Where is the boundary between New Zealand’s western and eastern provinces? A case study in describing terrane relationships using cladistic methods.

ABSTRACT

A cladistic analysis of 21 geological characters taken from the literature is used to determine the relationship between the Drumduan Terrane, Takaka Terrane (Western Province), and Brook Street Terrane (Eastern Province) of Nelson, New Zealand. We present the analysis as a case study to illustrate the utility of a cladistic approach to describing terrane relationships. The results of the analysis support the placement of the Drumduan Terrane as the eastern boundary of the Western Province providing the Pepin Group and Echinus Granite are the basement of the overlying Marybank Formation sediments rather than being tectonically juxtaposed. The importance of establishing whether the Drumduan Terrane and its proposed correlatives in Fiordland and Stewart Island are part of the Western or Eastern Province is that natural groupings can be identified and tectonic reconstructions can be free of artifice.

KEYWORDS:

• Drumduan Terrane
• natural groupings
• monophyly
• Eastern Province
• Western Province
• cladistics

Introduction

New Zealand’s pre-Cenozoic basement terranes are divided into a Western Province – composed of the Palaeozoic Buller and Takaka Terranes that had amalgamated by the Middle Devonian and sutured to Gondwana during the Tuhua Orogeny – and an Eastern Province formed in response to Permian – Cretaceous convergent margin tectonics and consisting of a number of tectonostratigraphic terranes that include volcaniclastic turbidites (greywacke terranes), marginal basins (Murihiku Terrane), island arcs (Brook Street Terrane), and slices of oceanic crust (Dun Mountain) (Mortimer et al. 2014). Much of the Eastern Province is exotic and was accreted to the Western Province foreland during the Lower Cretaceous Rangitata Orogeny. Although it has been established that the Eastern and Western Provinces are natural areas on a national scale (Michaux et al. 2018), the Drumduan Terrane was not included in the analysis. The Drumduan Terrane has previously been assigned to both Provinces making its position equivocal. By establishing natural (monophyletic) geological groupings, relationships can be described with some confidence and their tectonic histories more accurately reconstructed.

The nature of the boundary between the Western and Eastern Provinces has undergone a series of revisions since Landis and Coombs (1967) first introduced the concept of a Median Tectonic Line (MTL) separating the two. Landis and Coombs (1967) interpreted the MTL as a steeply dipping fault system separating a high temperature/low pressure metamorphic belt in the west (i.e. the modern Western Province) from a high pressure/low temperature belt in the east (i.e. the modern Eastern Province). Subsequent work showed the concept of a paired metamorphic belt to be oversimplified and that the MTL represented a tectonic boundary between terranes (Bishop et al. 1985). The term Median Tectonic Zone (MTZ) was introduced by Frost and Coombs (1989) when the MTL was reinterpreted as a complex magmatic belt rather than a narrow fault zone (Bradshaw 1993). The MTZ is composed of a suite of Palaeozoic to Mesozoic plutons and arc-related sedimentary terranes that are thought to represent a dismembered Carboniferous volcanic arc/pluton complex (Kimbrough et al. 1993). Fragments of the arc are found in Nelson (Rotoroa Complex, Drumduan Terrane, Teetotal Group (Figure 1)), Fiordland (Darran Complex, Loch Burn Formation, Largs Terrane) and Stewart Island (Paterson Group) (Scott 2013; Table 1). While the contact between the Jurassic metasedimentary units and Carboniferous basement is unseen in Nelson, the evidence for such a relationship is clearer in Fiordland and Stewart Island (Allibone and Tulloch 2004; Ewing et al. 2007; Scott et al. 2008). Drumduan Terrane fragments can be correlated across the Alpine Fault because each conforms to a generalised structure of arc-related sediments overlying Carboniferous granites and juxtaposed with Mesozoic plutons.

Figure 1. Locality map showing the basement terranes of Nelson referred to in the text. Western Province = stars; Buller Terrane = closed stars, Takaka Terrane = open stars. Eastern Province = diagonal hatching. Drumduan Terrane (s. l.) = horizontal hatching and names in bold. DFZ = Delaware Fault Zone, SPB = Separation Point Batholith.

