Granite typology categorizes granitoid rocks based upon distinguishing characteristics that are interpreted to indicate sources, conditions of generation, and, by implication, tectonic setting. Complexities of elemental and isotopic geochemistry, however, commonly preclude simple typological interpretation and suggest more complex petrogenetic histories. Granitoids from the Songpan-Ganzi terrane in the eastern Tibetan Plateau were emplaced within a short interval (~15 m.y.). They display mineralogical and geochemical characteristics that are consistent with a wide range of proposed typologies (I-, S-, and A-type; high Ba-Sr and adakitic variants). Despite their close spatial and temporal association, these granitoids exhibit diversity in geochemical characteristics that indicates a broad spectrum of contributing sources. Radiogenic isotope data reveal a continuum from primitive to evolved crustal compositions; i.e., 87Sr/86Sr(t) = 0.704–0.715 and εNd(t) = +2 to −11. All granitoid “types” have variable but commonly high zircon δ18O (+4.1‰ to +11.6‰) and low whole-rock Li-B-Mg isotopic ratios compared to mantle and/or seawater (δ7Li = +5.1‰ to −3.2‰; δ11B = −10.7‰ to −16.5‰; δ26Mg = −0.23‰ to −0.59‰). These stable isotopic compositions suggest that the Songpan-Ganzi granitic magmas of all “types” had contributions from sediment, ranging from minor to dominant. The highly variable isotopic compositions of the granitoids rule out a single homogeneous source for these diverse yet contemporaneous granitoids. Their compositional variability may have been significantly influenced by sedimentary contributions, and these results demonstrate the difficulty of straightforward assignment and interpretation of granitoids using conventional typology.

Granitoids (sensu lato, intermediate to felsic quartz-bearing intrusive rocks) are the predominant constituents of Earth's continental crust (Campbell and Taylor, 1983; Kemp and Hawkesworth, 2003). The widely used paradigm of I- and S-type granites divides granitic rocks into groups interpreted to be derived from metaigneous (I) and metasedimentary (S) sources, respectively (Chappell and White, 1974). Crustal sources were originally inferred for both I- and S-types, though substantial mantle contributions are now considered important for many I-type granitoids (e.g., Castro, 2020). The other widely cited “alphabetic” category, A-type, is not specifically linked to a source, rather being characterized by low H2O (“anhydrous”), alkalic composition and intraplate setting (“anorogenic”) (Loiselle and Wones, 1979). A-type granites are variously attributed to crustal igneous sources, mantle sources via fractional crystallization of mafic magma, or hybridization of mantle- and crustal-derived magma (e.g., Bonin, 2007; Frost and Frost, 2011).

The compositional diversity of Phanerozoic granitoids has fed a long-standing debate on the relative importance of initial melt source versus subsequent closed- and open-system processes (fractional crystallization, magma mixing, assimilation; Moyen et al., 2021). Recent studies have suggested that stable isotopic compositions may be useful tracers of magma sources (e.g., Tomascak, 2004; Kemp et al., 2007; Teng et al., 2009), complementing traditional radiogenic isotopic tracers of petrogenesis. Oxygen isotopes provide a widely accepted tool for petrogenetic interpretation of igneous rocks. Li, B, and Mg isotopic systems are promising but not as well understood, especially in potential magma source materials and under conditions of magma generation.

In this study, we applied O-Li-B-Mg isotope analysis to a prime example of diverse granitic rocks in the Songpan-Ganzi terrane (Tibetan Plateau). These intrusions, which were formed and emplaced during a geologically short time interval, span much of the global compositional range of granitoids, including examples of the most widely cited typologies (Fig. 1). We conclude that stable isotopes provide valuable petrogenetic insights. The continuum of compositions demonstrates that many, and probably most, of these magmas were hybrids rather than products of single sources. Importantly, there is evidence for contributions from sediment, ranging from minor to dominant, to all of the granite “types.”

The Songpan-Ganzi terrane (Fig. S1 in the Supplemental Material1), which comprises ~20% of the Tibetan Plateau, evolved during the Triassic from a remnant ocean basin of the Paleo-Tethys into a major orogenic belt (e.g., Xu et al., 2015). It consists of two main rock types, Triassic turbidites and Triassic–Jurassic granitoids, and thus bears lithological similarities to the accretionary Lachlan orogen of eastern Australia, where I-, S-, and A-type granites were originally defined (e.g., Collins et al., 2019). The plutonism was petrologically and geochemically highly diverse, encompassing granitoids that have been characterized as I-, S-, and A-type granitoids (Fig. 1; Table S1). The I-type granitoids include variants that could be categorized as high Ba-Sr variants (>2000 ppm Ba + Sr; e.g., Tarney and Jones, 1994; Fowler and Rollinson, 2012) and adakitic (Sr/Y >20, Y <20 ppm; e.g., Defant and Drummond, 1990; see Fig. 1). I-type plutons (including adakitic variants), which are the most areally extensive, consist of diorite, monzonite, syenite, granodiorite, and granite. A-type granitoids are dominated by granite, S-type granitoids are dominated by two-mica granite, and high Ba-Sr granitoids are dominated by monzonite and quartz monzonite. These diverse granitoids were emplaced during a brief interval of ~15 m.y. between 213 Ma and 198 Ma (Table S3).

