Insights into the evolution of the Thomson Orogen from geochronology, geochemistry, and zircon isotopic studies of magmatic rocks

Zircon U–Pb ages, eHf(t), and dO isotopic data together with geochemistry and limited Sm–Nd results from magmatic rocks sampled in deep-basement drill cores from undercover parts of the Thomson Orogen provide strong temporal links with outcropping regions of the orogen and important clues to its evolution and relationship with the Lachlan Orogen. SHRIMP U–Pb zircon ages show that magmatism of Early Ordovician age is widespread across the central, undercover regions of the Thomson Orogen and occurred in a narrow time-window between 480 and 470Ma. These rocks have evolved eHf(t)zrn ( 12.18 to 6.26) and eNd ( 11.3 to 7.1), and supracrustal dOzrn (7.01–8.50‰), which is in stark contrast to Early Ordovician magmatic rocks in the Lachlan Orogen that are isotopically juvenile. Two samples have late Silurian ages (425–420Ma), and four have Devonian ages (408–382Ma). The late Silurian rocks have evolved eHf(t)zrn ( 6.42 to 4.62) and supracrustal dOzrn (9.26–10.29‰) values, while the younger Devonian rocks show a shift toward more juvenile eHf(t)zrn, a trend that is also seen in rocks of this age in the Lachlan Orogen. Interestingly, two early Late Devonian samples have juvenile eHf(t)zrn (0.01–1.92) but supracrustal dOzrn (7.45–8.77‰) indicating rapid recycling of juvenile material. Two distinct Hf–O isotopic mixing trends are observed for magmatic rocks of the Thomson Orogen. One trend appears to have incorporated a more evolved supracrustal component and is defined by samples from the northern two-thirds of the Thomson Orogen, while the other trend is generally less evolved and from samples in the southern third of the Thomson Orogen and matches the isotopic character of rocks from the Lachlan Orogen. The spatial association of the Early Ordovician magmatism with the more evolved metasedimentary signature suggests that at least the northern part of the Thomson Orogen is underlain by older pre-Delamerian metasedimentary rocks. ARTICLE HISTORY Received 18 February 2018 Accepted 16 July 2018


Introduction
The Thomson Orogen of eastern Australia (as defined by Fergusson, Henderson, Hutton, & Withnall, 2013;Glen et al., 2013) covers $1,000,000 km 2 but is poorly understood and underexplored due to extensive sedimentary cover that in places reaches thicknesses of 4 km. This has necessitated a reliance on geophysical datasets to advance our understanding of this region and its relationships within the broader Tasmanides of eastern Australia (Burton, 2010;Glen et al., 2010Glen et al., , 2013Korsch et al., 2012;Roach, 2015). Outcropping rocks occur at the margins of the orogen in northern Queensland (Fergusson, Carr, Fanning, & Green, 2001;Fergusson, Henderson, Fanning, & Withnall, 2007a;Henderson, Innes, Fergusson, Crawford, & Withnall, 2011), northwestern New South Wales (Greenfield, Musgrave, Bruce, Gilmore, & Mills, 2011), and as scattered exposures associated with the Eulo Ridge in south-central Queensland (Bultitude & Cross, 2013;Purdy et al., 2016a). Structural, chemical, and geochronological investigations of rocks in these areas and the lack of sample material from the undercover regions have necessarily meant that interpretations from the outcropping regions have been extrapolated to the entire orogen.
A key to a better understanding of the Thomson Orogen is to sample the concealed basement rocks. Throughout the Thomson Orogen, $160 petroleum, stratigraphic, and mineral exploration drill holes have penetrated the cover and retrieved samples of metasedimentary and igneous basement rocks. Most of these basement intersections have been described in detail (Murray, 1994;Brown, Carr, & Purdy, 2012), but until recently, have not been analysed with modern geochronological and isotopic techniques.
Recent work has focused on the metasedimentary units where detrital zircon age spectra and maximum depositional ages from SHRIMP U-Pb dating have helped to characterise Thomson Orogen stratigraphy and provenance (Cross et al., 2015;Cross, Purdy, Bultitude, Brown, & Carr, 2016;Fraser et al., 2014;Kositcin, Purdy, Brown, Bultitude, & Carr, 2015a;Kositcin et al., 2015b;Purdy, Cross, Brown, Carr, & Armstrong, 2016b). However, crucial to understanding the development of the Thomson Orogen is knowledge of the nature, timing, and origin of magmatic events. Igneous rocks have the additional benefit of providing a geochemical and/ or isotopic glimpse of mid to lower crustal source rocks and can therefore provide a better understanding of the tectonic development of the Thomson Orogen and Tasmanides.
Available geochronology for igneous rocks in basement cores is limited to nine samples reported in abstract format (Draper, 2006). The SHRIMP U-Pb zircon ages for these samples range from middle Cambrian to Late Devonian. Although published without thorough peer review, the interpreted ages of these samples have propagated into $20 research papers. The extent to which these ages have been cited is testament to how important they are to any consideration of the Tasmanides.
In this contribution, we fully report the SHRIMP U-Pb geochronology of the nine samples mentioned by Draper (2006) and supplement this work with (1) new SHRIMP U-Pb geochronology from other basement cores and remote outcrops, (2) new zircon and (3) whole-rock geochemistry and Nd isotopic data. These data provide clues to the development of the Thomson Orogen, as well as reveal fundamental differences and similarities with the Lachlan Orogen.

Background geology
The Thomson Orogen formed along part of the Pacific margin of Gondwana from the Neoproterozoic to the Paleozoic. It is a major component of the Tasmanides (Figure 1) that records the growth of eastern Australia following break-up of the Rodinia supercontinent. The Delamerian, Lachlan, Thomson, Mossman, and New England orogens, which comprise the Tasmanides, are divided on the basis of geophysical patterns, structure, and age of component units. To the south, rocks deformed during the Delamerian Orogeny (514-490 Ma; Foden, Elburg, Dougherty-Page, & Burtt, 2006) are divided as a separate component (Delamerian Orogen) from the younger Lachlan Orogen (Champion, 2016, figure 3.10). To the north, this division has not been achieved and it is likely that the Thomson Orogen includes rocks emplaced both before and after the Delamerian Orogeny.
The majority of the Thomson Orogen is concealed below thick sedimentary cover. SHRIMP U-Pb zircon geochronology for metasedimentary rocks sampled from deep drill cores demonstrates strong similarities between the undercover part of the Thomson Orogen and outcropping regions and confirms the presence of at least two stratigraphic packages (Purdy et al., 2016b). The oldest metasedimentary succession is inferred to have been deposited during the late Neoproterozoic to early Cambrian. Outcropping examples occur in the Charters Towers Province (Fergusson et al., 2007a) and Anakie Province (Fergusson et al., 2001), while undercover probable correlatives have been grouped into the Machattie beds by Carr, Purdy, and Brown (2014) and occur near the western margin of the Thomson Orogen (Purdy et al., 2016b;Figure 2). Overlying these rocks is a widespread package of dominantly low-grade, quartz-rich turbiditic metasedimentary rocks deposited from the middle-late Cambrian to Early Ordovician. Outcropping units occur in the Charters Towers and Anakie provinces (Fergusson et al., 2001(Fergusson et al., , 2007a as well as in the undercover parts of the Thomson Orogen, where they are represented by the Thomson beds (Carr et al., 2014;Purdy et al., 2016b; Figure 2).
The magmatic history of the Thomson Orogen extends from the late Neoproterozoic through to the Triassic and several distinct pulses of activity are recognised. The earliest preserved magmatism is represented by mafic schists in the Bathampton Metamorphics (Anakie Metamorphic Group). These are interpreted as late Neoproterozoic in age and associated with a rifting event (Fergusson, Offler, & Green, 2009). As such, they may correlate with the Mount Arrowsmith Volcanics in the Koonenberry Belt of western New South Wales (Delamerian Orogen) (Crawford, Stevens, & Fanning, 1997;Greenfield et al., 2011).
Cambrian magmatism is recorded adjacent to the Thomson Orogen in the Koonenberry Belt of western New South Wales (Mount Wright Arc) and Stavely region of western Victoria (Stavely Arc). This magmatism occurred prior to or during the Delamerian Orogeny and has been attributed to a continental volcanic arc formed along the eastern margin of Gondwana (Cayley, 2011;Greenfield et al., 2011). Volcanic rocks of similar age occur in the subsurface Warburton Basin at the western extremity of the Thomson Orogen (Mooracoochie Volcanics). This has led some workers to support a continuation of the Mount Wright Arc into the Warburton Basin area as the Gidealpa Arc (Gatehouse, 1986;Scheibner & Veevers, 2000); however, the tectonic affinity of volcanic rocks in the Warburton Basin is contested (Purdy, Carr, & Brown, 2013). The relationship between the Warburton Basin and the Thomson Orogen is also unclear although it is likely that they are equivalent (Purdy et al., 2016b).
Following the Delamerian Orogeny, voluminous magmatism in the Thomson Orogen occurred during the extensional Larapinta Event in the late Cambrian and Early to Middle Ordovician (490-460 Ma;Fergusson et al., 2007b). Extensive volcanic and intrusive magmatism is recorded in the Charters Towers Province by the dominantly silicic Seventy Mile Range Group, older components of the Ravenswood Batholith and the Fat Hen Creek Complex (Fergusson et al., 2007a). In the Greenvale Province, it is recorded by the Balcooma Metavolcanic Group and the Lynwater Complex, while in the Anakie Province, it is represented by the Mooramin Granite and Coquelicot Tonalite. Igneous rocks of similar age also occur in the central undercover Thomson Orogen and comprise the informally named, Maneroo volcanics of Carr et al. (2014) and felsic intrusions intersected in AOD Budgerygar 1, LOL Stormhill 1, and AMX Toobrac 1 (Figure 2).
The next major phase of magmatism occurred in the mid Silurian to Devonian (430-395 Ma). In the Charters Towers Province, intrusions of this age are grouped into the Pama Igneous Association and represent the major component of the Ravenswood Batholith (Hutton, Rienks, Tenison-Woods, Hartley, & Crouch, 1994). In the southern part of the Thomson Orogen, late Silurian intrusions are also abundant and include the Tibooburra Suite, Hungerford Granite, and granites intersected in drill holes DIO Ella 1, TEA Roseneath 1, and DIO Wolgolla 1, which form a distinctive batholithic-scale belt of intrusions (Ella belt of Purdy, Hegarty, & Doublier, 2018). Another distinctive pulse of volcanism occurred from 408 to 403 Ma associated with a widespread extensional event resulting in opening of the Adavale Basin and other Devonian basins throughout the Thomson Orogen. This magmatism is recorded by volcanic deposits that comprise the Gumbardo Formation and is dated in drill holes PPC Carlow 1 and PPC Gumbardo 1.
The Middle Devonian (395-380 Ma) saw another period of magmatism in the Thomson Orogen that is recorded by the Retreat Batholith and Theresa Creek Volcanics in the Anakie Province, the Eulo Granite and Currawinya Granite in the southern Thomson Orogen, and crystal-rich ignimbrite in the central Thomson Orogen intersected in drill holes APC Thunderbolt 1 and AAE Towerhill 1. In the Late Devonian (380-360 Ma), voluminous S-type granites that form the Roma granites (Murray, 1994) were emplaced in the Roma Shelf region immediately east of the southern Thomson Orogen.
Granite as young as 235 ± 1.4 Ma (Burton, Trigg, & Black, 2007) and a suite of Permian to Triassic plugs, stocks, and dykes (Dwyer, Collins, Hack, Hegarty, & Huang, 2018) have recently been identified in the southern Thomson Orogen but these are individually small volume features.