Table 1. Geological characters used in this study.

Occurrence of Cambrian – Early Palaeozoic metasedimentary rocks Occurrence of Cambrian rocks Intruded by I- and S-type Devonian plutonic rocks Intruded by S-type Carboniferous granites Intruded by I-type Carboniferous granites Radiogenic composition of Carboniferous plutonic rocks Intruded by c220 Ma plutons (early Darran Suite) Intruded by <168 Ma plutons (late Darran Suite) Intruded by Separation Point Suite granitoids Pre-Ordovician inherited zircon in any plutonic rock Affected by Devonian and Carboniferous penetrative metamorphism Presence of Devonian metasedimentary rocks Presence of Permian metasedimentary rocks Presence of Triassic metasedimentary rocks Presence of Jurassic – Early Cretaceous metamorphosed volcanic and volcaniclastic rocks High T/Low P metamorphism High P/Low T metamorphism Precambrian – Lower Palaeozoic detrital zircons Devonian – Carboniferous detrital zircons Permian – Triassic detrital zircons Jurassic detrital zircons

Notes: Characters 1 – 17 from Scott (2013); Characters 18 – 20 (zircon age profiles) from Wysoczanski et al. (1997) (Takaka Terrane), Adams et al. (2002) (Brook Street Terrane), Ewing et al. (2007) (Loch Burn Formation) and Allibone and Tulloch (2004) (Paterson Group).

Mortimer et al. (1999) coined the term Median Batholith following coordinated research and fieldwork on rocks of the MTZ, and argued that as the MTZ is predominantly plutonic (92% of the total exposure) and less tectonised than previously thought, with much of the movement on faults occurring long after pluton emplacement, the use of the term batholith would be more appropriate. The MTZ occupies the eastern half of the Median Batholith, with the plutons of the Western Province and those stitching the MTZ to the Takaka Terrane, such as the Separation Point Batholith in Nelson (Figure 1), making up the western half. In the view of Mortimer et al. (1999) and others (e. g. Schwartz et al. 2017) the Median Batholith formed largely in situ. The genesis of the Median Batholith involved the episodic melting and emplacement of significant volumes of igneous rock from Late Devonian/Carboniferous times until an Early Cretaceous climax, a time span of at least 200 million years. While there are minor A-type granites, the majority of these plutons show subduction-related geochemistries (Kimbrough et al. 1993; Muir et al. 1998; Mortimer et al. 1999; Allibone et al. 2009; Scott et al. 2011). As Scott (2013) noted, the terms Median Batholith and MTZ are often not clearly distinguished in the literature, and often synonymised, but represent quite radically different conceptions viz an autochthonous suture zone versus an allochthonous terrane.

Apart from the igneous component of the Median Batholith, there are small volcaniclastic inliers within the MTZ. These include the Drumduan Terrane in Nelson, the Largs Terrane and Loch Burn Formation in Fiordland, and the Paterson Group in Stewart Island (Allibone and Tulloch 2004; Ewing et al. 2007; Scott et al. 2008; Scott 2013). Whether these terranes should be regarded as part of the Gondwanan foreland (i.e. part of the Western Province) or as part of the accretionary complex (i.e. Eastern Province) is still an unresolved issue. While many workers regard the Drumduan Terrane or its equivalents south of the Alpine Fault as part of the Eastern Province (Bishop et al. 1985; Coombs 1985; Scott et al. 2009; Scott 2013) there are others who have argued that it is part of the Western Province (Johnston et al. 1987; Kimbrough et al. 1993; Wandres and Bradshaw 2005). The purpose of this paper is to use cladistic methods to establish which Province the Drumduan Terrane is part of, thereby fixing the boundary between these two basements, and refining reconstructions of Gondwana’s Devonian/Carboniferous margin.