Previous radiogenic isotope (Sr-Nd) studies have revealed considerable heterogeneity and postulated contributions from both a mantle component (enriched and/or depleted) and a crustal component (metasedimentary and/or igneous) to account for the observed diversity (e.g., Yuan et al., 2010; de Sigoyer et al., 2014). The granitoids yield variable εNd(t) from −11.4 to +2.1 and 87Sr/86Sr(t) of 0.7039–0.7148 (Fig. 1C), affirming their petrogenetic diversity. However, details of the implied source heterogeneity and magma genesis have not been well constrained.

We present here a systematic investigation of zircon O and whole-rock Li-B-Mg isotopes from 24 samples (Figs. 2 and 3), shedding new light on the complexity of granitoid magma genesis. A-type granitoids show mantle-like δ7Li values of +1.9‰ to +2.3‰ and a large range in zircon δ18O values from +4.1‰ to +8.6‰ (Fig. 2A). Other granitoid types also show variable zircon δ18O values (high to very high: +6.6‰ to +11.6‰) and highly variable δ7Li values of +5.1‰ to −3.2‰ (Figs. 2A and 3A). All of the sampled granitoids show lower δ11B values (−10.7‰ to −16.5‰) than those characteristic of the upper mantle (~−7‰; Horst Marschall and Foster, 2018; see Figs. 2B and 3B). With the exception of high Ba-Sr granitoids, which have mantle-like δ26Mg values of −0.23‰ to −0.27‰, sampled granitoids have low δ26Mg values of −0.30‰ to −0.59‰ (Figs. 2B and 3C).

The large variation in the Songpan-Ganzi granitoid isotopic compositions requires diverse source materials. The Nd-Sr isotope ratios suggest contributions from ancient crust and permit small to large mantle, or juvenile crust, components. All Nd-Sr compositions fall within the range defined by the western South China and Songpan-Ganzi Paleozoic–Triassic sediments (Fig. 1C), suggesting possibly dominant contributions from the regional crust.

The oxygen isotopic composition of zircons from S-type granitoids (δ18O values = +7.9‰ to +11.6‰) reflects major incorporation of sediments. Their light Li isotopic compositions (δ7Li = −1.1‰ to +2.6‰) are similar to those typical of the middle to upper crust (~−1.2‰ to +4.0‰) and Lachlan S-type granite, while their lower δ11B values (−14.5‰ to −16.5‰) compared to mantle (~−7‰) or seawater (~+40‰) are consistent with a contribution from sediments (Romer et al., 2014; Horst Marschall and Foster, 2018; see Figs. 2C and 3A). The δ11B values also overlap with those of global S-type granitic rocks (δ11B = −11‰ ± 8‰; Horst Marschall and Foster, 2018). Their δ26Mg values (−0.30‰ to −0.34‰) overlap with subducting sediments (−3.65‰ to +0.52‰, dominantly ~0 to −0.5‰; Teng, 2017). Therefore, the stable isotopic compositions of these S-type granitoids support the presence of a dominant amount of sediment. Zircon analyses of I-type granitoids (both with and without adakite affinities) show high δ18O values of +6.6‰ to +9.84‰, also indicating substantial contributions from sediments. Generally low whole-rock Li-B-Mg ratios are also consistent with a sedimentary component (Figs. 2 and 3). The δ7Li values (−4.5‰ to +4.7‰) fall within the range of global oceanic subducting sediment (GLOSS-II; Plank, 2014) and shales from the Schwarzburg area in Europe (Fig. 3A; Romer et al., 2014), and the δ11B values of −12.2‰ to −15.1‰ are distinctly lower than those of global I-type magmas (δ11B = −9‰ to +12‰, median −3‰; Horst Marschall and Foster, 2018).

A few A-type granitoids show elevated zircon δ18O values (up to +8.6‰) with light whole-rock Li isotopic composition (δ7Li values = +1.9‰ to +2.3‰) compared with the mantle, consistent with contributions from sedimentary materials (Penniston-Dorland et al., 2017). However, some A-type granitoids also show mantle-like zircon δ18O and δ7Li values, which may reflect heterogeneous lithospheric mantle metasomatized by subducted sediment. These A-type granitoids are the most evolved, MgO-poor samples (Fig. 3C), and they show the lowest δ26Mg values of −0.53‰ to −0.59‰, probably indicating that detectable Mg isotopic fractionation occurred in these magmas (Wang et al., 2021). The high Ba-Sr granitoids show high MgO contents and mantle-like δ26Mg values (Fig. 3C). However, they have uniformly high zircon δ18O (+7.3‰ to +8.8‰) and low δ11B (−11.5‰ to −14.6‰) compared with the mantle (Figs. 2 and 3). These stable isotope compositions point to significant contributions of both sedimentary and primitive mantle materials to the high Ba-Sr magmas. The variable Li isotopic compositions (δ7Li = +5.1‰ to −0.7‰) of the high Ba-Sr samples do not show any significant correlation with the degree of magmatic differentiation, as inferred from geochemical parameters (e.g., SiO2; Fig. S2), indicating that the Li isotopic heterogeneity likely reflects heterogeneous deep (middle or lower crust, or mantle) sources (Fig. 2). Moreover, some high Ba-Sr granitoids also show elemental similarity to Miocene ultrapotassic-potassic rocks in south Tibet (Zhao et al., 2009), Archean sanukitoids (Martin et al., 2009), and the high Ba-Sr syenites of the northern Scottish Highlands (Fowler et al., 2008) with high zircon δ18O (Fig. S2), all of which are interpreted to require input of sediments to the melt (Zhao et al., 2009; Fowler and Rollinson, 2012).

