Matthews et al., 2016)

. Three major models showing the paleogeographic adjacency of Lhasa to (b) NW Australia (Audley-Charles, 1984;Wang et al., 2021;Zhu et al., 2011), (c) Greater India (i.e., northern India, Metcalfe, 1988;Gehrels et al., 2011), and (d) eastern Africa (Hu et al., 2021;Zhang et al., 2012). The ∼1.3-0.9 Ga orogenic belts related to the assembly of Rodinia are modified from Spencer et al. (2015). (e) Present-day map showing locations of all collated detrital zircon samples (Table S1 in Supporting Information S1).

XU ET AL.

10.1029/2022GL100160 3 of 11 Hu et al., 2021), which is supported by widespread Rodinia assembly related orogenic belts in Gondwana of that age (∼1.3-0.9 Ga, Figures 1b-1d; Spencer et al., 2015). DZ U-Pb age spectra of (meta-)sedimentary units have been used extensively to help constrain paleogeographic positions of continents (e.g., Cawood et al., 2013;G. Gehrels, 2014), but uncertainties in paleogeographic reconstruction of the Lhasa terrane within Gondwana reflect the fact that DZ age spectral comparisons commonly lead to multiple nonunique solutions. Previous studies have shown that the long-distance transport of sediments in Gondwana resulted in complex sedimentary recycling (cf., Andersen et al., 2016;Morón et al., 2019;Myrow et al., 2010;Zoleikhaei et al., 2022), during which the sediments from different source regions have likely mixed in varying degrees (e.g., Sharman & Johnstone, 2017), and the Lhasa DZ signature might also reflect such mixing (Figures 1b-1d). In this paper we compile DZ U-Pb data for the Permo-Carboniferous strata (glacial deposits, Craddock et al., 2019) from the Lhasa terrane and those for coeval or older rock units from the potential source regions (Figure 1e, Table S1 in Supporting Information S1, including NW Australia, the Himalaya, cratonic India, Arabia and Iran, and Africa). We use the inverse Monte Carlo modeling (Sundell & Saylor, 2017) to test which reconstruction models provide best-fit matches to the observed Lhasa DZ age spectra. We show that a source from NW Australia, and especially the Perth basin, provides the best fit for the Lhasa DZ age spectra. In combination with other geological evidence, this suggests the Lhasa terrane was located outboard between Australia and India prior to its rifting from Gondwana to open the Neo-Tethys Ocean.

Geological Background, Methods and Detrital Zircon U-Pb Age Compilation

Following the final amalgamation of Gondwana, several microcontinents separated from its northern margin (i.e., peri-Gondwanan dispersion) and successively drifted northwards via the opening and closing of a series of Tethys oceans from the middle Paleozoic to the Cenozoic (Metcalfe, 2021). During these events the longest-lived Phanerozoic icehouse event occurred in Gondwana, and the frequency of glacial advance and retreat reached a peak during the Permo-Carboniferous (Montañez & Poulsen, 2013). These repeated cycles of glacial advance and retreat formed transcontinental ice flow, which transported sediment thousands of kilometers and produced massive glacial deposit across Gondwana and its periphery (including the Lhasa and Southern Qiangtang terranes) at the time (Craddock et al., 2019). Hence, the associated DZ age spectra are particularly useful for paleogeographic reconstruction and tracing continental-scale drainage systems (e.g., Craddock et al., 2019;Morón et al., 2019;Zhu et al., 2011).

Once exposed on the surface, the denudation of an igneous pluton rich in zircon will generate local sediments (that may later become sedimentary rocks) rich in DZs originating from the igneous source, but that may also contain DZs from additional sources, such as plutons of other ages, or from recycling of zircon grains once part of older sedimentary rocks (e.g., Zoleikhaei et al., 2022;Figure S1 in Supporting Information S1). Erosion, transportation, and then redeposition of such material will produce a DZ age distribution characterizing the various sources within the sedimentary basin (Figure S1 in Supporting Information S1, Sharman & Johnstone, 2017). Attempts to quantitatively reconstruct this mixing process can be made through inverse Monte Carlo modeling of DZ age spectra (Sundell & Saylor, 2017). When applied to the Permo-Carboniferous sedimentary rocks of the Lhasa terrane, competing reconstruction models can be tested because different potential source regions predict different DZ age spectra and sediment transport pathways (Figures 1b-1d).

We compiled published U-Pb DZ data from (meta-)sedimentary rocks deposited on Gondwana, and these data fall into six structurally defined geographic domains, including the Lhasa terrane, NW Australia (i.e., the Canning and Perth basins), the Himalaya (part of Greater India), cratonic India, Arabia and Iran, and Africa (Figure 1e and Table S1 in Supporting Information S1). Data quality for each analysis was reevaluated based on a rigorous set of criteria in the Supporting Information S1. For simplicity, we group the filtered samples from every domain based on probability density plot Cross-correlation three-dimensional metric multidimensional scaling (3D MDS, Figures S2-S6 in Supporting Information S1), which provides the best summary representation of intersample similarity among DZ age spectra (Saylor & Sundell, 2016). We regard each group recognized through MDS as one potential source sample (Figure 2).

