Figure 3 summarizes our new U/Pb zircon and 40Ar/39Ar white mica ages, and the Si-in-white mica data. Probability density plots of each sample are shown with respect to stratigraphic age, geologic formation name, and geographic location. Because of analytical limitations and the rare white mica content in the samples, the number of analyzed grains per sample is less than the value (∼100 [Vermeesch, 2004]) necessary to guarantee that no fraction of the population comprising more than 5% of the total is missed at the 95% confidence level. However, this study aims to identify major source areas that fed the SGC; we base our discussion on linking major age populations of our samples with those of possible source areas and do not discuss single-grain ages. We accept that minor source areas might have contributed to the sedimentation but cannot be identified in this study.
 The data are interpreted in the context of the paleocurrent data of Weislogel et al.  shown in Figure 3 and published radiometric ages of potential source areas (Figures 2 and 4). Seven orogenic belts or segments of them and two crustal blocks were discriminated (Tongbai-Hong'an-Dabie, Qinling, east Kunlun, northern Qaidam, Qilian, Qiangtang, Yidun, north China, south China; Figures 2 and 4 and reference list in Text S1 in the auxiliary material). The literature compilation reflects the fact that bedrock thermochronology studies are heterogeneously distributed; e.g., 272 ages were published from the Hong'an-Dabie section of the QTHD orogen but only 22 ages from the Qiangtang metamorphic belt. This bias in the quality of the definition of age populations should be kept in mind when interpreting provenance.
5.1. Southern Songpan-Ganzi Complex
 The Ladinian sample (E30/02MTZ1) is dominantly (70% of all dated grains) composed of Silurian–Early Devonian (430–385 Ma) white mica ages that are consistent with ages from the compound Paleozoic orogenic belt that lines the southern margin of the north China block (northern Qinling-Tongbai-Hong'an, north Qilian, and east Kunlun, Figures 4c–4f). It also contains a population (23%) of Late Permian–Early Triassic mica ages (265–240 Ma), corresponding to ages in the Huwan mélange and the Qinling microcontinent in Hong'an [Xu et al., 2000; Ye et al., 1993; Niu et al., 1994] and the Triassic HP-UHP belt of the QTHD (Figure 4f) [Webb et al., 1999; Hacker et al., 2000]; it also corresponds to granitic gneisses from east Kunlun (Figure 4d) [Liu et al., 2005]. Zircon from this sample yielded primarily (37%) 880–720 Ma ages; such zircon ages occur frequently in the south China block (Figure 4g) and the metamorphosed south China crust found in the QTHD orogen [e.g., Hacker et al., 2000; Chen et al., 2003]. They also occur in the Qinling microcontinent (Figures 4c and 4f) [Ratschbacher et al., 2006]. Another significant zircon age component of ∼1.9 Ga (14%) is consistent with a derivation from the trans-north China block. The population of ∼440 Ma zircon could be derived either from the Qilian Shan (Figure 4e) [e.g., Gehrels et al., 2003; Song et al., 2005, 2006], east Kunlun (Figure 4d) [Chen et al., 2002; Cowgill et al., 2003], or the Qinling microcontinent (Figures 4c and 4f) [Reischmann et al., 1990; Kröner et al., 1993; Lerch et al., 1995].
 The early Carnian sample (E28) is located adjacent to and up section from E30 in the northeastern part of the southern SGC (Figure 3). White mica ages are almost exclusively (92%) Silurian-Carboniferous (420–325 Ma), corresponding to the Paleozoic orogenic belts along the southern margin of the north China block (Figures 2 and 4b–4f). No zircon age data from the lower Carnian SGC strata were acquired at this location; however, zircon ages from the Ladinian through Carnian rocks were located farther west (Figure 3) exhibit major populations of Permian (280–260 Ma) and Paleozoic (500–300 Ma) ages but lack a significant population of Precambrian zircons [Weislogel et al., 2006].
 The sediments represented by the Ladinian and early Carnian samples of the southern SGC were most likely fed from the south China block fragments that were accreted to the north China block during the Paleozoic in the QTHD orogen; i.e., the Qinling microcontinent, the Early Ordovician intraoceanic arc (Erlanping unit), the superimposed Devonian arc, the Devonian-Permian fore arc (Liuling, Nanwan, and Foziling units), and the Huwan mélange (Figure 2, 4c, and 4f).