Methods
Igneous rocks used in this study include 10 samples from petroleum and stratigraphic drill holes and three samples from surface outcrop (Table 1). SHRIMP U-Pb zircon analyses were carried out on the SHRIMP II ion microprobe at the Australian National University, Canberra, SHRIMP B at the Curtin University of Technology, Perth, Western Australia, and SHRIMP IIe at Geoscience Australia, Canberra. Zircon O and Hf isotopic analyses were completed for 12 of the samples. Oxygen isotopes were measured on SHRIMP II at the Australian National University (ANU) and the d 18 O values are reported relative to VSMOW (Vienna Standard Mean Ocean Water; Baertschi, 1976). The Lu and Hf isotopes were measured on the ANU ThermoFinnigan multi-collector LA-ICP-MS. Abundances of major and trace elements in whole samples were determined at Geoscience Australia, Canberra. Major and minor elements were determined by wavelength-dispersive X-ray fluorescence and selected trace elements and Rare Earth elements were analysed by Inductively Coupled Plasma Mass Spectrometry (ICPMS). Sm-Nd isotope analyses were carried out at the University of Melbourne using a NU Plasma multi-collector ICP-MS.
Full location and sample information, zircon images, analytical procedures, SHRIMP session information, and data for each method are compiled in the Supplementary papers.

Results
Discussed below are the results of zircon U-Pb, O, and Hf isotopic studies of 13 igneous samples from the Thomson Orogen (seven volcanic and six intrusive rocks). Zircons considered to have formed from mantle-like sources have d 18 O values within error of the normal mantle zircon range 5.3 ± 0.6‰ (2r) (Valley et al., 2005), and positive eHf(t) values. External uncertainties for oxygen isotope values reported have an upper limit of 0.50‰. We therefore classify zircons with d 18 O values between 4.2 and 6.4‰ as normal mantle-like zircon. Also discussed are results of wholerock geochemistry and Sm-Nd isotopic studies. Table 1 shows the details of the samples and Table 2 is the summary of the zircon, U-Pb, d 18 O, and initial eHf(t) and whole-rock eNd values for each of the samples. Geochemistry data are included in the Supplementary papers, Table S14.
The U-Pb zircon isotopic results reported here for three samples are a synthesis of full and detailed results reported in Cross et al. (2015;2122055 [Hungerford Granite]; 2122056 [Currawinya Granite]) and Kositcin et al. (2015a;2130082 [AOP Scalby 1]). All other samples were analysed between 2004 and 2005; the ages for nine of these were mentioned in an abstract by Draper (2006), while the results for sample 1574649 (Coquelicot Tonalite) have not previously been reported. SHRIMP U-Pb data are also available from Geoscience Australia's Geochronology Delivery database (http://www.ga.gov.au/geochron-sapubweb/).

(DIO Adria Downs 1)
Sample 1682886 (DIO Adria Downs 1) is an altered, finegrained, welded rhyolitic ignimbrite with geochemical characteristics (elevated Y, HREE, Zn, Zr, Ga/Al 2 O 3 ) that suggest A-type affinities. Zircons recovered comprise a homogeneous population of mostly broken fragments with average diameters of about 30 mm. Intact grains are euhedral with sharp terminations and have aspect ratios that range from 1 to 6. Cathodoluminescence (CL) images show simple oscillatory zoning. Sixty-one analyses were conducted on 61 zircons from this sample ( Figure 3a; Supplementary papers, Table S1.1). Seven grains have common 206 Pb proportions >1 wt% and were removed from interpretations. The remaining analyses have moderate uranium concentrations (median 314 ppm) and Th/U (median 0.77). Two grains are older than the main population and represent inheritance at ca 555 Ma. The remaining 52 analyses have the same age within their analytical uncertainties and yield a weighted mean 206 Pb/ 238 U age of 509.7 ± 3.8 Ma (MSWD ¼ 1.37, POF ¼ 0.04), which is interpreted as the eruptive age for this rock. Twenty-two oxygen and 20 hafnium isotopic analyses were undertaken ( Figure 4; Supplementary papers, Tables S1.2 and S1.3). The d 18 O zrn from 22 analyses ranges from 7.08 to 8.20‰ and is dispersed beyond uncertainty (MSWD ¼ 4.3, POF 0.025). The robust median for this grouping (which ignores the individual uncertainties) is 7.49 þ 0.22/À0.15‰. Initial eHf(t) zrn from 20 analyses range between À6.8 and À0.5 but is significantly scattered (MSWD ¼ 2.84, POF 0.025). Removing the two lowest values removes the scatter (MSWD ¼ 1.32, POF ¼ 0.17) and gives a weighted mean eHf(t) zrn value of À2.82 ± 0.52. The sample has a whole-rock eNd (t) of À7.2.

