Genesis of the Wilsons Promontory batholith of southeast Victoria, Australia: heterogeneities in the Proterozoic basement

Abstract The monzogranites and syenogranites of the Early Devonian (Emsian), post-tectonic, high-level, S-type Wilsons Promontory batholith contain magmatic garnet and cordierite, in addition to biotite, as well as biotite–quartz pseudomorphs after early orthopyroxene. The batholith forms the southern tip of the Australian mainland and consists of four penecontemporaneous plutons, emplaced at shallow crustal depth, intruding Ordovician to Lower Devonian lower greenschist-facies, quartz-rich metaflysch rocks in the Bassian tectonic zone. The magmas were initially H2O-undersaturated and at T > 800 °C, having been generated by partial melting of greywacke-dominated, Paleoproterozoic to Mesoproterozoic (ca 1.6 Ga) basement rocks in the Selwyn Block (the northern tip of the VanDieland microcontinent). However, the specific source rocks differ from those of the S-type granitic plutons in the neighbouring Melbourne tectonic zone. This probably reflects compositional variations within the Selwyn Block, which has been proposed to form the basement terrane beneath both the Melbourne and Bassian zones. The compositional zonation in the batholith is due to sequential intrusion of separate magma batches rather than in situ differentiation, and the main mechanism responsible for primary chemical variations in the rocks appears to have been peritectic assemblage entrainment. Some local differentiation also occurred, probably by filter pressing, to produce aplites and tourmaline-bearing pegmatites. Normal granitic rocks (i.e. not aplites or pegmatites) with SiO2 >76 wt% were not produced through differentiation of other granitic magmas in the batholith. At the global scale, this suggests that fractionation models for the origins of high-silica granitic and rhyolitic magmas may need revision. KEY POINTS S-type magmas of the Wilsons Promontory batholith were H2O-undersaturated and at T > 800 °C. They were generated by partial melting of greywacke-dominated, Paleoproterozoic to Mesoproterozoic (ca 1.6 Ga) basement. The sources differ from those of the plutons in the Melbourne Zone, suggesting heterogeneity in the Selwyn Block. Most compositional variation is due to intrusion of separate magma batches and peritectic assemblage entrainment, not differentiation.


Introduction
Except for some metatexitic envelopes around diatexites, the wall rocks intruded by granitic magmas seldom represent the magma sources. This is especially true of granitic magmas that are emplaced as upper-crustal plutons or erupted as rhyolites. The magma source rocks commonly lie in the deep crust and have no surface expression. Thus, the nature and locations of the source rocks for high-level granitic plutons are of considerable interest. Aside from geophysical imaging and occasional xenoliths in alkalibasaltic rocks, the granitic plutons represent the only probe into this deep-crustal terra incognita. Additionally, when a batholith is sheeted or layered, we have the question of whether that layering preserves source-rock heterogeneity. The S-type Wilsons Promontory batholith (WPB) provides an opportunity to explore these issues, in relation to the deep metasedimentary crust of southeast Victoria, in southeastern Australia. This contribution follows on from the mapping and petrographic study of Wallis and Clemens (2018) and the recent use of granitic rocks to probe the depths of the Bendigo and Stawell zones, to the west and northwest (Clemens, 2020).

Regional framework
The high-level, post-tectonic, S-type WPB is dated to 395 ± 4 Ma (SHRIMP U-Pb zircon; Elburg, 1996) and 396.4 ± 3.2 Ma (LA-SF-ICP-MS U-Pb zircon; Clemens et al., 2022) and intrudes lower Paleozoic, low-grade, metasedimentary rocks in the Bassian structural zone of southeast Victoria, in the Gippsland region ( Figure 1). The WPB forms the northern end of a belt of granitic bodies that extends southward through the islands of eastern Bass Strait (particularly on Flinders and Cape Barren islands) and into northeastern Tasmania, notably the ca 384 Ma Scottsdale and Blue Tier batholiths (Groves et al., 1977;Richards & Singleton, 1981). The exposed basement in central and southeast Victoria was intruded and contact metamorphosed by numerous Devonian granitic bodies, and some large volcanic complexes were also emplaced in this period ( Figure 2). The exposed country rocks are upper Cambrian to Middle Devonian, mainly deep-marine, sedimentary rocks that were deformed and metamorphosed to lower greenschist facies during the Late Ordovician to mid-Silurian Benambran and the mid-Devonian Tabberabberan orogenies. However, despite its pre-Tabberabberan age, the WPB carries no regional tectonic fabrics or structures, other than later brittle faults (Wallis & Clemens, 2018). Thus, this northernmost part of the Bassian Zone cannot have experienced appreciable effects of the Tabberabberan Orogeny, which terminated at about 385 Ma.
