Lithosphere thickness controls continental basalt compositions: An illustration using Cenozoic basalts from eastern China

Recent studies demonstrate that lithosphere thickness variation exerts the primary control on global seafloor basalt compositions. If the mechanism of such control, i.e., the lid effect, is indeed at work, lithosphere thickness variation must also influence basaltic compositions in continental settings. To test this hypothesis, we chose to study Cenozoic basalts in eastern continental China over a distance of ∼260 km along a southeast-to-northwest traverse with a steep topographic gradient (∼500 to ∼1500 m above sea level) mirrored with a steep lithospheric thickness gradient (∼80 to ∼120 km). The basalts erupted on the thinned lithosphere to the east are characterized by lower pressure (e.g., higher Si72, lower Mg72, Fe72, and [Sm/Yb]N; subscript “72” refers to corresponding oxides corrected for fractionation effect to Mg# = 72; N—primitive mantle normalized) and higher extent (e.g., low Ti72, P72, K72, Rb, Ba, Th, and ratios of moreto less-incompatible elements such as [La/Sm]N, Ba/Zr, and Zr/Yb) of melting than basalts erupted on the thickened lithosphere to the west. Importantly, these geochemical parameters all show significant correlations with both lithosphere thickness and topographic elevation. These first-order observations are a straightforward manifestation of the lid effect. Lithospheric contamination and mantle-source compositional variation can indeed contribute to the compositional variability of these continental basalts, but these latter effects are averaged out and are overshadowed by the lid effect. This finding emphasizes the importance of evaluating the lid effect before interpreting the petrogenesis of continental basalts and mantle dynamics. Our results also indicate that the continental surface elevation is isostatically balanced above a mantle depth that is deeper than the lithosphere-asthenosphere boundary. INTRODUCTION Basaltic magmas produced in continental settings have large compositional variations: petrologically from tholeiites to varying alkalirich basalts (e.g., Dupuy and Dostal, 1984; Bell and Peterson, 1991; Guo et al., 2016). While factors such as source compositional variation (e.g., Lum et al., 1989), fractional crystallization (e.g., Peterson, 1989), and crustal contamination (Dupuy and Dostal, 1984; Ingle et al., 2004) can all affect erupted basalt compositions, the lithosphere thickness effect on the compositional variation of continental basalts has been largely overlooked, despite some speculation about lithospheric affects on the abundance and  patterns of rare earth elements in oceanic basalts by Ellam (1992) and other studies, and implications in experimental petrology (e.g., Green and Ringwood, 1967). Recent studies of global seafloor basalts demonstrate that lithosphere thickness variation exerts the primary control on the compositions of these basalts, especially those erupted on intraplate ocean islands with varying lithosphere thickness at the time of eruption (Humphreys and Niu, 2009; Niu et al., 2011; Niu, 2016; Niu and Green, 2018). Seafloor basalts erupted on thicker lithosphere have geochemical characteristics of lower extent (F) and higher pressure (P) of melting, whereas basalts erupted on thinner lithosphere have geochemical signatures of higher F and lower P. This is because F ∝ P0 – Pf, where P0 is the initial depth of melting when the adiabatically upwelling asthenospheric mantle intersects the solidus, and Pf is the depth of melting cessation and melt extraction when the decompression-melting mantle encounters the lithosphere, that is, the depth of the lithosphereasthenosphere boundary (LAB). This is the concept of the “lid effect” (see Niu et al., 2011), and its mechanism is simply capping decompression melting at the LAB (Niu and Green, 2018). If the lithospheric lid affects oceanic basaltic magmatism, then the lid effect must also affect basaltic magmatism in continental settings, with erupted basalts recording the lid effect as the result of varying lithosphere thickness. To test the lid effect hypothesis for continental basaltic magmatism, and to evaluate the extent of this effect on the compositional variation of continental basalts, we chose to study Cenozoic basalts in eastern continental China in the Chifeng–Xilin Hot area along a southeast-to-northwest, ∼260 km, traverse with a steep topographic gradient (∼500 to ∼1500 m above sea level) corresponding to a steep lithospheric thickness gradient (∼80 to ∼120 km). Our finding is fully consistent with the lid effect. We note that source compositional variation and lithospheric contamination can contribute to the compositional variability of continental basalts, but these are secondary and are averaged out, with the mean compositions markedly reflecting the lid effect. GEOLOGICAL BACKGROUND Cenozoic basaltic volcanism is widespread in eastern continental China, from Wudalianchi in the northeast to Hainan Island in the south (Fan