(GSQ Maneroo 1) -Maneroo volcanics
Sample 1682888 (GSQ Maneroo 1) is an altered porphyritic rhyolite. Similar to the Adria Downs sample, the sample is significantly altered (lost CaO and Na 2 O) and is difficult to classify; however, the unit is most likely an I-type. The zircons from this rock are a homogeneous population of clear to weakly turbid and cracked euhedral prisms. Most have diameters of about 60 mm and aspect ratios between 2.5 and 4. In CL images, the grains have fine-scale oscillatory zoning and about 20% appear to have rounded cores. Fifty-nine analyses on 59 zircons were undertaken ( Figure 3b; Supplementary papers, Table  S2.1). Transient instrument instability affected two analyses that are not considered further. Eight analyses have ages younger than the main population and are considered to have suffered Pb-loss that is possibly related to the altered condition of this rock. There are also two analyses with ages that are older than the main population, one is a core with an age of ca 590 Ma and the other is an unzoned zircon fragment at ca 1105 Ma. The remaining 47 analyses have moderate uranium and Th/U (median 274 ppm and 0.66, respectively) and are all within uncertainty (MSWD ¼ 1.43, POF ¼ 0.03) combining to yield a weighted mean 206 Pb/ 238 U age of 472.0 ± 3.6 Ma. This is interpreted as the crystallisation age of this rock. The results of 20 oxygen isotope and 19 hafnium isotope analyses are displayed in Figure 4 and in Supplementary papers, Tables S2.2 and S2.3). Zircon d 18 O values range from 7.13 to 8.82‰ and do not form a single population (MSWD ¼ 7.9, POF 0.025). Their robust median is 8.10 þ 0.18/À0.14‰. Nineteen initial eHf(t) zrn values from 19 zircons range between À12.8 and À6.3 but are scattered beyond uncertainties (MSWD ¼ 2.8, POF 0.025). A statistically coherent grouping can be made by eliminating the highest and lowest value (MSWD ¼ 1.72, POF ¼ 0.04), which gives a weighted mean eHf(t) zrn of À10.46 ± 0.48. This compares with a whole-rock eNd (t) of À10.0.

(BEA Coreena 1) -Maneroo volcanics
Sample 1682890 (BEA Coreena 1) is a porphyritic dacite. The sample is one of the least altered of the Ordovician volcanic rocks. It has many similarities to 1682888 (GSQ Maneroo 1) (above), and is an I-type, based on its geochemistry. The zircons recovered are a homogeneous population of clear, euhedral and mostly fragmented grains, many with acicular inclusions. Grain diameters are mostly about 70 mm and aspect ratios for the few intact grains are generally between 2 and 3. Sixty analyses were made on 60 grains from this sample ( Figure 3c; Supplementary papers, Table S3.1). An analysis that was affected by instrument instability has been removed from the interpretation. One analysis with an age of ca 525 Ma is significantly older than the main population and represents inheritance. A further analysis is significantly younger than the dominant population and is interpreted to have suffered Pb-loss. The remaining 57 analyses have moderate uranium (median 159 ppm) and Th/U (median 0.61) and combine to yield a weighted mean 206 Pb/ 238 U age of 477. 8 ± 3.5 Ma (MSWD ¼ 1.02, POF ¼ 0.43) for this dacite. Twentytwo oxygen and hafnium isotope analyses were undertaken on zircons from this sample ( Figure 4; Supplementary papers, Tables S3.2 and S3.3). Oxygen isotope values (d 18 O zrn ) range between 6.40 and 7.86‰ and do not form a statistically coherent grouping (MSWD ¼ 4.2, POF 0.025). The robust median for all 22 analyses is 7.01 þ 0.12/À0.23‰. Initial eHf(t) zrn from 22 analyses range from À8.1 to À3.4 and combines to give a weighted mean value of À6.26 ± 0. 29 (MSWD ¼ 0.89, POF ¼ 0.60). The initial whole-rock eNd (t) is À7.1.

(PPC Carlow 1) -Maneroo volcanics
Sample 1574653 (PPC Carlow 1) is an altered, felsic autoclastic volcanic. The geochemistry of this sample is  equivocal. It is significantly less siliceous than other Ordovician volcanic samples (63 wt% SiO 2 ). The sample is altered and possibly mineralised (8600 ppm S). The mounted zircons can be separated into two broad types in roughly equal proportions: (1) fragments with crystal faces and more rarely intact grains and (2) rounded (sometimes pitted) and subhedral grains. In CL images, the euhedral population has a moderate luminescence with fine, faint oscillatory growth zoning; the rounded-subrounded grains are variable in their luminescence and internal zoning features. Fifty-three analyses were collected on zircons from this sample. Thirty-nine SHRIMP U-Pb analyses preferentially targeted the euhedral fragments and intact grains. Twelve analyses that are significantly older than the dominant population have a range of ages between ca 1603 and ca 542 Ma. The majority of these grains have rounded to subhedral morphologies and are interpreted to represent inheritance. Two further pre-1000 Ma grains are >10% discordant and removed from interpretation. The 39 U-Pb analyses that targeted the more euhedral fragments and intact grains have moderate uranium (median 236 ppm) and Th/U (median 0.66) and combine to give a weighted mean 206 Pb/ 238 U age of 479.7 ± 7.4 Ma (MSWD ¼ 1.37, POF ¼ 0.06, Figure 3; Supplementary papers, Table S4.1). Zircon d 18 O compositions from 20 analyses range from 7.52 to 9.04‰ and do not form a single population within their uncertainties (MSWD ¼ 9.9, POF 0.025). They have a combined robust median of 8.50 þ 0.14/À0.33‰. The eHf(t) zrn values from 18 grains range between À13.5 and À7.4. Three analyses are considered outliers and once removed, result in a grouping of 15 analyses (MSWD ¼ 1.6, POF ¼ 0.074) that give a weighted mean eHf(t) zrn value of À12.18 ± 0.50 ( Figure 4; Supplementary papers, Tables S4.2 and S4.3). Initial whole-rock eNd (t) is À11.3.

(AMX Toobrac 1)
Sample 1574651 (AMX Toobrac 1) is a biotite-muscovite monzogranite. Although altered, the unit has characteristics (e.g. elevated P 2 O 5 , low Zr, LREE, Th/U) that strongly suggest it is S-type (Champion & Bultitude, 2013). Only 52 zircons were separated from this sample and comprise a heterogeneous mix of sizes, shapes, clarity, and colour. Almost all grains are either broken fragments or rounded to subrounded and many are variably cracked. They are clear to partly turbid and colourless to light brown with grain diameters ranging between $20 and $100 mm. Luminescence is weak to strong and some grains have a weakly luminescent rim that is visible in reflected light images as zones of likely metamict zircon. Internal zoning patterns are predominantly oscillatory to planar and about half of the grains appear to have rounded cores. Of the 28 U-Pb analyses undertaken, seven were removed from interpretation. One analysis was affected by transient instrument instability, two pre-1000 Ma grains are >10% discordant and a further four analyses are significantly younger than the dominant population and are interpreted to have undergone Pb-loss. The overall poor quality of recovered zircons likely contributed to the relatively high proportion of grains interpreted to have suffered Pb-loss. Five analyses are older than the dominant population and have ages of ca 2378, ca 1154, ca 1013, ca 985, and ca 609 Ma, are interpreted as inheritance. The remaining 16 zircons have moderate uranium concentrations (median 305 ppm) and Th/U ratios (median 0.64) and contribute to a weighted mean 206 Pb/ 238 U age of 471.8 ± 9.1 Ma (MSWD ¼ 1.72, POF ¼ 0.04, Figure 3e; Supplementary papers, Table S5.1). This is interpreted as the crystallisation age of this rock. Owing to the scarcity and poor quality of grains recovered from this sample, only seven oxygen and hafnium isotope analyses were undertaken ( Figure 4; Supplementary papers, Tables S5.2 and S5.3). Of these, two were carried out on grains interpreted to have undergone Pb-loss and are not considered. The remaining five zircon d 18 O values range from 7.85 to 9.77‰ but are significantly scattered (MSWD ¼ 32, POF 0.025). These analyses have a robust median value of 8.30 þ 1.47/À0.44‰. Initial eHf(t) values have a wide range between À21.5 and À9.1 and are significantly dispersed (MSWD ¼ 48). The lowest value appears to be an outlier and after removing it, the remaining four are only slightly dispersed in their eHf(t) (MSWD ¼ 3.19, POF ¼ 0.023) and have a weighted mean eHf(t) value of À10.63 ± 2.72. Initial wholerock eNd (t) is À9.1.

(Coquelicot Tonalite)
Sample 1574649 (Coquelicot Tonalite) is a foliated biotitehornblende tonalite. This is the least altered Ordovician sample in this study. It has the lowest silica (59.5 wt%) of the samples in this study and most probably represents a fractionated mantle melt. The zircons recovered are a homogeneous population of clear, light brown grains with average diameters of about 60 mm and aspect ratios between 2 and 2.5. They have a moderate luminescence and are oscillatory zoned. Thirty-seven analyses were made on zircons from this sample. One analysis has anomalous common 206 Pb and is removed from interpretation. The remaining 36 analyses have moderate uranium (median 179 ppm) and Th/ U (median 0.66) and combine to yield a weighted mean 206 Pb/ 238 U age of 471.0 ± 3.8 Ma (MSWD ¼ 0.53, POF ¼ 0.99, Figure 3f; Supplementary papers, Table S6.1). This is interpreted as the crystallisation age of this rock.