Materials

Takaka Terrane (Figure 1)

The New Zealand Western Province and equivalent sequences in East Antarctica, Tasmania and South-Eastern Australia (Cooper and Tulloch 1992; Flottman et al. 1993; Wysoczanski et al. 1997; Münker and Crawford 2000; Vandenberg et al. 2000; Bradshaw et al. 2009; Gibson et al. 2011; Crispini et al. 2014; Dowding and Ebach 2016) formed part of the Cambrian-Carboniferous Gondwana margin upon which later Eastern Province terranes were accreted. While there is little doubt the Western Province, which formed when the Buller and Takaka Terranes amalgamated in Late Devonian times, is a natural geological unit, there are a number of features that remain puzzling. For example, compared to similar sequences in Tasmania and east Antarctica (Bradshaw et al. 2009) where the turbidites are outboard of the arc complex, the arrangement is reversed in Nelson where the arc complex (Takaka Terrane) is outboard of the turbidites (Buller Terrane).

The Takaka Terrane outcrops in northwest South Island and is separated from the adjacent Buller Terrane by the Anatoki Thrust (Cooper 1989; Jongens 2006). The Takaka Terrane has a complex, dismembered structure that includes Cambrian quartzofeldspathic turbidites and associated arc volcanics, and a Late Cambrian to Early Devonian passive margin succession (Cooper 1989; Pound 1993; Roser et al. 1996; Münker and Cooper 1999; Bradshaw 2000; Münker and Crawford 2000; Jongens 2006). The terrane structure is complicated with a minimum of 12 north-trending, fault-bound slices each with their own unique stratigraphy (Jongens 2006) leading some to suggest that the Takaka Terrane is composite (e. g. Cooper 1989). However, the overall coherence of the terrane’s post-amalgamation geological history justifies its treatment as a single terrane in this study, but emphasises the need for a robust characterisation of both the Western and Eastern Provinces.

Brook Street Terrane (Figure 1)

The Brook Street Terrane (BST) (Mortimer et al. 1999; Tulloch et al. 1999; Spandler et al. 2005; Wandres and Bradshaw 2005; McCoy-West et al. 2014) is a remnant of a Permian intra-oceanic arc complex composed of a 14–16 km thick sequence of Permian-aged metamorphosed submarine volcanics and volcanoclastic sediments, characterised by basaltic-andesitic with minor rhyolites and dacitic lithologies, mafic dykes, mafic-ultramafic cumulates and trondhjemite plutons. Overlaying the arc complex, but separated from it by a major unconformity, are the predominately non-volcanic Late Permian formations of the Productus Creek Group, which in turn are unconformably overlain by the Jurassic Barretts Formation (Landis et al. 1999). The Median Suit intrudes the Brook Street Terrane although the contact between the two is largely concealed by younger Early Cretaceous plutons of the Separation Point Batholith (Tulloch et al. 1999).

The BST has been correlated with the Gympie Terrane in Queensland Australia, and the Teremba Terrane of New Caledonia (Spandler et al. 2005). These correlations have been used to suggest that these terranes once formed a Permian arc complex that was disarticulated during or after its accretion to the Gondwanan margin (Spandler et al. 2005). Palaeomagnetic data indicate that the BST formed at lower latitudes than other Eastern Province terranes and its isotopic signature also distinguishes it from the adjacent volcanics (Frost and Coombs 1989; Haston et al. 1989; Adams et al. 2002; Spandler et al. 2005). A cladistic analysis by Michaux et al. (2018) clustered the BST with other ‘Central Arc Terranes’ of the Eastern Province.

Drumduan Terrane (sensu stricto) (Figure 1)

The Drumduan Terrane (Bishop et al. 1985) outcrops in Nelson, South Island, New Zealand and is bounded to the east by the BST and to the west by the Rotoroa Complex. The Delaware Fault separates the Drumduan terrane from the BST and has been interpreted as the boundary between the Western and Eastern Provinces (e. g. Johnston et al. 1987). The Rotoroa Complex is a predominantly dioritic plutonic assemblage dating from the Upper Jurassic (156 Ma) and was stitched to the Western Province Takaka Terrane by the Separation Point Batholith during the Lower Cretaceous (∼117–114 Ma) (Waight et al. 1998; Kimbrough et al. 1993). The Rotoroa Complex is geochemically distinct from the closest intrusive volcanic rocks of the Western Province (the S-type Devonian Karamea granite) and has the chemical characteristics of being generated in an island arc subduction zone. However, there are also I-type granites within the Buller Terrane, such as the Carboniferous Paringa Suite (Mortimer et al. 1999; Tulloch et al. 2009). Geochemically similar Palaeozoic plutons are common south of the Alpine Fault leading Kimbrough et al. (1993) to interpret the MTZ as a disrupted and geodispersed Late Palaeozoic arc. Scott (2013) included the Rotoroa Complex as part of the Drumduan Terrane, but for the purposes of this paper the term is used in its restricted sense to refer only to the volcaniclastic sediments and associated Carboniferous basement.