During the Benambran Orogeny, southern Australia is inferred to have docked with a Proterozoic to Cambrian terrane, known as the VanDieland microcontinent (Moore et al., 2016;Moresi et al., 2014). The northern tip of this terrane forms the true, structural basement in the Melbourne Zone, the western fringe of the Tabberabbera Zone, and the eastern edge of the Bendigo Zone. It also lies beneath the Bassian Zone, which extends southward into northeastern Tasmania (Figure 2). The northern part of the VanDieland microcontinent is known locally as the Selwyn Block (Cayley, 2011;Cayley et al., 2002). showing the boundaries between the various Paleozoic structural/metamorphic zones. From west to east, these are Bendigo (BZ), Melbourne (MZ) and Tabberabbera (TZ) zones, with the Bassian Zone (BaZ) in the south. The map also shows the locations of Devonian granitic intrusive bodies, with their ages (where well constrained) and whether they are dominantly S-type or Itype-in red and green, respectively. In many cases, both types are present, and the minor component is indicated in parentheses. Large caldera complexes of silicic ignimbrites are also present but have been omitted. This map was redrawn and modified from Wallis and Clemens (2018, figure 1). For further details, including the Late Devonian volcanic rocks, see Figure 2. In Victoria, there are just a few, very limited outcrops of the Neoproterozoic to Cambrian section of the Selwyn Block, but none of any underlying Paleoproterozoic and Mesoproterozoic portions. However, chemical and isotopic studies on the Devonian granitic and silicic volcanic rocks in central Victoria have shown that neither the Paleozoic rocks of the Melbourne Zone, nor the Neoproterozoic to Cambrian rocks in the Selwyn Block, represent suitable sources for the granitic magmas Clemens, Regmi et al., 2016;Clemens & Benn, 2010;Clemens & Bezuidenhout, 2014;Clemens & Birch, 2012;Clemens & Phillips, 2014;Clemens & Wall, 1984;Phillips et al., 1981). As shown in these cited works, the granitic rocks of the region have some Paleoproterozoic to mostly Mesoproterozoic Nd model ages (1.65-1.17 Ga). The inference from this is that the source terrane for the granitic magmas must lie in the deeper parts of the Selwyn Block (Clemens & Buick, 2019). The cited studies have also suggested that the most suitable crustal source rocks for the granitic magmas would be arc-related potassic andesites, dacites and volcaniclastic rocks (greywackes and pelites) with varying original clay contents and terrigenous components. A recent zircon inheritance study (Clemens et al., unpublished data) also shows that the Paleozoic country rocks made no discernible contribution to the Devonian granitic magmas, including those of the WPB.

Geology
The geology of the WPB is described in Wallis (1981) and Wallis and Clemens (2018). The batholith contains four penecontemporaneous plutons in two lobes ( Figure 3). In order of decreasing age, the western lobe is composed of the Glennie pluton (Sealers Cove cordierite monzogranite), the Oberon pluton and the Vereker pluton (Mount Wilson biotite monzogranite). The Oberon pluton contains three members. In order of decreasing intrusive age, as inferred from field relations, these are-the Mount Oberon leucogranite, the Norman Point biotite monzogranite and the Pillar Point biotite monzogranite. The eastern lobe is formed by the presumed youngest intrusive body-the large Singapore pluton (Mount Hunter leucogranite). All these unit names remain informal but are convenient to use here. Wallis and Clemens (2018) suggested that the elongate Oberon pluton formed the feeder zone for most of the granitic magmas emplaced in the western lobe of the batholith, and also established that the batholith was constructed from multiple sheet-like intrusions, on a variety of scales, although sheeted structure had been noted previously (e.g. . Overall, the batholith may have an eastward tilt, and the shallower-emplaced parts of each component pluton therefore crop out toward the east. Most of the exposed rocks in the batholith are monzogranites, with minor syenogranites (including leucogranitic types). Some rocks contain large K-feldspar phenocrysts and many, especially in the Oberon pluton, contain prominent flow foliations, lineations and magmatic layering. These features, and the emplacement of the units as extensive sheets, testify to the relatively fluid character of the magmas (Clemens, 2015). In the Norman Point biotite monzogranite, magma flow resulted locally in spectacular accumulations of enclaves and the formation of mafic schlieren that are rich in garnet and biotite, with notable concentrations also of apatite, ilmenite, zircon, monazite and xenotime. Clemens et al. (2020) attributed schlieren formation to pulsed magmatic flow segregation and showed that neither gravitational settling nor filter pressing/matrix compaction played significant roles in their formation. However, as expected from the work of Clemens et al. (2010), apart from these local flow cumulates, there are few signs of in situ, large-scale differentiation. There are small bodies of highly leucocratic pegmatites and aplites, as well as some tourmaline-quartz veins and nodules, and miarolitic cavities. These, features formed late in the solidification histories of the members that host them, and their common enrichment in tourmaline attests to their highly fractionated character. When examined in mutual contact, the various intrusive members commonly present variable relationships, with some members having both sharp, intrusive and gradational contacts. Many of the component sheets in the batholith coexisted in the magmatic or submagmatic (mushy) states, but no mechanical mixing between magma pulses is evident at the present level of exposure (Wallis & Clemens, 2018).

Petrography
The field characteristics of the various units of the WPB were described in Wallis and Clemens (2018) but only rudimentary petrographic information was included. An atlas of textures and minerals observed in thin-section, for most of the units of the batholith, is presented in the Supplemental data A1.