(DIO Ella 1) -Ella Granite
Sample 1586685 (DIO Ella 1) is a biotite-muscovite monzogranite. Like the older Toobrac 1 granite, the Ella 1 granite has characteristics, e.g. elevated ASI, P 2 O 5 , low Zr, LREE, Th/U, that strongly suggest it is an S-type. The recovered zircons are euhedral, clear to partly turbid and colourless to light brown. Average diameter is $40 mm and aspect ratios typically range from 3 to 4 but extend to 8. They have a moderate to strong luminescence, are mostly oscillatory zoned and the majority of grains appear to have rounded cores. Forty-seven analyses were undertaken on 45 zircons (Figure 3g; Supplementary papers, Table S7.1). Two analyses are significantly younger than the dominant population and are interpreted to have lost radiogenic Pb. There are also two pre-1000 Ma grains that are >10% discordant and have also been removed from further interpretation. Seven analyses of cores are significantly older than the dominant population and have ages of ca 1215, ca 1069, ca 1020, ca 620, ca 532 (n ¼ 2), and ca 456 Ma and are interpreted as inheritance. The dominant zircon population has moderate uranium concentrations (median 495 ppm) and moderate to low Th/U ratios (median 0.16). This grouping comprises 36 analyses that are the same within uncertainty (MSWD ¼ 1.51, POF ¼ 0.03) and combine to yield a 206 Pb/ 238 U age of 425.4 ± 6.6 Ma, which is interpreted as the crystallisation age of this rock. Results of 15 oxygen and 16 hafnium isotope analyses are displayed in Figure 4; Supplementary papers, Tables S7.2 and S7.3. The d 18 O zrn values range from 9.21 to 10.87‰ and are significantly scattered (MSWD ¼ 12.13, POF 0.025). Combined, they have a robust median value of 10.29 þ 0.09/À0.52‰. Initial eHf(t) from 16 zircons ranges between À9.7 and À3.4 and are also scattered beyond uncertainty (MSWD ¼ 4.42, POF 0.025). The scatter is eliminated by removing two analyses (MSWD ¼ 1.52, POF ¼ 0.10) resulting in a combined weighted mean eHf(t) zrn value of À6.42 ± 0.72.

(Hungerford Granite)
Sample 2122055 (Hungerford Granite) is a muscovitebiotite monzogranite. The low Na 2 O, high K 2 O/Na 2 O and elevated ASI of this relatively unaltered unit suggest it is an S-type. Zircons recovered are clear and colourless euhedral grains with aspect ratios between 2 and 4 and diameters between $40 and $120 lm. About 50% are broken fragments. They have a moderate luminescence and display oscillatory to planar zoning. Twenty-three analyses were carried out on 23 zircons from this sample. Uranium concentrations and Th/U ratios are moderate (medians 405 ppm and 0.30, respectively). All 23 zircons analysed have the same radiogenic 206 Pb/ 238 U within their analytical uncertainties (MSWD ¼ 0.94, POF ¼ 0.54, Figure 3h; Supplementary papers, Table S8.1) and combine to yield an age of 419.1 ± 2.5 Ma, which is interpreted as the crystallisation age of this rock. Results of 22 oxygen and 19 hafnium zircon analyses are displayed in Figure 4 and Supplementary papers (Tables S8.2

(PPC Carlow 1) -Gumbardo Formation
Sample 1682891 (PPC Carlow 1) is a rhyolitic ignimbrite. The geochemical classification of this sample is uncertain. It is slightly less siliceous than other volcanic samples although this in part reflects greater hydration and alteration. Zircons from this rock mostly comprise euhedral grains $40 mm in diameter and their broken equivalents. Aspect ratios range from 2 to 3 but range up to 5. They have a moderate to bright luminescence with faint oscillatory zoning and rare cores. Fifty-nine analyses were undertaken on 59 grains. One analysis is significantly younger than the dominant population and interpreted to record Pb loss. The 46 analyses from the dominant population have moderate uranium (median 301 ppm) and Th/U ratios (median 0.63) and combine to yield a 206 Pb/ 238 U age of 408.1 ± 3.1 Ma (MSWD ¼ 1.35, POF ¼ 0.06). This is interpreted to be the eruptive age for this ignimbrite. There is also a single grain at ca 551 Ma and another population of inherited grains (n ¼ 11) that have similar morphological and internal zoning patterns as well as uranium and Th/U ratios (medians 207 ppm and 0.61, respectively) to the younger dominant population. These 11 analyses are the same within uncertainty (MSWD ¼ 1.82, POF ¼ 0.05) and combine to give a 206 Pb/ 238 U age of 474.6 ± 5.4 Ma ( Figure  3i; Supplementary papers, Table S9.1). Hafnium and oxygen isotope analyses were undertaken on 15 zircons from the dominant population (Figure 4; Supplementary papers, Tables S9.2 and S9.3). The d 18 O zrn values from 14 analyses range from 5.69 to 6.67‰ but are significantly scattered (MSWD ¼ 4.29, POF 0.025) and have a robust median value of 6.30 þ 0.24/À0.26. An additional analysis of an inherited grain with an age of ca 466 Ma has a d 18 O value of 6.27 ± 0.33‰. Hafnium compositions were determined on 13 zircons from the magmatic population and two zircons from the inherited grouping. Initial eHf(t) values for the magmatic grouping ranges between À5.6 and À0.2, and when combined, are heterogeneous (MSWD ¼ 2.10, POF 0.025). The heterogeneity can be removed by eliminating the lowest analysis and leaves a weighted mean eHf(t) zrn value of 1.73 ± 0.82 (MSWD ¼ 1.70, POF ¼ 0.07). In contrast, the two Hf isotopic values on zircons from the inherited ca 475 Ma grouping have eHf(t) zrn values of À9.6 and À10.5.

(PPC Gumbardo 1) -Gumbardo Formation
Sample 1682892 (PPC Gumbardo 1) is a rhyolitic ignimbrite. Geochemistry of this unit is equivocal. Although the ASI suggests it may be an S-type, this may simply reflect some redistribution of CaO and Na 2 O. The mounted zircons are a homogeneous population of euhedral, clear, colourless, and generally equant grains with average diameters of about 90 mm. All grains have fine oscillatory growth zoning. Eighty analyses were undertaken on 80 zircons (Figure 3j; Supplementary papers, Table S10.1). One analysis is significantly younger than the dominant population and interpreted to have undergone some Pb-loss. There are also three grains at ca 420 Ma that are significantly older than the dominant population and represent inheritance. The remaining 76 analyses have moderate uranium concentrations and Th/U ratios (medians 182 ppm and 0.86, respectively) and form a single population with a weighted mean 206 Pb/ 238 U age of 402.9 ± 2.9 Ma (MSWD ¼ 1.21, POF ¼ 0.11). This is interpreted as the eruptive age of this rock. The results of 19 hafnium and oxygen isotope zircon analyses are displayed in Figure 4 and in Supplementary papers (Tables S10.2 and S10.3). Nineteen d 18 O zrn values range between 5.64 and 6.55‰. Removing the highest analysis leaves a single population (MSWD ¼ 1.67, POF ¼ 0.04) that gives a mean value of 6.01 ± 0.15‰. Nineteen initial eHf(t) zrn values range from À0.6 to þ1.9 and form a single population (MSWD ¼ 1.17, POF ¼ 0.27) with a weighted mean value of 0.51 ± 0.32.

(AAE Towerhill 1) -Unnamed unit
Sample 1585223 (AAE Towerhill 1) is a rhyolitic ignimbrite. The elevated Na 2 O and low K 2 O of this sample suggest it is an I-type. Zircons recovered comprise a homogeneous population of euhedral, clear, colourless to light-brown grains, many with acicular inclusions. Average grain diameters are about 60 mm and the majority are broken fragments. The few intact grains have aspect ratios between 2 and 3, but larger broken grains indicate aspect ratios of >6. They have a moderate luminescence and oscillatory growth zoning. Two analyses are significantly younger than the dominant population and are interpreted to have undergone Pb-loss. The remaining 45 U-Pb zircon analyses have moderate uranium (median 173 ppm) and Th/U ratios (median 0.71) and have identical radiogenic 206 Pb/ 238 U within their uncertainties (MSWD ¼ 1.09, POF ¼ 0.32) and yield an age of 381.7 ± 5.7 Ma, which is interpreted as the eruptive age of this rock (Figure 3k; Supplementary papers, Table S11.1). The results of 20 oxygen and 21 hafnium isotope analyses for this sample are shown in Figure 4 and in the Supplementary papers (Tables S11.2 and S11.3). Zircon d 18 O values range between 4.97 and 6.28‰ and do not form a single population (MSWD ¼ 6.09, POF 0.025). The robust median for all values is 5.78 þ 0.20/À0.14‰. Hafnium isotope values have eHf(t) values that range from À3.4 to þ1.6 but together do not constitute a single population (MSWD ¼ 2.0, POF 0.025). Removing two analyses removes the scatter (MSWD ¼ 1.39, POF ¼ 0.13) and gives a weighted mean initial eHf(t) zrn value of À1.72 ± 0.36. This unit has a whole-rock eNd (t) value of À3.5.