 The two late Carnian samples (E22/02UTZ4, E26) are located in the central part of the southern SGC (Figure 2). Similar to sample E28, 94% of the detrital mica ages of E22 are Silurian-Carboniferous (420–325 Ma), corresponding to the Paleozoic belts along the southern margin of the north China block (Figures 4b–4f). Only two micas yielded Late Permian–Early Triassic (250–242 Ma) ages. Paleocurrent data from this area indicate a supply from the northeast [Weislogel et al., 2006]; the intermediate to high Si-in-white mica values of E22 correspond to Paleozoic phengites reported from Hong'an (Figure 4f) [Xu et al., 2000] or north Qilian (Figure 4e) [Song et al.  (see below). Zircon ages from this sample location are primarily 900–660 Ma (23%) and 495–325 Ma (23%); these ages correspond to crystalline basement ages of the south China block and its reworked counterparts in the QTHD orogen [e.g., Hacker et al., 2000, 2004; Chen et al., 2003] and to north Qinling, east Kunlun, Qaidam, and Qilian [e.g., Lerch et al., 1995; Xue et al., 1996; Gehrels et al., 2003] (Figures 4b–4f). Another significant component (15%) at 2.0–1.8 Ga stems from the north China block. The population of 295–250 Ma zircon could be derived either from east Kunlun (Figure 4d) [Cowgill et al., 2003], Hong'an-Dabie (Figure 4f) [e.g., Hacker et al., 1998; Ayers et al., 2002], or north Qaidam (Figure 4b) [Gehrels et al., 2003]. The Archean–Early Proterozoic (3.2–2.4 Ga, 10%) zircon ages correspond to ages of the north China block (Figure 4h). The broad peak of Paleozoic mica ages is in contrast to the tighter and ∼30 Ma older peak of zircon ages; this suggests slow cooling within the Paleozoic orogen. Zircon data from the time-equivalent Zhuai Formation farther to the north (UTZh7 of Weislogel et al. ; Figure 3) show a broad age range between 440 and 250 Ma, corresponding to the Paleozoic orogenic belts (east Kunlun, north Qinling, Qilian, and Qaidam) located to the north of this sample and the QTHD belt (Figures 4b–4f); this sample, however, does not contain Precambrian zircons.
 In sample E26, 34% of mica ages are younger than its stratigraphic age. The sample location is close to a granite body and was probably thermally effected. Thin sections of a sample derived from the same outcrop shows grow of new white micas, which supports a thermal overprint. We therefore do not interpret the ages of E26 geologically.
 We suggest that the late Carnian samples mainly share their source with the Ladinian to early Carnian samples, i.e., the Paleozoic QTHD belt. The likely greater supply of Triassic detritus in the late Carnian than in the Ladinian–early Carnian samples suggests an eastward and southward shift of the drainage area to the Triassic metamorphic belt of the QTHD orogen; rapid exhumation in Hong'an-Dabie probably caused a more restricted drainage area [e.g., Amidon et al., 2005].
 The Middle (E16) and late Norian (E13) samples are located in the southeastern part of the southern SGC (Figure 3). The white micas comprise a broadly distributed Triassic age population (250–207 Ma; ∼50%), corresponding to ages in the Triassic HP-UHP metamorphic Hong'an–Dabie Shan (Figure 4f) [e.g., Webb et al., 1999; Hacker et al., 2000] and the Triassic basement domes of the south China block [Ratschbacher et al., 2003]. The Triassic white mica ages also correspond to ages form the west Yidun metamorphic and magmatic belt [Reid et al., 2005] and the Qiangtang belt [Kapp et al., 2000, 2003; Li et al., 2006] located to its west (Figure 2, 3, and 4a). However, the east Yidun volcanic arc, active in the Middle and Late Triassic, likely prohibited sediment supply from areas to its west and probably did not contain white mica-bearing rocks (Figure 3). Samples E16 and E13 also contain Paleozoic white mica age populations at 370–330 Ma (48%) and 390–370 Ma (20%), respectively. These ages correspond to ages in the Qinling (Figure 4c) [Mattauer et al., 1985] and in the east Kunlun belt (Figure 4d) [Liu et al., 2005]. Zircons from locality E16 (UTYa3 of Weislogel et al. ) primarily contain Neoproterozoic (∼750 Ma) and Paleoproterozoic (∼1.9 Ga) age populations. The