and Hooper, 1991). Most of these volcanic rocks are alkali-rich varieties (e.g., Guo et al., 2016; Sun et al., 2017), with tholeiites also present in several locations (Zhi et al., 1990; Zou et al., 2000; Xu et al., 2005). Studies on these continental basalts reveal that they are isotopically depleted relative to the bulk silicate earth, with εNd(t) > 0 and εHf(t) > 0, *Current address: Institute of Oceanology, Chinese Academy of Sciences, Nanhai Road 7, Qingdao, Shandong 266071, China; E-mail: guopy@qdio.ac.cn. Published online 22 November 2019 Downloaded from https://pubs.geoscienceworld.org/gsa/geology/article-pdf/48/2/128/4926775/128.pdf by Durham University user on 03 February 2020 Geological Society of America | GEOLOGY | Volume 48 | Number 2 | www.gsapubs.org 129 but highly enriched in incompatible elements and also enriched in the progressively more-incompatible elements (e.g., Guo et al., 2016; Sun et al., 2017), resembling present-day ocean island basalts (OIBs). The origin of these basalts has been explained as the result of asthenospheric mantle upwelling and decompression melting, which was ultimately caused by the western Pacific wedge suction–induced eastward asthenosphere flow beneath eastern continental China (Niu, 2005, 2014). The Cenozoic basaltic volcanism in the Chifeng–Xilin Hot area is a type example of Cenozoic volcanism in the region, with its eruption age ranging from ca. 23.8 Ma to ca. 0.19 Ma (Ho et al., 2008; Wang et al., 2015). These basalts spread over a spatial distance of ∼260 km across the “Great Gradient Line” (GGL; Niu, 2005), a steep gradient in gravity, elevation, topography, crustal thickness, lithosphere thickness, and heat flow between the high plateaus to the west and the hilly lowland plains in the east (Fig. 1A). Previous studies demonstrated that the distinct contrasts in these geological observations were the result of lithosphere thinning in the region to the east of GGL in the Mesozoic (e.g., Niu, 2005, 2014; Zhu et al., 2011). As shown in Figures 1B and 1C, Chifeng is to the east of the GGL, and Xilin Hot is to the west. Regionally, high-resolution seismic tomography reveals significant changes in the depth of the LAB beneath the Chifeng–Xilin Hot area, ranging from ∼80 km beneath Chifeng to ∼120 km beneath Xilin Hot (Fig. 1C). This LAB depth also correlates well with the surface elevation (Fig. 1C), reflecting a first-order isostatic equilibrium. The Chifeng–Xilin Hot Cenozoic basalts thus offer a prime opportunity to test the lid effect hypothesis in a continental setting. SYSTEMATIC COMPOSITIONAL VARIATIONS OF THE CHIFENG–XILIN HOT BASALTS We analyzed bulk-rock major element, trace element, and Sr-Nd-Hf isotope on 19 fresh basalt samples from three locations (solid symbols in Fig. 1B). The analytical methods and results are given in the GSA Data Repository1. We also used recently published data on 41 basaltic samples from the Chifeng–Xilin Hot area (half-filled symbols in Fig. 1B; Wang et al., 2015; Guo et al., 2016; Pang et al., 2019). In order to remove the effects of fractional crystallization, we corrected major element compositions of all these samples to Mg# = 0.72, the minimum value to be in equilibrium with mantle olivine, following the method of Humphreys and Niu (2009) (see the Data Repository). Spatially, from the southeast to northwest, these basalts change gradually from tholeiite (quartz normative) to transitional basalts (hypersthene normative) to alkali basalts (nepheline normative) (Fig. DR1 in the Data Repository). Figure 2 plots major element compositions corrected to Mg# = 0.72 (notated using the subscript “72” on element symbols) as a function of distance relative to the location of the most southeastern sample (sample CF14-02) calculated using the great-circle distance (e.g., Niu and Batiza, 1993), and shows that Si72 decreases while Mg72, Fe72, Ti72, P72, and K72 increase toward the northwest. Such consistent spatial trends are also obvious for incompatible elements (Fig. DR2), for ratios of highly to moderately 1GSA Data Repository item 2020xxx, methods, fractionation correction procedure, Figures DR1– DR9, and Tables DR1–DR3, is available online at http://www.geosociety.org/datarepository/2020/, or on request from editing@geosociety.org. Figure 1. (A) Topographic map of East Asia (data from Amante and Eakins, 2009). “Great Gradient Line” (GGL) is indicated as purple dashed line, which contrasts highelevation and thickened lithosphere to the west from low-elevation and thinned lithosphere to the east. Study area is indicated with a rectangle, with the A-B traverse used in subsequent figures. (B) Distribution and sample locations of Chifeng–Xilin Hot Cenozoic basalts. Solid blue triangles, solid green diamonds, and solid red circles represent sample locations in this study. Half-filled diamonds and half-filled circles represent transitional basalt locations from the literature (Wang et  al., 2015; Guo et  al., 2016; Pang et al., 2019). (C) Top: Topographic profile along the A-B section in A.


