Introduction

Cordilleran arcs form within continental crust above convergent plate margins as trench-parallel belts of voluminous calc-alkaline magmatism. They are important locations for continental crustal growth1, with the bulk of the igneous rocks being emplaced in episodes of short (5–20 my) duration high magmatic flux, called magmatic flare-ups that have a periodicity of 30–70 my2,3,4,5,6,7,8. The production of such large volumes of intermediate to felsic melt requires ~50% contribution of partially melted mafic lower continental crust and lithospheric mantle, and ~50% from melt sourced from the mantle wedge above the subducting oceanic slab5, 7,8,9,10,11. As the ratio of plutonic to volcanic rock is typically 30/1 or higher, plutonism is the dominant expression of a flare-up event12. A dense root of ultramafic residues/cumulates is produced and eventually delaminates and founders into the mantle13,14,15,16, possibly balancing the difference between dominantly basaltic island arc crust and the average andesitic composition of continental crust1, 17, 18. Long-lived Cordilleran arcs record repetitive thickening via under-thrusting and shortening of the foreland, with partial melting of this fertile crust facilitating the magmatic flare-up event (75–100 km3/my per km of arc). Cyclicity is recorded by repeated short-lived (5–20 my) high volume magmatic flux events involving magmas with high Sr/Y ratios and isotopic excursions to evolved signatures that punctuate background “steady state” magmatic flux rates less than 30 ± 10 km3/my per km of arc involving magmas with low Sr/Y ratios7, 12, 19. It is posited that the removal of the residue/cumulate root through foundering induces a new cycle5. The ensuing influx of asthenosphere returns the arc to background levels of flux and juvenile isotopic compositions.

In this contribution, we utilise 1:250,000 scale mapping20, geochemistry and geochronology21,22,23,24,25,26,27,28,29,30 to identify a flare-up event in a 200 km long segment of a Mesozoic arc formed along the Pacific margin of Gondwana (Cordillera Zealandia: a title introduced by A. Tulloch in conference presentations). Two contrasting models proposed for this segment of the arc involve allochthonous29, 31,32,33 or autochthonous34,35,36 older (>c. 130 Ma) arc material. New in-situ zircon Hf isotope data and a reappraisal of published Sr-Nd-Hf data support a relationship between the older and younger arc components.

Geological Setting

New Zealand comprises Permian to Cretaceous forearc terranes (Eastern Province) that had accreted to the Pacific margin of Gondwana (Western Province) by the Triassic (Fig. 1). Both provinces are intruded by the Median Batholith (>10,000 km2), which resulted from episodic Cambrian to Early Cretaceous (500–105 Ma) subduction-related plutonism21, 22, 37.

Figure 1
figure 1

(A) Tectonic reconstruction of the Palaeo-Pacific Gondwana Margin at c. 115 Ma (modified from Mortimer,73). The Whitsunday Volcanic Province has been viewed as a correlative to the Cretaceous Flare up in New Zealand74, however some features have been interpreted as a large igneous province75. NEO = New England Orogen. The yellow shaded region represents the Zealandia orogenic belt prior to rifting and fragmentation at 85 Ma. HP = Hikurangi Plateau (approximate location). (B) Geological map of New Zealand showing main plutonic suites; WFO (Western Fiordland Orthogneiss), SPS (Separation Point Suite) and DS (Darran Suite). The ARC (Arthur River Complex) is comprised of an older Carboniferous magmatic component and, and a younger Jurassic to Cretaceous magmatic component that overlaps with the Darren Suite in age25, 53, 54, 76. This study includes Cretaceous ARC samples that have not been formally assigned to a suite however are considered a correlative of the Darren Suite based on age and geochemistry25, 53, 54. A further sample defined as Darren Suite is also included. Modified from Turnbull et al. 20, available under CC BY 4.0 license (http://creativecommons.org/licenses/by/4.0/). Inset: Palinspastic reconstruction of Median Batholith prior to dextral displacement along Alpine Fault.

Full size image

There are three volumetrically important Mesozoic components of the Median Batholith (Fig. 1): (i) Triassic–Early Cretaceous (>129 Ma) calc-alkaline rocks of the Darran Suite characterised by dominantly low Sr/Y ratios (Fig. 2; defined as less than 4038); (ii) Early Cretaceous (<125 Ma) alkaline-calcic granitoid of the Separation Point Suite; and (iii) Early Cretaceous (<125 Ma) pyroxene ± hornblende diorite and monzodiorite of the Western Fiordland Orthogneiss. Both the Separation Point Suite and Western Fiordland Orthogneiss are characterised by dominantly high Sr/Y ratios and represent upper and lower crustal components of the flare-up event respectively (Fig. 2) 21,22,23, 32, 36, 39.