Enclaves
The Norman Point biotite monzogranite member of the Oberon pluton is rich in enclaves, including the dominant igneous microgranular enclaves (IMEs) and the subordinate xenoliths of schistose metapelites, milky polycrystalline vein quartz and metapelitic hornfelses. The schistose xenoliths are inferred to have been derived from the middle crust (probably in the Neoproterozoic to upper Cambrian parts of the underlying Selwyn Block), whereas the hornfels xenoliths were derived from Ordovician to Lower Devonian metasedimentary wall rocks, at near-emplacement levels. The IMEs originated through quenching of separate, mingled, more mafic magmas that contain some mantle components (Clemens, Stevens, et al., 2017;. However, as demonstrated by  these appear to represent high-level magma mingling features; magma mixing played no role in shaping the variations in the host granitic magmas.

IMEs
The IME-focused work of ,  and Elburg (1996) concluded that the IME magmas in the WPB accompanied (or preceded) emplacement of the main granitic magmas. Thin sheets of these high-level, quenched IME magmas are inferred to have been disrupted by subsequent magma movements and reincorporated into the host plutons as the IMEs . Although some have isotopic and elemental indications of a mantle connexion, the IME magmas may not have been derived directly from the mantle but were already heavily hybridised, with crustal materials, at their level of generation. Alternatively, the mantle here is likely to be significantly enriched through ancient (Proterozoic) subduction processes. In either case, further hybridisation with the host granitic magmas occurred, mainly by incorporation of host-derived crystals. Chemical interactions (diffusive and reactive) between the IMEs and their host magmas were minimal at emplacement levels, and the IME magmas did not form parts of their host-magma lineages. The issues surrounding the petrology and petrogenesis of the IMEs in this region of Victoria were examined comprehensively in  and are not developed further here. Instead, the focus of the present work is on the chemical and isotopic characteristics of the mediumto coarse-grained granitic rocks of the batholith and what these reveal about its genesis, the composition and location of its crustal source terrane, the construction of the batholith and controls on the chemical compositions of the magmas. Clemens and Wall (1988) presented experimentally determined, isobaric T-melt H 2 O content phase diagrams for garnet-biotite monzogranite sample 2116 from the Norman Point biotite monzogranite in the Oberon pluton. The Supplemental data A2 provides the details of these experiments because the data are reused here and presented in modified form. Also, this is the first time that full descriptions of the experimental procedures have been given. The geological constraints suggest that the magma that formed rock 2116 reached its emplacement level carrying only a small to moderate amount of crystalline material, and that the magma was in a relatively fluid state. The Norman Point biotite monzogranite rocks generally lack evidence for any high-P crystallisation history; this is likely to have been erased by reactions during magma ascent and low-P (emplacementlevel) crystallisation. The only indication of magma temperature at a relatively early stage of crystallisation is from garnet-biotite thermometry (Ferry & Spear, 1978). Assuming an emplacement pressure of 100 MPa, this thermometer yields a temperature estimate of 850 C for coexistence of the 'early' garnet and biotite. The uncertainty on this calibration is probably of the order of ± 75 C (e.g. Cox, 1992). However, from the work of Gulbin (2012) in H 2 O-undersaturated systems, T estimates from this thermometer are likely to represent minima. Textural features in the rocks suggest a crystallisation sequence of ilmenite (and most other accessory minerals), quartz, garnet and orthopyroxene-biotite-plagioclase-cordierite-K-feldspar. Clemens and Wall (1988) concluded that the strongly H 2 Oundersaturated Norman Point biotite monzogranite magma was emplaced at T > 800 C. This probably represents a reliable estimate of minimum magma T. Note that these earlymagmatic conditions are virtually the same as those inferred for early crystallisation in the far more magnesian magmas of the Mount Wombat pluton of the Strathbogie batholith . This suggests that the chemical differences in these batholiths are related to source chemistry rather than to melting temperature or aH 2 O.

Granitic rocks
T-XH 2 O phase diagrams for Norman Point biotite monzogranite sample 2116 at 100 and 500 MPa (1 and 5 kbar) are presented in Figure 4, redrawn from Clemens and Wall (1988, figure 1), with melt H 2 O contents corrected using the model equations of Tamic et al. (2001). The rocks of the WPB lack mineral assemblages that can be used to tightly constrain the pressure of magma origin. However, as with the other Devonian granitic rocks of this region, it is likely that the source terrane lies in the Selwyn Block. By analogy with other Devonian S-type rocks in the region (e.g. Clemens & Phillips, 2014;, 1984, the WPB magmas were most probably formed by partial melting of rocks in the Selwyn Block at P > 400 MPa (4 kbar). Thus, the phase relations at 500 MPa ( Figure 4a) are used as a model for early, midcrustal crystallisation of a typical Norman Point biotite monzogranite magma, showing which minerals might have formed the peritectic and near-liquidus magmatic crystals that could have been carried with the melt to nearemplacement level in the shallow crust.
