feldspar with values ranging between 0.65 and 14.5% (Zuffa et al. 1980). K-feldspar is absent in the amphibolite facies schists of the Chugach Metamorphic Complex while it is present in the migmatites in the central parts of the complex (Hudson and Plafker 1982). The low abundance of K-feldspar matches a Chugach and Yakutat terrane provenance but cannot be used to further define the provenance of the sediments.
The rock fragment spectra of the samples show a slight decrease of metamorphic grains from the Miocene to the Pliocene and even lower numbers of metamorphic grains in all Pleistocene samples from both sites. This indicates an increased derivation during the Pliocene of material from the non-metamorphic and low-grade metamorphic units now exposed in the near-shore region on the Yakutat terrane. Metapelites, metagraywacke, shists and migmatites are the main lithologies of the inboard Chugach Metamorphic Complex whereas greenschist facies phyllites occur outside of it (Bruand et al. 2014). The phyllites have a uniform petrography of chlorite, epidote, muscovite, plagioclase, quartz and biotite (Bruand et al. 2014).
The trend of increasing input from the areas closer to the coast with younger depositional age and during increasing degrees of glaciation and advancing glaciers has also been indicated by zircon U/Pb age data (Huber et al. 2018a). The overall decrease of metamorphic rock fragments from site U1417 to site U1418 reflected in this study has also been observed regarding rock fragments > 2 mm by IODP 341 Scientists (2014; Fig. 9).

Heavy minerals
Pyroxene is the most abundant ferromagnesian heavy mineral with a relatively uniform composition at Sites U1417 and U1418. Augite appears together with pumpellyite, titanite, albite and quartz which is typical for the pumpellyite facies (Coombs 1989) supporting direct input to all samples from the low-grade metamorphic rocks of the Chugach Metamorphic Complex. Some of this detritus may also have been supplied indirectly by recycling through the Miocene-Present Yakataga Formation (Plafker et al. 1994;Enkelmann et al. 2008) now exposed in the fold and thrust belt on the western Yakutat Terrane (Fig. 1b). The higher amount of pyroxene present in the sediments at the proximal site is again in line with a derivation mainly from the Chugach Metamorphic Complex via the Bering-Bagley and Malaspina glaciers and the Icy Leg channel (Fig. 1).
Epidote together with albite and quartz are important components of the metabasalt belt cropping out along the southern border of the Chugach Metamorphic Complex (Lull and Plafker 1985;Bruand et al. 2011). The higher abundance of pyroxene at the distal site from the Miocene to the Pliocene and especially in the Pleistocene is in line with an increasing input from this part of the orogen. In the Pleistocene, the mineral associations containing epidote and pyroxene match input from the lower-grade metamorphic rocks on the Yakutat terrane to the proximal Site U1418, and from the higher-grade Chugach Metamorphic complex to the distal Site U1418. Increased erosion rates and changes and differences in the petrographical composition of the deposits are in line with the contemporaneous advance and expansion of the glaciers covering the St. Elias orogen.
The suites of transparent heavy minerals of both sites show marked differences (Fig. 14). The site U1417 sediments are rich in amphibole, plotting in the typical area for a clastic wedge or greenschist facies to amphibolite facies axial-belt provenance according to . The site U1418 sediments show higher amounts of epidote, placing them in the middle of the unroofing trend typical for undissected arc settings . This underlines the difference in provenance of the two sites being fed by different glacial systems and different parts of the orogen.

