Geochemical evidence for provenance of Ordovician cherts in southeastern Australia

Early to Middle Ordovician cherts of the Girilambone and Adaminaby groups are widespread in the Lachlan Orogen in central New South Wales. Their ages are well constrained biostratigraphically by conodonts ranging from the late Tremadocian to earliest Sandbian. Broadly contemporaneous cherts are exposed in the Narooma terrane (latest Cambrian to Darriwilian), and in allochthonous blocks (of late Middle and Late Ordovician age) of the New England Orogen at Port Macquarie. In the Kiandra–Tumut region of southern NSW Darriwilian to earliest Gisbornian cherts are interbedded with volcaniclastics of the Macquarie Volcanic Province. To determine their provenance and compare depositional settings, 60 chert samples representative of these regions were analysed for selected major, trace and rare earth elements (REE). Al2O3/TiO2 ratios enable recognition of two dominant sources from which the sediment fraction of the cherts was derived at different times in the evolution of the Tasmanides, one indicative of old continental crust and the other sourced from a juvenile continent or plateau. All cherts analysed, regardless of geological province, carry continental margin signatures demonstrated by high Al2O3/Fe2O3 ratios, LREE enrichment, small negative Ce anomalies, prominent negative Eu anomalies, low total REEs and near-chondritic Y/Ho ratios. The late Cambrian and Early Ordovician Narooma terrane cherts display a clastic component derived chiefly from a juvenile continent or plateau, whereas cherts of the Hermidale and Albury-Bega terranes contain detritus of mixed origin. During the Early to Middle Ordovician magmatic hiatus in the Macquarie Volcanic Province, all cherts regardless of tectonic affiliation incorporate terrigenous detritus solely of Gondwana origin. Upon resumption of magmatism in the Macquarie Volcanic Province, a mixed terrigenous source is recorded in all analysed cherts, before a dominant Gondwana source is re-established in the late Darriwilian for cherts of the Hermidale, Albury-Bega and Kiandra regions.


INTRODUCTION
Cherts and associated siliceous siltstones attained their maximum development in southeastern Australia during the Early and Middle Ordovician, when they became widespread in the Lachlan Orogen of central and southern New South Wales (NSW), extending south into the northeast corner of Victoria ( Figure 1). For convenience in referring to the various regional components of the Lachlan Orogen, we utilise (informally) a modification of the terrane terminology introduced by Glen et al. (2009). The Albury-Bega terrane incorporates turbiditedominated successions assigned to the Adaminaby and Wagga groups, and overlying Upper Ordovician black shales of the Bendoc Group, west and east of the Macquarie Volcanic Province (MacVP). In central NSW, the Hermidale terrane is represented by the Girilambone Group. The Narooma terrane (Glen et al. 2004), exposed on the far south coast of the state, received predominantly siliceous deposition from the Furongian (latest Cambrian) to late Darriwilian (latest Middle Ordovician). The Port Macquarie Block of the New England Orogen, located on the mid north coast of NSW, includes Late Ordovician cherts as a minor component; these are younger than those of the Lachlan Orogen and are interpreted as allochthonous in origin. The geographic distribution in NSW of Ordovician rocks that include chert horizons, and their relation to adjacent terranes, is depicted in Figure 1 Ordovician cherts from southeastern Australia and their included biota have been studied over the past three decades mainly by microscopic examination of thicker than normal petrological sections, and polished rock sections, pioneered in Australia by Ian Stewart of Monash University (Victoria). The Geological Survey of NSW utilises the thick section technique, with approximately 3200 sections (each about 50 microns thick) having been prepared from chert samples collected during regional mapping programs in the Lachlan Orogen over the past 15 years. These sections are cut parallel to bedding planes in order to maximise the chance of intersecting accumulations of pelagic or neritic biota (Percival 2012) that have been fossilised in siliceous sediments on the sea floor. Despite limitations on identification of specimens encountered in a twodimensional thick section, conodonts are the most useful fossils to underpin a biostratigraphic subdivision of Ordovician deep-water cherts. A preliminary conodont zonation developed over the past decade in the Lachlan Orogen (Percival 2006a) recognises two zones in the Early Ordovician and a further two in the Darriwilian (late Middle Ordovician), as follows: Paracordylodus gracilis assemblage zone, of late Lancefieldian to early Bendigonian age (i.e. early Floian, in global biostratigraphic terminology), lacking Oepikodus evae, which is characteristic of the overlying zone; Oepikodus evae assemblage zone, of late Bendigonian to early Castlemainian age (equivalent to late Floian), commonly associated with Periodon flabellum; Paroistodus horridus-Spinodus spinatus assemblage zone, of early to mid Darriwilian age (in some areas substituted by a zone characterised by Histiodella sp.), with Periodon aculeatus a common associate; and Pygodus serra assemblage zone, of late Darriwilian age, frequently occurring with Periodon aculeatus.
The slightly younger zone of Pygodus anserinus, spanning the MiddleÀLate Ordovician boundary, is rarely encountered in cherts from the Kiandra region and in the Narooma terrane. Conodont faunas preserved in all these cherts are representative of the Open-Sea Realm , which occurs around all continents and oceanic terranes known in the Ordovician, thus facilitating global correlations. Complete documentation of the conodont fauna will be published elsewhere;