(Currawinya Granite)
Sample 2122056 (Currawinya Granite) is a biotite monzogranite. The Currawinya Granite is geochemically quite similar to the volcanic rock 1585223 (AAE Towerhill 1), although has lower Na 2 O and higher K 2 O. It is also most likely an I-type. The zircons recovered are clear and colourless euhedral prisms and their broken equivalents. They have a wide range in size from $20 to $120 lm in diameter with aspect ratios from 1 to 9. They have a moderate luminescence and are predominantly oscillatory zoned. Common Pb proportions are generally below 1 wt%; however, the grain with the highest 206 Pb c (2.8 wt%) also has the lowest 206 Pb/ 238 U ratio and is therefore been removed from further consideration. One of the remaining 21 analyses is of a distinct core that is significantly older (ca 409 Ma) than the dominant population and is interpreted as inherited. The remaining 20 analyses have moderate uranium concentrations and Th/U ratios (medians 262 ppm and 0.50, respectively). These analyses have the same radiogenic 206 Pb/ 238 U within their analytical uncertainties (MSWD ¼ 0.53, POF ¼ 0.95) and combine to give an age of 381.5 ± 2.5 Ma, which is interpreted as the crystallisation age of this rock (Figure 3l; Supplementary papers, Table  S12.1). Twenty oxygen and 17 hafnium isotope analyses were undertaken on zircons from this sample (Figure 4; Supplementary papers, Tables S12.2 and S12.

(AOP Scalby 1) -Scalby Granite
Sample 2130082 (AOP Scalby 1) is a biotite-muscovite granite. The Scalby Granite has characteristics (e.g. elevated ASI, K 2 O/Na 2 O, P 2 O 5 , low LREE, Th/U) that indicate that it is an S-type. Zircons recovered are clear and colourless, euhedral, with subordinate subhedral and rounded grains. Grains range from $80 up to $300 lm in length with aspect ratios commonly between 1 and 2 but ranging up to 4. Almost all have some crystal faces and some have pointed prismatic terminations, some of which surround a rounded core. Fewer grains are almost completely rounded and display faint signs of pitting visible in transmitted light. Most grains have a moderate luminescence, although the full range extends from strong to weak, and all grains either display sector or oscillatory zoning. Thirty-four SHRIMP U-Pb analyses were undertaken on 30 zircons from this sample. Of these, one analysis was adversely affected by instrumental instability during data acquisition and is not considered further. Another three with 206 Pb c > 1.5 wt% are also removed from further consideration. A further analysis with a pre-1000 Ma age is >10% discordant and also removed from interpretation. Three analyses of zircon cores are significantly older than the main population and represent inheritance at ca 1240, ca 570, and ca 540 Ma. The remaining 26 analyses have moderate uranium concentrations (median 520 ppm) and low Th/U (median 0.10). All 26 analyses are within uncertainty (MSWD ¼ 1.33, POF ¼ 0.12) and have a 206 Pb/ 238 U age of 368.4 ± 2.5 Ma, which is interpreted to record the igneous crystallisation age of this rock (Figure 3m; Supplementary papers, Table S13.1). Nineteen oxygen and 17 hafnium isotopic analyses were undertaken on zircons from this sample (Figure 4; Supplementary papers, Tables S13.2 and S13.3). d 18 O zrn values range from 7.08 to 10.38‰ and are dispersed beyond their uncertainties (MSWD ¼ 9.73, POF 0.025). Together they yield a robust median value of 8.77 þ 0.12/À0.30‰. The 17 hafnium isotopic analyses are spread between negative and positive eHf(t) values (À4.7 to 1.0) and do not form a single population within uncertainty (MSWD ¼ 6.86, POF 0.025). They have a robust median initial eHf(t) zrn value of 0.01 þ 0.46/À2.0.

SHRIMP U-Pb reinterpretation of mounts Z4497 and Z4638
The SHRIMP U-Pb zircon ages for the nine samples mentioned by Draper (2006) and reprocessed with SQUID 2.50 as a part of this work, are all within analytical uncertainty of the original age interpretations. Minor differences between the two sets of age interpretations and their uncertainties are the result of different (1) SHRIMP U-Pb data reduction software, (2) steps used for the propagation of uncertainties related to multiple 206 Pb/ 238 U calibration sessions, (3) propagation of uncertainties associated with establishing 95% confidence limits, and (4) human interpretation of complex zircon age patterns. Importantly, the original ages noted by Draper (2006) were processed using the MS Excel-based, SQUID processing package of Ludwig (2001), which in some cases, underestimates uncertainties associated with single spots by factors of between 1.2 and 1.3 (Ludwig, 2009a). Using SQUID 2.50 (Ludwig, 2009b) corrects this issue and establishes better comparability to existing data from the Thomson Orogen (Cross et al., 2015;Cross, Purdy, Bultitude, Brown, & Carr, 2016;Fraser et al., 2014;Kositcin et al., 2015a;Kositcin et al., 2015b;Purdy et al., 2016b).
Where available, whole-rock initial eNd values correlate very well with zircon eHf values, as is to be generally expected (Vervoort, Blichert-Toft, Patchett & Albar ede, 1999;Vervoort, Plank, & Prytulak, 2011), although agreement is closest using the earlier model of Vervoort, Blichert-Toft, Patchett, and Albar ede (1999). In the two samples with most disagreement (1682886 DIO Adria Downs 1 and 2122056 Currawinya Granite), zircon eHf(t) values are more juvenile than expected, by $4 epsilon units. Interestingly, apart from one sample 2122056 (Currawinya Granite), there is a very good negative linear correlation (correlation coefficient of 0.94) between the available whole-rock eNd (Ordovician and one Devonian sample) and zircon d 18 O values (Table 2). Notably, the units with the most mantle-like oxygen (the Devonian I-type units [1585223 AAE Towerhill 1 and 2122056 Currawinya Granite] with 5.78 and 7.45‰, respectively) still have significantly evolved Nd, with eNd values of À3.5 to À3.4, a point we return to below.
There is also a reasonable correlation between d 18 O zrn values and assigned granite types, which given the obvious alteration in many samples, is best viewed as preliminary. Iand A-type igneous rocks range from 5.78 to 8.10‰ (highest values in the Ordovician rocks), in comparison to 8.30 to 10.29‰ for the S-type rocks. There is one exception. The Devonian sample 1682892 (PPC Gumbardo 1), thought to possibly be an S-type, has mantle-like d 18 O value (6.01‰) and so the suggested S-type geochemistry may actually reflect alteration. Regardless, the spread in d 18 O values, especially for the Ordovician I-type units, suggests involvement of significant supracrustal material in their genesis (as discussed in more detail below).

Regional correlations
The U-Pb ages for the 13 magmatic samples reported range from the middle Cambrian to the Late Devonian and correlate well with other magmatic events recorded in the Thomson Orogen and adjacent parts of the Delamerian Orogen.

Cambrian
The oldest sample reported is a rhyolitic ignimbrite with an age of 509.7 ± 3.8 Ma from southwestern Queensland 1682886 (DIO Adria Downs 1). This age is within uncertainty of volcanism recorded in basal volcanics of the Warburton Basin (Mooracoochie Volcanics 517 ± 9 Ma, PIRSA, 2007), along the Cambrian Delamerian margin in the Gnalta Group of western New South Wales (ca 510 Ma, see Greenfield et al., 2011) and Mount Stavely Volcanic Complex of western Victoria (510-500 Ma; Lewis, Taylor, Cayley, Schofield, & Skladzien, 2015;Lewis, Cayley, Duncan, Schofield, & Taylor, 2016). Carr et al. (2014) assigned the rhyolitic ignimbrite from DIO Adria Downs 1 as an extension of the Mooracoochie Volcanics, but the relative isolation of this sample ( Figure  1) renders any correlations tenuous. Although components of the Mount Wright Arc in northwest NSW are outlined in geophysical images (Greenfield et al., 2011), little is revealed about the continuation of any such feature into the Warburton Basin area and further into the western Thomson Orogen. The age of the ignimbrite in DIO Adria Downs 1 is consistent with belonging to the Delamerian Orogeny but the interpreted A-type geochemistry suggests an extensional non-arc environment, as suggested by Champion (2016) for this unit and also the Mooracoochie Volcanics. The negative eHf(t) zrn and whole-rock eNd values together with the supracrustal d 18 O zrn values from sample 1682886 (DIO Adria Downs 1) ( Table 2) indicate derivation from an evolved supracrustal source. These values are broadly similar to the limited zircon eHf(t) and d 18 O data reported for other Delamerian rocks, presented by Kemp, Hawkesworth, Collins, Gray, and Blevin (2009) and suggest derivation from a similar crustal domain (Figure 4).