Figure 2
figure 2

Sr/Y (ppm) over time displays the shift to a garnet rich, plagioclase free source at <c. 130 Ma (Geochemical data from Allibone et al. 21, 22 and references therein).

Full size image

Crustal Profile

Crustal profiles of the magmatic arc before and after c. 130 Ma (Fig. 3) were reconstructed by incorporating published seismic velocity and gravity geophysical data, Ce/Y ratios for mafic rocks40, Sr/Y ratios for intermediate rocks41,42,43 and maximum pressures recorded throughout the arc. Sr/Y data were filtered by MgO (1–6 wt%) and SiO2 (55–70 wt%) content, and then Sr/Y outliers were removed via the modified Thompson tau statistical method, similar to the methods described by Chapman et al.41. Median values were then obtained for the Darran Suite, Separation Point Suite and Western Fiordland Orthogneiss respectively.

Figure 3
figure 3

(A) Estimated crustal profile of the magmatic arc before and after c. 130 Ma. See Fig. 1. for abbreviations. (source data: Tulloch and Challis44; Eberhart-Phillips and Reyners45; Scott, et al. 29; De Paoli et al. 47. Figure adapted from Tibaldi et al. 77). (B) Schematic cross section of Cordillera Zealandia showing the inferred under-thrust Gondwana margin converted to an eclogitic root following partial melting during a flare-up event, which produced the voluminous WFO and SPS suite magmatism.

Full size image

Steady-State (Pre-c.130 Ma: Darran Suite)

Hornblende-Al geobarometry records an emplacement depth of 15–22 km for the Darran Suite in Fiordland29, 44 providing a minimum estimate of the crustal thickness before c. 130 Ma. Utilizing the whole rock geochemical data of Allibone et al. 21, gabbroic rocks of the Darran Suite have maximum Ce/Y ratios of 2.5, equating to a crustal thickness in excess of 35 km40. The Darran Suite whole rock data also has median Sr/Y ratios of 28, equating to a crustal thickness of ~40 km41, 42. Seismic velocity data delineates the Moho at approximately 60 km depth beneath the whole of Fiordland45. However, it is unclear to what degree subsequent arc magmatism and the recent transpressional plate tectonic setting46 and subduction initiation modified the pre-c. 130 Ma crustal profile. Geochemical data are consistent with arc material older than c. 130 Ma having been emplaced in an arc approximately 40 km thick (Fig. 3).

Flare-up (Post-c. 130 Ma: Separation Point Suite and Western Fiordland Orthogneiss)

The maximum-recorded emplacement depth of the Western Fiordland Orthogneiss is 18 kbar47, indicating a crustal thickness in excess of 60 km at c. 130 Ma, confirming early inferences of crustal thickness up to 70 km48. Utilizing the whole rock geochemical data of Allibone et al. 22, gabbroic rocks from these plutons have maximum Ce/Y ratios >5, extending the crustal thickness estimates for Fiordland of Mantle and Collins40 to greater than 60 km. Median Sr/Y ratios of 68 for the Western Fiordland Orthogneiss whole rock data set of Allibone et al. 22 estimate a thickness in excess of ~70 km41, 42. The Separation Point Suite’s median Sr/Y ratios of 108 fall outside the correlations established by Profeta et al. 42 and Chapman et al. 41 and may reflect higher degrees of differentiation43. Seismic velocity data extends the current surface geology to at least 10 km depth45. Higher velocity material, inferred by Eberhart-Phillips and Reyners45 to be dense residues/cumulates left behind in the source region of the Western Fiordland Orthogneiss, extends to more than 40 km depth where it is juxtaposed against the newly subducting Australian plate. These observations suggest that the crust may have been more than 80 km thick after c. 130 Ma (Fig. 3). Ducea13 and Ducea and Barton6 posit the residue/melt ratio in a batholith as 1/1 to 3/1. This implies that the 30 km of residues/cumulates imaged by Eberhart-Phillips and Reyners45 beneath the Western Fiordland Orthogneiss is smaller than expected for the predicted 80 km thick arc. The missing residue/cumulates may have either delaminated after flare-up or been tectonically removed during the recent transpressional plate tectonic setting and subduction initiation. The preservation of residue/cumulate materials49,50,51 is unusual as they commonly founder due to having densities greater than the mantle. The root of residue/cumulates may have been preserved by a buoyant subducting slab, and partially removed by recent subduction similar to the Laramides13, 16.