AIMS AND SCOPE OF THE PROJECT
In the only study previously conducted into the geochemistry of cherts from eastern Australia, Aitchison & Flood (1990) analysed major and selected trace-element compositions of cherts from the Djungati and Anaiwan terranes of the New England Orogen, ranging in age from ?Silurian to Early Carboniferous, to provide insights on depositional settings. However, Ordovician cherts from the Lachlan and New England orogens and the Narooma terrane have not been analysed geochemically. Elsewhere in the world (e.g. Scotland) considerable research has been carried out into the REE components of cherts and implications for their depositional environments. Our aims in undertaking this study were first to compare and contrast our findings with research Figure 2 Stratigraphy of Ordovician turbidites depicting main chert occurrences. Note the relationship of these cherts to informal conodont zonation discussed in the text. elsewhere to identify, where possible, general tectonic and depositional relationships among Ordovician cherts from southeastern Australia. Numerous tectonic models have been formulated to explain the geological complexity of this region, forming part of the Tasmanides east of the Gondwana cratonic margin; see Gray & Foster (2004) and Glen (2005) for reviews of pertinent literature to 2004; some more recent interpretations include those of Fergusson (2009Fergusson ( , 2014, Glen et al. (2009) andGlen (2013). However, all these models draw tectonic implications from rocks other than cherts. We focused our investigations on whether significant differences were evident in cherts from different terranes, such as the Albury-Bega and Hermidale terranes distinguished by Glen et al. (2009). In particular, the project sought to determine whether an exotic origin could be determined for the Narooma and Port Macquarie cherts, which are postulated in some tectonic models (e.g. Glen et al. 2004) to have been deposited remote from the Gondwana craton. Finally, utilising the well-constrained biostratigraphic control provided by conodonts in the cherts, we also hoped to identify any temporal changes in the geochemical parameters within a particular terrane.
Sixty samples were selected for this project, representative of the geographic spread and age range of Ordovician cherts in southeastern Australia. Three-quarters of these came from the two largest areas of Early to Middle Ordovician turbidite deposition in the Lachlan Orogen: 38 cherts from the Albury-Bega terrane and seven from the Hermidale terrane. A further three samples were analysed from cherts interbedded with volcaniclasticdominated rocks on the margins of the Kiandra Volcanic Belt of the southern MacVP. Two cherts from the Port Macquarie Block of the New England Orogen were included. The Narooma terrane provided 10 chert samples. Ages of the samples ranged from latest Cambrian (Narooma terrane), through Floian (Early Ordovician) and Darriwilian (Middle Ordovician) cherts that are present in most terranes, to Late Ordovician samples, found in the Kiandra region and at Port Macquarie. Locality details for each sample are provided in the Appendix.

Albury-Bega terrane
In the CoomaÀBegaÀMallacoota region, Lower and Middle Ordovician turbiditic sandstones and cherts are assigned to the Adaminaby Group (Glen 1994). Cherts are present at two main levels, the older consisting of thinly bedded cherts, somewhat sporadic in outcrop and continuity, that contain conodont faunas of Floian age, dominated by Paracordylodus gracilis and Oepikodus evae. In contrast, the prominent and widespread upper horizon, known as the Numeralla Chert, is generally more than 50 m thick. It contains latest Darriwilian (Pygodus serra Zone) conodonts, and consists of individual beds of parallel-sided ribbon chert up to 50 cm thick, interbedded with cleaved slate and siltstone, and rare sandstone beds.
Contemporaneous strata in the GoulburnÀTaralga area (between Sydney and Canberra) of similar turbiditic origin are referred to as the Abercrombie Formation of the Adaminaby Group. Three chert members are defined within the formation (Thomas & Pogson (Murray & Stewart 2001). Also in the vicinity of Oberon, but higher in the stratigraphy, the Mozart Chert correlates with the Nattery Chert Member. However, correlatives of the Peach Tree Chert Member have not been identified in this region.
Cherts are of comparatively minor occurrence within the Lower to Upper Ordovician Wagga Group, which consists predominantly of quartz-rich turbidites. Detailed mapping of these rocks in the Lake Cargelligo region (Colquhoun et al. 2005) assigned the turbidites to the Clements Formation, composed of beds of quartzrich sandstone grading to siltstone and slate. Two discontinuous chert horizons occur in the Clements Formation, the older Milby Chert Member containing the late Bendigonian to early Castlemainian conodont Oepikodus evae, and the younger Doongala Chert Member characterised by a Pygodus serraÀPeriodon aculeatus conodont assemblage (Percival & Zhen 2007) that establishes its age as latest Darriwilian. The Doongala Chert Member, estimated to be less than 10 m thick by Colquhoun et al. (2005), is much thinner than the Numeralla Chert and Nattery Chert Member with which it is correlated. Both chert members in the Clements Formation are relatively poorly exposed and are impersistent laterally, forming lenticular bodies that pass into sandstone, mudstone and shale of the turbiditic succession. Cherts of early Darriwilian age, comparable with those recognised in the Adaminaby Group, are not presently known from the Wagga Group.

Hermidale terrane
Ordovician turbidite-derived rocks of the Hermidale terrane, forming the Girilambone Group, occupy a belt extending north and east of Cobar (Burton et al. 2012). The Lower Ordovician Narrama Formation is dominated by thick to thin bedded quartz-rich sandstone locally grading to siltstone, in which thin (although sometimes laterally extensive) chert horizons are present, containing the conodonts Paracordylodus gracilis and Oepikodus evae of Floian age (Percival 2006b. Oepikodus evae has also been found within inter-pillow chert of the Mount Dijou Volcanic Member of the Narrama Formation, which includes amygdaloidal pillow basaltic and trachytic lavas. The overlying Ballast Formation consists of interbedded turbidites and chert, as well as thick persistent packages of ribbon chert tens of metres thick in the upper part of this formation termed the Whinfell Chert Member. These cherts contain conodonts including Pygodus serra and Periodon aculeatus, indicative of a late Middle Ordovician age (upper Darriwilian 3 to top Darriwilian 4) (Stewart & Glen 1986;Iwata et al. 1995;Percival & Zhen 2007).

