Geochemical and isotopic characterisation of trench sediment at the Hikurangi Margin from IODP sites U1518 and U1520

ABSTRACT

The composition of trench sediment at convergent margins exhibits strong compositional links to the corresponding arc magmas. To test for the existence of such links at the Hikurangi margin, we provide a first geochemical characterisation of the IODP sites U1518 and U1520 drilled at the upper and incoming plate respectively. The identification of a décollement in the trench sediments allows distinction of the accreting clastic sediment above the décollement and the subducting pelagic and volcaniclastic material below it. The accreted material and that located above the décollement show homogenous Pb, Sr and Nd isotope ratios with bulk compositions of ²⁰⁶Pb/²⁰⁴Pb: 18.8953, ²⁰⁷Pb/²⁰⁴Pb: 15.6552, ²⁰⁸Pb/²⁰⁴Pb: 38.8371, ⁸⁷Sr/⁸⁶Sr: 0.70935, ¹⁴³Nd/¹⁴⁴Nd: 0.51241. In contrast, the material located below the décollement shows more heterogeneous Pb-Sr-Nd isotope ratios and bulk compositions of ²⁰⁶Pb/²⁰⁴Pb: 20.4826, ²⁰⁷Pb/²⁰⁴Pb: 15.7355, ²⁰⁸Pb/²⁰⁴Pb: 39.9833, ⁸⁷Sr/⁸⁶Sr: 0.70752, and ¹⁴³Nd/¹⁴⁴Nd: 0.51286. Tectonic erosion in the Hikurangi margin implies that accreted material could subsequently be subducted from the underside of the upper plate. Therefore, this characterisation is of prime importance for future work to constrain the role that subducted materials play in the compositional variations of the arc magmatism associated with subduction along the Hikurangi margin.

KEYWORDS:

• Subducting sediments
• bulk composition
• tectonic erosion
• décollement
• isotopes

Introduction

The chemical characterisation of local trench sediments is essential to constrain the chemical budget of arc magmatism globally (e.g. Plank 2014). The Global Subducting Sediments (GLOSS-I and GLOSS-II), an average of global trench sediment has emphasised the links between subducting sediments and the composition of arc magmatism worldwide (Plank and Langmuir 1998; Plank 2014). The occurrence of tectonic erosion implies that materials from the upper plate forearc and accretionary prism can also affect the chemical budget of arc magmatism globally (e.g. Jicha et al. 2004; Risse et al. 2013; Straub et al. 2015; Vannucchi et al. 2016; Straub et al. 2020). Regardless of the subduction process, subducted materials play a significant role in magma generation processes and the formation of new or recycling of existing crust, not only in subduction zones (e.g. Plank and Langmuir 1998; Kay et al. 2005; Risse et al. 2013; Holm et al. 2014; Plank 2014; Parolari et al. 2018; Straub et al. 2020; Parolari et al. 2021; Cornet et al. 2022) but also in intra-plate settings and in relation to alkaline volcanism globally (e.g. Eisele et al. 2002; Chauvel et al. 2008; Rapp et al. 2008) through the role of these components metasomatising the mantle.

Locally, in the Tonga-Kermadec-Hikurangi subduction system, a general increase in subducted material input into the mantle towards the south is displayed, with a large increase along the Hikurangi margin (Gamble et al. 1996). In the southern Kermadec arc, a mélange of subducting sediments, tectonically eroded forearc crust, and material from the Hikurangi Plateau has been proposed to affect the chemical composition of arc volcanism (Timm et al. 2014). The presence of mantle heterogeneities caused by mantle-slab interactions, particularly the addition of subducting sediments, has been proposed to also have an effect in the Taupo Volcanic Zone (TVZ) (Gamble et al. 1993; Graham et al. 1995; Gamble et al. 1996; Rooney and Deering 2014). As with the Kermadec arc, tectonic erosion of forearc crust and subsequent interaction with the mantle source has been suggested to affect the chemical composition of volcanic products of the TVZ (Corella Santa Cruz et al. 2023).

Prior to 2018, the closest drillhole locations to the Hikurangi trench were Ocean Drilling Program (ODP) sites 1123 and 1124, located c. 500 km to the east (Figure 1) and separated from the trench by the Hikurangi channel. These sites were used to estimate the bulk geochemical composition of Hikurangi margin subducting sediments by Plank (2014) and integrated into the composition of the estimated Global Subducting Sediments II (GLOSS-II). Materials and units of the Hikurangi Margin were identified by Davy et al. (2008) based on seismic and ODP site 1124 logging data. They reported six units varying from sediments to the basaltic basement, described as follows. The first unit corresponds to Cenozoic sediments with interbeds of tephra and clays, and an age range of 32 to 0 Ma. The second unit corresponds to their Sequence Y, comprising chalks and mudstones with an age range of 70 to 32 Ma. Below this unit are laminar Mesozoic sediments ranging from 100 to 70 Ma containing volcaniclastic sediments with basalt interbeds aged 100 to 90 Ma. The fifth unit corresponds to Cretaceous volcaniclastic sediments with an uncertain age range, but likely to be 125 to 100 Ma old. Finally, the basaltic basement corresponds to the sixth unit and has an age range of 125 to 120 Ma.

Figure 1. Location map of ODP sites 1123 and 1124, and IODP sites U1518 and U1520. For reference, the Taupo Volcanic Zone (TVZ), Ruatoria debris avalanche (R), Chatham Rise, Chatham Islands, the Hikurangi Plateau, and the Hikurangi-Kermadec Trench are shown. Cross section of the trench and location of sites U1518 and U1520 after Saffer et al. (2019), discontinuous line indicating the formation of the décollement after Barnes et al. (2020).

For Plank’s (2014) estimation of the bulk composition of Hikurangi trench sediment, the composition was taken to be 50% carbonate-rich Cenozoic sediments and 50% volcaniclastics from the Hikurangi Plateau. The composition of the carbonate-rich Cenozoic sediments was calculated based on geochemical analyses (major and trace elements, Sr-Nd-Pb isotopes) of ODP Sites 1123 and 1124, supplemented by geochemical analyses of surface sediments from the Hikurangi Plateau. As samples of the volcaniclastic sequence of the Hikurangi Plateau were not available, the geochemical composition of the volcaniclastic section was based on the volcaniclastic sediments of the genetically related Manihiki Plateau (Deep Sea Drilling Project (DSDP) Site 317) and other intraplate and basement volcaniclastic sediments (ODP Site 801, DSDP Site 417) as well as basalts reported by Hoernle et al. (2010).