INTRODUCTION
Basaltic magmas produced in continental settings have large compositional variations: petrologically from tholeiites to varying alkalirich basalts (e.g., Dupuy and Dostal, 1984;Bell and Peterson, 1991;Guo et al., 2016). While factors such as source compositional variation (e.g., Lum et al., 1989), fractional crystallization (e.g., Peterson, 1989), and crustal contamination (Dupuy and Dostal, 1984;Ingle et al., 2004) can all affect erupted basalt compositions, the lithosphere thickness effect on the compositional variation of continental basalts has been largely overlooked, despite some speculation about lithospheric affects on the abundance and patterns of rare earth elements in oceanic basalts by Ellam (1992) and other studies, and implications in experimental petrology (e.g., Green and Ringwood, 1967).
Recent studies of global seafloor basalts demonstrate that lithosphere thickness variation exerts the primary control on the compositions of these basalts, especially those erupted on intraplate ocean islands with varying lithosphere thickness at the time of eruption (Humphreys and Niu, 2009;Niu et al., 2011;Niu, 2016;Niu and Green, 2018). Seafloor basalts erupted on thicker lithosphere have geochemical characteristics of lower extent (F) and higher pressure (P) of melting, whereas basalts erupted on thinner lithosphere have geochemical signatures of higher F and lower P. This is because F ∝ P 0 -P f , where P 0 is the initial depth of melting when the adiabatically upwelling asthenospheric mantle intersects the solidus, and P f is the depth of melting cessation and melt extraction when the decompression-melting mantle encounters the lithosphere, that is, the depth of the lithosphereasthenosphere boundary (LAB). This is the concept of the "lid effect" (see Niu et al., 2011), and its mechanism is simply capping decompression melting at the LAB (Niu and Green, 2018). If the lithospheric lid affects oceanic basaltic magmatism, then the lid effect must also affect basaltic magmatism in continental settings, with erupted basalts recording the lid effect as the result of varying lithosphere thickness.
To test the lid effect hypothesis for continental basaltic magmatism, and to evaluate the extent of this effect on the compositional variation of continental basalts, we chose to study Cenozoic basalts in eastern continental China in the Chifeng-Xilin Hot area along a southeast-to-northwest, ∼260 km, traverse with a steep topographic gradient (∼500 to ∼1500 m above sea level) corresponding to a steep lithospheric thickness gradient (∼80 to ∼120 km). Our finding is fully consistent with the lid effect. We note that source compositional variation and lithospheric contamination can contribute to the compositional variability of continental basalts, but these are secondary and are averaged out, with the mean compositions markedly reflecting the lid effect.