Kiandra Volcanic Belt
The Kiandra Volcanic Belt is the southernmost component of the MacVP that is otherwise confined to the central NSW region of the Lachlan Orogen. Atypically for lithological associations of the MacVP, cherts in the Kiandra Volcanic Belt are intermingled with volcaniclastic units. Quinn et al. (2014) have reassessed the stratigraphy of this belt, describing the oldest part of the succession as consisting of turbidites and quartz sandstones (cf. Adaminaby Group), with rare volcaniclastic debris flows and pillow basalts (L. Wyborn 1977;D. Wyborn et al. 1990), intercalated with thin chert beds of Darriwilian 2À3 age. Succeeding thick bedded cherts of late Darriwilian age are overlain by volcanics, with latest Darriwilian to earliest Gisbornian thin chert layers (containing Pygodus anserinus) interbedded with volcaniclastic rocks. Chert is absent from younger Ordovician strata of the Kiandra Volcanic Belt, which consist of more extensive volcaniclastic deposits interbedded with black shale.

Narooma terrane
The Narooma terrane, exposed on the NSW south coast in the immediate vicinity of Batemans Bay and Narooma, consists of a dismembered succession of late Cambrian to Late Ordovician oceanic sedimentary rocks interpreted as an exotic terrane (Glen et al. 2004). These rocks are assigned to the Wagonga Group, which includes the Narooma Chert and the overlying argillaceous Bogolo Formation. The alternative interpretation of these rocks is that they represent an accretionary complex (Prendergast 2007;Prendergast & Offler 2012, and references therein), but the stratigraphy is essentially the same in both models. The lower part of the Narooma Chert consists of slates interbedded with ribbon cherts that contain a diverse conodont fauna (illustrated in Glen et al. 2004, figure 3) ranging in age from late Cambrian (Furongian) to late Darriwilian. Typical lithologies in the lowermost Narooma Chert are pale yellow translucent cherts (particularly in the upper Cambrian section) up to 10 cm thick, varying to dark grey, brown and black opaque varieties as silt content increases up section. In the upper Narooma Chert, individual chert beds up to 30 cm thick alternate with shale and siltstone that contain Eastonian (Late Ordovician) graptolites.

Port Macquarie Block, New England Orogen
On the mid north coast of NSW, the Watonga Formation (described by Och et al. 2007a) is well exposed along the coastline immediately south of Port Macquarie, where it comprises a mostly broken formation inferred to result from disruption of a once-stratified sequence of basalt, ribbon chert, siliceous mudstone, siltstone, sandstone and conglomerate. Most exposures of ribbon chert are discontinuous and lensoidal with individual beds 5 mm to 200 mm thick (rarely to 1 m), intercalated with thinner recessive black mudstone units. Ghosts of radiolarians are preserved in some cherts whereas others consist solely of fine anhedral quartz grains with a dusting of iron oxide minerals. Conodonts visible in thin-sections of cherts establish a maximum age range for these rocks extending from late Darriwilian to possibly as young as the end of the Eastonian (Och et al. 2007b).

Allochthonous rocks of ocean-floor derivation
The Jindalee Group, best developed in the TumutÀCootamundra region, comprises a fragmented ophiolitic sequence of now-serpentinised harzburgites and other ultramafic rocks, gabbros, pillow basalts and bedded and inter-pillow cherts, and jaspers within a sedimentary matrix of probable Silurian age (Percival et al. 2011, based on unpublished data of C. D. Quinn). This group also incorporates the Brangan Volcanics and Hoskins Chert (the latter with a sparse Darriwilian to Late Ordovician conodont fauna) that are exposed north and south of Grenfell. Allochthonous chert blocks within the postulated Silurian sequence near Cootamundra are similar in age, containing late Middle to Late Ordovician conodonts together with numerous cyanobacterial filaments (Lyons & Percival 2002).
Apart from rare examples from Port Macquarie mentioned previously, allochthonous cherts of ocean floor derivation were not analysed in this initial study, but we anticipate that geochemical signatures recognised in the current project will be used in future research to determine the origin of such clasts.