Taking the l00 MPa phase relations ( Figure 4b) as a model for emplacement-level crystallisation, the inferred crystallisation sequence can be matched for melt H 2 O contents >3.2 but <3.9 wt%. At H 2 O contents <3.2 wt%, plagioclase would crystallise before biotite, and at H 2 O contents >3.9 wt%, cordierite would not crystallise at all. A noteworthy feature of the phase relations is that orthopyroxene would have been present in the hypersolidus assemblage at high T (>800 C) but, on cooling, would have completely reacted out by 785 C-between 40 and 90 C above the solidus, depending on the melt H 2 O content. Thus, at emplacement conditions, the 'primary' crystal cargo in such a magma (with an initial melt H 2 O content of 3 to 4 wt%) would have consisted of garnet, orthopyroxene, quartz, plagioclase and minor ilmenite. From the work of Clemens and Stevens (2012) we know that parental granitic melts, which are formed through crustal anatexis, cannot contain anything like enough dissolved Fe and Mg to account for the overall magma/rock compositions. Thus, much of this crystalline material is likely to have been derived as peritectic mineral grains entrained into the magma at source levels. The 100 MPa phase relations show that most of the early, higher-pressure crystal cargo would have reacted to cordierite and biotite, although garnet should reappear in the stable assemblage at some temperature between 760 and 800 C. This lower-T, emplacement-level garnet would have a different composition from the original peritectic garnet. Specifically, it would contain higher concentrations of the almandine and spessartine end-members.

Whole-rock chemistry
Aside from samples analysed in the present work, chemical data for rocks of the WPB have been taken from Wallis (1981), Elburg (1995),  and Clemens, Elburg and Harris (2017). Wallis (1981) did not analyse his samples for most trace elements. So, having obtained some original rock powders, these were analysed for a full trace-element suite. For three previously unanalysed samples from Wallis (1981), we analysed these for both major and trace elements. Supplemental data A3 gives details of the methods used for the major-and traceelement analyses (XRF and LA-ICP-MS in the Central Analytical Facilities at the University of Stellenbosch) and the Sr and Nd elemental and isotope ratio determinations (by solution quadrupole ICP-MS and MC-ICP-MS, respectively, in the Department of Geological Sciences at the University of Cape Town).
The purpose of examining the chemical variations among the rocks is to understand what kinds of processes could and could not have been involved in producing the observed variations. Various forms of differentiation, which involve separation of crystalline from liquid fractions of magmas, are perhaps the most well-regarded variation mechanisms for producing chemical variation. This class of mechanism includes such models as differentiation by crystal fractionation (either Rayleigh distillation or simple unmixing of early-magmatic or restitic crystals).
At this point, it is necessary to understand a fundamental similarity between these mechanisms. They all involve a parent magma and a set of crystals that must be in isotopic equilibrium with that melt. Therefore, a group of magmas related by such mechanisms should all have quite similar Sr and Nd isotope characteristics. Thus, it seems appropriate to first examine the isotopic characteristics of the rocks from the WPB since, if there were significant isotopic variation within a given unit and within the batholith as a whole, this would effectively rule out large-scale crystal-liquid differentiation as the process that produced the various magma fractions that have been mapped in Figure 3. Of course, given the complex, sheeted structure of the batholith, it would remain possible that differentiation of this sort could have occurred within individual, relatively small magma batches. In any case, the isotope data place the rest of the petrogenetic discussion in context.

Sr and Nd isotope data
Sr and Nd isotope data for the Wilsons Promontory granitic rocks are presented in Table 1 and Figure 5, with initial ratios calculated at a reference age of 395 Ma. The analyses from Elburg (1996),  and Waight et al. (2000) were carried out by TIMS and, for the present work, by solution MC-ICP-MS (Supplemental data A3). Broadly speaking, the initial Sr isotope ratios vary widely between 0.707 and 0.714, with most values >0.710. Since these variations are very much larger than the uncertainties in the isotope ratio, this suggests that rather heterogeneous, chemically evolved, crust dominated the source region for the WPB magmas. In interpreting the Sr isotope data, it is useful to recall that a source rock that undergoes partial melting by mica breakdown can yield a melt that has a higher initial Sr isotope ratio than the original source. This is because Rb concentrates in micas and, over time, results in a local buildup of radiogenic 87 Sr, which then passes into the melt phase. Similar effects can occur with Nd isotopes, due to disequilibrium dissolution of apatite and monazite. Readers interested in these aspects are referred to Ayres and Harris (1997), Zeng et al. (2005) and Farina and Stevens (2011). In any case, this means that the original source rocks for the WPB may have been somewhat less crustally evolved than suggested by the Sr-and Nd-isotope characteristics of the granitic magmas, although it is difficult to quantify the magnitude of this potential effect. Given the S-type mineralogical characteristics of the WPB, metasedimentary source rocks probably dominated,  Clemens (1981) and Clemens and Wall (1988) and are shown as T-X Fl H 2 O plots at (a) P ¼ 500 MPa (5 kbar) and (b) 100 MPa (1 kbar), contoured for melt H 2 O contents of 2, 3, 4 and 5 wt%. The positions of these isopleths were corrected from those in the original work using the Tamic et al. (2001) a-X relationships for H 2 O in the near-binary H 2 O-CO 2 experimental fluid phases. The fO 2 was approximately 1 log unit below that of the fayalite-magnetite-quartz buffer (i.e. FMQ-1). The relations shown in (a) are used to model nearliquidus crystallisation at the pressure of magma genesis in the deep crust, and those in (b) to model phase relations at shallow-crustal emplacement conditions. The solidi and the various saturation boundaries are colour-coded and labelled with the respective mineral abbreviations as recommended by Warr (2021). A given mineral is stable either between its upper and lower temperature boundaries (e.g. Opx and Crd) or on the low-T side of its saturation curve. Note that, in (b), the upper stability limit of orthopyroxene was not well constrained by the experiments and so is plotted as a dashed curve. The positions of the solidi are calculated rather than experimental (Johannes & Holtz, 1996). See Supplemental data A2 for experimental details and Clemens and Wall (1988) for details on the derivation of the phase diagrams. but minor meta-igneous rock types may also have been present (e.g. . More specifically, the Sr isotope data suggest sources dominated by metagreywackes or metavolcaniclastic rocks rather than metapelites. The Sr isotope results effectively rule out large-scale differentiation as the origin of the variations across the batholith. In addition, the scatter among the initial Sr isotope ratios for individual units (e.g. the low-volume Norman Point biotite monzogranite with 87 Sr/ 86 Sr 395 ¼ 0.7095 to 0.7137) suggests that little of the variation could have been due to differentiation on any scale, except very locally (see e.g. Clemens et al., 2020). Also, the absence of any mixing trends in Figure 5a excludes the operation of magma mixing, for which there is no textural evidence in any case. The following descriptions and discussion of the chemical variations should be read with this overall context in mind.