Considerations of climate and tectonics
The data obtained from sands and silts recovered by IODP Expedition 341 by various studies yield a consistent scenario for the development of tectonic-climate interactions at the southern Alaska margin. Thermochronological data obtained from the distal Site U1417 (Fig. 1) indicate that rapid, strong, and deep-seated exhumation happened between ca. 11 and 8 Ma (Dunn et al. 2017). Site U1418 and more proximal sites on the southern Alaska shelf record detritus input from the general St. Elias orogen and minor mixing on the glaciated margin since 1.2 Ma (Dunn et al. 2017;Huber et al. 2018a). Depending on model parameters and the applied age model, exhumation rates could have been as high as between 2.6 and 5 km/My (Dunn et al. 2017;Huber et al. 2018a). Considering the area underneath the Bagley Ice Field, through which most of the detritus at Site U1418 has been delivered, ca. 3-4 km of rock exhumation has taken place since 4-3 Ma (Dunn et al. 2017).
A general and potentially abrupt change in provenance in response to the intensification of glaciation at the mid-Pleistocene transition is not reflected in our data. Heavy mineral spectra, the compositions of amphibole, garnet, pyroxene and epidote as well as the zircon age data and the framework petrography are uniform throughout the Pleistocene stratigraphy (Huber et al. 2018a, b). This implies that climatic changes at the mid-Pleistocene transition did not cause significant changes in the principal areas under erosion, with glaciers eroding most of the material in their topographically predefined positions . Similar uniform heavy mineral spectra and amphibole and garnet geochemical compositions have been observed also in the middle Pleistocene to Holocene sediments from sites U1419 and U1420 (Drewer 2016).
Examination of the silt-sized fraction (15-63 μm) of surficial sediments collected throughout the Gulf of Alaska and of the mid-Pleistocene sediments recovered by IODP Expedition 341 confirms the establishment of persistent erosion centers controlled by the pronounced interaction of tectonics and the glacial cover at least since the mid-Pleistocene transition Penkrot et al. 2017Penkrot et al. 2018. Since the mid-Pleistocene transition at ca. 0.7 Ma, sediment sources are inferred to have been situated within a narrow band of ~ 200-1500 m elevation that coincides with the last glacial maximum equilibrium-line altitude (ELA) range during the mid-late Pleistocene, feeding sediments of relatively homogenous composition into the Gulf of Alaska through the Bering-Bagley and Malaspina glacial systems Penkrot et al. 2017). The age information obtained from detrital minerals (Dunn et al. 2017;Huber et al. 2018a,b) of Surveyor Fan sediments as well as the light mineral component and heavy mineral distributions do not provide evidence for the mass redistribution after the mid-Pleistocene transition having had a significant effect on the stress regime of the orogen.
The multiproxy dataset on detrital U-Pb zircon age and Hf isotope data, and Ar-Ar age data on detrital amphibole together with framework component petrographic and single-grain geochemical data on amphibole, garnet, pyroxene and epidote offers consistent constraints on the ultimate source lithologies of the Surveyor Fan sediments. The principal sources since the late Miocene are the Chugach Metamorphic Complex and the Paleocene-Eocene Sanak-Baranof plutonic belt (Fig. 1), both of which are included in the Mesozoic subduction complex comprising the Prince William and Chugach terranes (Fig. 1). This detritus has been reworked at least partly through the sedimentary formations deformed in the Cenozoic fold and thrust belt on the Yakutat Terrane as the lower-grade and non-metamorphic rocks closer to the coast have been incorporated into the mass redistribution processes ( Fig. 1; Huber et al. 2018a,b). Increasing glaciation at the mid-Pleistocene transition did not result in a significant change in the provenance, implying the drainage area and glaciers to have been nested in their topographically defined positions throughout this combined tectonic-climatic evolution Huber et al. 2018a, b).

Conclusion
The framework component composition of the sands and silts of Miocene to Pleistocene age recovered by IODP Expedition 341 in the Gulf of Alaska uniformly reflects the expected active margin provenance most likely connected to erosion of exposed subduction complexes representing the Chugach and Prince William terranes. Pyroxene and epidote also have a consistent geochemical composition in all Miocene to Pleistocene samples. This indicates a persistence of sources and provenance. Detailed analysis of pyroxene and epidote grains indicates the Sanak-Baranof Plutonic Belt, the Chugach Metamorphic Complex and the metabasite belt at its southern border as the main sources of pyroxene and epidote, and the low-grade metamorphic rocks on the Yakutat terrane as additional sources. Increasing amounts of epidote from the Miocene to the Pleistocene at the distal site are in line with progressive input from the metabasite belt at the southern edge of the Chugach Metamorphic Complex. This confirms a general shift in erosion from the central part of the Chugach Metamorphic Complex to the southern flanks of the orogen with increasing glaciation.

3
The higher amount of epidote in the Pleistocene sediments from the proximal site is in line with higher contributions from the formations exposed in fold and thrust belt on the western Yakutat terrane. The different positions of the sites in relation to the Surveyor Fan tributary system very likely caused different parts of the orogen to contribute variable amounts of material. This has also been found in zircon and other heavy mineral data (Dunn et al. 2017;Huber et al. 2018a,b). In general, framework component, pyroxene and epidote data as well as the abundance and origin of different larger rock fragments are in line with a dominant contribution from relatively near-shore sources in the Pleistocene. This concurs with the increasing glaciation since the Miocene to have shifted erosion towards the coast, a trend enhanced by the onset of Northern Hemisphere glaciation in the Pleistocene.
Acknowledgements This study was funded by German Research Foundation (DFG) Grant BA 1011/43-1 and -2 and is part of the research subsequent to IODP Expedition 341 Southern Alaska Margin. We thank M. Dröllner, Münster, for helping to process the samples and J. Berndt and B. Schmitte, Münster, for assistance with acquiring the microprobe data. We appreciate the comments of P. Castillo, Münster, on an early version of this manuscript. The reviews by Sergio Andò, Milan, and Ken Ridgway, West Lafayette, IN, were very helpful and are highly appreciated.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.