The compiled Permo-Carboniferous sedimentary samples from the Lhasa terrane mainly contain zircons ranging in age from 2000 to 480 Ma, with two major age populations of ca. 1,250-1,050 and ca. 620-500 Ma (Figure 2a). The former was considered diagnostic of a link between the Lhasa terrane and NW Australia (e.g., Wang et al., 2021;Zhu et al., 2011). The 283 filtered samples from the potential source regions were assigned into 61 groups by the MDS, suggesting that the general Gondwana DZ signature is highly variable (Figure 2). With the exception of the Arabia-Iran samples, many samples from NW Australia, the Himalaya, cratonic northern India, and central and southern Africa contain abundant middle-late Mesoproterozoic DZs (Figures 1e and 2b-2g), also supporting the paleogeographic proximity of the Lhasa terrane to either Greater India or eastern Africa (cf., Gehrels et al., 2011;Hu et al., 2021;Zhang et al., 2012).

Detrital Zircon Mixing Modeling Constraining the Detrital Provenance of the Lhasa Terrane

The compiled DZ age data reveal a wide distribution of middle-late Mesoproterozoic DZs in Gondwana, providing a unique opportunity to test the reconstruction models (Figures 1b-1d) via the inverse Monte Carlo modeling (Sundell & Saylor, 2017). In each reconstruction model, we treat all Lhasa DZ samples as a single, mixed sample, and we perform 20,000 model trials with the best-fitting 0.5% selected to determine whether synthetic DZ age spectra match the natural Lhasa DZ age spectra by varying the contributions from all possible sources (Figure 3). The Lhasa-NW Australia link suggests that the Canning and Perth basins were the major source regions that transported detritus to the Lhasa terrane (Figures 1b, Wang et al., 2021;Zhu et al., 2011), and the inverse Monte Carlo modeling shows similar age spectra and yields matches with coefficients of determination (R 2 ) of ∼0.896 and Kuiper test V values of ∼0.082 (Figure 3a). The Lhasa-Greater India link considers cratonic India and Greater India as the potential source regions (Figure 1c, cf., Gehrels et al., 2011), but the model results (R 2 = ∼0.645, V = ∼0.118) apparently cannot match the Lhasa sample age distribution (Figure 3b). In terms of the Lhasa-eastern Africa link, it requires a complicated source system, and the possible source regions include cratonic India, Greater India, Africa, Arabia, and Iran (Figure 1d, cf., Hu et al., 2021;Zhang et al., 2012). Similarly, the DZ mixing modeling yields poor matches with R 2 = ∼0.680 and V = ∼0.118 (Figure 3c). Taken together, although factors such as sample size have significant influence on model results, inverse Monte Carlo results based on large-n DZ data set for the Lhasa-NW Australia link yield very good model fits with high R 2 and low V values, suggesting that the most likely detrital provenance of the Lhasa terrane was NW Australia (cf., Sundell & Saylor, 2017).

The traditional Lhasa-NW Australia connection suggests that the Canning and Perth basins supplied detritus to eastern and western Lhasa (Figure S7a in Supporting Information S1) through two different sediment dispersal pathways, respectively (Figure 1b, cf., Morón et al., 2019;Wang et al., 2021;Zhu et al., 2011). To test the possibility of one-to-one sediment dispersal pathway, we consider the eastern and western Lhasa DZ samples as two independent, mixed samples (Figure 2a). However, the inverse Monte Carlo model results for the eastern Lhasa-Canning pathway (maximum R 2 = 0.780, minimum V = 0.128) and the western Lhasa-Perth pathway (maximum R 2 = 0.783, minimum V = 0.081) are far from perfect fits (Figures 3d and 3e); particularly, the dominant source in the eastern Lhasa-Canning pathway (Figure 3d, i.e., Group 02, 80% contribution) consists of only limited DZ samples in the central Canning basin (Figure S7b in Supporting Information S1). We thus infer that both the Canning and Perth basins supplied detritus to both eastern and western Lhasa, but in different proportions. The DZ mixing modeling supports this inference, and the new model results indeed yield better fits for eastern Lhasa (Figure 3f, maximum R 2 = 0.896, minimum V = 0.097) and western Lhasa (Figure 3g, maximum R 2 = 0.864, minimum V = 0.070). It is noted that, from eastern Lhasa to western Lhasa, the major Perth source (i.e., Group04) that contains DZ samples widespread in the Perth basin (Figure S7b in Supporting Information S1) contributed constantly (∼40%-50%), whereas the major Canning source varied and contributed decreasingly (Figures 3f-3g), probably indicating that the Perth basin was the main source for Lhasa detritus.