The Mesozoic arc-related sediments and Carboniferous basement rocks are described in Beresford et al. (1996). The contact between the sediments and the basement rocks is not exposed in Nelson so there is some doubt as to whether the boundary is an unconformity, i. e. the basement is an eroded surface upon which the overlying sediments were deposited, or a tectonic boundary indicating it is an exotic fragment unrelated to the overlying sediments. In Fiordland (Ewing et al. 2007; Scott et al. 2008) and Stewart Island (Allibone and Tulloch 2004) the contacts between Carboniferous granitic basements and Jurassic sediments are exposed and show the basements there are part of the terrane. The basement rocks include the Pepin Group and the Echinus Granite. The Pepin Group is composed of low-grade metamorphic clastics of the Fall Formation and the Platform Gneiss. The Pepin Group is intruded by the Echinus Granite, a Carboniferous I-type granite generated in an arc environment and dated at 310 Ma (Kimbrough et al. 1993). The age of these basement rocks suggests that they are related to the Western Province (Kimbrough et al. 1993).

The overlying sediments include the Marybank and Botanic Hill Formations. The Botanic Hill Formation is composed of tuffs and breccias and is in fault contact with the Marybank Formation. The Marybank Formation includes a variety of terrestrial volcanoclastic sediments and associated lavas. The sediments are metamorphosed and include lawsonite, a mineral that forms in high pressure/low temperature environments, indicating burial to depths of around 15 km and a low geothermal gradient (Galvez et al. 2012). Johnston et al. (1987) ascribed this metamorphism to under-thrusting beneath the Brook Street Terrane at the height of the Cretaceous Rangitata Orogeny, but a more usual interpretation is of burial in a subduction zone (Miyashiro et al. 1982). Remarkably, these sediments contain rare, but well-preserved Jurassic plant macrofossils of Podocarpaceae, Araucariaceae, and Pteridophyta (Johnston et al. 1987; Galvez et al. 2012). A Jurassic age for the Marybank Formation is also supported by a minimum U/Pb zircon age of 142 Ma for the Cable Granodiorite that intrudes the Marybank Formation (Kimbrough et al. 1993).

Material and methods

Data

Scott (2013; Table 2) provided a comparison between the Western Province Takaka terrane, the Eastern Province Brook Street Terrane, and the Drumduan Terrane of Nelson based on 15 different geological characters. These characters and an additional six based on metamorphic style (Scott 2013) and detrital zircon age profiles (Wysoczanski et al. 1997; Adams et al. 2002; Allibone and Tulloch 2004; Ewing et al. 2007) are described in Table 1. Simple binary coding of these characters is shown in Table 2A–C which records only presence or absence and cannot distinguish between states such as ‘common’ and ‘minor’ that are used by Scott (2013). A 0/1/2 coding scheme was also used to describe these graduated characters and is shown in Table 2B.

Table 2. Character states and character state matrices.

A. Scott’s original data for characters 1–15 (Scott 2013; Table 2) with presence (1) or absence (0) coding and character 6 omitted. Outarea coded all 0
Outarea	00000000000000
Takaka	11111011111110
Drumduan	00001111000001
Brook Street	00000110000100
B. 20 characters included with 0/1/2 coding as follows:
characters 1-4, 9-12, 14-15, 17–20 presence (1) absence (0)
characters 5, 8, 13 common (2) minor (1) absent (0)
character 6 negative δNd (2) positive δNd (1) missing (?)
character 16 high T/low P metamorphism (2) high P/low T metamorphism (1)
Outarea	00000000000000000000
Takaka	11112202111111021110
Drumduan	00001112100000110111
Brook Street	00000?21000020010011
C. All 21 characters included with presence (1) absence (0) or missing (?) coding
Outarea	000000000000000000000
Takaka	111111011111110101110
Drumduan	000011111000001010111
Brook Street	00000?110000100010011