CHERT GEOCHEMISTRY CONCEPTS
The chemistry of chert is largely controlled by diagenesis, whereby the transformation of the silica phases increases the dissolved silica content, effectively diluting the relative concentration of the other elements. In addition, many elements are fractionated, transported and redeposited during this process. However, some elements are immobile [e.g. Al, Ti, Fe and the rare earth elements (REEs)] and ratios of immobile elements are useful indicators in determining depositional environment (Murray 1994). In cherts, the Al and Ti components are terrigenous in origin and are carried by the detrital aluminosilicate phase associated with the clay fraction of the rock (Sugisaki 1984;Toyoda & Masuda 1990;Murray et al. 1992). Thus, the Al 2 O 3 /TiO 2 ratio can be used to infer source-rock compositions of the clastic component (Girty et al. 1996;Hayashi et al. 1997;Huang et al. 2013). High Fe content is usually associated with metalliferous or hydrothermal input (Murray et al. 1991;Murray 1994) although the signature may also carry, at least in part, a terrigenous component (Murray 1994;Huang et al. 2013).
Th/Sc and La/Sc ratios are commonly used as provenance indicators, particularly in shale (Nagarajan et al. 2007 and references therein) and turbidite sequences (e.g. Bhatia & Crook 1986). Such ratios have also been applied to classify the terrigenous component of cherts (e.g. Girty et al. 1996;Tanner et al. 2013), the rationale being that these elements are only brought in by the aluminosilicate phase, and processes such as diagenesis, seawater adsorption and metalliferous input have negligible effects.
The REEs have become increasingly important in providing greater clarity of chert depositional environments than can be derived from the major elements alone. According to Murray et al. (1991Murray et al. ( , 1992; and references therein), REE variation in marine cherts is dependent on: (a) adsorption from seawater (itself dependent on burial rate); (b) the composition of terrigenous input; and (c) the composition of metalliferous input. The Ce anomaly (Ce/Ce Ã ) is of particular importance in defining cherts from ridge proximal, open ocean and continental margin environments (Murray et al. 1990) and has been used in numerous studies (Murray et al. 1991(Murray et al. , 1992Armstrong et al. 1999;Owen et al. 1999a, b;Huang et al. 2013). In oxygenated seawaters, Ce(III) is oxidised to insoluble Ce(IV), which drops out of the water column resulting in seawater with distinct negative Ce/Ce Ã values (Elderfield & Greaves 1982). Near ridge plumes, the Ce/Ce Ã values of seawater are even lower as Fe and Mn particulates from hydrothermal activity scavenge relatively more Ce (Klinkhammer et al. 1983). In continental margin settings, river waters are the dominant REE carrier to the oceans and they typically display little or no Ce fractionation (Goldstein & Jacobsen 1988).
In terms of overall chondrite-normalised REE patterns, McLennan (1989) and McLennan et al. (1990) demonstrated that continental margin mudrocks typically display light REE enrichments, pronounced negative Eu anomalies and flat heavy REEs. This signature appears robust regardless of whether the sediments are derived from an old continent or a fractionated volcanic arc (Girty et al. 1996;Owen et al. 1999a). However, Eu anomalies are small or absent where there has been a significant contribution from an unfractionated active margin volcanic source (fore-arc, continental arc, back-arc; McLennan et al. 1990). Given that the REEs are largely derived from terrigenous sources in continental margin cherts, a similar REE pattern to continental margin mudrocks can be expected.
Total REE abundances (SREE) are low in continental margin cherts owing to the large-scale removal of REEs in estuaries (Murray et al. 1991) as well as a fast burial rate, limiting their exposure time to seawater (Murray et al. 1990). In the open ocean, cherts receive little if any terrestrial input, and their REE chemistry is dominated by adsorption of REEs from seawater (see Murray et al. 1992; and references therein) combined with a slow burial rate. Thus, total REE abundances tend to be high.
In silicate systems, Y behaves like an REE (in particular Ho, because of a similar ionic radius). Hence, in most rocks the Y/Ho ratio is near chondritic (»28; Jochum et al. 1986) but this is not the case in seawater or its chemical precipitates, which have Y/Ho ratios between 50 and 60, considerably higher than chondrite (Minami et al. 1998). It would thus be expected that continental margin cherts should have near-chondritic Y/Ho ratios because of a significant terrigenous input. Open ocean cherts, despite being chemical precipitates of seawater, are more complex with Y/Ho ratios ranging from as low as »15 in ferromagnesian nodules and crusts to suprachondritic in some bedded cherts. It is probable that, in this environment, the mineral phase(s) where the REE (and Y) resides is important in determining the extent of YÀHo fractionation in aqueous solutions (Minami et al. 1998). The Y/Ho ratio cannot determine depositional environment by itself (cf. Huang et al. 2013), but it can provide further supporting evidence when combined with other chemical criteria. Table 1 provides a succinct summary of key parameters and their implications for tectonic settings.

ANALYTICAL METHODS
The chert samples were analysed at the University of Queensland. The submitted rock fragments were crushed in a steel jaw-crusher before being pulverised in a tungsten carbide mill. For major, REE and other traceelement analysis, samples were digested using ultrapure HFCHNO 3 in Teflon beakers on a hot plate. The procedure was repeated for a subset of samples in high-pressure bombs to confirm complete digestion. The dissolved samples were diluted using ultra-pure water and spiked with a multi-element internal standard solution ( 6 Li, 61 Ni, 103 Rh, 115 In, 187 Re, 209 Bi and 235 U) in 2% HNO 3 . Analyses were carried out by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) on a Thermo X7, following the procedure described in Eggins et al. (1997). Duplicate hot-plate digestions of US Geological Survey W-2 standard material (diabase) were used as the calibration standard. Rock standard AGV2 was run as an unknown for accuracy check. Results were not blank-subtracted because detection limits and procedural blanks were negligible.

RESULTS
Selected major, trace and REE of cherts from the Lachlan Orogen as well as from Port Macquarie in the New England Orogen and from the Narooma terrane are reported in Tables 2 and 3. Using chemical criteria set out by Murray (1994), most cherts, regardless of terrane or age, overlap (or are close to) the field of cherts from continental margins ( Figure 4). A continental margin depositional environment for the NSW cherts is also suggested by the LaÀScÀTh ternary diagram, which was used by Tanner et al. (2013) to plot the composition of Loch Ard and North Glen Sannox marginal cherts from the Highland Border Complex of southern Scotland ( Figure 5). Similarly, deposition of the NSW cherts close to a continental margin is also suggested by chondrite-normalised Ce anom and Eu anom as well as (La/Yb) n values, SREEs and Y/Ho ratios.    We use the Al 2 O 3 /TiO 2 ratio as a proxy for the provenance of the terrigenous component of the cherts rather than Th/Sc or La/Sc values for several reasons. First, as can be seen in Figure 6a, there is no correlation between Th/Sc and Al 2 O 3 /TiO 2 ratios; the same is true of La/Sc (not shown). Al and Ti are well known to be affiliated with aluminosilicate phases in cherts and unaffected by diagenetic processes (Murray 1994). The application of Th, Sc and La ratios as provenance indicators, while seemingly robust in sedimentary rocks, is less well known in the literature in chemical sediments such as cherts. Second, there is an excellent correlation between Al 2 O 3 and TiO 2 values for the cherts (within the 95% confidence level; Figure 6b), whereas the correlation between Th and Sc (Figure 6c) or La and Sc (not shown) is less robust. This suggests that the latter elements are less likely to reflect the source composition than Al and Ti, and may be affected by other processes. Third, the Al 2 O 3 /TiO 2 ratios for the NSW cherts provide a consistent, interpretable data set. The same cannot be said of the Th/Sc and La/Sc ratios, in which compositional differences in timeÀspace are random.