The eNd t values vary from about À4.4 to À2.7, which, although firmly crustal in character, suggest that the source rocks for the WPB had not resided in the crust for as long as the sources of some other S-type rocks in the region. Two-stage Nd model ages (t 2DM ) calculated for the rocks (Table 1) are plotted against their SiO 2 contents in Figure 6. These ages are dominantly Mesoproterozoic and range from 1.62 to 1.38 Ma. If the source rocks in the Selwyn Block were deposited in an arc setting, just after their parent igneous rocks had been formed, this would suggest that the WPB magmas were derived through partial melting of the Mesoproterozoic (middle) part of the block. Also, note that the model ages become generally younger for the rocks with higher SiO 2 contents. The two exceptions are the schliere (WPB7), which represents a former cumulate mush, and W96H2, a very-high-silica Mount Oberon leucogranite sample, which appears to have had a slightly older Paleoproterozoic source. On the assumption that the thermal anomaly responsible for the crustal melting migrated upward with time, this general trend suggests that there was significant source control on magma composition and that the more silicic magmas may have been derived from rocks at higher structural levels within the Selwyn Block. The possible meaning of this is discussed below.

Major-and trace-element variations
Chemical analyses of the Wilsons Promontory granitic rocks, with major oxides normalised to 100 wt% volatilefree and all Fe expressed as FeO T , are presented in the Supplemental data A4. All the sampled rocks plot in the high-K calcalkaline field on the K 2 O-SiO 2 diagram (Figure 7a), and all are peraluminous, with apatite-corrected ASI values (mol. Al 2 O 3 /[CaO-3.33P 2 O 5 þNa 2 O þ K 2 O]) mainly >1.10 (Figure 7b). Along with the presence of accessory garnet, cordierite, monazite and ilmenite, these characteristics mean that the batholith is unequivocally a reduced Stype and therefore derived through partial melting of mainly metasedimentary source rocks. In this context,  Clemens (1981) carried out high-P-T experiments to saturate Wilsons Promontory monzogranite 2116 with sillimanite and quartz, and found that, under all investigated P, T and aH 2 O conditions, aAl 2 O 3 in the 2116 magma was insufficient to stabilise a quartz-Al 2 SiO 5 mineral assemblage. As for similar experiments reported by Clemens and Wall (1981), the conclusion is that the granitic magmas were produced by partial melting of a metasedimentary source dominated by rocks that were less aluminous than typical sillimanite-bearing metapelites. Metagreywackes would be suitable candidates. Clemens, Stevens, et al. (2011) showed that, in a mixed source of meta-andesitic rocks and metagreywackes, S-type melt chemistry and residual mineral assemblages could only be obtained in cases where the source rocks contain a metasedimentary component of !80%. Thus, it is highly likely that the sources for the WPB magmas were dominated by metagreywackes, but with probable minor metapelitic rocks. If present at all, andesitic to dacitic metavolcanic rocks could only have been very minor source components.
