The role of arc migration in Cordilleran orogenic cyclicity

Continental arc rocks located further away from the trench are typically characterized by more evolved radiogenic isotopic compositions. Episodic arc migration away from the trench would produce a temporal record distinguished by episodic shifts to more evolved compositions. In most continental arcs, temporal shifts to evolved isotopic compositions correlate with magmatic high-flux events, which are the basis for cyclicity models in Cordilleran orogenic systems. Landward arc migration into more melt-fertile regions of the continental lithosphere can explain both episodic shifts in isotopic composition and highflux events without requiring underthrusting of retroarc lithosphere. If arc migration is predominantly autocyclic, controlled by intra-arc processes, then arc migration itself may drive orogenic cyclicity. Conversely, periodic arc migration may be controlled by allocyclic processes like slab anchoring or folding in the mantle transition zone. In either case, arc migration may be the key to understanding what drives high-flux events and cyclicity in Cordilleran orogens. INTRODUCTION Models for orogenic cyclicity in Cor di­ lleran systems espouse teleconnections between forearc, intra­arc, and retroarc deformation; mag­ matic arc tempos and composition; crustal thick­ ening; uplift, erosion, and sedimentation; and subcrustal mass transfer like arc root growth, arc root founder ing, lithospheric delamination, and tectonic underplating (Ducea, 2001; Ducea and Barton, 2007; DeCelles et al., 2009; Cao et al., 2015; DeCelles and Graham, 2015; Ducea and Chapman, 2018). The foundation of orogenic cy­ clicity models is the episodic nature of continen­ tal arc magmatism, manifest as arc flare­ups or high­flux events (generally occurring every 30– 70 m.y.) separated by magmatic lulls (Armstrong and Ward, 1993; DeCelles et al., 2009; Pater son and Ducea, 2015; Ducea et al., 2015; Kirsch et al., 2016). A breakthrough in deciphering the under­ lying cause of orogenic cyclicity came from the recognition that the radiogenic isotopic composi­ tion of continental arc rocks commonly become more evolved during high­flux events (Ducea, 2001; Ducea and Barton, 2007). These temporal shifts to evolved compositions have been called isotopic pull­downs because initial εNd values become more negative (DeCelles et al., 2009). There is no consensus on what drives high­flux events and isotopic pull­downs, but a promi­ nent hypothesis is that shortening in the retro arc (retro wedge) thrust belt under thrusts continental lower crust and mantle into the melt source re­ gion beneath the arc ( Ducea, 2001; Ducea and Barton, 2007; DeCelles et al., 2009; DeCelles and Graham, 2015; Lee and Lackey, 2015). The logic behind the hypothesis is that the influx of melt­fertile lower crust produces the high­flux event, and the antiqui ty of the crust produces the isotopic pull­down. However, not all Cordilleran batholiths that exhibit high­flux events show isotopic pull­ downs (e.g., the Coast Mountains batholith in British Columbia [Canada] and Alaska [United States]; Girardi et al., 2012; Cecil et al., 2018). This suggests that other factors may play a role or that in these instances there is not enough isotopic contrast to produce a pull­down. An­ other complexity is that some Cordilleran arcs record high­flux events despite limited amounts of retro arc underthrusting (e.g., the southern Andean arc; Kirsch et al., 2016; Horton, 2018). Existing models for producing isotopic pull­ downs have generally treated the position of the arc as static, but if the locus of magmatism shifts across lithosphere of variable age and isotopic composition, then isotopic pull­downs could be related to arc migration and need not be caused by retroarc underthrusting. A common feature of igneous rocks in Cor­ dilleran orogens is a spatial isotopic trend in which arc rocks located further from the trench (up to a few hundred kilometers) are more radio­ genically evolved (Kistler and Peterman, 1973; Farmer and DePaolo, 1983; Chapman et al., 2017). Controls on this trend include (1) crustal assimilation (DePaolo, 1981), and (2) the na­ ture of the mantle in the melt source region: asthenospheric (juvenile) near the trench, and lithospheric (evolved) away from the trench (Chapman et al., 2017). As a result of this spa­ tial isotopic trend, landward arc migration may produce an isotopic pull­down, which would be reflected in any temporal records of mag­ matism (e.g., detrital zircon εHf data sets). By inference, arc­migration processes may also be linked to periodic high­flux events in Cordilleran orogenic systems. DATA AND METHODS To explore this idea, arc magmatism is ex­ amined in the Sierra Nevada (western United States), the continental arc originally used to develop Cordilleran orogenic cyclicity con­ cepts (Ducea, 2001; Ducea and Barton, 2007; DeCelles et al., 2009). We compiled a variety of data sets relevant to the evolution of the batho­ lith, which are summarized in Figure 1 and pre­ sented in the GSA Data Repository1. Although not examined here, associations between arc mi­ gration and isotopic pull­downs (e.g., central Andean arc; Haschke et al., 2002; Kay et al., 2005) and arc migration and high­flux events (e.g., Gangdese batholith [Tibet]: Chapman and Kapp, 2017; Coast Mountains batholith: Cecil et al., 2018) can also be observed in other mod­ ern and ancient Cordilleran systems. 