In late 2017–2018, Expedition 372B/375 of the International Ocean Drilling Program (IODP) took place at the Hikurangi Trough, drilling sites U1518, U1519, U1520, and U1526 in a trench-perpendicular transect, including both the subducting plate and the accretionary prism (Barnes et al. 2019; Saffer et al. 2019; Wallace et al. 2019). Thus, material closer to the trench became available along with the volcaniclastic sequence (Figure 1), enabling a more accurate estimation of the geochemical composition of the trench sediment. In addition, recent tectonic and structural observations became relevant to determine subducting and accreting materials, due to the identification of the development of a décollement zone in the subducting plate (Barnes et al. 2020).

According to Hori and Sakaguchi (2011), a décollement is a shear localisation structure formed by variations in the strength of the subducting plate material during the formation of the accretionary prism. This works as a plate boundary fault, where the material above it collides with and is accreted to the overriding plate, whereas the material below it stays in the subducting plate (Westbrook et al. 1988; Moore et al. 1998). Thus, accreting material is located above the décollement zone, whereas subducting material is located below it. However, the occurrence of tectonic erosion in the Hikurangi margin (Collot et al. 1996; Collot and Davy 1998; Lewis et al. 1998; Barker et al. 2009; Pedley et al. 2010), implies that accreted material can also be ultimately subducted. Variations in the structure of the Hikurangi margin indicate that tectonic erosion occurs in the northern portion (−39.3°S), while the southern portion corresponds to a more accretionary margin (Schwarze and Kukowski 2022). The presence of the décollement zone can help to distinguish between subducting material, and that accreting and prone to ultimate tectonic erosion in the Northern Hikurangi margin. Bulk major and trace element concentrations, and Sr-Pb-Nd isotope compositions are presented in this study, to geochemically characterise the drilled sequences above and below the décollement zone, which may impact the composition of the magmas erupted in the TVZ.

General description of the lithostratigraphic units at the Hikurangi margin

Lithostratigraphic units found in IODP Expedition 372B/375 can be divided into three main groups: hemipelagic and terrigenous facies (Units I–III), pelagic facies (Unit IV) and the Hikurangi Plateau volcaniclastic facies (Units V and VI). Based on expedition reports (Barnes et al. 2019; Saffer et al. 2019; Wallace et al. 2019), Unit I, a trench-wedge facies with a Quaternary age (<2.58 Ma) is composed of silty clay to clayey silt interbedded with silt and sand, indicating turbidity current activity and hemipelagic settling during deposition. Unit II with a Quaternary age (<2.58 Ma) is from a distal edge of the Ruatoria debris avalanche, which formed from a collapse resulting from seamount subduction (Collot et al. 2001). It is composed of silty clay to clayey silt interbedded with silt. Unit III, a trench-wedge facies with a Quaternary–late Miocene age (<11.63 Ma), is composed of mud and mudstone interbedded with silt, indicative of hemipelagic settling and dilute turbidity currents during deposition. Unit IV, a pelagic facies of early Paleocene to late Miocene age (66–5.3 Ma), is composed of marl, calcareous mudstone, and chalk, and represents pelagic sedimentation. Minor horizons of airborne volcanic ash are found within Units I to IV. Unit V, a volcaniclastic facies with a late Cretaceous age (100.5–66 Ma), is composed of volcaniclastic conglomerate associated with the Hikurangi Plateau. Unit VI has mixed lithologies with a Cretaceous age (145–66 Ma) and is composed of volcaniclastic conglomerate, siltstone, limestone, mudstone, and basalt.

A good correlation is found in the drilled units and those identified by Davy et al. (2008): Units I, II, and III from the IODP Expedition 372B/375 correlate with the Cenozoic sediments, Unit IV correlates with sequence Y, and Units V and VI may correspond to the basement and volcaniclastic units. Barnes et al. (2020) identified the formation of a décollement zone (Figure 2) based on seismic line data and deformation in drill sites U1520 and U1526. They reported that the Hikurangi Plateau, its volcaniclastic sequence, and the lowest portion of the pelagic sequence are located below the plate boundary fault, and thus are being currently subducted. In contrast, the terrigenous sequence and the rest of the pelagic sequence are located above the plate boundary, and therefore accreted to the forearc complex. However, the accreted material may subsequently be subducted, as the accretionary forearc is tectonically eroded by the seamount-studded subducting plate (Collot et al. 1996; Collot and Davy 1998; Lewis et al. 1998; Barker et al. 2009; Pedley et al. 2010; Schwarze and Kukowski 2022).

Figure 2. Site recovery columns after Saffer et al. (2019). The selected samples are shown in squares. Red squares represent samples with only major elements reported; blue squares represent samples with major and trace elements as well as selected isotopic ratios. Décollement formation after Barnes et al. (2020). Details are in Table 1.

Methods

Sample selection

The subducting and accreted sediments were sampled from IODP drill holes U1518E and F, and U1520C and D (Figures 1 and 2; Table 1). A total of 28 samples were selected which fully represent the lithological units present in the subducting complex as well as the accretionary prism. Twelve samples from drill holes U1518E and F correspond to Units IA-B, II, and IIIA-B and represent the accreted material, with associated lithological descriptions given in Saffer et al. (2019). At site U1518, Unit IA (1 sample) has a thickness of 197.7 m and Unit IB (4 samples) is 106.8 m thick, giving a total thickness of 304.5 m for Unit I. Unit II has a thickness of 65.9 m and is represented by 4 samples. Unit IIIA (2 samples) has a thickness of 105.30, and Unit IIIB (1 sample) is 16.6 m thick, giving a minimum thickness of 121.86 m for Unit III. Sixteen samples from drill holes U1520D and C correspond to units I, II, III, IV, V, and VI with unit descriptions reported in Barnes et al. (2019) and representing material from the subducting plate above and below the décollement based on Barnes et al. (2020). Here, thicknesses of 110.5 m (1 sample) for Unit I, 109.5 m (1 sample) for Unit II, 289.8 m (3 samples) for Unit III and 338.6 m (5 samples) for Unit IV are given. Unit V has a thickness of 137.8 m (3 samples), and Unit VI has a minimum thickness of 29.5 m (1 sample). Samples were carefully selected to avoid the inclusion of volcanic ash layers.

Table 1. Selected samples from IODP sites U1518 and U1520.

