Although peridotite xenoliths are common in alkaline basalts, it is still unclear whether the chemical compositions of their host rocks have been affected by these mantle fragments and, if so, what processes are involved in this alteration of the host basalts. Here, we document a kind of xenolith-rich alkaline basalts from the Nanjing basaltic field, eastern China. These basalts contain lower concentrations of CaO (4.1–7.8 wt %) and Sc (3.3–17.8 ppm) and have lower Ca/Al (0.3–0.6) and higher Na/Ti ratios (2.8–11.2) than other Cenozoic basalts in this area. These xenolith-rich basalts show good correlations between elemental ratios (e.g., Lu/Hf and Ca/Al) and εHf values, which are indicative of mixing of two distinct components during the genesis of the magmas that formed these basalts: a high-εHf end-member (with low Lu/Hf and Ca/Al ratios) and the primitive melt-related low-εHf end-member. In addition, peridotite xenoliths hosted in these basalts have distinct core-mantle textures, with the margins having higher modal olivine abundances (70%) than the xenolith cores (52%). Within the xenolith margins, some orthopyroxenes are enclosed in the olivines, and all clinopyroxenes are sponge textured. These sponge-textured clinopyroxenes have higher CaO and Sc concentrations, higher Ca/Al ratios, and lower Na/Ti ratios than clinopyroxenes within the cores of the xenoliths, suggesting that the xenoliths underwent low-pressure melting within the host magma. This indicates that xenolith-rich magmas within the study area were contaminated during ascent by melts derived from mantle xenoliths within the magmas, transforming the magmas into the low-Ca alkaline basalts.
 Widespread intraplate basalts are generally geochemically diverse, with the origin of this diversity remaining a matter of considerable debate. In addition to fractional crystallization and differing degrees of partial melting, mantle source heterogeneity is an important control on the geochemical compositions of such basalts, which is largely due to the recycling of crustal materials [Hofmann, 1997, 2003; Prytulak and Elliott, 2007; Willbold and Stracke, 2010]. It is also difficult to exclude the effects of the lithosphere that continental basaltic magma passed through during ascent from the mantle, with continental basalt compositions potentially modified by crustal [Carlson et al., 1981; Koszowska et al., 2007] and/or lithospheric mantle [Niu, 2008; O'Reilly and Zhang, 1995; Tang et al., 2006; Xu et al., 2005] contamination. Analysis of xenolith-rich alkaline continental basalts (e.g., alkali basalts, basanites, and nephelinites) is an ideal way to assess the composition of the subcontinental asthenospheric mantle in an area [Basu et al., 1991; Zou et al., 2000], primarily because these alkaline magmas ascend rapidly and as such have little chance to assimilate material prior to eruption or emplacement within the crust. However, evidence from peridotite xenoliths [Kushiro, 1975; Shaw, 1999; Shaw et al., 2006] suggests that mantle xenolith compositions (e.g., clinopyroxene CaO, Na2O, and Al2O3 concentrations) may be modified by melting and/or reaction during ascent. If xenolith-derived melts are released into the host magma, this may cause changes in the geochemical composition of the host basalt, a process that has been neglected during previous research. Consequently, the potential influence of peridotite xenoliths on the geochemistry of alkaline basalts needs to be evaluated, especially as the geochemistry of these basalts is commonly used as a proxy for the composition of the mantle and the associated mantle processes. Here, we provide a case study of interaction between mantle xenoliths and magmas during the formation of Cenozoic alkaline basalts in the Nanjing basaltic field, eastern China.
 Cenozoic alkaline basalts in eastern China are typical intracontinental alkaline basalts and provide an ideal opportunity to study the genesis of continental basalts [Basu et al., 1991; Chen et al., 2009; Liu et al., 1994, 2008; Peng et al., 1986; Song et al., 1990; Tang et al., 2006; Wang et al., 2011; Xu et al., 2005; Zeng et al., 2010, 2011; Zhang et al., 2009; Zhi et al., 1990; Zhou and Armstrong, 1982; Zou et al., 2000]. The majority of these basalts are thought to represent melts derived from the asthenospheric mantle, as evidenced by relatively depleted Sr-Nd isotopes and ocean island basalt (OIB)-like trace-element signatures in primitive-mantle-normalized diagrams (e.g., Nb, Ta, and light rare earth element (LREE) enrichments, with negative K and Pb anomalies). Here, we present major and trace element, and Sr-Nd-Hf isotope data for alkaline basalts of the Miocene Panshishan and Tashan volcanoes in the Nanjing basaltic field, as well as petrological observations on peridotite xenoliths hosted by basalts from Tashan volcano. The data presented here suggest that the compositions of alkaline basalts from these two volcanoes were modified during ascent by low-Ca melts released from peridotite xenoliths.