Albury-Bega terrane
A total of 27 cherts, spanning the conodont zones Oepikodus evae to Pygodus serra, were analysed from the Goulburn region. The majority of samples overlap the field of continental margin cherts (Figure 4) with three exceptions (8728-AYW-20, 8829-ODT-543.1 and 8828-ODT-203.1). These particular samples have greater Ce anomalies (Ce/Ce Ã D 0.57À0.70) compared with the remainder of the Goulburn cherts but similar Al 2 O 3 /(Al 2 O 3 C Fe 2 O 3 ) ratios, and also record no discernible difference in Eu/ Eu Ã , (La/Yb) n , SREEs or Y/Ho values. All but one of the suite of cherts from Goulburn have Al 2 O 3 /TiO 2 ratios ranging between 25 and 35, clustering around the average upper continental crust value of 30 (Taylor & McLennan 1995). The exception is sample 8828-AJJ-33.1 which records an Al 2 O 3 /TiO 2 ratio of 22, a value more akin to the average andesitic crust (Taylor & McLennan 1995). This chert also has total REE concentrations 2À9 times greater than in any other sample as well as the lowest (La/Yb) n ratio (4.2). Th/Sc ratios range from 0.72 to 1.37 with a median value of 0.97 and La/Sc D 1.74À4.22.
Plots of chondrite-normalised REEs for Goulburn samples, grouped according to biostratigraphic age, are  Ten chert samples were analysed from the Braidwood area, ranging in age from the Paracordylodus gracilis to Paroistodus horridus conodont zones. All but three samples (8827-JAF-109.1, 8827-ODT-1019.1 and 8827-ODT-924.1) overlap the field of continental margin cherts on the discrimination diagram of Murray (1994), as shown in The Budhang Chert sample (8830-SM Budhang) from the Oberon region plots within the field defined by continental margin cherts (Figure 4) on the discrimination diagram of Murray (1994). It has an Al 2 O 3 /TiO 2 ratio of 33, a value similar to the average upper continental crust (Taylor & McLennan 1995). This sample has a low Th/Sc ratio (0.51) but a comparable La/Sc value (3.13) with that of the other cherts of the Albury-Bega terrane. A chondrite-normalised REE plot for this sample (Figure 9)

Hermidale terrane
A total of seven chert samples were analysed from the SussexÀByrock area, four from the Oepikodus evae zone (Early Ordovician) and three with a Pygodus serra age (latest Middle Ordovician). All but two of the samples lie within the field of continental margin cherts as defined by Murray (1994; Figure 4). One of these exceptions, sample 8135-SJT-427.1 from the Pygodus serra zone, is slightly more Fe enriched with respect to Al. The other sample from the Oepikodus evae zone (8135-GRB-461.1) has a prominent Ce anomaly (Ce/Ce Ã D 0.42) but is low in Fe with respect to Al. The ratio of Al 2 O 3 /TiO 2 for all the Plots of chondrite-normalised REEs, grouped according to biostratigraphic age, are displayed in Figure 10. One of the cherts from the Oepikodus evae zone (8135-SJT-270.1) has comparatively very high total REEs (SREEs D 397 ppm, cf. 15À65 ppm) with LREE enrichment and HREE depletion [(La/Yb) n D 75.9]. This chert also records the highest Al 2 O 3 /TiO 2 ratio (35.1). Conversely, a further sample from this zone (8136-SJT-25.2) is considerably less enriched in LREEs [(La/Sm) n D 2.22] than its contempories [(La/Sm) n D 5.67À7.33], and has the lowest Al 2 O 3 /TiO 2 ratio (25.2). It also has a negligible Eu anomaly (Eu/Eu Ã D 0.92, cf. 0.64À0.81). Cherts from the Pygodus serra zone appear to have similar REE patterns. However, sample 8135-GRB-497.1 differs in that it has lower LREE enrichment [(La/Sm) n D 3.77] than the other two samples [(La/Sm) n D 5.81À7.67] combined with a lower Al 2 O 3 /TiO 2 ratio (25.7, cf. 29.6À30.5) and a smaller Eu anomaly (Eu/Eu Ã D 0.91, cf. 0.57À0.64).

Kiandra Volcanic Belt
Of the three chert samples analysed from the Kiandra region, only one (8527-CDQ-382, containing diagnostic conodonts of the Pygodus anserinus zone) was able to be definitively dated. On the depositional environment discrimination diagram of Murray (1994; Figure 4), this sample lies below the field of continental margin cherts. This is due entirely to a significantly positive Ce anomaly (Ce/Ce Ã D 1.46). Of the two other samples (8527-CDQ-255 and 8626-CDQ-HH2), only the latter plots entirely within the field of continental margin cherts. The remaining chert (8527-CDQ-255) is slightly more enriched in Fe with respect to Al. Although not definitive, conodont elements in thick section suggest the same age range for 8527-CDQ-255 as 8527-CDQ-382 (i.e. late DarriwilianÀearly Gisbornian) whereas a slightly older age (Darriwilian 3 or 4) is proposed for 8626-CDQ-HH2. The Al 2 O 3 /TiO 2 ratios for the younger samples (8527-CDQ-382 and 8527-CDQ-255) are 34 and 55, respectively, which suggests the terrigenous components of these cherts were sourced from material more felsic than the average continental crust (Al 2 O 3 /TiO 2 D 30; Tay