Sample	Lithology	Age (Ma)	Unit	Position	Depth interval (mbsf)
U1518E
26X 6W 69/71	Sandy silt	Quaternary	IA	FA	123–133
U1518F
5R 1W 12/14	Clayey silt	Quaternary	IB	FA	226–236
9R 5W 24/26	Sandy silt	0.63	IB	FA	264–274
11R 2W 47/49	Silty clay	Quaternary	IB	FA	284–294
12R 2 18/20	Silty clay	0.53–2.17	IB	FA	294–304
13R 1W 59/61	Clayey silt	Quaternary	II	FA	304–313
14R 1W 133/135	Clayey silt	Quaternary	II	FA	313–323
16R 4W 24/26	Silty clay	Quaternary	II	FA	332–342
19R 3W 33/35	Clayey silt	Quaternary	II	FA	360–370
24R 2W 66/68	Silty clay	Quaternary	IIIA	FA	408–418
27R 3W 8/10	Sandy silt	Quaternary	IIIA	FA	437–447
32R 7 29/31	Silty clay	0.126–0.53	IIIB	FA	485–495
U1520D
2H 7W 67/69	Silty sand	0.011	I	AD	14.15–15.15
4H 5W 104/106	Silty sand	0.011–0.29	I	AD	30–32.5
8H 5W 55/57	Sand	0.011–0.53	I	AD	68–69
14H 3W 54/56	Silty clay	Quaternary	II	AD	121–122
28F 4W 33/35	Sandy silt	Quaternary	III	AD	236.7–238
35X 1W 66/68	Silty clay	0.53	III	AD	366.6–368
44X 1W 83/85	Silty clay	0.53–0.62	III	AD	443.2–444.7
U1520C
5R 1W 72/74	Marl	10.66	IV	AD	670.4–671.9
5R 1W 105/107	Marl	10.66	IV	AD	670.4–671.9
13R 2W 51/53	Marl	24.67	IV	AD	748.6–750.1
19R 3W 34/36	Muddy chalk	54.4	IV	BD	807.6–809
21R 1W 71/73	Chalk	55	IV	BD	823.9–825.41
24R 1W 91/93	Silty claystone	< 72	V	BD	852.6–853.6
31R 1W 58/60	Coarse sandstone	Cretaceous	V	BD	919.7–920.6
41R 1W 119/121	Siltstone	100	VI	BD	1015.7–1017.2
44R 1W 116/118	Basalt	Cretaceous	VI	BD	1044.5–1045.45

Numeric age is based on the fossils present in the samples; the period is present when no approximation is available. The position corresponds to forearc (FA), above the décollement (AD), and below the décollement (BD) based on Barnes et al. (2020). Sediment description, age, unit, and depth interval after Barnes et al. (2019); Saffer et al. (2019). Bold italicised samples correspond to those selected for trace element concentrations and Sr-Nd-Pb isotope analyses in addition to major elements.

All samples, provided as 10 cm³ splits of the IODP drillcores, then sub-sampled and were powdered using an agate mortar and pestle, and analysed for their major element compositions. A subset of ten representative samples was selected for trace element concentration and Sr-Pb-Nd isotope analysis, including three samples from the accretionary prism, two samples from above the décollement, and five samples from below the décollement (Figure 2; Table 1).

Analytical methods

Major element analyses (Table 2) were carried out with a Wavelength Dispersive X-ray Fluorescence (WD-XRF) S8 Tiger spectrometer from Bruker-AXS (Germany) at Massey University, New Zealand, using glass beads made with a sample:flux ratio of 1:10 ratio, with 12:22 X-Ray flux (35.3% Lithium Tetraborate and 64.7% Lithium Metaborate). Interference-corrected spectral intensities were converted to oxide concentrations using calibration curves consisting of natural standards. The long-term reproducibility of the oxide analyses was assessed using the basaltic OREAS 24c and granodioritic OREAS 24b reference materials (OREAS, Australia). The 1σ reproducibility is typically ±0.5%–1% for major oxides and better than ±3% for minor oxides, except for P₂O₅ (±5%–10%).

Table 2. Major elements results from IODP sites U1518 and U1520. All results are in wt.%

Sample name	Unit	SiO2	Al2O3	TiO2	MnO	Fe2O3	MgO	CaO	Na2O	K2O	P2O5	LOI	Total
U1518E 26X 6W 69/71	IA	57.71	13.85	0.61	0.05	4.49	1.93	6.70	2.92	2.53	0.19	9.06	100.10
U1518F 5R 1W 12/14	IB	59.63	13.90	0.60	0.06	4.48	1.82	5.95	2.93	2.49	0.20	8.10	100.20
U1518F 9R 5W 24/26	IB	62.23	13.53	0.56	0.05	4.12	1.66	4.79	3.10	2.43	0.18	7.21	99.90
U1518F 11R 2W 47/49	IB	62.76	13.09	0.53	0.05	3.89	1.63	5.39	2.94	2.40	0.17	7.18	100.08
U1518F 12R 2 18/20	IB	59.30	16.64	0.72	0.07	5.93	2.34	2.88	3.09	3.30	0.22	5.60	100.14
U1518F 13R 1W 59/61	II	61.34	15.94	0.70	0.07	5.21	2.17	2.80	3.24	2.99	0.22	5.33	100.04
U1518F 14R 1W 133/135	II	59.65	14.20	0.61	0.05	4.51	1.88	5.72	3.03	2.54	0.18	7.77	100.22
U1518F 16R 4W 24/26	II	58.03	14.60	0.63	0.05	4.67	1.89	6.29	2.98	2.64	0.18	8.23	100.25
U1518F 19R 3W 33/35	II	58.80	14.38	0.63	0.06	5.16	1.99	5.06	2.98	2.61	0.19	8.33	100.25
U1518F 24R 2W 66/68	IIIA	58.80	14.78	0.65	0.06	5.40	2.06	4.75	3.00	2.66	0.20	7.97	100.37
U1518F 27R 3W 8/10	IIIA	55.26	14.36	0.62	0.06	4.85	1.99	7.42	2.88	2.66	0.19	9.97	100.29
U1518F 32R 7 29/31	IIIB	59.24	14.78	0.66	0.07	4.94	1.96	5.11	3.16	2.74	0.20	7.23	100.17
U1520D 2H 7W 67/69	I	72.06	11.08	0.38	0.04	2.57	1.03	3.45	3.31	2.07	0.19	3.96	100.17
U1520D 4H 5W 104/106	I	72.83	10.94	0.39	0.04	2.56	1.05	3.23	3.24	1.99	0.13	3.82	100.23
U1520D 8H 5W 55/57	I	72.58	10.80	0.32	0.04	2.31	0.98	3.98	3.25	1.98	0.12	4.04	100.45
U1520D 14H 3W 54/56	II	58.70	15.25	0.66	0.06	5.05	2.14	4.51	3.17	2.88	0.19	7.66	100.30
U1520D 28F 4W 33/35	III	61.76	14.39	0.56	0.05	4.13	1.72	4.56	3.39	2.91	0.17	6.72	100.38
U1520D 35X 1W 66/68	III	70.08	12.31	0.47	0.04	3.07	1.05	2.98	3.68	2.14	0.16	3.58	99.71
U1520D 44X 1W 83/85	III	61.75	15.44	0.68	0.06	4.75	2.00	3.49	3.35	2.93	0.21	5.39	100.10
U1520C 5R 1W 72/74	IV	26.37	7.66	0.32	0.09	2.95	1.29	31.79	1.40	0.74	0.14	27.24	100.00
U1520C 5R 1W 105/107	IV	24.59	7.11	0.29	0.08	2.74	1.20	33.26	1.37	0.62	0.14	28.31	99.71
U1520C 13R 2W 51/53	IV	30.13	8.14	0.41	0.10	3.11	1.55	28.94	1.25	1.32	0.14	24.92	100.02
U1520C 19R 3W 34/36	IV	1.86	0.49	0.02	0.04	0.22	0.30	51.79	0.07	0.02	0.07	42.91	97.83
U1520C 21R 1W 71/73	IV	2.38	0.58	0.02	0.05	0.26	0.35	51.63	0.06	0.02	0.09	42.66	98.12
U1520C 24R 1W 91/93	V	51.01	14.55	3.31	0.24	13.07	4.12	1.84	2.43	2.08	0.94	6.79	100.38
U1520C 31R 1W 58/60	V	30.30	10.02	2.18	0.15	9.12	3.73	21.92	1.88	1.58	0.93	18.60	100.45
U1520C 41R 1W 119/121	VI	56.51	15.05	0.87	0.03	6.53	5.22	3.41	2.54	1.76	0.30	7.73	100.04
U1520C 44R 1W 116/118	VI	55.33	18.15	1.36	0.05	5.23	1.53	4.66	4.44	5.69	0.95	3.37	100.76