New Magmatic Flux Rates and Isotopes

Richard Jongens (GNS Science) utilized a 1:250 000 digital map20 to calculate the outcrop area of each of the three volumetrically important Mesozoic components of the Median Batholith. The Darran Suite is exposed over approximately 2700 km2, the Separation Point Suite over approximately 1050 km2 and the Western Fiordland Orthogneiss Suite over approximately 2700 km2. The available geochronological data are consistent with the majority of the Darran Suite have been emplaced over approximately 40 my, and the Separation Point and Western Orthogneiss plutons having been emplaced coevally over approximately 10 my21,22,23, 28, 31, 32, 36, 52. The strike length of this segment of the arc is approximately 200 km.

To calculate the flux rate (km3/my per km of arc), the volume in km3 is estimated by utilising published data on crustal thickness (see Crustal Profile above, pre-130 Ma = 40 km thick, post-130 Ma => 60 km thick) and calculations of the area of the suites in km2 (see above). This is then divided by the length of the arc in the study (200 km) and further divided by the period for each suite emplacement (see above). This enables a direct comparison of flux rates estimated for the pre-c. 130 Ma Darran Suite and the combined post-c. 130 Ma Separation Point Suite and Western Fiordland Orthogneiss.

The Darran Suite magmatic flux rate is 14 km3/my per km of arc. As the Separation Point Suite and Western Orthogneiss were coeval, their combined magmatic flux rate is greater than 100 km3/my per km of arc and possibly up to 150 km3/my per km of arc depending upon the thickness of the arc at the time. Individual plutons within these suites were assigned ages based on available geochronology or field relationships to create a histogram of age versus exposed area (Fig. 4).

Figure 4
figure 4

Hfi evolution over time for the three main Mesozoic plutonic suites of Fiordland combined with the flux rates (km2) over time. The excursion to evolved compositions is concomitant with the flare-up, attesting to the crustal component in these magmas. Circles: Western Fiordland Orthogneiss28, diamonds: Separation Point Suite and Western Fiordland Orthogneiss24), Triangles: this study, squares29. Inset of Sri and εNd27, 32, 33 over time (110–170 Ma) display complementing trends of involvement of an evolved crustal source post c. 130 Ma. See Fig. 1 for abbreviations, DM = depleted mantle, CHUR = Chondritic uniform reservoir.

Full size image

Zircon grains from five rock samples of Darran Suite and their corellatives25, 53, 54 (Fig. 1), were previously dated via SHRIMP by Hollis et al. 25 and selected for Lu-Hf analyses via MC-LA-ICP-MS at Macquarie University, Sydney, Australia. See Supplementary Information for raw data and Milan et al. 28, for methods and operating conditions for Lu-Hf data collection.

The initial Hf isotopic ratio (Hfi) of the Darran Suite samples is consistent with recycling of a common ancient source with a c. 500 Ma average model age for all samples (this study, Scott et al. 29) during a 40 my period of low magmatic flux. The Hfi of the Separation Point Suite24, and Western Fiordland Orthogneiss28 shows an excursion to less radiogenic values over time, during a brief pulse of high-flux magmatism. Sri and εNd data (Fig. 3 inset) from previous studies27, 32, 33 show an identical excursion towards an evolved component in the source region.