On Harker-type plots, most of the units within the WPB show limited degrees of variation, with analytical points that cluster in data clouds with low degrees of correlation (Figures 7 and 8). Note that the linear trend lines in Figure  8 are for the Norman Point biotite monzogranite unit only. This Norman Point biotite monzogranite member of the Oberon pluton is distinctive in having relatively low SiO 2 , Al 2 O 3 (Figure 8c), and K 2 O (Figure 7a), and high TiO 2 (Figure 8a), FM (FeO T þMnO þ MgO; Figure 5d) and CaO (Figure 7f), without any appreciable changes in Mg# across the SiO 2 range (Figure 8e). These features suggest that, rather than crystal fractionation, the quasi-linear variation shown among the data points for this member, on most plots, could be due to variations in mechanical concentration of magmatic biotite, plagioclase, ilmenite and garnet. However, within this member, there is no correlation between K 2 O and FM concentrations (Figure 9b), so biotite accumulation cannot have been the cause of the increase in FM contents. There is also no evidence for magma mixing here, despite the presence of IMEs, which are commonly misinterpreted as evidence for such a process (see e.g. . Thus, for the great majority of the magmas that formed the WPB, as a whole, we infer that the bulk of the variation was created, at source, through the mechanism of peritectic assemblage entrainment (PAE; Clemens & Stevens, 2012;Stevens et al., 2007). The ferromagnesian minerals involved would have been orthopyroxene, garnet and ilmenite, together with peritectic plagioclase. Indeed, Wallis and Clemens (2018) noted the presence of biotite-quartz pseudomorphs after orthopyroxene in the rocks of the WPB (e.g. Supplemental data A1, figure A1.12). The overall TiO 2 variation (Figure 8a) appears curvilinear rather than linear, which could be interpreted as a signature of crystal fractionation involving ilmenite. However, note that the distribution is strongly affected by a single, unusually high-SiO 2 analysis (WPB13, with 75.91 wt% SiO 2 ). This sample probably does represent the product of local crystal fractionation from a Norman Point biotite monzogranite magma, with the rest of the rocks defining a relatively flat trend, more consistent with PAE.
Another somewhat chemically distinctive member of the batholith is the Mount Oberon leucogranite, which also forms part of the Oberon pluton. This member is rather silicic and has relatively elevated K 2 O (Figure 7a) and Al 2 O 3 (Figure 8c), with Mg# values that scatter widely but extend to values higher than any other analysed rocks in the batholith, especially at the high-SiO 2 end of the compositional spectrum (Figure 8e). These characteristics suggest that the Mount Oberon leucogranite magmas did not originate through fractionation of the magma of any other member of the WPB but arose as a compositionally distinct but heterogeneous magma pulse. This inference is supported by the fact that the Mount Oberon leucogranite is the earliest-emplaced member in the Oberon pluton, and one of the earliest in the batholith (Wallis & Clemens, 2018). Although also a leucogranitic unit, the Mount Hunter leucogranite is part of the Singapore pluton and has characteristics that clearly distinguish it from the Mount Oberon leucogranite. Specifically, the Mount Hunter leucogranite has a higher Na 2 O content and consistently low Mg# (Figure 10).

Formation of the chemical variations in the main series of WPB granitic magmas
Reservations regarding the effectiveness of all differentiation mechanisms in granitic magmas were first expressed by Barbey (2009) and Clemens et al. (2010), with the latter concluding that commonly invoked mechanisms such as crystal fractionation and restite unmixing cannot have been responsible for the bulk of the chemical variations among series of granitic rocks. Instead, the mechanism of PAE (Clemens & Stevens, 2012;Stevens et al., 2007) has been invoked as the most credible explanation for the main variations. Essentially, PAE involves the entrainment of the small crystals of peritectic minerals (formed in magma source rocks during partial melting reactions) together with tiny crystals of accessory minerals, into a segregating and ascending melt-dominated granitic magma fraction. In a given granitic rock series, the entrained minerals and their relative proportions are inferred to remain constant, but their overall proportion in an individual magma batch can vary. This endows the magmas in such a series with close kinship but with source-inherited compositional contrasts.
As shown above, the WPB magmas were most probably formed by high-T partial melting of mainly metasedimentary source rocks. Based on this premise, Figure 11 provides some constraints on the key factors that controlled the magma compositions. Figure 11a is a log-log plot of Ba vs Pb concentrations in the WPB rocks. This type of diagram was introduced by Finger and Schiller (2012) and illustrates the distinct trends expected for magmas produced by fractionation and partial melting. It is evident that the rocks of the WPB form a data cloud that indicates the dominance of partial melting over fractionation. This coincides with our inferences from other lines of evidence and lends some support to the idea that PAE played a major role in creating the variations among most of the WPB magmas.
Molal concentrations of Al, plotted against M (¼Fe þ Mn þ Mg) in units of moles/100 g of rock, are shown in Figure 11b. The use of such molal plots allows for simple calculation of vectors for addition or subtraction of mineral phases to or from the magmas. This is because variations in Mg# do not affect the slopes of stoichiometric mineral phases; see e.g. Clemens, Stevens, et al. (2017). Vectors indicate how a liquid magma (with very low mafic-oxide content) would evolve with the addition of plagioclase, garnet and biotite. From this figure, we can rule out any strong role for plagioclase because the trend for the rocks does not have any significant elongation in that direction. On the other hand, biotite and garnet accumulation would be compatible with the rock trend. A similar plot, but with K as the abscissa, is shown in Figure 11c. If biotite accumulation had been a significant cause of variation in the magmas, this plot would show a positive correlation between K and M. In fact, the correlation is negative, suggesting that biotite was not involved, and that the main carrier of the K signal in the magmas (prior to late K-feldspar saturation) would have been the silicate liquid phase.