Neoproterozoic population is evidently derived from the rift-related rocks of this age that are common in the south China block (Figure 4g); Neoproterozoic ages occur in the Qinling microcontinent but mostly in the Triassic belt of the QTHD orogen. They also occur in the basement domes of the Longmen Shan that is closest to our sample locality (Kangding area of Figure 3 [Liu et al., 2003]). Paleoproterozoic ages were reported from the south China block basement but such ages are typical for the trans-north China block [e.g., Guo et al., 2005; Kröner et al., 2006], and for volcanic rocks along the southern margin of the north China block (Figure 4h) [Zhao et al., 2004]. The Permo-Triassic zircon age population (260–219 Ma; 6%) tightly overlaps the mica age population (250–207 Ma) and thus suggests a supply from a rapidly exhuming orogen. The overall meager mica content in these samples indicates an additional source that lacks white mica, or destruction of mica due to extreme environments. Such sources could have been the Permo-Triassic Emeishan magmatic province that is located to the southeast, at the western margin of the south China block (Figure 2) and yielded whole rock, biotite, hornblende, and plagioclase 40Ar/39Ar ages of 256–246 Ma [Boven et al., 2002; Lo et al., 2002] and the east Yidun arc that provided 240–216 Ma zircon ages [Reid et al., 2005; Liu et al., 2006]. Paleocurrent data from the localities of samples E16 and E13 suggest a source area in the southeast [Weislogel et al., 2006] (Figure 3). However, these southeastern directions in the southernmost SGC likely need to be rotated counterclockwise by ≤50° in the light of GPS and paleomagnetic data that indicate clockwise rotation around the eastern Himalayan syntaxis in this area [e.g., Shen et al., 2005; Sato et al., 1999; Otofuji et al., 1990]. Taking this into account, a source from the northeast, i.e., the QTHD belt, is likely.
 The zircon ages indicate derivation from combined south China and north China block sources, i.e., from the entire QTHD belt (Figures 4c and 4f). Locality E16/UTYa3 records a significant increase in Triassic mica ages and the first occurrence of Late Triassic zircons [Weislogel et al., 2006]. The overlap of Late Permian–Late Triassic mica and zircon ages indicates that the southern SGC deposystem was most likely fed by the rapidly exhuming Triassic QTHD orogen. A supply from this orogen implies a long and narrow drainage system following the strike of the orogen but also significant southward channeling of the sediments along the western margin of the south China block (Figure 5). We suggest that the triple-junction configuration, constituted by the east trending Kunlun-Anyemagen-Qinling suture/orogen and the south-southwest trending margin of the south China block played a significant role for channeling the sediments into the southeastern part of the southern SGC [e.g., Wang et al., 2001].
Figure 5. Summary of suggested source areas for the southern Songpan-Ganzi complex in the (top) Ladinian–early Carnian, (middle) late Carnian, and (bottom) the northern and southern Songpan-Garzi complex in the Norian and the northeastern Sichuan basin in the Rhaetian.
Download figure to PowerPoint
5.2. Northern Songpan-Ganzi Complex
 In contrast to the Norian samples of the southern SGC, no Permo-Triassic white mica ages occur in the two Norian samples (E40/02UTZu1 and E46) of the northern SGC (Figure 3). Most ages are Silurian-Devonian (435–380 Ma; 40–70%), corresponding to the Paleozoic belts along the southern margin of the north China block (Figures 2 and 4h), the east Kunlun [e.g., Gehrels et al., 2003], Qilian [Song et al., 2006], and Qinling-Tongbai-Hong'an (Figures 4c–4f) [e.g., Xu et al., 2000; Ratschbacher et al., 2003]. Additionally, both samples yielded ages of 370–320 Ma (13–25%). Such Late Devonian–Carboniferous mica ages have rarely been reported; two ages from the Kuangping and Qinling units of northern Qinling (Figure 4c) [Mattauer et al., 1985; Zhang et al., 1991], one age from south Qilian (Figure 4e) [Liu et al., 2003], and one from east Kunlun (Figure 4d) [Liu et al., 2005] are known.