Early Ordovician
Five of the igneous rocks analysed here have Early Ordovician crystallisation ages that range from 480 to 470 Ma. Three volcanic samples from the undercover central Thomson Orogen (1682888 GSQ Maneroo 1, 1682890 BEA Coreena 1 and 1574653 PPC Carlow 1; Table 2) have been assigned by Carr et al. (2014) to the informally named Maneroo volcanics and are similar in age and lithology to exposed felsic volcanics in the Charters Towers Province (Seventy Mile Range Group) and Greenvale Province (Balcooma Metavolcanic Group). The Seventy Mile Range Group and Balcooma Metavolcanic Group have SHRIMP U-Pb ages of 479 ± 5 Ma (Perkins, McDougall, & Walshe, 1993) and 480 ± 4 Ma (Rea, 2000;cited in Jell, 2013), respectively, and host significant copper-lead-zinc-gold-silver-bearing volcanic massive sulfide style deposits at the Thalanga, Balcooma and other mines. Importantly, the age obtained for altered felsic volcanic sample 1574653 (PPC Carlow 1) (479.7 ± 7.4 Ma) confirms the presence of Early Ordovician magmatic Thomson Orogen rocks below the Adavale Basin (Draper, 2006).
The biotite-muscovite granite, sample 1574651 (AMX Toobrac 1) in the undercover central Thomson Orogen and the foliated biotite-hornblende tonalite sample 1574649 (Coquelicot Tonalite) from the outcropping Anakie Province, have ages of 471.8 ± 9.1 and 471.0 ± 3.8 Ma, respectively. These new ages are indistinguishable from SHRIMP U-Pb zircon ages from two other granites intersected in drill holes in the central Thomson Orogen (AOD Budgerygar 1: 476.9 ± 3.5 Ma; LOL Stormhill 1: 477.9 ± 2.7 Ma; Si egel et al., 2018). Crustal involvement is also evident by the five inherited zircons analysed from 1574651 (AMX Toobrac 1), which range from ca 2378 to 609 Ma indicating magma interaction with the Thomson beds or similar sedimentary source (Purdy et al., 2016b). This is consistent with the interpreted S-type geochemistry.
The Ordovician extrusive and intrusive I-and S-type rocks documented here and by Si egel et al. (2018), extend the known distribution of magmatism in the Thomson Orogen during this time. Similar aged (latest Cambrian to Early Ordovician) I-and S-type intrusive magmatism is also known in the Charters Towers and Lolworth batholiths of the northeastern Thomson Orogen (Fergusson et al., 2013). S-type magmatism is also documented in the Anakie Province, in the same general region as the Coquelicot Tonalite. Early Ordovician I-type tonalites and granites also occur in the Greenvale Province and surrounds (Fergusson et al., 2013). Magmatism during this time was synchronous with deposition of widespread turbiditic deposits (e.g. Warratta Group, Greenfield, Gilmore, & Mills, 2010;Thomson beds, Purdy et al., 2016b).
Both sample 1586685 (DIO Ella 1) and sample 2122055 (Hungerford Granite) have negative eHf(t) (À6.42 and À4.62, respectively) values coupled with significantly elevated d 18 O zrn (10.29 and 9.26‰, respectively) values consistent with the interpreted S-type nature of these granites. Interestingly, zircon eHf(t) values are not as evolved as those seen in the Ordovician igneous units, suggesting source components for the Ordovician and Silurian igneous rocks were not the same (Figure 4a).

Devonian
The Gumbardo Formation is a sequence of volcanic and volcaniclastic rocks that forms the lowermost unit of the entirely subsurface Adavale Basin (see Champion, 2016, figure 1.2), (McKillop, McKellar, Draper, & Hoffman, 2005. The U-Pb geochronology of two rhyolitic ignimbrite samples from the Gumbardo Formation, 1682891 (PPC Carlow 1), 408.1 ± 3.1 Ma and 1682892 (PPC Gumbardo 1), 402.9 ± 2.9 Ma, constrain the lower age and initial phases of basin opening to approximately Pragian (Draper 2006). These ages are within error of other Gumbardo Formation data recently acquired by Asmussen, Bryan, Allen, and Purdy (2018). The inherited zircon population in sample 1682891 (PPC Carlow 1) at ca 475 Ma may have been derived from a local volcanic or plutonic source that was mixed with the ignimbrite during eruption or the later phases of magmatic evolution. These inherited grains have eHf(t) values similar to that of the other Early Ordovician zircons analysed, but an oxygen isotope analysis from only one inherited grain has a d 18 O zrn value of 6.27‰, which is at the lower end for the Early Ordovician grains analysed ( Figure 4). In comparison, magmatic grains from this sample have a mantle-like d 18 O zrn value (6.30‰) and a zircon eHf(t) value (À1.73) that is close to the CHUR line and supports a derivation from a predominantly juvenile magma source (Figure 4). Magmatic zircons in sample 1682892 (PPC Gumbardo 1) have a very similar d 18 O zrn (6.01‰) value and an even more juvenile eHf(t) zrn (0.51) value. These samples show a distinct shift toward a derivation from more juvenile sources during the late Early Devonian in comparison to the late Silurian to earliest Devonian granitic samples from 1586685 (DIO Ella 1) and 2122055 (Hungerford Granite) where zircon eHf(t) and d 18 O values demonstrate derivation from evolved source magmas (Figure 4).
The rhyolitic ignimbrite, 1585223 (AAE Towerhill 1) lies in the northeast of the undercover Thomson Orogen (Figure 1) and its age of 381.7 ± 5.7 Ma is similar to another nearby, undercover, rhyolitic ignimbrite, which has an SHRIMP U-Pb age of 392.9 ± 2.7 Ma (APC Thunderbolt 1; Kositcin et al., 2015a). These samples lie below the mid Carboniferous to late Middle Triassic Galilee Basin and Mesozoic Eromanga Basin and represent a Devonian felsic volcanic province in this region (Kositcin et al., 2015a).
Zircons from sample 1585223 (AAE Towerhill 1) have the lowest d 18 O zrn values from the sample suite presented here, marginally negative eHf(t) zrn values and a negative whole-rock initial eNd value. Further south, sample 2122056 (Currawinya Granite) (an outcropping unit on the Eulo Ridge basement high) has a similar magmatic age of 381.5 ± 2.5 Ma. This sample has positive eHf(t) zrn values but supracrustal d 18 O zrn values. Intriguingly, it has a negative whole-rock initial eNd value of À3.42 that suggests a decoupling between the eNd and eHf(t) zrn values for this sample.
The Late Devonian age of 368.4 ± 2.5 Ma for 2130082 (AOP Scalby 1) has a strong temporal relationship with similar granites in the nearby Roma Shelf region that Murray (1994) interpreted to have crystallised between 360 and 355 Ma. Zircons from this sample have eHf(t) values that straddle the CHUR line, and supracrustal d 18 O values. These features are similar to the results from a nearby basement granite from the Roma Shelf, intersected in SSL Javel 2 and analysed by Si egel (2015) (Figure 4).