Trace element concentration and Sr-Pb-Nd isotope analysis

Trace element concentration and isotopic analyses were performed at the Centre for Trace Element Analysis, University of Otago, New Zealand. Selected samples were analysed for their trace element concentrations using inductively coupled plasma mass spectrometry with a quadrupole mass analyser (Q-ICP-MS) and a 7900 instrument (Agilent Technologies, USA). Strontium and Pb isotope analyses were performed by multiple-collector ICP-MS (MC-ICP-MS), using a Nu Plasma-HR instrument (Nu Instruments Ltd, UK). For all samples, aliquots of powdered material of ∼0.5 g were taken for combined trace element concentration and Sr-Pb-Nd isotope analyses. All samples were chemically processed within ISO 4 ducted laminar flow workstations (ISO 5 for Nd separation) housed in an ISO 5 metal-free cleanroom. All reagents used (HF, HNO₃, and HCl) were purified in-house using Teflon and Quartz sub-boiling distillation, and ultra-high purity H₂O was dispensed from a Milli-Q Element water purification system (Millipore, Ltd, USA).

Each sample was digested sequentially on a hotplate using HCl-HNO3-HF following routine procedures (Stirling et al. 2005, 2006, 2007; Corella Santa Cruz et al. 2023) for the concentration determination of all reported elements except Zr and Hf, due to their mass balance being dominated by zircon that is not digested to completion using standard hotplate procedures. Zr and Hf were instead quantified on melted ultra-clean lithium metaborate flux discs. Total procedural replicates gave concentrations that were in agreement with expected values, within their analytical uncertainties (Appendix A). Values for limits of detection (LOD) were 5 µg/g for Ni, Zn, Ga, Rb, Sr, Zr, Cd, Sn, Sb, and Ba, 3 µg/g for V and Cr, 2 µg/g for Co, Y and Nb, 1 µg/g for Li, 0.5 µg/g for Cs, Pb, Th, and U, 0.1 µg/g for Be and Ta, 0.06 µg/g for REE, Hf and W.

Separation of Sr and Pb from the sample matrix was performed using ion exchange chromatographic columns loaded with 250 µl of Sr.spec resin (100–150 µm) (Eichrom Technologies, USA) modified from Shaw et al. (2011); Pin et al. (2014); Scholes et al. (2016), while Nd separation was performed following a protocol modified from Pin and Zalduegui (1997); Baker et al. (2002). Procedural blanks were <60, <10 and <50 pg, all of which are negligible compared to the 300, 10 and 30 ng load sizes for Sr, Pb and Nd, respectively. For Sr isotope analysis, all ⁸⁷Sr/⁸⁶Sr ratios were corrected for instrumental mass fractionation using the ‘true’ ⁸⁶Sr/⁸⁸Sr value of 0.1194 and the exponential mass fractionation law (Hart and Zindler 1989; Habfast 1998), as well as for potential isobaric interferences from Kr and Rb across the Sr mass range. All ⁸⁷Sr/⁸⁶Sr results were normalised to the NIST SRM 987 Sr reference value of 0.710248 (McArthur et al. 2001). For Pb isotopic analysis, the ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb ratios were corrected for instrumental mass fractionation using thallium (Tl) normalisation and the exponential mass fractionation law using a NIST SRM-997 Tl value for ²⁰⁵Tl/²⁰³Tl of 2.3875 and a Pb/Tl ratio of 5. All Pb isotope ratios are reported relative to reference values for NIST SRM 981 Pb of ²⁰⁶Pb/²⁰⁴Pb = 16.9406 ± 0.0003, ²⁰⁷Pb/²⁰⁴Pb = 15.4957 ± 0.0002, and ²⁰⁸Pb/²⁰⁴Pb = 36.7184 ± 0.0007 (Yuan et al. 2016). All ¹⁴³Nd/¹⁴⁴Nd values were corrected for instrumental mass fractionation adopting a ‘true’ ¹⁴⁶Nd/¹⁴⁴Nd ratio of 0.7219 and applying the exponential mass fractionation law, as well as for potential interferences from Sm across the Nd mass range. All corrected Nd isotopic ratios were normalised to the international reference material JNdi-1 using a ¹⁴³Nd/¹⁴⁴Nd ratio of 0.512115 after Tanaka et al. (2000). All uncertainties for individual samples are reported as ± 2 SE, unless stated otherwise, and include the analytical uncertainties of the sample and any normalising, standards combined in quadrature, using standard error propagation techniques.

Analytical performance for Sr, Pb and Nd isotope analysis was initially assessed on the basis of replicate analyses of the primary standards SRM 987 Sr, SRM 981 Pb and JNdi-1, respectively, yielding average compositions of 0.710247 ± 0.000013 (2 SD, n = 18) for ⁸⁷Sr/⁸⁶Sr, 36.7184 ± 0.0032, 15.4957 ± 0.0013 and 16.9406 ± 0.0012 (2 SD, n = 20) respectively for ²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb and ²⁰⁶Pb/²⁰⁴Pb, and 0.512115 ± 0.000009 (2 SD, n = 14) for ¹⁴³Nd/¹⁴⁴Nd. The matrix-matched BHVO-2, BCR-2, AGV-2 (USGS, United States), JR-2, and JG-3 (GSJ, Japan) reference materials were also used for quality control purposes, giving Sr, Pb and Nd isotopic compositions that are identical, within error, to reported values (Appendix A). The results for all trace element concentrations and Sr-Pb-Nd isotope compositions are presented in Table 3.