Port Macquarie Block, New England Orogen
Two chert samples containing different conodont assemblages were analysed from the Watonga Formation of Port Macquarie. The older chert (9435-Pmtk-436 of late DarriwilianÀearly Gisbornian age) overlies the field of continental margin cherts (Murray 1994; Figure 4), whereas the younger chert (9435-Ptmk-432; Gisbor-nianÀearly Eastonian in age) is slightly more Fe enriched with respect to Al. The Al 2 O 3 /TiO 2 ratios of the two samples differ, with the older chert (9435-Pmtk-436) recording a clastic contribution more akin to the

Narooma terrane
Ten cherts spanning the latest Cambrian (Furongian) to earliest Late Ordovician (Pygodus anserinus conodont Zone) interval were analysed from the Narooma terrane. All samples overlie the field of continental margin cherts as defined by Murray (1994), with one exception (8925-N35), which differs in being slightly richer in Fe with respect to Al (Figure 4). Seven of these cherts have Al 2 O 3 /TiO 2 ratios ranging from 15.9 to 22.9 with a median value of 20.9. This value is slightly more mafic than the average andesitic crust ( REE plots normalised to chondrite and grouped according to biostratigraphic age are displayed in Figure 13. Overall, the cherts are light REE enriched [(La/Sm) n ] D 1.75À4.51], with Ce anomalies (Ce/Ce Ã ) ranging from 0.79 to 1.02. Eu anomalies are prominent (Eu/ Eu Ã D 0.57À0.86) and the total REEs are low (SREEs D 5.6À40 ppm). Some fractionation of Y from Ho has occurred with Y/Ho D 24À31.