1GSA Data Repository item 2019223, tables of compiled data with data sources, is available online at http:// www .geosociety .org /datarepository /2019/, or on request from editing@ geosociety .org. CITATION: Chapman, J.B., and Ducea, M.N., 2019, The role of arc migration in Cordilleran orogenic cyclicity: Geology, v. 47, p. 627–631, https:// doi .org /10 .1130 /G46117.1 Manuscript received 25 September 2018 Revised manuscript received 11 April 2019 Manuscript accepted 11 April 2019 https://doi.org/10.1130/G46117.1 © 2019 The Authors. Gold Open Access: This paper is published under the terms of the CC-BY license. Published online 13 May 2019 Downloaded from https://pubs.geoscienceworld.org/gsa/geology/article-pdf/47/7/627/4775187/627.pdf by guest on 29 June 2019 628 www.gsapubs.org | Volume 47 | Number 7 | GEOLOGY | Geological Society of America Sierra Nevada Batholith The Sierra Nevada displays prominent spa­ tial isotopic trends in which radiogenic isotopes become more evolved landward of the trench (Fig. 1). The Sr87/Sr86 spatial isotopic trend is the most well known because the Sr87/Sr86 = 0.706 isopleth (“0.706 line”) has been used as a proxy for the western edge of North American basement (Kistler and Peterman, 1973, 1978), outboard of which are a series of isotopically juvenile accreted oceanic and ophiolitic terranes (Snow and Scherer, 2006). The spatial distribu­ tion of εNd mimics that of Sr87/Sr86 (negative cor­ relation) and shows increasingly evolved values to the east (Fig. 1). Similar isotopic trends are known from within the underlying continen­ tal mantle lithosphere (Ormerod et al., 1988; Wenner and Coleman, 2004). The radiogenic isotopic composition of magmatism through­ out the history of the batholith has followed the spatial isotopic trend. The Sierra Nevada experienced three high­flux events during the Mesozoic, centered on ca. 225 Ma, ca. 160 Ma, and ca. 95 Ma (Ducea, 2001; Cecil et al., 2012; Paterson and Ducea, 2015). All three high­flux events are associated with isotopic pull­downs, and all three occurred when magmatism was concentrated furthest from the trench (Ducea and Barton, 2007; Kirsch et al., 2016) (Fig. 1). As a result, the Triassic, Jurassic, and Creta­ ceous arcs are largely spatially coincident and form a composite batholith. The final high­flux event is diachronous from north to south, but it occurred during the Late Cretaceous in the southern and central Sierra Nevada where it is the most well documented (Paterson and Ducea, 2015). During the mag­ matic lull preceding the Late Cretaceous high­ flux event, magmatism was located closer to the trench (modern­day western Sierra Nevada and Great Valley) and produced rocks with more ju­ venile isotopic compositions (Fig. 1). From 130 to 80 Ma, magmatism migrated away from the trench at 2–5 mm/yr and produced rocks with in­ creasingly evolved isotopic compositions (Chen and Moore, 1982; Cecil et al., 2012; Ardill et al., 2018) (Fig. 1). Arc migration started during a magmatic lull, ca. 130 Ma. The Late Cretaceous high­flux event occurred when the arc was lo­ cated in what is today the eastern Sierra Nevada, 30–40 m.y. after arc migration had started (Fig. 1). The data suggest not only that the Nd isotopic pull­downs and high­flux events in the Sierra Nevada were produced by delivery of more iso­ topically evolved and melt­fertile material into a fixed melt source region beneath the arc, but that the melt source region and arc itself had migrated landward into a different part of the lithosphere. Higher Sr/Y and La/Yb suggest an increase in crustal thickness during this time (Karlstrom et al., 2014; Profeta et al., 2015; Kirsch et al., 2016), and an east­west spatial gradient in Sr/Y (Ardill et al., 2018) indicates that this thicken­ ing was concurrent with arc migration (Fig. 1). Other spatial geochemical gradients that were produced during eastward arc migration include increasing alkalinity (at approximately constant weight percent SiO2; Bateman, 1992), increasing oxygen fugacity (Dodge, 1972), and decreas­ ing Dy/Yb (Ardill et al., 2018). Xenolith data indicate that the mantle lithosphere was also thickened during this time, at least in part by melt entrapment (Chin et al., 2012). In the cen­ tral to northern Sierra Nevada, melt depletion and/or a thickened lithosphere that impinged on the subducting slab may have terminated the high­flux event and all subduction magmatism at ca. 80 Ma (Ducea, 2001; Chin et al., 2012; Karlstrom et al., 2014).