GEOLOGICAL BACKGROUND
Cenozoic basaltic volcanism is widespread in eastern continental China, from Wudalianchi in the northeast to Hainan Island in the south (Fan and Hooper, 1991). Most of these volcanic rocks are alkali-rich varieties (e.g., Guo et al., 2016;Sun et al., 2017), with tholeiites also present in several locations (Zhi et al., 1990;Zou et al., 2000;Xu et al., 2005). Studies on these continental basalts reveal that they are isotopically depleted relative to the bulk silicate earth, with ε Nd(t) > 0 and ε Hf(t) > 0, but highly enriched in incompatible elements and also enriched in the progressively more-incompatible elements (e.g., Guo et al., 2016;Sun et al., 2017), resembling present-day ocean island basalts (OIBs). The origin of these basalts has been explained as the result of asthenospheric mantle upwelling and decompression melting, which was ultimately caused by the western Pacific wedge suction-induced eastward asthenosphere flow beneath eastern continental China (Niu, 2005(Niu, , 2014. The Cenozoic basaltic volcanism in the Chifeng-Xilin Hot area is a type example of Cenozoic volcanism in the region, with its eruption age ranging from ca. 23.8 Ma to ca. 0.19 Ma (Ho et al., 2008;Wang et al., 2015). These basalts spread over a spatial distance of ∼260 km across the "Great Gradient Line" (GGL; Niu, 2005), a steep gradient in gravity, elevation, topography, crustal thickness, lithosphere thickness, and heat flow between the high plateaus to the west and the hilly lowland plains in the east (Fig. 1A). Previous studies demonstrated that the distinct contrasts in these geological observations were the result of lithosphere thinning in the region to the east of GGL in the Mesozoic (e.g., Niu, 2005Niu, , 2014Zhu et al., 2011). As shown in Figures 1B and 1C, Chifeng is to the east of the GGL, and Xilin Hot is to the west. Regionally, high-resolution seismic tomography reveals significant changes in the depth of the LAB beneath the Chifeng-Xilin Hot area, ranging from ∼80 km beneath Chifeng to ∼120 km beneath Xilin Hot (Fig. 1C). This LAB depth also correlates well with the surface elevation ( Fig. 1C), reflecting a first-order isostatic equilibrium. The Chifeng-Xilin Hot Cenozoic basalts thus offer a prime opportunity to test the lid effect hypothesis in a continental setting.

SYSTEMATIC COMPOSITIONAL VARIATIONS OF THE CHIFENG-XILIN HOT BASALTS
We analyzed bulk-rock major element, trace element, and Sr-Nd-Hf isotope on 19 fresh basalt samples from three locations (solid symbols in Fig. 1B). The analytical methods and results are given in the GSA Data Repository 1 . We also used recently published data on 41 basaltic samples from the Chifeng-Xilin Hot area (half-filled symbols in Fig. 1B; Wang et al., 2015;Guo et al., 2016;Pang et al., 2019). In order to remove the effects of fractional crystallization, we corrected major element compositions of all these samples to Mg# = 0.72, the minimum value to be in equilibrium with mantle olivine, following the method of Humphreys and Niu (2009) (see the Data Repository).

EVALUATION OF CRUSTAL MATERIAL CONTAMINATION
Continental crustal contamination during magma ascent is inevitable, but crustal contamination proxies, such as SiO 2 /MgO, K 2 O/TiO 2 , K 2 O/P 2 O 5 , Ce/Pb, Nb/Th, Ta/U, 87 Sr/ 86 Sr, 143 Nd/ 144 Nd, and 176 Hf/ 177 Hf, show no coherent correlations between each other in our study, suggesting that this effect is negligible (see Fig. DR5). Furthermore, the occurrence of mantle xenoliths carried in Xilin Hot alkaline basalts (Fig. DR6) indicates that the magma ascended rapidly with limited interaction with the crust, which is supported by recent studies Guo et al., 2016;Sun et al., 2017;Pang et al., 2019). Therefore, the observed major element and trace element compositional systematics in these basalts (Figs. 2-4) largely reflect those of primary magmas parental to the basalts, as the result of varying source composition or varying extent and pressure of mantle melting.

EVALUATION OF MANTLE SOURCE COMPOSITIONAL VARIATIONS
Generally, the studied basalts have OIB-like incompatible element compositions with high [La/ Sm] N (1.2-3.5) and [Sm/Yb] N (2.4-9.2), and are progressively more enriched in the progressively more incompatible elements (Fig. DR7), indicating (1) their derivation from varying low-degree melting of prior metasomatically enriched sources, and (2) the partial melting occurring in the sublithospheric mantle garnet stability field, with garnet as a residual phase. These basalts also show OIB-like depleted Sr-Nd-Hf isotope compositions (Fig. DR3) and elevated [Nb/Th] N and [Ta/U] N (Fig. DR8), differing distinctively from those of the >110 Ma alkali basalts in eastern China (Figs. DR3 and DR8), which were derived from melting of the continental lithospheric mantle (e.g., Niu, 2014;Meng et al., 2015). Therefore, these basalts originated from partial melting of metasomatized asthenospheric (versus lithospheric) mantle. Furthermore, the lack of temporal and spatial variation of Sr-Nd-Hf isotopes of these basalts (Fig.  DR4) indicates that these basalts have shared a similarly heterogenous asthenospheric mantle source in the Cenozoic.