Zircon Hf-O time trends
The zircon U-Pb, Hf, and O isotopes presented here for 13 samples from the Thomson Orogen span from 510 to 368 Ma, and allow for some preliminary interpretations of the apparent, broad isotopic time trends as well as comparison with the adjacent Delamerian and Lachlan orogens.
The negative eHf(t) zrn and whole-rock eNd isotope values together with the elevated d 18 O zrn values from sample 1682886 (DIO Adria Downs 1) indicate derivation from an evolved supracrustal source. These values are broadly similar to the limited zircon eHf(t) and d 18 O values reported for other Delamerian rocks, presented by  Tables 1 and 2) and suggest derivation from a similar crustal domain. In the Delamerian Orogen, Cambrian magmatism is interpreted to be associated with a west-dipping subduction zone that was formerly continuous along the eastern Gondwana margin (Foden et al., 2002). With only one Delamerian-aged sample in our dataset, any correlation between sample 1682886 (DIO Adria Downs 1) with the Delamerian Orogen cannot be properly assessed.
Following the Delamerian Orogeny, a widespread extensional event affected the Tasmanides and extended into central Australia during the late Cambrian and Early Ordovician. Research focused on the outcropping Tasmanides associates this extension with slab roll back and a broadly back-arc setting (Fergusson et al., 2007a(Fergusson et al., , 2007bGlen, Vaughan, Leat, & Pankhurst, 2005;Henderson, 1986;Stolz, 1995). In the Thomson Orogen and central Australian basins this is known as the Larapinta Event. During this time, protoliths of the vast low-grade metasedimentary rocks, which typify much of the Thomson Orogen, were deposited (Purdy et al., 2016b). Igneous rocks emplaced during this time vary substantially in whole-rock eNd(t), eHf(t) zrn and d 18 O zrn values between the northern (Thomson Orogen) and southern Tasmanides (as represented by the Macquarie Arc, Lachlan Orogen).
Samples from the Lachlan Orogen emplaced during the Larapinta Event have distinctly juvenile eHf(t) zrn , whole-rock eNd(t), and mantle-like d 18 O zrn values (Figure 4). This is consistent with their spatial location and emplacement within the Macquarie Arc or rift, a series of structural belts comprising volcanic, intrusive, volcaniclastic, and carbonate rocks variably interpreted as an oceanic island arc accreted to the eastern Gondwana margin (Glen et al., 2005;Glen, Walshe, Barron, & Watkins, 1998;Glen, Poudjom Djomani, Korsch, Costelloe, & Dick, 2007) or a zone of localised rifting within an established marginal or back-arc basin (Quinn, Percival, Glen, & Xiao, 2014;Wyborn, 1992).
In stark contrast are four Early Ordovician magmatic samples reported here from the undercover central Thomson Orogen that have evolved eHf(t) zrn and elevated, supracrustal d 18 O, as well as negative initial whole-rock eNd(t) values, demonstrating derivation from evolved crustal sources (Figure 4; Table 2). Thus, while magmatism in the southern Tasmanides was clearly sourcing juvenile material, magmatism throughout the central Thomson Orogen at this time involved a much greater metasedimentary component. This likely derived from a pre-Thomson beds, sedimentary pile thickened during Delamerian orogenesis that was heated and reworked during Ordovician extension (Collins & Richards, 2008) and suggests that at least part of the Thomson Orogen contains pre-Delamerian material, as suggested by Champion (2016). This postulated older package could be the Machattie beds, which Purdy et al. (2016b) suggest represents an extension of the Centralian Superbasin into eastern Australia during the late Neoproterozoic to early Cambrian. Fundamentally, the isotopes confirm significant contrasts in the composition of the deeper crust between the Thomson and Lachlan orogens. This distinction, however, is not universal. More juvenile Ordovician magmatism is also recorded in the northeastern Thomson Orogen, in both the Greenvale (Champion, 2013) and northeastern Charters Towers (Stolz, 1995) regions, interpreted to be arc or back-arc related (Fergusson et al., 2013;Stolz, 1995 (Figure 4). This apparent trend toward near-juvenile to juvenile eHf(t) zrn is also shown by the Late Devonian samples, 1585223 (AAE Towerhill 1), 2122056 (Currawinya Granite) and 2130082 (AOP Scalby 1), which have eHf(t) zrn values between À1.72 and 1.92. However, except for 1585223 (AAE Towerhill 1), which has distinctly mantle-like d 18 O zrn (5.78‰) value, 2122056 (Currawinya Granite) and 2130082 (AOP Scalby 1) have supracrustal d 18 O zrn (7.45-8.77‰) values. This uncommon isotopic signature of juvenile to near-juvenile eHf(t) zrn but supracrustal d 18 O zrn values may be explained by the reworking of juvenile sediments sourced from nearby, Early Devonian juvenile magmatism, a phenomenon that has previously been reported for S-type granites in the New England Orogen Jeon, Williams, & Bennett, 2014).
Overall, our eHf(t) zrn data for the Thomson Orogen from the late Silurian to Late Devonian show a gradual increase from distinctly evolved in the late Silurian to juvenile and near-juvenile from the late Early Devonian to Late Devonian (Figure 4a). A trend, together with supporting eNd(t) wr and d 18 O zrn data was also noted for Lachlan Orogen samples presented by   (Figure 4). These researchers suggested that periods of dominantly evolved eHf(t) zrn and whole-rock eNd(t) as well as supracrustal d 18 O zrn indicate contractional orogenic episodes where higher relative proportions of metasedimentary material are incorporated into magmas. Conversely, periods of dominantly juvenile eHf(t) zrn , whole-rock eNd(t) and mantle-like d 18 O zrn indicate increased mantle input during extensional episodes.
An extensional event that resulted in opening of the Adavale Basin and emplacement of the Gumbardo Formation (samples, 1682891 PPC Carlow 1 and 1682892 PPC Gumbardo 1) affected the Thomson Orogen in the Early Devonian. Rhyolitic ignimbrite emplaced at this time (1682892 PPC Gumbardo 1) has the most juvenile compositions of this age in the Thomson Orogen (eHf(t) zrn value of 0.51 and mantle-like d 18 O value of 6.01‰) highlighting the importance of this extensional event that resulted in a very widespread array of Devonian basins (e.g. Barrolka Trough, Warrabin Trough, Adavale Basin). This apparent trend of increasing eHf(t) zrn and decreasing d 18 O zrn values of Early Devonian rocks relative to broadly Benambran-aged magmatism (430-420 Ma), was also observed in Lachlan Orogen samples reported by   (Figure 4).
The Hf-O isotopic signature of the Late Devonian samples appears to deviate from the observed trend seen in the Early Devonian samples (Figure 4). Although all samples have juvenile to near-juvenile eHf(t) zrn values, 158223 (Towerhill 1), located in the central-northern Thomson Orogen is distinctive by mantle-like d 18 O zrn values whereas three samples in the south of the orogen, are characterised by supracrustal d 18 O zrn values (Figure 4). This could suggest that in the southern Thomson Orogen, Late Devonian magmatism was characterised by rapid crustal recycling of juvenile material possibly related to the 390-380 Ma Tabberabberan Orogeny.