DISCUSSION
Albury-Bega terrane cherts from the Goulburn, Braidwood and Oberon regions are typical of those deposited in a moderate-sized basin surrounded by passive and/or active margins receiving variable amounts of terrigenous sediment (e.g. Murray et al. 1992;Armstrong et al. 1999). They are LREE-enriched and have small negative Ce anomalies, moderately prominent negative Eu anomalies, relatively low total REEs and near-chondritic Y/ Ho ratios. A small number of samples do not quite conform to the above pattern but still carry the overall geochemical fingerprint of continental margin cherts. These samples provide additional information about their formation. For example, sample 8827-JAF-109.1, which is depleted in light REEs with high total REE abundance and enrichment in Fe, implies some seawater adsorption combined with a more mafic detrital input. This is consistent with a relatively low Al 2 O 3 /TiO 2 ratio ( Based on Al 2 O 3 /TiO 2 ratios, two terrigenous sources contributing to chert formation can be recognised. Early Ordovician cherts record a detrital source component more mafic than the average continental crust (S1: 21 < Al 2 O 3 /TiO 2 27) while the majority of Middle Ordovician cherts contain clastics similar to the average continental crust composition (S2: Al 2 O 3 /TiO 2 >27), although terrigenous input into some late Middle Ordovician  cherts records S1 chemistry. In addition, two mid-dleÀlate Darriwilian cherts of S2 affinity (8828-DJP-285.1 and 8828-ODT-201.1) are interpreted to record a clastic contribution from an immature active volcanic margin with negligible Eu anomalies (Eu/Eu Ã D 0.95).
Geochemical characteristics of Hermidale terrane cherts are similar to those of the Albury-Bega terrane, implying deposition in an enclosed moderatesized basin receiving terrigenous input from a continental margin. Both terranes display a comparable change in Al 2 O 3 /TiO 2 ratios through the Early and Middle Ordovician. Overall, the Ce anomalies (»1) are typical of continental margin cherts, the one exception (8135-GRB-461.1) possibly reflecting interaction with oxygenated seawater or chemical weathering. Similarly, the enriched light REEs and negative Eu anomalies are characteristic of continental margin chert. Excessive SREEs in sample 8135-SJT-270.1 are probably related to provenance of the clay fraction of the chert, as Al 2 O 3 /TiO 2 ratios and LREE enrichment suggests a more felsic source. The very small Eu anomalies in the Early Ordovician chert (8136-SJT-25.2) and the late Middle Ordovician chert (8135-GRB-497.1) combined with lower Al 2 O 3 /TiO 2 ratios suggest that these two samples may have tapped calcic plagioclase from a mafic volcanic source.
We interpret the Kiandra cherts to have been deposited proximal to a continental margin as overall they are LREE-enriched [(La/Yb) n D 2.2À12.3] and have small negative to positive Ce anomalies (Ce/Ce Ã D 0.84À1.46) and low total REE abundances (SREEs D 5.5À45.2 ppm). Sample 8626-CDQ-HH2 also has a prominent Eu anomaly (Eu/Eu Ã D 0.62). In terms of their Al 2 O 3 /TiO 2 ratios (30À55) S2 chemistry is evident. The prominent positive Ce anomaly in sample 8527-CDQ-382 (Ce/Ce Ã D 1.46) is unusual but by no means unique, as similar anomalies have been found in cherts from marginal basins that lie between a continental margin and a trench (Armstrong et al. 1999). Also unusual in the younger cherts is the extremely high Al 2 O 3 /TiO 2 ratio (55) of sample 8527-CDQ-255 and, to a lesser extent, 8527-CDQ-382 (34), as these are at odds with their relatively low LREE enrichment and negligible Eu anomalies, which tend to rule out a highly felsic source. It is possible that their Al 2 O 3 / TiO 2 ratio has been affected by a metallic input such as the surprisingly high contribution of hydrothermal Al found along the crest of the East Pacific Rise (Murray 1994). Such a hydrothermal input would be consistent with the high Fe 2 O 3 /TiO 2 , MnO/TiO 2 and Mo/TiO 2 ratios of 8527-CDQ-255 (46.7, 1.2 and 32.5, respectively) compared with 17.9, 0.1 and 2.0 of the older chert (8626-CDQ-HH2). However, it should be noted that Mn may be mobile during diagenesis (Murray et al. 1992;Murray 1994). The two younger samples (8527-CDQ-382 and 8527-CDQ-255) also differ from the slightly older chert in that they have virtually no Eu anomalies, which would suggest a source contribution from an unfractionated volcanic arc system.
Both cherts from Port Macquarie are interpreted to have been deposited in a continental marginal basin as they are enriched in LREEs [(La/Yb) n D 3.8À8.4] with low total REE abundances (SREEs D 8.2À53.1 ppm) and have virtually no Ce anomalies (Ce/Ce Ã D 0.94À1.04). The slightly younger chert (9435-Pmtk-432) possibly formed outboard of the older chert (9435-Pmtk-436) as its total REE content is significantly higher (most likely reflecting a slower burial rate and thus seawater adsorption). Sample 9435-Pmtk-432 also records S1 chemistry. In contrast, sample 9435-Pmtk-436 was probably deposited adjacent to the continental margin, with very low total REE content suggesting a fast burial rate with clastics derived from S2-type source. Volcanic detritus from an immature active margin appears to be present in this chert given that it records a small negative Eu anomaly (Eu/Eu Ã D 0.89).
Narooma terrane cherts are interpreted to have been deposited in a continental marginal basin as suggested by LREE enrichment [(La/Sm) n ] D 1.75À4.51], low total REE abundances (SREEs D 5.6À40 ppm), small Ce anomalies (Ce/Ce Ã D 0.79À1.02) and distinctive Eu anomalies (Eu/Eu Ã D 0.57À0.86). The data indicate that there were three possible sources supplying clastic detritus to the cherts. Late Cambrian and some Early Ordovician cherts contain a terrigenous component more mafic than the S1-type cherts from the other terranes (S0: 15<Al 2 O 3 /TiO 2 21). Other Early and Middle Ordovician cherts record S2 chemistry whereas late Middle Ordovician cherts comprise detritus typical of S1 composition.
Small variations in the near-chondritic Y/Ho ratios in all cherts regardless of tectonic affiliations can be attributed to scavenging by natural marine particulates (Zhang et al. 1994).
Importantly, all the cherts analysed record consistent changes in clastic composition throughout their depositional history. This is best illustrated on a TiO 2 % vs Al 2 O 3 % diagram (Figure 14), which is based around the Early to Middle Ordovician magmatic hiatus in the MacVP of the Lachlan Orogen (Percival & Glen 2007), during which some cherts from the Hermidale, Albury-Bega and Narooma terranes were deposited. The geochemical fingerprint of detrital input into cherts from the Albury-Bega and Hermidale terranes prior to the magmatic hiatus is interpreted as a mixture (S1) of two source components: an average Post-Archean Australian Shale (PAAS; S0) source and an average Upper Continental Crust (UCC; S2) source. In contrast, terrigenous input into cherts from the Narooma terrane during this period only has a PAAS source. During the magmatic hiatus, any evidence of a PAAS source in the Narooma cherts disappears and is replaced solely by the UCC source. At this time, the UCC source is also the sole contributor to cherts of the Hermidale and Albury-Bega terranes. Following the magmatic hiatus, terrigenous chert composition from all tectonic regions reverts back to a mix of the two sources up until the early Pygodus serra zone (1À2 Ma after the end of the hiatus) when detrital input into the Hermidale and Albury-Bega terrane cherts is again sourced solely from the UCC component. This is also the source for the Kiandra cherts and for one of the allochthonous cherts from Port Macquarie that was deposited not long afterwards. In contrast, cherts of the Narooma terrane (and another, younger, allochthonous chert from Port Macquarie) continue to have a bimodal clastic source composition.