LITHOSPHERE THICKNESS EFFECT ON BASALT COMPOSITIONS
In the Chifeng-Xilin Hot area, the lithosphere gradually thickens northwestward (dotted line labeled LAB in Figure 1C). This is mirrored by the increasing surface elevation as the result of isostatic equilibrium; the surface elevation positively correlates with the LAB depth. If the lid effect hypothesis is valid and applies here, we should see systematic variation in the compositions of these basalts as a function of the lithosphere thickness and surface elevation. This is indeed the case (Figs. 2-4). Figures 2 and 3 show that the Chifeng-Xilin Hot basalts display decreasing Si 72 but  A-B traverse (Fig. 1B). Relative distance is calculated following the method of Niu and Batiza (1993). Black circles represent average composition, with error bars of 2 standard deviations. Subscript "72" refers to corresponding oxides corrected for the fractionation effect to Mg# = 72 (Humphreys and Niu, 2009) so as to facilitate discussion of mantle sources and processes. Three bands with different colors indicate thick, medium, and thin lithosphere at the time of basalt eruption. Symbols are as in Figure 1B. increasing Mg 72 , Fe 72 , and [Sm/Yb] N with increasing lithosphere thickness toward the northwest, which is consistent with increasing pressure (depth) of melt extraction, P f (Fig. 1D). P f corresponds to the thickened lithosphere and is physically the LAB depth (Niu and Green, 2018). Also, toward the northwest, the basalts show an increase in [La/Sm] N , Rb/Hf, Hf/Lu, Ba/Zr, Zr/Yb, Ti 72 , P 72 , K 72 , and other incompatible elements (Figs. 2 and 3; Fig. DR2), which is consistent with the decreasing extent of melting. This is also consistent with the northwestwardthickening lithosphere with a deepening LAB which caps decompression melting and results in melt extraction at greater depths (Fig. 1D). Figure 4 shows the significant correlations of basalt compositions with surface elevation, which further illustrates the lithosphere thickness control (see above). All of these coherent and systematic changes in basalt compositions with lithosphere thickness variation, as well as surface elevation variation, demonstrate the in-fluence of the lid effect on basaltic magmatism in continental settings.
Note that lithosphere thickening due to conductive heat loss to the surface is a natural process (see Niu and Green, 2018). Thus, the lithosphere in the Chifeng-Xilin Hot area would gradually thicken in the Cenozoic, and the erupted magmas would have systematic geochemical variations over the eruption age due to the lid effect, as is shown in Figure DR9. However, a lithosphere with an initial thickness of ∼80 km would only thicken by ∼13 km due to conductive cooling over the time span of ∼20 m.y.; or, to thicken from ∼80 km to ∼120 km would require ∼65 m.y. (see Niu and Green, 2018). Therefore, the ∼40 km difference in lithosphere thickness on both sides of the GGL (Fig. 1C) is not caused by conductive cooling. In fact, there is no evidence that the geotherm beneath the region has changed significantly over the past ∼22 m.y. (Huang and Xu, 2010). In brief, the systematic compositional variation of the Chifeng-Xilin Hot basalts over time (Fig. DR9) is not caused by lithosphere thickening, but is the the result of lithosphere thickness variation along the southeast-to-northwest traverse across the prominent GGL (Figs. 1C and 1D).

CONCLUSIONS
Studies show that global OIBs vary significantly in their compositions, but the lithosphere thickness (i.e., the depth of the LAB) at the time of OIB eruption exerts the primary control on OIB compositions in terms of the extent and pressure of melting (Humphreys and Niu, 2009;Niu et al., 2011;Niu, 2016;Niu and Green, 2018). This is the "lid effect"; i.e., the lithosphere lid caps the upwelling and decompression melting of mantle, resulting in melts erupted on thin lithosphere having geochemical signatures of a high extent and low pressure of melting, whereas melts erupted on thick lithosphere show the inverse. If the lid effect is globally significant, it should also be true for continental basalts. Our results demonstrate that the lid effect also operates in continental settings. The continental lithosphere thickness (i.e., the LAB depth) exerts the primary control on basaltic compositions in terms of the extent and pressure of melting, and thus is critically important for understating upper mantle dynamics and continental geology. Our results also provide strong evidence for isostatic equilibrium in the crust-mantle system above a mantle depth deeper than the LAB.