Table 3. Trace element concentrations and Pb-Sr-Nd isotopic ratios for IODP sites U1518 and U1520.

	Unit I	Unit II	Unit III	Unit III	Unit IV	Unit IV	Unit V	Unit V	Unit VI	Unit VI
	U1518F 11R 2W 47/49	U1518F 12R 2 18/20	U1518F 27R 3W 8/10	U1520D 44X 1W 83/85	U1520C 5R 1W 72/74	U1520C 21R 1W 71/73	U1520C 24R 1W 91/93	U1520C 31R 1W 58/60	U1520C 41R 1W 119/121	U1520C 44R 1W 116/118
Li (µg/g)	47.79	61.95	57.59	60.73	30.06	8.23	48.79	26.60	80.08	14.50
V	75.99	101.81	96.99	91.13	46.08	<3	86.59	76.83	47.26	40.84
Cr	38.88	53.83	46.28	46.83	14.06	<3	151.63	115.34	19.69	<3
Co	5.57	8.74	5.61	8.64	5.14	2.79	38.13	21.02	8.66	6.35
Ni	15.64	22.03	16.88	21.36	13.30	6.54	98.58	83.78	25.56	6.51
Zn	58.70	83.58	68.34	80.09	35.64	24.02	200.69	93.80	140.05	53.57
Ga	16.37	21.88	18.68	20.55	8.54	<5	25.16	18.87	23.78	27.06
Rb	89.45	127.88	104.95	115.65	39.50	<5	39.39	26.93	30.97	69.04
Sr	296.96	263.26	372.44	314.88	969.66	854.90	152.83	409.09	176.39	364.02
Y	15.90	21.41	18.35	21.76	13.84	10.93	40.37	28.51	21.82	24.50
Zr	152.97	135.59	117.91	157.25	46.06	5.86	531.44	399.62	288.82	499.40
Nb	6.96	10.59	8.56	10.34	3.13	<2	111.90	86.54	26.21	96.03
Cs	4.76	7.57	6.23	6.56	2.22	<0.5	0.83	0.54	0.81	0.68
Ba	458.35	589.43	463.89	548.22	442.22	1109.15	98.92	215.74	2207.21	561.72
La	18.48	28.46	22.02	29.10	11.16	7.72	91.04	63.76	25.73	82.79
Ce	39.08	58.70	46.65	60.35	23.75	10.23	160.54	120.91	48.65	143.99
Pr	4.59	6.66	5.40	7.03	2.57	1.71	19.95	14.12	6.64	15.09
Nd	17.42	25.47	20.42	26.78	10.52	6.97	76.76	54.63	26.57	51.41
Sm	3.39	5.21	3.97	5.39	2.22	1.56	13.99	9.90	5.35	8.14
Eu	0.82	1.12	0.92	1.12	0.54	0.40	4.18	2.91	1.41	2.45
Gd	3.14	4.56	3.59	4.66	2.41	1.53	11.73	8.34	4.75	6.39
Dy	2.77	3.85	3.22	3.94	2.17	1.46	8.18	5.92	3.79	4.58
Ho	0.56	0.74	0.64	0.77	0.44	0.31	1.49	1.05	0.76	0.84
Er	1.64	2.11	1.91	2.18	1.34	0.85	3.93	2.82	2.13	2.55
Tm	0.23	0.30	0.26	0.31	0.21	0.13	0.52	0.36	0.29	0.35
Yb	1.57	1.91	1.75	2.02	1.30	0.82	3.21	2.27	1.93	2.09
Lu	0.25	0.28	0.27	0.30	0.18	0.12	0.48	0.34	0.30	0.36
Hf	3.91	4.30	3.43	4.25	1.24	0.14	10.26	8.07	7.22	11.29
W	1.03	1.57	1.23	1.57	0.64	0.24	0.51	0.89	0.58	0.63
Pb	11.38	18.62	13.10	18.29	2.46	1.38	4.19	3.32	15.79	4.72
Th	7.48	11.21	9.05	11.02	3.37	0.69	9.92	7.75	9.21	12.13
U	1.80	2.40	2.19	2.55	0.63	<0.5	1.10	1.69	1.31	2.90
^206Pb/^204Pb	18.8876	18.8746	18.8881	18.8942	19.0054	18.8162	20.7920	21.0496	18.9035	21.0324
(±2SE)	0.0023	0.0023	0.0025	0.0016	0.0021	0.0022	0.0028	0.0026	0.0019	0.0026
^207Pb/^204Pb	15.6511	15.6568	15.6538	15.6567	15.6572	15.6616	15.7549	15.7687	15.6353	15.7691
(±2SE)	0.0011	0.0011	0.0013	0.0007	0.0009	0.0022	0.0011	0.0023	0.0008	0.0012
^208Pb/^204Pb	38.8219	38.8258	38.8343	38.8322	39.0067	38.8885	40.3204	40.3032	38.8172	40.5693
(±2SE)	0.0040	0.0040	0.0043	0.0027	0.0035	0.0056	0.0044	0.0057	0.0031	0.0044
^87Sr/^86Sr	0.709092	0.710953	0.709694	0.710082	0.708923	0.707843	0.708057	0.704792	0.707026	0.703878
(±2SE)	0.000020	0.000022	0.000020	0.000023	0.000021	0.000020	0.000019	0.000019	0.000025	0.000017
^143Nd/^144Nd	0.512510	0.512382	0.512456	0.512400	0.512479	0.512358	0.512886	0.512907	0.512688	0.512803
(±2SE)	0.000023	0.000021	0.000024	0.000015	0.000013	0.000015	0.000019	0.000016	0.000020	0.000014
ϵNd	−2.5	−5.0	−3.5	−4.5	−3.1	−5.5	4.8	5.2	1.0	3.2

All trace element concentrations are given in µg/g. Due to the presence of zircon, Zr and Hf were measured on melted ultra-pure flux discs, reported in italics. The ±2SE uncertainties for the Pb-Sr-Nd isotope results are indicated in parentheses. Present-day ϵNd relative to CHUR value of 0.512638.