A role of feldspar fractionation or accumulation can be gauged from the REE spectra ( Figure 12). The rocks show a range of negative Eu anomalies (Eu/Eu Ã ¼ Eu= ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Sm Á Gd p ) and the anomalies deepen within each series. This means that there must be a role for feldspar in the creation of the chemical variations in the WPB. Since K-feldspar is a latecrystallising mineral in these magmas and there is no Kfeldspar accumulation signal in Figure 11c, plagioclase is likely to be the only feldspar involved; its role is further investigated below. A fractionation overlay on the primary peritectic assemblage entrainment signal As noted above, the main mechanism by which the WPB magmas attained their compositional heterogeneity is inferred to have been PAE. However, there is also evidence that there was some crystal fractionation, which operated as an overlay on the primary, source-inherited variations. In the normal granitic rocks of the WPB, the main indication of this is in the chemical variations within the Norman Point biotite monzogranite, as already described. We recall that this member of the Oberon pluton is the only part of the WPB that shows chemical trends that strongly indicate that fractionation might have operated; see, for example, Figures 7a, 8, 9 and 12. These plots mostly show effects attributable to PAE but the REE spectra ( Figure 12) show a distinct deepening of the negative Eu anomaly, and this is amplified in Figure 13, which shows Eu/Eu Ã plotted against SiO 2 and CaO. From these plots it is evident that there was some fractionation of plagioclase feldspar in the rocks of the Norman Point biotite monzogranite, and possibility in the Mount Wilson biotite  The aplites and pegmatites that are common in several members of the batholith represent segregated magmas that were highly evolved chemically. This is evident from the fact that these bodies commonly contain the mineral tourmaline, in some samples in quite high abundance (several modal %), which is vastly in excess of the normal concentration of this mineral in the batholith. The aplites and pegmatites were derived from the final dregs of H 2 O-rich liquid magma left in the crystallising parent magmas. The tourmaline patches and tourmaline-quartz nodules were formed through precipitation from these hydrous but Brich terminal fluids. The mechanisms that are likely to have caused segregation and emplacement of these late, fractionated liquids would include upward percolation of lowdensity liquids and fluids, and possibly dilation during shear caused by late magma flow (e.g. Petford et al., 2020). It is currently not possible to estimate how much parental magma fractionated to produce these features but, although prominent in the field, they are minute in overall volume, compared with the batholith as a whole.
Other field-observable features that were produced by local magma fractionation are the garnet-biotite-rich schlieren. Like the tourmaline-bearing rocks, these are minuscule in volume compared with the overall batholith. However, the field relations suggest that these represent local mechanical accumulations of magmatic minerals. As such, they represent a minor and local sign of more-or-less in situ differentiation by flow segregation. However, as shown by Clemens et al. (2020) the processes involved in the formation of the schlieren were not allied to those that caused the variations in the main volume of the Norman Point biotite monzogranite magmas; see also Figure 9. The conclusion to be drawn here is that the main variation mechanism among the magmas of the WPB appears to have been PAE. Minor contributions came through plagioclase fractionation, local crystal segregation in the flowing magmas, and a late contribution that involved the local expression of small volumes of highly evolved magma to form the aplites and pegmatites. It is possible that the magmas for some of the Mount Oberon leucogranite and Mount Hunter leucogranite rocks with SiO 2 contents >76 wt% were also produced by a degree of crystal fractionation. However, experimentally produced partial melts of crustal rocks exhibit a wide range of compositions, with SiO 2 contents ranging up to 79 wt%, so differentiation need not be invoked simply because some rocks in the series have high-silica compositions. We interpret the Mount Oberon leucogranite and Mount Hunter leucogranite as having crystallised from parental magmas that had very low contents of entrained crystalline material.
The batholith in its regional context The Bassian Zone and the adjoining Melbourne Zone are inferred to share the same basement terrane-the Selwyn Block, which is regarded as the northern tip of the VanDieland microcontinent (Cayley, 2011;Cayley et al., 2002). This crustal block is known to be heterogeneous (Moore et al., 2016), so it is worth investigating whether the source rocks for the magmas of the WPB are the same as those that formed the sources of the younger (Late Devonian) S-type granitic plutons in the adjacent Melbourne Zone. The isotope variations suggest that there are differences, and the isotopically anomalous position of the WPB has already been noted by Clemens and Elburg (2016). This is particularly apparent in Figure 14d. This anomalous status can be tested further using chemical data. Some Harker-type variation diagrams, with Melbournezone S-type rocks plotted as small grey squares and WPB samples as red dots, are illustrated in Figure 14a-c. Each of these plots reveals the existence of compositional contrasts between the two groups of S-type granitic rocks in central and southeast Victoria. Specifically, the WPB has lower Al 2 O 3 and Mg# and higher Zr, initial Sr isotope ratio and initial eNd. To provide a somewhat crude statistical test of these relationships, the analyses of Melbourne-zone S-type rocks were averaged over the range of SiO 2 contents that exist among the analysed rocks of the WPB (69 to 78 wt%). The means and standard deviations were then used in a Student's t-test to compare the two datasets. For all the tested parameters, apart from Zr content, the means are significantly different at the 99% confidence level. The Zr means are significantly different at the 90% level (Table 2).