TOWARDS A TECTONIC MODEL
All cherts analysed from the Lachlan and New England orogens, as well as those from the Narooma terrane, contain a clay-sized fraction of terrigenous material sourced from a nearby continental margin. Their chemistry suggests that these cherts, spanning an interval from the late Cambrian to early Late Ordovician, were deposited in a moderate-sized basin enclosed by either passive or active margins.
One model that would accommodate the above scenario is a back-arc basin, or a series of horst and graben style sub-basins, underlain by juvenile crust of average andesitic composition, which began rifting in the late Cambrian/Early Ordovician as a result of outboard subduction of the proto-Pacific oceanic plate ( Figure 15). ]. Extension was bounded by old continental crust of the Gondwana margin to the west and ultimately the convergent proto-Pacific plate in the east, and further constrained in between by uplifted fragments of juvenile crust underlying emergent plateaux that were actively eroding. Clastic contents of cherts from the Hermidale and Albury-Bega terranes suggest that they received clay-sized detritus from both sources, whereas the Narooma terrane cherts were deposited much closer to a juvenile plateau from which its clastics are predominantly derived. Probable rollback of the subducting plate during the late Early Ordovician propagated back-arc rift magmatism oceanward, resulting in cessation of magmatism in the MacVP and isolating the juvenile plateau, thereby effectively removing it as a clastic source to cherts from late Bendigonian to mid Darriwilian times. Advancement of the subducting plate brought the locus of the marginal basin back near its original position, initiating Phase 2 magmatism in the MacVP and enabling the juvenile plateau to once again contribute clastics to chert deposition within the basin. However, as rifting re-established itself during the late Darriwilian, cherts of the Hermidale, Albury-Bega and Kiandra terranes and the older chert from Port Macquarie were deposited sufficiently close to the Gondwana margin to receive detritus solely from it. Only cherts from the most outboard terranes within the basin (i.e. Narooma, and the younger chert from Port Macquarie) continued to receive clastics from both margins.
Evidence of primitive arc-related detritus is present in cherts formed from Early to Late Ordovician time but is a relatively minor contribution. Unfractionated arcrelated detritus is found in »16% of all analysed samples from the Hermidale and Albury-Bega terranes and the Kiandra and Port Macquarie regions but is completely absent from the Narooma cherts. This suggests that the active margin source of this volcanic detritus was in fact  the back-arc rather than the arc itself, which was closest to the depositional locus of cherts of the Narooma terrane.
In analysing the distribution of inherited zircon grains in MacVP rocks, with ages indicative of derivation from the Delamerian Orogen, Ross Orogen and East Antarctica, Glen et al. (2011) surmised that the arc lay closer to the continental margin during the Early Ordovician than through the Middle and Late Ordovician, when an intervening marginal basin developed and captured much of the sediment being eroded from the Gondwana craton. We agree that the arc was probably closer to the craton in the Early Ordovician than it was for most of the Middle Ordovician, but our interpretation of the chert geochemistry suggests a landward-stepping of the arc in the late Darriwilian (Da3ÀDa4). Glen et al.
(2011) also discovered that the age populations of inherited detrital zircon grains with a Gondwana signature, recovered from volcaniclastic rocks deposited during the evolution of the MacVP, were identical to zircons present in the coeval quartz-rich Ordovician turbidites. Hence, it is likely that-despite the lack of provenance mixing of volcaniclastic material into the turbidite succession (except in the Kiandra Volcanic Belt)-the turbidite-dominated terranes were not distal to the volcanics, as they tapped much the same source material.

CONCLUSIONS
The main points arising from this first geochemical investigation of Ordovician cherts in southeastern Australia are: (1) The depositional setting interpreted for cherts from the Albury-Bega, Hermidale and Narooma terranes, Kiandra Volcanic Belt, and Port Macquarie Block lay within a marginal basin offshore to the Gondwana craton and underlain by juvenile continental crust with emergent plateaux, both of which contributed terrigenous detritus to the cherts. (2) There are no geochemically discernible differences between Girilambone Group cherts of the Hermidale terrane and Adaminaby Group cherts of the Albury-Bega terrane, suggesting both regions shared identical sources and depositional regimes, occupying near-identical tectonic settings in a back-arc basin. (3) Unfractionated volcanic arc-related contributions to cherts of the Albury-Bega and Hermidale terranes are very minor, with only two samples of late Middle Ordovician (Da3) age from the former region, and an Early Ordovician chert and one of Da3 age from the Girilambone Group showing any suggestion of this. (4) Cherts of the Narooma terrane contain terrigenous sediment similar in composition to detritus of the Albury-Bega and Hermidale terrane cherts during the Early to Middle Ordovician magmatic hiatus in the MacVP, but are otherwise too distal to the Gondwana cratonic margin to receive sediments from it. (5) Probably in part owing to the more continuous depositional history of cherts forming in the Narooma terrane, this area displays clear evidence for consistent changes in terrigenous source component type from late Cambrian (S0) through Early to Middle Ordovician (S2) to latest Middle Ordovician time (S1). (6) Narooma terrane cherts lack a volcanic source component characteristic of an unfractionated arc, although a greater mafic terrigenous contribution was found in the late Cambrian to earliest Ordovician interval. (7) Cherts from the Kiandra area display the strongest evidence for a source contribution from an unfractionated volcanic arc (consistent with field evidence for cherts interbedded with volcaniclastic rocks of the Kiandra Volcanic Belt), but this volcanic component is not evident in the oldest Kiandra chert analysed. (8) One allochthonous chert from Port Macquarie displays similarities to those from Narooma, whereas the other Port Macquarie chert more closely resembles the two younger chert samples from the Kiandra Volcanic Belt, including evidence of a significant input of volcanic detritus from an unfractionated source.
Previous concepts of the Albury-Bega and Hermidale terranes being two distinct and geographically separated regions of Ordovician turbidite deposition (Glen et al. 2009) must now be modified as the chert geochemistry demonstrates widespread uniformity of terrigenous source during Early and Middle Ordovician times across these extensive areas of the Tasmanides. Although significant differences in Late Ordovician deposition remain apparent between these terranes, their Early to Middle Ordovician depositional history and derivation is remarkably similar. Quinn et al. (2014) came to an identical conclusion on the basis of coincidence of timing of significant depositional and magmatic events, demonstrated by an enhanced conodont-based biostratigraphy determined by integrating the occurrence of species found in cherts and contemporaneous limestones in shallow water successions. There is no evidence from this study to support the concept of cherts of the Narooma terrane being deposited on an oceanic plate remote from any continental source of sediment, as suggested in part by the oceanic terrane model of Glen et al. (2004); nor does it support the accretionary prism model originally proposed by Powell (1983), which relied on off-scraping from the advancing oceanic plate. Instead, the chert geochemistry tends to confirm the alternate view put forward by Fergusson (2009, p.184) that the Narooma terrane cherts were deposited in a deep ocean basin on the outboard part of the Gondwana plate.
Finally, it must be recognised that our contribution to the complex development of the Tasmanides is restricted to a detailed (and, we believe, thorough) geochemical analysis of Ordovician cherts that are very well constrained biostratigraphically by conodonts. We do not pretend that the tectonic model synthesised from these data is complete; nor will it resolve longstanding arguments in the literature. However, future tectonic models of the region will need to take the chert story into account.