Although the Songpan-Ganzi granitoids might be thought of as discrete, distinctive types, each of which has a unique origin (e.g., de Sigoyer et al., 2014), the broad ranges and overlap of isotopic compositions of all granitoid “types” indicate that an array of materials contributed to all the magmas. Notably, even though a majority of the granitoids appear clearly to have dominantly igneous sources (crust and/or mantle; A-type and I-type, including adakitic and high Ba-Sr), all “types”—not just S-type—reveal isotopic evidence for sedimentary contributions.

Supracrustal components may be incorporated into granitoid magmas by various processes, including primary generation (anatexis of metasedimentary rocks in the crust); contamination of juvenile magma by assimilation of sedimentary rocks; subarc mantle metasomatism by fluids released by sedimentary material in a subducting slab; and partial melting of subducted and underplated sediments. All of these processes are possible during the period during and immediately after ocean basin closure and continent collision, as experienced by the Songpan-Ganzi terrane in the Triassic. By the time the granitoids were generated and emplaced in the Late Triassic to Early Jurassic, closure of the Paleo-Tethys and subsequent continental collision were complete (Pullen et al., 2008; Roger et al., 2010). Thus, the underlying lithospheric column in which magmas were generated and contaminated likely contained diverse metasedimentary and sediment-modified rocks at multiple levels (Fig. 4).

Our new O-Li-B-Mg isotopic data imply that most of the diverse granitoids incorporated variable but significant amounts of sedimentary material (Figs. 2 and 3). Most of the granites with S-type characteristics appear to have been generated in large part from Songpan-Ganzi–like metasedimentary rocks, based on radiogenic (Fig. 1C) as well as O-Li-B-Mg isotopes, at moderate crustal levels (most Sr/Y <20; Table S1). In the other granitoids, the sedimentary component, though generally evident, is relatively small, giving rise to their seemingly I- and A-type characteristics. Furthermore, many have high Sr/Y ratios (20–110), suggesting much deeper origins (adakitic, high Ba-Sr, many I-type). Finally, the stable and radiogenic isotopic compositions of the non–S-type granitoids, and a minority of the S-types, require an important primitive component, presumably either mantle-derived magma or juvenile mafic crust.

These isotopic constraints are consistent with magma generation within the geochemically complex crust-mantle column produced by the tectonic history that preceded the generation of the granitoids. In this setting, contributions from sedimentary material—by direct anatexis, assimilation, and sedimentary fluid–induced metasomatism—can be expected in magmas for which their initial source was the upper mantle, underplated juvenile crust, or older, less mafic crust, as well as in those considered to be conventional S-type magmas (Fig. 4).

  1. New O-Li-B-Mg isotope data indicate highly variable but significant contributions from sedimentary materials to the extremely wide array of the almost coeval Songpan-Ganzi granitoids, only a small proportion of which would be considered S-type based on accepted criteria.

  2. Rapid transition from ocean basin closure to continental collision to postcollisional tectonics produced a complex crust-mantle column; sediments, or rocks modified by sedimentary fluids, were present in various parts of this column, and magma generation occurred at a wide range of depths and temperatures. This diversity of source materials and conditions of generation led to production of magmas that spanned much of the known range of granitoid “typologies,” and to the widespread geochemical influence of sediment.

  3. We suggest that (a) potentially important contributions of sediment to many magmas may be most easily identified through analysis of multiple stable isotopes, and (b) rigid application of “typology” to interpretation of granitoid magma sources should be viewed with caution—multiple sources are likely, and they may only be discerned, if at all, through very thorough petrochemical investigations.

1Supplemental Material. Analytical methods, Figures S1 and S2, and Tables S1 and S2. Please visit https://doi.org/10.1130/GEOL.S.17118791 to access the supplemental material, and contact editing@geosociety.org with any questions.

We sincerely thank Benxun Su and Ming Tang for discussions regarding initial development of this manuscript, and Jinlong Ma for his help during B isotopic analyses. Thorough reviews by Federico Farina, I.N. Bindeman, and an anonymous reviewer, and insightful comments by editor Urs Schaltegger, on an earlier version of our manuscript were extremely helpful. This study was financially supported by projects of the National Natural Science Foundation of China (grants 41772232 and 41888101). This is a contribution to International Geoscience Programme (IGCP) 662 “Orogen Architecture and Crustal Growth from Accretion to Collision.”

Gold Open Access: This paper is published under the terms of the CC-BY license.