Analyses

The data matrices were analysed with the cladistic software TNT v1.5 (Goloboff et al. 2008), using a heuristic search algorithm (ienum) to find the most parsimonious solution(s). This algorithm provides an efficient method of finding exact solutions for small data sets. Trees were rooted using an all zero outgroup. The trees produced by the analyses of the data matrices shown in Table 2 are shown in Figure 2.

Figure 2. Results of the Analysis. A. Single shortest tree from 0/1 coding of Scott (2013) characters (Data matrix Table 2A). Length (l) = 16, Consistency Index (CI) = 0.88, Retention Index (RI) = 0.60. Characters 5 and 9 are synapomorphies. B. Single shortest tree from 0/1/2 coding of all characters (Data matrix Table 2B) l = 27, CI = 0.96, RI = 0.67, characters 8, 9, 18 are synapomorphies. C. Two shortest trees from 0/1 coding of all characters (Table 1) (Data matrix 2C) (i) l = 25, CI = 1.0, RI = 1.0, characters 5, 9, 19 are synapomorphies. (ii) l = 25, CI = 0.84, RI = 0.43, characters 7, 17, 21 are synapomorphies.

Results and discussion

Two of the three analyses produced a single shortest tree that show the Drumduan Terrane and Takaka Terrane are more closely related to each other than either is to the Brook Street Terrane, indicating that the Drumduan Terrane is part of the Western Province. The result of the analysis of the original data of Scott (2013) showed that the characters linking the Drumduan Terrane to the Takaka Terrane were the presence of I-type Carboniferous granites (5) and intrusion by Separation Point granitoids (9) (Figure 2A). The presence of Separation Point granitoids records the amalgamation of the two terranes, not their respective origins, and is consequently not relevant to this and subsequent analyses. The presence of Carboniferous granites is therefore the critical characteristic on which the relationship is based. If the Echinus Granite and Pepin Group is the basement of the Jurassic Marybank Formation, and comparison with Fiordland (Scott et al. 2008) and Stewart Island (Allibone and Tulloch 2004; Ewing et al. 2007; Tulloch et al. 2009) correlatives suggest it is, then the presence of Carboniferous granites is a valid synapomorpy linking the Drumduan Terrane to the Western Province.

The inclusion of additional characters (16 [metamorphic grade] and 17–20 [zircon age-profile]), combined with a finer grained approach to character coding yielded a tree with the same topology (Figure 2B). Note that this coding resulted in the reduction of character numbers from 21 to 20 because ‘metamorphic style’ codes as a single character (16) rather than two (16, 17). Characters 17–20 refer to zircon age-profiles. Characters supporting a Western Province/Drumduan Terrane are the common occurrence of Cretaceous-age late Darran Suite intrusives (8) and the presence of Devonian/Carboniferous detrital zircons (18). The occurrence of late Darran Suite intrusives, as discussed previously, is probably an invalid character because their emplacement likely post-dates amalgamation. Late Jurassic early Darran Suite rocks were originally described from Fiordland but Scott (2013) regarded the Rotoroa Complex and Tasman Intrusives of Nelson as equivalents. The 156 Ma (Late Jurassic) age of the Rotoroa Complex predates both its suturing to the Takaka Terrane and the juxtaposition of the Brook Street Terrane to the Drumduan Terrane and is, therefore, relevant to the terrane’s pre-amalgamation history. The presence of Devonian/Carboniferous age detrital zircons is based on their occurrence in the Loch Burn Formation of Fiordland (Ewing et al. 2007) and the Paterson Group of Stewart Island (Allibone and Tulloch 2004). These data were used in the analysis because no zircon age profiles are available for the Marybank sediments and both have been correlated with the Drumduan Terrane volcaniclastics (e.g. Wandres and Bradshaw 2005; Scott 2013).