Cyclicity in Cordilleran Arcs

Cordilleran arcs are characterized by cyclical trends in the flux and isotopic composition of the magma supplied to the upper plate5. The Mesozoic component of the Median Batholith in Fiordland, New Zealand has remarkable similarities to key features of one cycle identified in Cordilleran arcs of the Americas, including the coincidence of a high-flux event and excursions in both Sr/Y ratios36 and isotopic data. There is no evidence for a second Mesozoic high-flux event in Fiordland, suggesting that the Cordilleran cycle we have identified was the only cycle that took place there in the Mesozoic. The cycle halted when plate motions changed, leading to the collapse of Cordillera Zealandia and fragmentation of the margin at c. 105 Ma55. A paucity of exposed rock older than 170 Ma in the Median Batholith suggests that the arc had a low magmatic flux during this time period and precludes the possibility of a second cycle. Rare clasts of c. 200–180 Ma Median Batholith rocks in conglomerates56 have Ce/Y ratios consistent with the crust having been approximately 35 km thick at this time. These observations suggest the crust was 35–40 km thick from the Permian to c. 130 Ma. A good geological record in Fiordland exists from 170–114 Ma. The first ~40 my stage is characterised by low magmatic flux rates of the Darran Suite, consistent with dominantly mantle derived ‘background’ levels of arc magmatism5, 12, 19. The last 15 my stage is characterised by an increase of magmatic flux rates by one order of magnitude coupled with a significant excursion in Sr-Nd-Hf isotopic ratios and the modification of pre-existing arc crust through melt-rock interaction54, 57, 58. Flare-ups and isotopic excursions have been explained in the Americas via the underthrusting of a melt-fertile, lower crustal foreland into the base of the arc5, 6. In Fiordland, zircon inheritance age spectra in the Western Fiordland Orthogneiss are dominated by c. 2480, 770 and 555 Ma peaks which are consistent with the underthrusting or burial, and partial melting of an amalgam of foreland Gondwana margin crust28. The excursion to more evolved isotopic signatures, coupled with the high-flux event (Figs 3 and 4) attests to the importance of this crustal contribution to the magmatic flux.

Arc Migration Over Time

Exhumed examples of long-lived Cordilleran arcs reveal they are constructed from a series of vertically aligned, trench-parallel igneous belts of varying age7. This wide footprint is produced by the migration of the active magmatic front inboard or outboard relative to a fixed position on the upper plate7. Inboard migrations are well documented in the American Cordilleras7, 59, 60, but outboard migrations in arcs also occur7.

For the Mesozoic Cordilleran Zealandia arc, the inboard migration of the active magmatic front (westward) during the flare up (Fig. 1) has been attributed to flat slab subduction36. The location of pre c. 130 Ma Darran Suite has led to competing tectonic interpretations. The most recent model (Scott et al. 29) calling for an allochthonous history of the Darran Suite and excision of a back-arc basin is based upon: (i) contrasting isotopic signatures either side of c. 130 Ma; (ii) the restriction of Gondwana-derived metasedimentary rocks to the west; (iii) a proposed crustal suture; and (iv) the presence of A-type granites. Here we argue against each of these points in turn and reaffirm an autochthonous relationship between the Darran Suite and western parts of the Mesozoic arc.

The apparent contrast in isotopic signatures are resolved in this study by new data defining a smooth excursion towards an evolved character, reflecting the underthrusting or burial of a Gondwanan lower crustal component, as required by the scale and geochemistry of the flare-up event. Though exposures of metasedimentary units in the proposed ‘allochthonous terrane’ lack Gondwana-derived detrital zircon, the very restricted nature of these units suggests that they represent small intra-arc basins formed during oblique subduction. In addition, significant Gondwana inheritance has been observed in Mesozoic plutons of the proposed ‘allochthonous terrane’ (e.g. Pomona Island and Clark Hut granites21, 61). The proposed crustal suture is unusually narrow (300 m wide), and lacks ophiolitic material that might support the argument for basin closure. The same structure has been interpreted (along strike) to reflect transpression within a Cordilleran setting34. Furthermore, intrusive relationships between the Western Fiordland Orthogneiss and parts of the Darran Suite are preserved in northern Fiordland48, 62.

Though Scott et al. 29 ascribe A-type granites in the Darran Suite to a rift setting, these are restricted to just ~42 km2 and a number of tectonic settings have been proposed for A-type granites elsewhere63,64,65,66. Furthermore, these rocks have similar Sr and Nd isotope ratios32 to the entire Darran Suite and intruded at a time of thickened crust inconsistent with a rift setting.

Generation of Crust in Arcs

The flux from mantle to crust is basaltic (e.g. Davidson and Arculus67) in contrast to the andesitic composition of average continental crust1, 68, 69. The Fiordland flare-up event represented by the Separation Point Suite and Western Fiordland Orthogneiss equates to two thirds of the total volume of newly created arc crust throughout the Mesozoic in ‘Cordillera Zealandia’ and highlights the significance of short-lived high-flux episodes in Cordilleran arcs. Long-lived Cordilleran arcs of the Americas record repetitive flare-up contributions as high as 85–90% of the total volume of the arc5,6,7, 12, 70,71,72. These high-flux episodes require a significant component of the arc magma to be derived from melt-fertile lower crust. These observations diminish the contribution of mantle-derived melts in Cordilleran arcs and in addition to the foundering of dense residues/cumulates from the roots of arcs13,14,15,16, contribute to resolving why the average continental crust is andesitic in composition.