There is no consensus on what drives highflux events and isotopic pulldowns, but a promi nent hypothesis is that shortening in the retro arc (retro wedge) thrust belt under thrusts continental lower crust and mantle into the melt source re gion beneath the arc ( Ducea, 2001;Ducea and Barton, 2007;DeCelles et al., 2009;DeCelles and Graham, 2015;Lee and Lackey, 2015). The logic behind the hypothesis is that the influx of meltfertile lower crust produces the highflux event, and the antiqui ty of the crust produces the isotopic pulldown.
However, not all Cordilleran batholiths that exhibit highflux events show isotopic pull downs (e.g., the Coast Mountains batholith in British Columbia [Canada] and Alaska [United States]; Girardi et al., 2012;Cecil et al., 2018). This suggests that other factors may play a role or that in these instances there is not enough isotopic contrast to produce a pulldown. An other complexity is that some Cordilleran arcs record highflux events despite limited amounts of retro arc underthrusting (e.g., the southern Andean arc; Kirsch et al., 2016;Horton, 2018). Existing models for producing isotopic pull downs have generally treated the position of the arc as static, but if the locus of magmatism shifts across lithosphere of variable age and isotopic composition, then isotopic pulldowns could be related to arc migration and need not be caused by retroarc underthrusting.
A common feature of igneous rocks in Cor dilleran orogens is a spatial isotopic trend in which arc rocks located further from the trench (up to a few hundred kilometers) are more radio genically evolved (Kistler and Peterman, 1973;Farmer and DePaolo, 1983;Chapman et al., 2017). Controls on this trend include (1) crustal assimilation (DePaolo, 1981), and (2) the na ture of the mantle in the melt source region: asthenospheric (juvenile) near the trench, and lithospheric (evolved) away from the trench (Chapman et al., 2017). As a result of this spa tial isotopic trend, landward arc migration may produce an isotopic pulldown, which would be reflected in any temporal records of mag matism (e.g., detrital zircon ε Hf data sets). By inference, arcmigration processes may also be linked to periodic highflux events in Cordilleran orogenic systems.

DATA AND METHODS
To explore this idea, arc magmatism is ex amined in the Sierra Nevada (western United States), the continental arc originally used to develop Cordilleran orogenic cyclicity con cepts (Ducea, 2001;Ducea and Barton, 2007;DeCelles et al., 2009). We compiled a variety of data sets relevant to the evolution of the batho lith, which are summarized in Figure 1 and pre sented in the GSA Data Repository 1 . Although not examined here, associations between arc mi gration and isotopic pulldowns (e.g., central Andean arc; Haschke et al., 2002;Kay et al., 2005) and arc migration and highflux events Chapman-G46117.1 1st pages

Sierra Nevada Batholith
The Sierra Nevada displays prominent spa tial isotopic trends in which radiogenic isotopes become more evolved landward of the trench (Fig. 1). The Sr 87 /Sr 86 spatial isotopic trend is the most well known because the Sr 87 /Sr 86 = 0.706 isopleth ("0.706 line") has been used as a proxy for the western edge of North American basement Peterman, 1973, 1978), outboard of which are a series of isotopically juvenile accreted oceanic and ophiolitic terranes (Snow and Scherer, 2006). The spatial distribu tion of ε Nd mimics that of Sr 87 /Sr 86 (negative cor relation) and shows increasingly evolved values to the east (Fig. 1). Similar isotopic trends are known from within the underlying continen tal mantle lithosphere (Ormerod et al., 1988;Wenner and Coleman, 2004). The radiogenic isotopic composition of magmatism through out the history of the batholith has followed the spatial isotopic trend. The Sierra Nevada experienced three highflux events during the Mesozoic, centered on ca. 225 Ma, ca. 160 Ma, and ca. 95 Ma (Ducea, 2001;Cecil et al., 2012;Paterson and Ducea, 2015). All three highflux events are associated with isotopic pulldowns, and all three occurred when magmatism was concentrated furthest from the trench (Ducea and Barton, 2007;Kirsch et al., 2016) (Fig. 1). As a result, the Triassic, Jurassic, and Creta ceous arcs are largely spatially coincident and form a composite batholith.