Bulk composition calculations

The bulk composition of the subducting and accreting material was obtained following the method of Plank (2014), where the major and trace element concentrations together with the Sr-Pb-Nd isotopic compositions determined in the present study are integrated with lithological descriptions, logging data and geophysical observations. This was based on published data for sediment unit thicknesses and descriptions (Barnes et al. 2019; Saffer et al. 2019; Wallace et al. 2019) and on the identification of a décollement zone reported by Barnes et al. (2020). The bulk composition of sites U1518 and U1520, alongside the composition of material located below and above the current décollement of site U1520 were obtained. Subsequently, the material located in the accretionary prism and above the décollement were combined to represent the material that may be recycled via tectonic erosion. The results of these calculations are presented in Table 4. However, we note that the décollement zone is a dynamic feature and as such, it may have migrated over the evolution of the Hikurangi subduction system. Thus, there remains some uncertainty about the material located below and above the décollement in the past.

Table 4. Bulk compositions for IODP sites U1520 and U1518 as well as for the composition of the material below the décollement (BD) and above the décollement (AD) at site U1520, and the tectonically erodible material composite (EC).

	U1520	U1518	BD	AD	EC	New Zealand
SiO2	52.15	64.75	38.36	57.77	60.48	42.21
Al2O3	12.44	15.26	11.30	13.07	13.92	12.01
TiO2	0.93	0.65	2.13	0.54	0.58	0.74
MnO	0.10	0.06	0.17	0.08	0.07	0.15
Fe2O3^T	5.27	4.96	9.12	4.09	4.42	6.77
MgO	2.20	2.02	3.76	1.74	1.85	5.24
CaO	21.88	6.05	30.84	17.36	12.98	12.55
Na2O	2.68	3.25	1.96	2.97	3.08	1.70
K2O	2.02	2.78	1.60	2.21	2.43	1.85
P2O5	0.33	0.21	0.76	0.19	0.19	0.13
Li	45.19	52.20	32.32	49.85	50.76	32.13
V	70.66	84.83	59.01	75.15	78.90	170.00
Cr	48.21	42.77	87.74	35.20	38.13	285.00
Co	10.67	6.00	20.69	7.40	6.86	36.80
Ni	29.34	16.81	62.40	18.50	17.85	101.50
Zn	76.11	64.47	112.99	64.33	64.39	115
Ga	16.51	19.77	17.78	16.29	17.64	-
Rb	73.35	98.53	28.52	88.65	92.48	46.80
Sr	512.32	312.08	408.59	547.10	456.01	557.00
Y	21.06	17.26	27.69	18.95	18.29	24.80
Zr	174.41	141.59	352.92	117.81	127.03	59
Nb	23.48	7.86	72.06	7.78	7.81	4.75
Cs	3.91	5.51	0.58	5.02	5.21	2.15
Ba	521.00	477.06	530.64	510.62	497.62	577.00
La	31.49	20.72	58.73	29.10	25.85	11.90
Ce	62.74	43.63	105.63	47.36	45.92	26.50
Pr	7.24	5.07	12.79	5.45	5.30	3.29
Nd	27.85	19.26	49.03	21.02	20.33	13.30
Sm	5.40	3.78	8.93	4.26	4.07	3.32
Eu	1.33	0.88	2.63	0.91	0.90	0.93
Gd	4.76	3.44	7.54	3.86	3.70	3.37
Dy	3.82	3.03	5.42	3.31	3.20	3.38
Ho	0.74	0.61	0.99	0.77	0.71	0.71
Er	2.07	1.77	2.68	1.88	1.84	1.98
Yb	1.86	1.66	2.21	1.76	1.72	1.95
Lu	0.27	0.26	0.33	0.26	0.26	0.30
Hf	4.13	3.84	3.55	3.18	3.44	1.62
W	1.07	1.15	0.59	1.24	1.20	-
Pb	10.60	12.81	4.25	12.70	12.75	7.40
Th	7.99	8.38	7.24	8.30	8.33	3.36
U	1.69	1.98	1.19	1.87	1.91	1.14
^206Pb/^204Pb	19.2719	18.8852	20.4826	18.9018	18.8953	18.8910
^207Pb/^204Pb	15.6753	15.6529	15.7355	15.6567	15.6552	15.6360
^208Pb/^204Pb	39.1417	38.8260	39.9833	38.8442	38.8371	38.7540
^87Sr/^86Sr	0.70892	0.70949	0.70752	0.70935	0.70939	0.70928
^143Nd/^144Nd	0.51250	0.51247	0.51286	0.51241	0.51243	0.51259

Major elements are in wt.% and normalised to 100% on an anhydrous basis. Trace elements are reported in µg/g. Data from the New Zealand margin are from Plank (2014).

Results

For simplicity, the results are presented in three groups instead of individual units, as hemipelagic and terrigenous facies (Units I–III), pelagic facies (Unit IV), and the Hikurangi Plateau associated volcaniclastic facies (Units V and VI). For the compilation of results for the individual samples within each unit, the reader is referred to Tables 2 and 3. The major oxide content is reported as weight % and in the text is referred to the data re-calculated relative to 100% on a volatile free basis.

Even though the terrigenous facies is subdivided into three different units, their geochemical characteristics are similar due to analogous detrital lithologies and depositional environments (Figure 3). SiO₂ content varies from 60 to 80 wt.% but commonly is around 65 wt.%. The content of Al₂O₃ varies from 11–18 wt.%, while TiO₂ and MgO contents are both low with values of <1 wt.% and ≤2 wt.%, respectively. The concentrations of K₂O and Na₂O vary from 2 to 4 wt.% and 3 to 4 wt.% respectively (Figure 3). The terrigenous facies also has the least variable trace element concentrations between samples, with the highest concentrations being observed for Rb and Cs, and intermediate concentrations being recorded for the remaining elements (Figure 4). The Sr-Pb-Nd isotope signatures are also well constrained, with ratios from 18.87 to 18.89 for ²⁰⁶Pb/²⁰⁴Pb, from 15.65 to 15.66 for ²⁰⁷Pb/²⁰⁴Pb, and from 38.82 to 38.83 for ²⁰⁸Pb/²⁰⁴Pb, while ⁸⁷Sr/⁸⁶Sr varies from 0.70909 to 0.71095, and ¹⁴³Nd/¹⁴⁴Nd ranges from 0.51238 to 0.51251 (ϵNd from −5 to −2.5) (Figure 5).

Figure 3. Major element content (wt.%) against depth in mbsf. All oxides are plotted normalised to 100% on a volatile free basis. Décollement formation after Barnes et al. (2020).

Figure 4. Trace and major element concentrations of the three different facies normalised to the reported bulk composition of the Hikurangi margin by Plank (2014).

Figure 5. Pb-Sr-Nd isotope ratios of the samples and calculated bulk compositions. Black circles correspond to samples from site U1518 (accretionary prism), and grey circles correspond to samples from site U1520 (subducting plate). The bulk composition of sites U1520 and U1518, as well as above and below décollement (AD and BD, respectively) of the subducting plate, as well as GLOSS-II and the bulk composition of the Hikurangi Margin after Plank (2014), are shown as lines for reference. Analytical uncertainty is within symbol size.