 Similar to the mica age distribution, the E40/02UTZu1 zircon age distribution also lacks Permo-Triassic ages but consists of 34% Ordovician-Silurian zircons (peak at 440 ± 10 Ma), corresponding to the HP-UHP metamorphic belts and the granitoids in northern Qaidam (Figure 4b) [e.g., Gehrels et al., 2003; Song et al., 2005, 2006] and northern Qilian (Figure 4e) [e.g., Cowgill et al., 2003; Song et al., 2004]; only one zircon age from this range is known from the east Kunlun [Chen et al., 2002], a few from the Huwan mélange [Jian et al., 2001; Gao et al., 2002], and the UHP unit in the Dabie Shan (Figure 4f) [Rowley et al., 1997; Ma et al., 2005]. The remaining populations are composed in about equal proportions of Archean (2.8–2.6 Ga; 11%), Mesoproterozoic (1.0–0.9 Ga; 16%), and Neoproterozoic grains (775–730 Ma; 11%). The Archean zircon grains are most likely derived from the north China block [e.g., Kröner et al., 2005; Gao et al., 2006]. The Mesoproterozoic and Neoproterozoic ages correspond to crystalline blocks that are involved in the Paleozoic orogenic belts of Qilian [Mao et al., 2000; Tseng et al., 2006], Qaidam [Gehrels et al., 2003], and Qinling [Chen et al., 1992; Ratschbacher et al., 2003] (Figures 4e, 4b, and 4c, respectively).
 The major zircon age population is ∼30 Ma older than the major mica age population, indicating that the source area is composed of rocks that cooled slowly during the Silurian-Devonian. This is consistent with derivation from north Qaidam and Qilian rocks produced by the amalgamation of the Qilian and Qaidam microcontinents with the north China block. Derivation from the Qinling rocks is less likely due to the paucity of Late Ordovician–Silurian zircon ages in the QTHD belt. The ∼80 Ma gap between the depositional age of samples E46 and E40/02UTZu1 and the youngest mica age population (370–320 Ma) indicates that the source rocks of these micas had been at shallow crustal depths for a significant time prior to exhumation and erosion.
 Zircons from four Late Triassic (Ladinian through Norian) samples of the northern SGC, located west (UTKa1, UTZh1, UTZho2) and east (UTNa2) of locality E40/02UTZu1 (Figure 3) show significant Late Permian–Early Triassic (280–245 Ma) age populations [Weislogel et al., 2006]. The occurrence of detrital zircon U/Pb ages that are younger than the detrital white mica 40Ar/39Ar ages can be explained either by a volcanic/magmatic source that did not contain white mica, recycling of a sedimentary source, or weathering during erosion and transport that destroyed mica in the sediment load. The Late Permian–Early Triassic east Kunlun arc and the Early Triassic granitoids of west Qinling are suggested to be the major source for the northern SGC. The lack of those zircon ages in sample E40/02UTZu1 is probably due to its location between the granitoids of west Qinling and east Kunlun (see Figure 3), or due to changes in the source area. We conclude that the northern SGC was fed by the Paleozoic orogenic belt along the southern margin of the north China block (east Kunlun, north Qaidam and Qilian, northern units of Qinling) and by the Late Permian–Early Triassic Kunlun arc (Figure 5). This implies that major exhumation and erosion in the Late Triassic was not only taking place along the QTHD belt due to the collision of the north China and south China blocks but occurred also north of the SGC, along the Paleozoic belts west of the QTHD belt; the latter might reflect northward subduction beneath an increasingly mature wedge-arc complex north of the Kunlun-Anyemagen-Qinling suture.
5.3. Sichuan Basin
 We analyzed mica grains from one Rhaetian (Late Triassic) sandstone sample (E38) located in the northwestern Sichuan basin. The white mica data show two major age populations: (1) Late Permian–Late Triassic (260–220 Ma, 59%) corresponding to ages from the Triassic HP-UHP belt of the QTHD orogen (Figures 4c and 4f), the southern margin of the Qinling microcontinent that was affected by a Late Permian–Triassic thermal event (Liuling and Nanwan units, Huwan mélange [e.g., Hacker et al., 2000; Ratschbacher et al., 2006]), and the east Kunlun (Figure 4d) [Liu et al., 2005]. (2) Silurian–Middle Devonian (440–380 Ma; 35%) corresponding to the Erlangping, Qinling, and Liuling units, and the Huwan mélange of the QTHD belt [Xu et al., 2000; Zhang et al., 1991; Ratschbacher et al., 2003] and the east Kunlun arc [Chen et al., 2002; Wang et al., 2003]. An east Kunlun source is improbable due to the sample location in the Sichuan basin that was likely separated from sources to the (north)west by an ocean basin as well as by the evolving Longmen Shan thrust belt in the Late Triassic (Figures 2 and 3).
 We suggest that the northern Rhaetian Sichuan basin was sourced by the QTHD orogen (Figure 5). The low Si contents in the micas suggest that the HP-UHP rocks of the Triassic Hong'an-Dabie orogen were not exposed at the surface in the Late Triassic (see below).