Paleogeographic Reconstruction of the Paleozoic Lhasa Terrane

Our DZ mixing modeling using a large-n DZ data set indicates that glaciers and rivers transported sediments from NW Australia (mainly the Perth basin) along long-distance pathways to the Lhasa terrane during the Permo-Carboniferous. We thus propose the paleogeographic proximity of the Lhasa terrane to western Australia (Figure 4). It is generally accepted that the strata in the Lhasa, and Southern Qiangtang terranes, and Greater India record a marked Cambrian-Ordovician angular unconformity (cf., Myrow et al., 2010;Hu et al., 2021), suggesting the paleogeographic links between them. In contrast, the Sibumasu terrane strongly resembles NW Australia (i.e., the Canning basin) in the lowermost Ordovician stratigraphic succession from clastics to carbonates, and in the latest Cambrian sedimentary succession and shared elements of its trilobite fauna (Wernette et al., 2020a(Wernette et al., , 2020b(Wernette et al., , 2021)). This history allies Sibumasu and NW Australia at this time, but the status of the Baoshan terrane, commonly considered part of Sibumasu, remains to be clarified. Baoshan locally shows a Cambrian-Ordovician unconformity (Huang et al., 2012) and has a somewhat different stratigraphy to that of other part of Sibumasu. However, the similar lower Permian stratigraphy and faunas also suggest paleogeographic affinity of the Lhasa terrane to the west end of the Sibumasu terrane (i.e., the Tengchong terrane, Zhang et al., 2013). These studies support our inference that the Lhasa terrane spanned the boundary interval between Australian Gondwana and Indian Gondwana, as Metcalfe ( 2013) suggested (Figure 4). DZ provenance analyses thus show that the Paleozoic strata of the Lhasa and Southern Qiangtang terranes were sourced from NW Australia (herein) and cratonic India/Greater India (Myrow et al., 2010), respectively, suggesting the east-west position of these two terranes on the northern margin of Gondwana (Figure 4). Note that the Southern Qiangtang and Baoshan terranes are uniquely characterized by Permo-Carboniferous intracontinental rift-related mafic magmatism (Liao et al., 2015;Xu et al., 2016), which is also reported in western Greater India (Shellnutt et al., 2011). According to this view, both the Southern Qiangtang and Baoshan terranes were likely adjacent to western Greater India (Figure 4), and Baoshan was separated from other part of Sibumasu by the intervening Lhasa terrane. Such a separation might explain why Baoshan is the only part of Sibumasu terrane to show a local Cambrian-Ordovician unconformity (cf., Wernette et al., 2021). These observations suggest that during or prior to Permo-Carboniferous time, several adjacent continental fragments made up an elongate strip located peripheral to the eastern Gondwanan core from west to east, including the Southern Qiangtang, Baoshan, Lhasa, and Sibumasu terranes (Figure 4). Finally, the outboard continental strip successively rifted from eastern Gondwana to open the Neo-Tethys Ocean during the Permian (cf., Metcalfe, 2021;Xu et al., 2020).

The inverse Monte Carlo modeling has been successfully used to track multiple detrital provenances of sedimentary basins (e.g., the Chinese Loess Plateau, Licht et al., 2016;Sundell & Saylor, 2017). Here we conduct the similar modeling to investigate the detrital source regions of the Lhasa terrane, as well as its origin. Our study indicates that this method can be used in paleogeographic reconstructions of continents, particularly when it comes to multiple detrital source regions. For example, the latest Mesoproterozoic-earliest Neoproterozoic (ca. 1,000 Ma) North China Craton has been variously positioned adjacent to India and Siberia (D. A. D. Evans, 2009), Laurentia and Baltica (Liu et al., 2020), and northern Australia (Merdith et al., 2021). Future studies may approach other similar problems through DZ mixing modeling.

Conclusions

The late Mesoproterozoic DZ age population widely occurs in several Gondwanan components (i.e., NW Australia, India, and Africa), all of which probably transported detritus to form Permo-Carboniferous sediments in the Lhasa terrane through the long-distance northward glacial transport during the longest-lived Phanerozoic icehouse event. The inverse Monte Carlo modeling based on a large-n DZ data set from the potential source regions, including NW Australia, the Himalaya, cratonic India, Arabia and Iran, and Africa, provides us an opportunity to reconstruct the associated sedimentary recycling and mixing process. Our modeling results suggest that NW Australia (mainly the Perth basin) rather than Greater India or eastern Africa transported detritus to the Lhasa terrane during the Permo-Carboniferous, thus supporting that the Paleozoic Lhasa terrane was adjacent at the boundary between Australia and India. This study presents an example of reconstructing the paleogeographic positions of continents using DZ mixing modeling rather than traditional DZ age spectral comparisons.