The final highflux event is diachronous from north to south, but it occurred during the Late Cretaceous in the southern and central Sierra Nevada where it is the most well documented . During the mag matic lull preceding the Late Cretaceous high flux event, magmatism was located closer to the trench (modernday western Sierra Nevada and Great Valley) and produced rocks with more ju venile isotopic compositions (Fig. 1). From 130 to 80 Ma, magmatism migrated away from the trench at 2-5 mm/yr and produced rocks with in creasingly evolved isotopic compositions (Chen and Moore, 1982;Cecil et al., 2012;Ardill et al., 2018) (Fig. 1). Arc migration started during a magmatic lull, ca. 130 Ma. The Late Cretaceous highflux event occurred when the arc was lo cated in what is today the eastern Sierra Nevada, 30-40 m.y. after arc migration had started (Fig.  1). The data suggest not only that the Nd isotopic pulldowns and highflux events in the Sierra Nevada were produced by delivery of more iso topically evolved and meltfertile material into a fixed melt source region beneath the arc, but that the melt source region and arc itself had migrated landward into a different part of the lithosphere.
Higher Sr/Y and La/Yb suggest an increase in crustal thickness during this time (Karlstrom et al., 2014;Profeta et al., 2015;Kirsch et al., 2016), and an eastwest spatial gradient in Sr/Y (Ardill et al., 2018) indicates that this thicken ing was concurrent with arc migration (Fig. 1). Other spatial geochemical gradients that were produced during eastward arc migration include increasing alkalinity (at approximately constant weight percent SiO 2 ; Bateman, 1992), increasing oxygen fugacity (Dodge, 1972), and decreas ing Dy/Yb (Ardill et al., 2018). Xenolith data indicate that the mantle lithosphere was also thickened during this time, at least in part by melt entrapment (Chin et al., 2012). In the cen tral to northern Sierra Nevada, melt depletion and/or a thickened lithosphere that impinged on the subducting slab may have terminated the highflux event and all subduction magmatism at ca. 80 Ma (Ducea, 2001;Chin et al., 2012;Karlstrom et al., 2014).

Melt Fertility
Generating a highflux event requires in creased melt fertility, which is primarily depen dent on rock composition, pressure, temperature, and water content. Any mechanism to explain the highflux events in the Sierra Nevada should also produce more isotopically evolved compositions, occur on an episodic basis, and be associated with landward arc migration. One possibility is that landward arc migration simply hastens the delivery of lowercrustal material to the arc and adds to the rate of retroarc underthrusting. Shallowing of the subduction angle (causing landward arc migration) is commonly associ   Chapman-G46117.1 1st pages ated with increased plate coupling and crustal shortening, which may explain the correlation between highflux events in the Sierra Nevada and kinematic jumps (new frontal thrust sheets) in the Sevier retroarc thrust belt ( DeCelles et al., 2009;DeCelles and Graham, 2015).
A second possibility is that highflux events are produced when an arc migrates into a more meltfertile crustal province, regardless of retro arc dynamics. The accreted oceanic and ophio litic terranes that compose basement in the western Sierra Nevada are relatively more mafic (Snow and Scherer, 2006) and presumably less melt fertile (cf. Clemens and Viezeuf, 1987) than the intermediate igneous and quartzofeldspathic rocks that compose (cratonic) North American crust in the eastern Sierra Nevada (Kistler and Peterman, 1978). However, the degree of crustal assimilation in the Sierra Nevada remains con tested (see review by Nelson et al. [2013]), and some researchers have suggested that the mantle exerts a firstorder control on isotopic compositions, particularly in the eastern Sierra Nevada-the locus of the Late Cretaceous high flux event (Wenner and Coleman, 2004;Lackey et al., 2008;Jagoutz and Klein, 2018).