Discussion and conclusions
The evidence that we present above suggests that the monzogranitic and syenogranitic magmas of the WPB were developed through high-T partial melting of quartz-rich metaflysch source rocks. Their compositions were largely set at source level, through the mechanism of PAE. Differentiation played a small role, mainly in producing some of the extreme rock compositions (mafic schlieren, aplites and pegmatites). This differentiation seems to have been related to flow segregation of crystals from silicate liquid and aqueous fluid magma components.
The apparent differences in elemental and isotopic compositions between the rocks of the WPB and S-type granitic rocks in the Melbourne Zone (Figure 14d) may mean either that the WPB magmas were not formed from the same formations within the Selwyn Block or that, contrary to present models, the Selwyn Block does not form the basement of the northwestern sector of the Bassian Zone. The geographical anomaly in tracer-isotope ratios has already been noted in Clemens and Elburg (2016). The Sr isotopic contrasts imply that the sources of the WPB magmas were more chemically evolved (originally more clayrich) than the metagreywackes that formed the sources of the Melbourne-zone S-type magmas. The higher eNd, however, suggests that these Bassian-zone source rocks were probably younger than the sources of the Melbourne-zone S-type magmas. To these compositional contrasts, we can add the difference in magmatic emplacement ages-Early Devonian for the WPB as against Late Devonian for the granitic plutons of the Melbourne Zone. It therefore seems likely that the magmas of the WPB were generated in a partial melting event that was unrelated to that which took place in the Melbourne Zone, and that the source rocks for the WPB had compositions dissimilar to those of the Selwyn Block in the Melbourne Zone. Furthermore, since the WPB does not appear to have been affected by the deformation in the Tabberabberan Orogeny, it seems probable that the section of the Bassian Zone in which the WPB lies had a separate tectonic and metamorphic history from the Selwyn Block and the whole of the now adjacent Melbourne Zone.
It is noteworthy that the volcanic rocks of the Tolmie Igneous Complex , which crops out astride the boundary between the Melbourne and Tabberabbera zones, share some of the compositional and isotopic peculiarities of the WPB. Relative to the S-type rocks of the Melbourne Zone, the Tolmie Igneous Complex and the WPB are both relatively garnet-rich, ferroan, high in Zr and have higher eNd t , although all these differences Figure 14. Harker-type plots (a) to (c) illustrating chemical and isotopic differences between the S-type rocks of the Melbourne Zone (small grey squares) and the rocks of the Wilsons Promontory batholith (red dots) that lies in the Bassian Zone but is inferred to share the same basement terrane-the Selwyn Block: (a) Al 2 O 3 , (b) Mg#, and (c) Zr. Plot (d) shows epsilon Nd plotted against Sr isotope ratio, both recalculated to a reference age of 370 Ma. Data for the Melbourne-zone rocks were taken from electronic appendix EA2 of Clemens, Elburg and Harris (2017), with additional data for the Wilsons Promontory batholith as given in Table 1  are more pronounced for the Tolmie Igneous Complex, probably because this complex should be considered to consist of strongly peraluminous A-type rocks (e.g. Clemens & Stevens, 2021). Nevertheless, both the Tolmie Igneous Complex and WPB occur at the eastern fringes of the Melbourne Zone and, assuming that their peculiar mineralogical, chemical, and isotopic characteristics are at least partly source-derived (Clemens et al., 2009), this suggests that the source rocks in the far northeastern and southeastern Selwyn Block differ from those elsewhere (see also Clemens & Elburg, 2016).
An additional perspective on the magma source comes from recent work on the zircon inheritance patterns of Devonian silicic igneous rocks in central Victoria (Clemens et al., unpublished data). Data for three rocks from the Oberon pluton (Mount Oberon leucogranite sample 2340, Norman Point biotite monzogranite sample W96H2 and Norman Point biotite monzogranite schliere WPB7) come from Elburg (1996) and Clemens et al. (unpublished data). In these samples there is only weak inheritance of any zircon older than earliest Mesoproterozoic, and there is no Archean zircon. Inheritance varies between the different units of the Oberon pluton, with 40% of the zircon inherited in the Mount Oberon leucogranite but <20% in the Norman Point biotite monzonite, including the very mafic schliere sample. This runs counter to the expectation that more mafic rocks, with higher Zr contents, ought to have higher degrees of zircon inheritance. There appears to be strong source control over the degree of inheritance and the overall inheritance pattern. Concerning the magma source rocks, although it seems probable that the magmas were formed by partial melting of rocks within the Selwyn Block, there is almost no trace of a Selwyn-block signature in the inherited zircon populations (i.e. no ca 1400 ± 50 Ma zircon, which is considered most diagnostic). Since this Selwyn-block signature applies to the Neoproterozoic to Cambrian upper sections of the Selwyn Block, where it crops out in Victoria, the Bass Strait islands and in Tasmania, this suggests magma formation from the unexposed deeper parts, which are considered to be Paleoproterozoic to Mesoproterozoic in age. A similar conclusion applies to the Late Devonian S-type granitic and volcanic rocks in central Victoria (e.g. Clemens & Buick, 2019).