The final analysis of all 21 characters using a coarse presence/absence coding produced two shortest trees which are shown in Figure 2C. The tree in Figure 2: C(i) links the Drumduan Terrane to the Western Province, while the other (Figure 2C(ii)) links it to the Eastern Province. The characters supporting this latter relationship are the presence of Late Jurassic early Darran Suite intrusives (7), High T/Low P metamorphism (17) and the presence of Jurassic detrital zircons (21). While early Darran Suite intrusives are common in the Brook Street Terrane they are minor components in the Drumduan Terrane, and the metamorphic styles are also quite distinct with High P/Low T metamorphism confined to the Drumduan Terrane. The linkage of the Drumduan Terrane with the Eastern Province in Figure 2C (ii) may therefore be an artifact of the coding. The consensus tree, which is not shown in Figure 2, is an unresolved trichotomy.

If the Drumduan Terrane is placed within the Western Province, then how is it related to the Takaka and Buller Terranes? Our results support a model of the Drumduan Terrane as a Carboniferous (or earlier) arc that was either part of or adjacent to the Gondwana foreland (and subsequently accreted during the final stages of the Tuhua orogeny). Given that the isotopic composition of the basement granite is distinct from that of similar aged I-type granites of the Takaka Terrane, it is more probable that the Drumduan Terrane was a separate terrane adjacent to the Gondwanan margin in the Carboniferous. The lack of typical Gondwanan U-Pb zircon peaks at 500–600 Ma and 1000–1200 Ma in the volcaniclastic correlates of the Drumduan Terrane indicates that this arc may have received detrital zircons from erosion of Western Province volcanics rather than typical Gondwanic source areas. We envisage erosion of a marginal range formed during the later stages of the Tuhua Orogeny and exposure of the buried granites on which the later Jurassic Marybank Formation was deposited. Subsequently, as the Eastern Province terranes approached, plutons were generated (e. g. Rotoroa Complex) above an active subduction zone culminating in the amalgamation of the Brook Street Terrane and the emplacement of the Separation Point Batholith. High P/Low T metamorphism of the Drumduan Terrane would, as Johnston et al. (1987) have suggested, resulted from this collision.

Conclusions

The position of the boundary between the Western and Eastern Provinces of New Zealand has been a point of contention and debate since the 1980s. The results of this study support the placement of the Drumduan Terrane (s.s.), and by implication the Teetotal Group in southeast Nelson Province (Figure 1), the Largs Terrane and Loch Burn Formation in Fiordland, and the Paterson Group in Stewart Island, in the Western Province. Defining the boundary between New Zealand’s two distinctive basement suites is a consequence of identifying natural groupings, and natural groupings allow interpretations that are not artifactual. Thus, reconstructions of the austral Gondwana margin should include a Carboniferous marginal arc and recognise that the Brook Street Terrane is the western border of the Eastern Province.

Finally, we hope that this case study in the application of cladistic methods to describe terrane relationships will encourage other workers to analyse their comparative data using these methods. We see identifying natural groupings as an essential first step in any comparative study because this can lead to greater stability in nomenclature and these groups can be confidently arranged in a hierarchy of relationships (Ebach and Michaux 2017). In this study we have established that the Drumduan Terrane of Nelson is part of the Western Province using published data and cladistic methods. However, we have made untested assumptions and left other questions unasked. For example, is the Carboniferous arc a single structure (as we have tacitly implied in this study) or are the Drumduan Terrane and Teetotal Group (Figure 1) separate from the Largs Terrane and Loch Burn Formation in Fiordland, and the Paterson Group in Stewart Island? Or what are the relationships of the Western Province to other Cambrian-Carboniferous Gondwana-margin terranes in East and West Antarctica, the Campbell Plateau, Tasmania, and southeast Australia, which we only obliquely referred to? We argue that cladistic methods provide the means to address these and other comparative questions.

Acknowledgements

We would like to thank James Scott and an anonymous referee for their detailed and thoughtful reviews of an earlier draft which greatly helped improve the present paper. EMD wishes to thank the University of New South for awarding her a Research Training Program scholarship.

Disclosure statement

No potential conflict of interest was reported by the authors.