If the mantle is an important contributor to Sierra Nevada arc rocks, then a third possibility to consider is whether processes that increase melt fertility in the mantle, specifically hydra tion ± (re)fertilization, could be responsible for producing highflux events. There is some evidence to suggest that this may be the case. Chin et al. (2012Chin et al. ( , 2014 documented refertil ized mantle (peridotite and garnet pyroxenite) xenoliths from the Sierra Nevada and suggested that refertilization occurred contemporaneously with the Late Cretaceous highflux event and that refertilization increased with depth in the mantle lithosphere. Compared to the convecting asthenospheric (depleted) mantle wedge, a thick and cool mantle lithosphere allows for efficient melt entrapment, incompatible element chroma tography, and longterm storage of water in hy drous phases and hydrated nominally anhydrous minerals (O'Reilly and Griffin, 2013). Many of the temporal and spatial geochemical gradients in the Sierra Nevada that correlate with landward arc migration-increased Sr/Y, La/Yb, and oxy gen fugacity, and decreasing Dy/Yb (Fig. 1)are consistent with exceptionally hydrous melt conditions and the stabilization of amphibole ± garnet (Müntener et al., 2001;Davidson et al., 2007). A landward increase in weight percent K 2 O in the Sierra Nevada (Fig. 1), and other arc systems globally, has been called the Kh (where h is depth) relationship (Dickinson, 1975). One of the many plausible explanations for the Kh relationship is partial melting of fluid and/or meltmetasomatized mantle litho sphere (e.g., phlogopiteperidotite), which is a common explanation for alkaline magmas in general (Menzies and Murthy, 1980). HighK magmas from the Sierra Nevada, erupted dur ing the Pliocene, have been related to melting of (delaminated) fluidmetasomatized mantle lithosphere (Farmer et al., 2002).
We envision a scenario in which meltfertile, subductionrelated, metasomatic products accu mulate over time in the mantle lithosphere, land ward of the arc axis (Fig. 2). Landward migra tion of the arc and mantle wedge into hydrated and fertile parts of the lithosphere may trigger a highflux event. This mechanism allows for a slow buildup (perhaps during magmatic lulls) but rapid delivery of meltfertile material to a melt source region. The midCenozoic ignim brite flareup in the North American Cordillera (Lipman, 1992) may be partially analogous to the highflux mechanism proposed above. Low angle subduction of the Farallon plate resulted in widespread refrigeration and hydration of the North American lithosphere (Humphreys et al., 2003). After the Farallon slab foundered (or rolled back), hydrated North American mantle lithosphere was heated by upwelling astheno sphere and produced the midCenozoic flareup (Farmer et al., 2008). Metasomatic accumulation and storage in the deep lithosphere followed by rapid heating may be a relatively common way to generate arc flareups and highflux events.

Arc Migration
Cyclical arc migration has been observed in longlived Cordilleran orogenic systems (e.g., central Andes; Haschke et al., 2002), but what drives this behavior is debated. Periodicity rules out many noncyclical factors like subducting plate age, plate thickness, and rates of motion, which could control slab dip. Explanations for periodic arc migration generally fall into two categories: autocyclic processes (occurring within the arc system) or allocyclic processes (externally driven). Examples of autocyclic pro cesses include magmatic and tectonic thickening of the arc root and arc lithosphere, which could deflect the mantle wedge or locus of pluton em placement landward (Karlstrom et al., 2014) or decrease slab dip by increasing mantle wedge suction (O'Driscoll et al., 2009). Examples of allocyclic processes include mechanisms like alternating periods of slab anchoring and slab breakoff (Faccenna et al., 2013) or periodic slab folding in the mantle transition zone (Gibert et al., 2012), which affect slab dip and plate cou pling. Regardless of the exact processes (auto cyclic or allocyclic) involved, we suggest that episodic arc migration through a spatial isotopic trend may be a contributing factor to producing isotopic pulldowns in the Sierra Nevada.

CONCLUSIONS
In many continental arcs, relatively subtle changes in the spatial distribution of magma tism influence radiogenic isotopic compositions.  Chapman-G46117.1 1st pages range from +5 to -5 (westtoeast gradient) in <100 km across strike of the arc. Nd isotopic pulldowns and highflux events in the Sierra Nevada correlate to periods when the locus of magmatism shifted landward, away from the trench, and into more evolved lithospheric prov inces and enriched uppermantle melt source re gions. Subductionrelated metasomatism (refer tilization and hydration) prior to highflux events may gradually increase meltfertility of the deep lithosphere in subduction zones. Arc migration into these cumulatively metasomatized regions could spark a highflux event. Highflux events may be generated by multiple processes, but the role and contribution of arc migration should be considered in models of Cordilleran orogenic cyclicity.