The pelagic facies (Unit IV) shows a large variation in major element concentrations due to differences in the types of sediments comprising this unit (Figure 3). An increasing detrital input causes SiO₂ to increase from around 5 wt.% for calcareous sediments to between 30 and 40 wt.% for siliceous sediments, while Al₂O₃ increases from around 1.5 wt.% for the former to 11 wt.% for the latter. Similar distributions are seen for TiO₂, MgO, Fe₂O₃^T, K₂O, and Na₂O, while CaO has the highest concentrations of the entire sequence. Trace element concentrations also show larger variations in the pelagic facies compared to the terrigenous facies, and give the lowest concentration of the sequence for virtually all elements, except Sr, which has the highest concentrations in the sequence. In Pb isotope space, the pelagic facies is slightly more radiogenic but similar in composition to the terrigenous facies, with values ranging from 18.816 to 19.005 for ²⁰⁶Pb/²⁰⁴Pb, 15.657 to 15.661 for ²⁰⁷Pb/²⁰⁴Pb and 38.888 to 39.007 for ²⁰⁸Pb/²⁰⁴Pb (Figure 5). Both ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd ratios are slightly less radiogenic in the pelagic facies than in the terrigenous sequence, varying from 0.70784 to 0.70892 and 0.51235 to 0.51248 (ϵNd from −5.5 to −3.1), respectively (Figure 5), with the Nd isotopic compositions spanning the variation shown by the terrigenous sequence.

The oldest units correspond to the sequences associated with the Hikurangi Plateau (Units V and VI) where SiO₂ content varies from 35 to 65 wt.%, Al₂O₃ varies from 12 to 19 wt.%. TiO₂, MgO, and Fe₂O₃^T are all higher in Unit V than Unit VI and correspond to the highest values of the entire sequence, while Na₂O and K₂O show the opposite behaviour and are higher in Unit VI than Unit V (Figure 3). This facies has the highest trace element concentrations of the sequence, except for Ba and Sr, which have the lowest contents (Figure 4). In Pb isotopic space, this facies is extremely variable and more radiogenic than the pelagic and terrigenous facies, with values ranging from 18.816 to 21.050 for ²⁰⁶Pb/²⁰⁴Pb, 15.635 to 15.769 for ²⁰⁷Pb/²⁰⁴Pb, and 38.817 to 40.569 for ²⁰⁸Pb/²⁰⁴Pb (Figure 5). Only one sample from Unit VI, corresponding to a siltstone located at the upper limit of the mafic material of the Hikurangi Plateau, has a similar Pb isotope composition to the pelagic and terrigenous facies. Both ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd show large variations with values ranging from 0.70388 to 0.70806 and 0.51235 to 0.51291 (ϵNd from +1 to +5.2), respectively (Figure 5).

Discussion

The thickness of the units on the subducting plate found in site U1520 differ from those used by Plank (2014), causing differences in the calculated geochemical bulk composition (Table 4 and Figures 5 and 6). The estimation by Plank (2014) assumed that 50% of the material in the subducting plate is made of the volcaniclastic sequence, but this is only around 30% in site U1520. Detrital units correspond to around 50% of the U1520 sequence and carbonate units to around 20%. Higher contents of SiO₂, CaO, Na₂O, and K₂O along with lower Fe₂O₃^T and MgO in site U1520 are due to the smaller proportion of the volcaniclastic material when compared to the estimate of Plank (2014).

Figure 6. Multi-element diagram of the calculated bulk composition normalised to the Hikurangi bulk composition by Plank (2014). The bulk composition of the sites U1518 and U1520 are shown alongside the composition of the material above and below the décollement in the subducting plate (AD and BD, respectively), the composition of material prone to tectonic erosion (AD and U1518 composite, shown as EC), and GLOSS-II after Plank (2014).

Trace element concentrations are also different. The elements most abundant in the volcaniclastic sequence, such as Cr, Co, and Ni, are significantly lower than in the Plank (2014) composite, whereas others that are more abundant in the pelagic and terrigenous facies, such as Rb, Cs, Ce and Li, are all higher in U1520 (Figure 4 and 6). Other elements, like Ba and Sr, have similar concentrations in U1520 and in the Plank (2014) composite. Isotopically, ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb are more radiogenic compared to the estimate of Plank (2014) (Tables 3 and 4, Figure 5). Large and variable isotopic compositions are also seen in ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd, but their weighted averages of 0.70892 and 0.51250 respectively, are slightly less radiogenic when compared to the corresponding Plank (2014) estimated values of 0.70928 and 0.51259 (Figure 5).

Forearc site U1518, has higher SiO₂ and Al₂O₃ than the trench site U1520 due to the influence of the detrital sequence, with lower TiO₂ and CaO from the volcanic material and carbonated sequences that are not involved in the accretionary prism (Figures 3 and 4). For the same reason, Cr, Co, Ni, Sr, and Ba are less abundant at site U1518 (Figure 4). However, site U1518 sediments have much more homogenous and less radiogenic Pb isotope ratios (²⁰⁶Pb/²⁰⁴Pb of ∼18.885, ²⁰⁷Pb/²⁰⁴Pb of ∼15.653, and ²⁰⁸Pb/²⁰⁴Pb of ∼38.826), while also having more radiogenic Sr and less radiogenic Nd isotope ratios (⁸⁷Sr/⁸⁶Sr of ∼ 0.70949, ¹⁴³Nd/¹⁴⁴Nd of ∼ 0.51247) compared to the estimate of Plank (2014) and U1520 (Figure 5). Although the elemental concentrations between sites U1520, U1518, and GLOSS-II behave similarly (Figure 6), isotopically, GLOSS-II has a signature which is closer to that of the accretionary prism (U1518) and different to that of the subducting plate (U1520) (Figure 5). Thus, the use of GLOSS-II as an equivalent to the subducting sediments in the Hikurangi margin is not recommended.