5.4. Comparison of Zircon and Mica Grain Age Data
 The combination of detrital zircon and white mica age data allows general statements regarding the contribution of first-cycle and recycled sediment sources to the SGC deposits. In addition, the Si-in-white mica content allows discrimination of HP-UHP phengitic white mica from low-pressure metamorphic or igneous muscovite [e.g., Massonne and Szpurka, 1997; Grimmer et al., 2003].
 Mica ages are predominantly Paleozoic, whereas zircon ages are primarily Precambrian; Precambrian ages are only sparsely represented in the mica age data. This suggests that Precambrian micas vanished during recycling of sedimentary rocks derived from the original Precambrian sources or were reset by later metamorphism. Where zircon ages exhibit a cluster at ∼445–430 Ma, mica ages are younger by approximately 30 Ma. Assuming the same source rocks for the micas and zircons, this age difference is interpreted to be indicative of slow cooling in the source region, likely along the Paleozoic orogenic belts that formed by accretion of microcontinents (Qinling, Qaidam, Qilian) against the southern margin of the north China block [Wang et al., 2005; Ratschbacher et al., 2003]. The Late Paleozoic–Triassic zircon and mica age clusters are less separated; this suggests more rapid exhumation in the hinterland, consistent with geochronometric data from the Triassic HP-UHP metamorphic belt of the QTHD orogen [Li et al., 2000; Hacker et al., 2000, 2004].
5.5. Does the SGC Contain HP-UHP Rocks From the Hong'an-Dabie Shan?
 Zircon U/Pb ages of crustal rocks of the Hong'an–Dabie Shan record protracted UHP metamorphic growth at 240–225 Ma [e.g., Hacker et al., 2006]. Together with Ar/Ar phengite cooling ages at ∼235–225 Ma, they suggest rapid exhumation from mantle to lower crustal depths; exhumation through the crust at ∼225–195 Ma was less rapid [e.g., Eide et al., 1994; Hacker et al., 2000; Ratschbacher et al., 2006]. Studies published so far indicate that the UHP rocks reached the surface not earlier than Jurassic [Grimmer et al., 2003; Wang et al., 2003; Wan et al., 2005; Li et al., 2005].
 Samples of our study show in general low Si-in-white mica content and thus exclude supply from a major HP-UHP source area; this conforms to a post-Triassic surface exposure of the Hong'an–Dabie's UHP rocks. Southern SGC sample E22 (upper Carnian) yielded a small population of intermediate to high Si content micas and mica age populations of 325–425 Ma and 240–250 Ma. Ar/Ar ages of phengite and muscovite of the UHP Hong'an–Dabie belt are younger (240–195 Ma, Figure 4f [e.g., Eide et al., 1994; Hacker et al., 2000; Ratschbacher et al., 2006]) and thus can most probably be excluded as a source. Similarly, a source from the westerly located central Qiangtang metamorphic belt, with phengite ages of ∼220 Ma [Kapp et al., 2003], can be excluded. Paleozoic phengite cooling ages were reported from eclogites in the Huwan mélange in the QTHD orogen (350–430 Ma, Figures 2 and 4f [Xu et al., 2000]) and from the north Qilian belt (400 Ma, Figure 4e [Song et al., 2006]). As the central southern SGC was mainly sourced from the northeasterly QTHD belt, we suggest that the intermediate to high Si content micas in the southern SGC were derived from the Huwan mélange. Furthermore, our results indicate that detritus of the Hong'an–Dabie HP–UHP rocks was neither deposited in the Middle and Late Triassic SGC basin nor in the uppermost Triassic strata of the Sichuan basin. That conforms to recent provenance studies, suggesting that the HP–UHP detritus arrived during the Jurassic in the foreland basins to the south and east [Grimmer et al., 2003] and to the north of the Dabie Shan (Hefei basin [e.g., Li et al., 2005]). These provenance studies imply that the vast amounts of Triassic sediments deposited in the SGC basin were derived from the Paleozoic QTHD orogen and the orogenic belts farther west (Qilian, Qaidam, east Kunlun). During the Late Triassic and with progressing collision of the south China and the north China blocks, the southern part of the QTHD and the roof of the HP–UHP rocks, representing south China block, were increasingly included into erosional denudation. The inference gained from the available provenance studies (little erosional denudation of the Hong'an–Dabie Shan in the Triassic but a substantial one in the Jurassic) support structural and geochronologic models that attribute Triassic exhumation of the HP–UHP rocks of the Hong'an–Dabie Shan to tectonic rather erosional denudation [e.g., Hacker et al., 2004; Ratschbacher et al., 2006].