Thomson and Lachlan orogens -Hf-O and Nd-O isotopic trends
The new Hf-O isotopic data fall into two subgroups lying along two discrete mixing trends in Hf-O space ( Figure 5). These subgroups still hold when Hf isotopic signatures are adjusted to the one age ca 370 Ma (i.e. older units corrected for radiogenic growth of 176 Hf/ 177 Hf, assuming a 176 Lu/ 177 Hf ratio of 0.015; e.g. Griffin et al., 2002; Figure 5; Supplementary papers, Figure S15). The one exception to this is 1682886 (DIO Adria Downs 1); it is noted that in this sample, the zircon Hf signature appears to be decoupled from the whole-rock Nd signature (as the sample lies on the Nd-O trend; Supplementary papers, Figure S16). A variety of possible, simple, mixing models are displayed on Figure 5 and for Nd-O in Supplementary papers, Figure  S16. Although somewhat simplistic, this modelling does illustrate a number of points: (1) one subgroup appears to have incorporated a significantly more evolved (i.e. more negative eHf(t) zrn and eNd wr ), most likely supracrustal, component (Trend 1), (2) the other subgroup can be successfully modelled using an average Lachlan Orogen sediment (average from Kemp et al., 2008) (Trend 2), and (3) juvenile endmembers are more equivocal and may represent mantle-derived material or, for the more isotopically evolved metasediment trend, possibly pre-existing basement such as that exemplified by the Coquelicot Tonalite.
It is noteworthy that granite samples from the Thomson Orogen analysed by Si egel et al. (2018) also fall along the same two trends on Figure 5. The relevance of the two trends, and the applicability of the Lachlan Orogen sediment end-member (Trend 2), becomes more apparent when the geographic locations of the samples (Figure 6) are considered. The samples that form Trend 2 (the possible mixing trend with Lachlan-like sediment) all occur in the southern third of the Thomson Orogen and are geographically adjacent to the Lachlan Orogen. These are geographically separate from the Trend 1 samples with the modelled more evolved Hf sediment end-member, which fall into a domain concentrated in the northern two-thirds of the Thomson Orogen. Recent Hf-O data from magmatic rocks from the southwestern part of the Thomson Orogen (Waltenberg et al., 2018) also fall along the more evolved, Trend 1 ( Figure 5). The latter samples (Tibooburra and Cuttaburra) are geographically separate from the samples analysed here. They occur close to the southwest margin of the Thomson Orogen ( Figure 6) and correspond to the Warratta Group domain of Waltenberg et al. (2018). It is not clear if this domain is geographically isolated from the main Trend 1 domain to the north.
The trends and the mixing models suggest that the felsic igneous rocks of the Thomson Orogen probably interacted with two spatially-and isotopically-distinct metasedimentary protoliths (which may or may not represent rocks of similar age). The interpretation of the two mixing trends and their spatial distinction is equivocal. On the basis of the isotopic data alone it is difficult to discriminate between models that suggest the two metasedimentary end-members represent distinct protoliths (e.g. different age and/or different crustal blocks) or simply just reflect a provenance change within the one general sequence. We contend that the best constraints for the interpretation of the Trend 1 sediments in the northern half of the Thomson Orogen come from their spatial association with the ca 480-470 Ma igneous magmatism. The presence of Early Ordovician S-type magmatism in this region is considered unlikely unless the sedimentary protolith had been through a prior crustal thickening event, similar to that advocated for magmatism in the Lachlan Orogen to the south (Collins & Richards, 2008). This is particularly the case if tectonic models that advocate an oceanic crust basement for (parts of) the Thomson Orogen  are correct. In this scenario, metasediments in the northern Thomson Orogen underwent thickening during the slightly earlier Delamerian Orogen (ca 514-490 Ma; e.g. Glen, 2013;Foden, Elburg, Dougherty-Page, & Burtt, 2006). These metasediments underwent subsequent partial melting in the Early Ordovician, probably as a result of the widespread extension interpreted at this time in the northern Thomson Orogen (Fergusson et al., 2007b). Our preferred model, therefore, suggests that the northern half of the Thomson Orogen, at least, is underlain by   Figure 6). In this latter region it is clear that Delamerian and older crust, which outcrops in the nearby Koonenberry region (Greenfield et al., 2010), underlies that part of the Thomson Orogen. In this model, the absence of Early Ordovician magmatism (extrusive or intrusive) within the Trend 2 isotopic domain is interpreted to indicate that the latter metasediments form part of a younger, post-Delamerian sequence. Given the similarity with Lachlan Orogen metasediments (which can be successfully modelled as the Trend 2 metasediment end-member; Figure 6) we speculate that these metasediments form an extension of the Ordovician metasediments of the Lachlan Orogen. Of course, the Trend 2 metasediments may indeed be older, i.e. pre-Delamerian Orogeny. Ireland, Fl€ ottmann, Fanning, Gibson, and Preiss (1998), for example, show that a provenance switch from local-derived (Australian craton) detritus to the Gondwana-provenance typical of the Lachlan Orogen metasediments is recorded in the early Cambrian Adelaidean metasediments in southeastern South Australia, prior to the Delamerian Orogeny. A similar scenario is also feasible for the Trend 1 and 2 domains in the Thomson Orogen, such that the Trend 2 domain represents the influx of the Gondwana detritus into a slightly younger sequence of the same pre-Delamerian metasedimentary package. Although further work is required to discriminate between the different possible scenarios for the Trend 2 metasediments, we suggest that the spatial distribution of Trend 1 metasediments, particularly where associated with Early Ordovician magmatism, can be taken as the minimum distribution of pre-Delamerian Orogeny basement in the Thomson Orogen.
The association of the Trend 1 sediments in the northern half of the Thomson Orogen, and the occurrence of the Ordovician ca 480-470 Ma igneous magmatism may suggest that the northern part of the orogen forms a distinct, older basement terrane relative to the southern part, which may be more akin to the Lachlan Orogen  (Champion, 2016). One key to further constrain, or otherwise, this possibility would be by a better definition of the respective isotopically juvenile end-members for the two mixing trends. Unfortunately, the isotopic data are equivocal. The modelling is permissive of either mantle-derived material or crustal reworking of such material. This is especially the case for Trend 2. Trend 1, however, is just as explicable as reworking of older basement material, as for example, possibly represented by the Coquelicot Tonalite (although no Hf or Nd data are available for this unit). In the above discussion, we have inferred that the distinct evolved protoliths identified in the Hf-O and Nd-O models are most likely metasedimentary in origin. This is largely based on both the S-type nature of the more evolved granites and their elevated 18 O signatures. We cannot, however, unequivocally rule out that these identified protoliths may be wholly or partly infracrustal in nature, that is, represent nonsedimentary basement. Assuming the latter is correct only reinforces our conclusions regarding the possible different basement blocks in the northern and southern Thomson Orogen.

Conclusions
Zircon U-Pb ages and Hf-O isotopes as well as whole-rock geochemistry and Sm-Nd isotopic results presented here provide important constraints for the development of the Thomson Orogen as well as reveal fundamental differences and similarities with the Lachlan Orogen through time.
The new U-Pb ages for the undercover parts of the Thomson Orogen closely match rocks in the outcropping regions of the orogen. Five of the igneous rocks analysed have Early Ordovician crystallisation ages that range from 480 to 470 Ma and comprise both S-and I-type granites. Volcanic and intrusive rocks outcrop in the Charters Towers, Greenvale, and Anakie provinces have similar ages. Magmatism throughout the Thomson Orogen during this time is interpreted to have occurred in an overall extensional tectonic regime during the Larapinta Event (Fergusson et al., 2007b). Two granitic samples, 1586685 (DIO Ella 1) and 2122055 (Hungerford Granite) have latest Silurian to earliest Devonian ages of 425.4 ± 6.6 and 419.1 ± 2.5 Ma, respectively, and are a part of a batholithicscale belt of intrusions termed the Ella belt . Magmatism of this age is related to the Benambran Orogeny (435-425 Ma;Collins, 2002) and occurs in the southern Thomson, Charters Towers Province, and Lachlan orogens. Following this period of compression, the Thomson Orogen returned to an extensional regime as shown by Gumbardo Formation ignimbrites in samples 1682891 (PPC Carlow 1) and 1682892 (PPC Gumbardo 1), which lie at the base of the Adavale Basin and have ages of 408.1 ± 3.1 and 402.9 ± 2.9 Ma, respectively. These ages are similar to granites of the Amarra Suite of the Lolworth Batholith in the Charters Towers Province. Rhyolitic ignimbrite from sample 1585223 (AAE Towerhill 1) in the northern Thomson Orogen has a Middle Devonian age of 381.7 ± 5.7 Ma. This is identical within uncertainty of sample 2122056 (Currawinya Granite; 381.5 ± 2.5 Ma) in the southern part of the orogen as well as outcropping granites of the Retreat Batholith in the Anakie Inlier. The youngest sample investigated, 2310082 (AOP Scalby 1; 368.4 ± 2.5 Ma) is an S-type granite in the southeastern Thomson Orogen and is probably associated with tectonics related with the New England Orogen to the east. The oldest sample investigated and the most westerly (1682886 DIO Adria Downs; 509.7 ± 3.8 Ma), is probably not related to the Thomson Orogen but is potentially a component of the Delamerian Orogen.
Hafnium, Nd, and O isotopes for the Early Ordovician samples from the central, undercover Thomson Orogen are in stark contrast to similarly aged rocks in the Lachlan Orogen. These Thomson Orogen samples have distinctly evolved eHf(t) zrn and eNd wr values as well as elevated d 18 O zrn values, suggesting significant interaction with a supracrustal component, whereas similar aged rocks from the Macquarie Arc in the Lachlan Fold Belt reported by , have distinctly juvenile eHf(t) zrn , wholerock eNd and mantle-like d 18 O zrn values ( Figure 4; Table 2). These contrasting isotopic characteristics suggest a major difference in the composition of the deep crust between these two orogens. This is further supported by modelling of the available Thomson Orogen Hf-O isotopic data that suggests that the felsic igneous rocks from the central and northern Thomson Orogen incorporated more evolved metasedimentary sources than those in the southern third of the Thomson Orogen ( Figure 5). The spatial association of the more evolved sediments and the ca 480-470 Ma igneous magmatism provides further constraints. The requirement from the Early Ordovician felsic magmatism for prior crustal thickening (to allow subsequent melts) suggests that the more evolved metasedimentary endmember represents a stratigraphically lower, that is, older pre-Delamerian package. Therefore, we speculate that the northern part of the Thomson Orogen may form a distinct, older basement terrane relative to the southern part, which may be more akin to the Lachlan Orogen.
The Hf-O isotope time-signature of the late Silurian to Late Devonian samples can be linked with the evolution of the Thomson Orogen. Significant crustal reworking associated with the Benambran Orogeny (435-425 Ma) is evident in the two late Silurian samples (1586685 DIO Ella 1; 2122055 Hungerford Granite) that have evolved eHf(t) zrn (À6.42 to À4.62) and supracrustal d 18 O zrn (9.26-10.29‰) values. These values contrast with the Early Devonian samples (1682891 PPC Carlow 1; 1682892 PPC Gumbardo 1) that have distinctly higher eHf(t) zrn (À1.73 to 0.51) and more mantle-like d 18 O zrn (6.01-6.30‰) values that likely reflects an increased mantle input during an extensional episode related to the development of the Adavale Basin. Intriguingly, two Late Devonian samples from the southern Thomson Orogen (2122056 Currawinya Granite, 2130082 AOP Scalby 1) have juvenile eHf(t) zrn (0.01-1.92) but supracrustal d 18 O zrn (7.45-8.77‰) values. This uncommon Hf-O isotope signature may reflect rapid recycling of juvenile material related to contraction during the Tabberabberan Orogeny.