The development of a décollement zone has been reported at site U1520 by Barnes et al. (2020). Dividing this site into the material located above and below the décollement may have a greater significance in the present-day tectonic distribution than the bulk composition of the site (Figure 6). This can help to elucidate if the subducted material, the tectonically eroded material, or both, are involved in magma generation of the associated volcanic arc of the TVZ. The composition of the material located below the décollement is controlled by both the pelagic and volcaniclastic units. For example, the low SiO₂ and high CaO (38 and 31 wt.%, respectively) are controlled by the carbonate unit, while the high TiO₂ (2.13 wt.%) is influenced by the volcaniclastic sequence (Figure 3). In general, the trace element concentrations are dominated by materials from the volcaniclastic sequence, which have the highest contents of Ni, Co, and Cr (Figure 4). The material located below the décollement zone is thought to be currently subducting, but large discrepancies are evident when compared to the Plank (2014) composite (Figure 6). The Pb isotope composition of the material below the décollement is highly variable, but the bulk composition is around 20.4826 for ²⁰⁶Pb/²⁰⁴Pb, 15.7355 for ²⁰⁷Pb/²⁰⁴Pb and 39.9833 for ²⁰⁸Pb/²⁰⁴Pb, i.e. much more radiogenic than the Plank (2014) estimate (Table 4, Figure 5). Both Sr and Nd isotope compositions are highly variable and less radiogenic than those reported by Plank (2014) (Table 4, Figure 5).

On the other hand, the material located above the décollement has a weighted average composition that is SiO₂-rich (58 wt.%), with lower CaO and TiO₂ (17 and 0.54 wt.%, respectively) due to the more detrital nature of these units, with less carbonate material and without input of material from the Hikurangi Plateau. The trace elements related to mafic volcanism, like Ni, Co, and Cr, have lower concentrations here compared to those below the décollement, but higher levels of Large-Ion Lithophile Elements (LILE), such as Li, Rb, Cs, and Pb (Figure 6). The material above the décollement has a much more limited and less radiogenic range in Pb isotope ratios (²⁰⁶Pb/²⁰⁴Pb of ∼18.902, ²⁰⁷Pb/²⁰⁴Pb of ∼15.657 and ²⁰⁸Pb/²⁰⁴Pb of ∼38.844) than the weighted average of the material below the décollement. On the other hand, the ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd isotope ratios are more radiogenic than the estimate by Plank (2014) and the material below the décollement (Figure 5).

Since the material located above the décollement is eventually accreted to the accretionary prism and may be subsequently eroded tectonically, a composite of the accretionary prism (U1518) and the material above the décollement in site U1520 may be useful (Figure 6; Table 4, referred to here as EC for erodible crust). The material above the décollement shares the same major oxide and trace element characteristics as the material in the accretionary prism. The biggest difference is a higher concentration of CaO and Sr in the material above the décollement, thus the material prone to tectonic erosion is similar to both sequences in all elements and major oxides other than these two (Figure 6). This composite has a weighted average of ∼18.895 for ²⁰⁶Pb/²⁰⁴Pb, ∼15.655 for ²⁰⁷Pb/²⁰⁴Pb, ∼38.837 for ²⁰⁸Pb/²⁰⁴Pb, ∼0.70939 for ⁸⁷Sr/⁸⁶Sr and ∼0.51243 for ¹⁴³Nd/¹⁴⁴Nd (Figure 5). The homogeneity seen in the major oxide and trace element characteristics of the materials prone to tectonic erosion is also observed in the Sr-Pb-Nd isotope characteristics, contrasting with the large variations of the material located below the décollement. Although the décollement zone is a dynamic feature that is prone to migrate over time, the geochemical homogeneity seen in the material that has already been accreted (U1518) and material that is yet to be accreted (above décollement) (Figures 3–5), indicates that the structural characteristics of the subducting plate in the Hikurangi margin may not have significantly changed over time. However, this remains a hypothesis that cannot be tested based on the data of the present study.

ODP sites 1123 and 1124, used by Plank (2014) to estimate the bulk composition of the subducting sequence, are located up to 500 km away from the trench. This newly calculated estimate is based on IODP sites U1518 and U1520, located on the Hikurangi trench, and it is thus a more accurate representation of the trench sediments relative to the previous estimate. The information about the lithological units and their thicknesses in these two sites is also well-constrained and slightly different from that of sites 1123 and 1124, particularly in the proportion of the volcaniclastic sequence, which is smaller than previously estimated. Further, in 2014, there was no information regarding which portion of the sequence was accreted to the Hikurangi margin and which portion was in the process of being subducted. The information provided by the identification of the décollement by Barnes et al. (2020) in site U1520 helps to identify and separate the material that is going to be accreted from that which is going to be subducted. Due to the lack of information on the décollement when Plank’s estimate was calculated, using it to test recycling of subducted sediments in the TVZ may lead to erroneous results. Instead, with these new data it is now possible to test if recycling is occurring in the source of the TVZ in the form of subducting sediments from the subducting plate, tectonically eroded material from the accretionary prism, or both. These interactions can occur by the release of fluids and/or melts from the subducting material, and/or bulk mixing between the mantle and subducting material. The data presented here can also help to better constrain differences between the different volcanic zones in the Tonga-Kermadec-Hikurangi subduction system.

Conclusions

New geochemical datasets for major oxide compositions, trace element concentrations, and Sr-Pb-Nd isotope ratios of the material in the Hikurangi margin were obtained for drill sites U1518 and U1520 retrieved during IODP Expedition 372B/375. The material in the subducting plate can be divided into two groups, namely material that is being subducted directly, and material that is accreted and may only be subducted subsequently by tectonic erosion, improving previous estimates. These two groups show important differences in their compositional ranges for trace element concentrations and particularly their isotopic compositions. The material below the décollement shows large variations in Sr-Pb-Nd isotope signatures, whereas the material prone to tectonic erosion is much more homogeneous and less radiogenic in Sr-Pb-Nd isotope space. The geochemical differences between material above and below the décollement presented here may be employed for assessment of the composition and relative contribution of the subducted components in studies of elemental cycling through arc magmatism at the Hikurangi margin.

Acknowledgements

We thank the scientists involved in the IODP Expedition 372B/375 for all the information publicly available. Samples were obtained from the IODP Gulf Coast Repository in College Station/TX, and we gratefully acknowledge the help of the curatorial staff (sample request number 081373IODP). We thank Masumi Mikoshiba from the Geological Survey of Japan for providing the JNdi-1 standard oxide. We thank Sophie Gangl for her help with laboratory work. We gratefully acknowledge Katherine Maier for editorial handling and feedback, and Tod Waight and an anonymous reviewer for their constructive reviews. CRCSC gratefully acknowledges the doctoral scholarship granted by Consejo Nacional de Ciencia y Tecnología (CONACyT, Número de apoyo 739571). GFZ acknowledges support through a Catalyst Seeding Grant (CSG-MAU1901). SMS acknowledges support from US NSF grant EAR 19-21624.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Appendix A containing the data quality of this study is openly available in figshare at https://doi.org/10.6084/m9.figshare.24483997.v1.

Additional information

Funding

This work was supported by Consejo Nacional de Ciencia y Tecnología [grant number: 739571]; National Science Foundation [grant number: EAR 19-21624]; Royal Society Te Apārangi [grant number: CSG-MAU1901].