2 Geological Setting
 This research focuses on the central part of eastern China, east of the Tan-Lu Fault (Figure 1). Geographically, this area is located in eastern Anhui and northern Jiangsu provinces, close to the city of Nanjing; given this proximity, for convenience we hereafter refer to this area as the Nanjing basaltic field. Cenozoic basalts in this area were predominantly emplaced and erupted between the Paleocene and the Miocene [Liu et al., 1992] and can be divided into two groups based on the scale of magmatism. Group 1 basalts are widespread and are associated with large volcanoes that are densely distributed in a narrow area near the Tan-Lu Fault (Figure 1b); these basalts are generally alkali olivine basalts and tholeiites. Group 2 basalts are associated with small isolated volcanoes that are scattered in areas distal from the Tan-Lu Fault and outcrop closer to Nanjing (Figure 1b); these basalts are generally alkali olivine basalts with minor basanites/tephrites. In addition, basanite-dominated Nushan volcano, the only Quaternary volcano within the Nanjing basaltic field, is located close to the northwestern edge of the study region (Figure 1b).
3 Sample Description
 This study of the genetic relationship between peridotite xenoliths and host basaltic magmas focuses on Panshishan and Tashan volcanoes in the Liuhe district of Nanjing, primarily as the basalts that form these volcanoes contain abundant peridotite xenoliths; some outcrops contain more xenoliths (50–80 vol %) than lava (Figure 2a). All samples from these volcanoes analyzed during this study are dark to dark gray in color and are massive. The basalts are fresh and contain minor olivine phenocrysts (<10% modal abundance) set in a groundmass of olivine, Ti-magnetite, plagioclase, and volcanic glass; no samples contain pyroxene or plagioclase phenocrysts. Olivine xenocrysts are common within all basalt samples barring sample 08PSS01, the sample with lowest concentrations of CaO. Crustal xenolith fragments and xenocrysts of crustal material are rare to absent in these samples. We also sampled alkali olivine basalts with minor or rare mantle xenoliths from the Matoushan, Fangshan, and Guabushan volcanoes to compare with the xenolith-rich basalts of the Panshishan and Tashan volcanoes (Figure 1b).
3.2 Peridotite Xenoliths
 Lavas of the Panshishan and Tashan volcanoes contain many peridotite xenoliths (5–80 cm diameter) (Figure 2a). These xenoliths have spinel lherzolite compositions, contain olivine (ol), orthopyroxene (opx), clinopyroxene (cpx), and spinel (sp) and usually have distinct core-mantle textures (Figures 2b and 2c). This is exemplified by xenolith sample TS-X03, which has a core with a modal composition of 52% ol, 30% opx, 17% cpx, and 1% sp and a margin with 70% ol, 14% opx, 15% cpx, and 1% sp, as determined by point counting in a thin section (Figure 2d). Differences between the core and the margin of TS-X03 were identified using the major element compositions of minerals within this xenolith (Table 1 in the supporting information).1
 Clinopyroxenes within xenolith cores have distinct core-mantle texture (Figures 3a and 3c) and contain both nonspongy core and sponge-textured margin. In contrast, clinopyroxenes within xenolith margins show sponge texture throughout the entire crystal (Figures 3b and 3d). Ten clinopyroxenes were analyzed to determine the geochemical differences between clinopyroxenes with and without sponge texture; three of the clinopyroxenes have sponge texture throughout the entire crystal. The sponge-textured clinopyroxenes have higher CaO (22.4–23.6 wt %) and lower Al2O3 (3.6–5.2 wt %) and Na2O (0.5–0.7 wt %) concentrations, higher Ca/Al ratios (5.9–8.9), and lower Na/Ti ratios (0.8–1.6) than nonspongy clinopyroxenes (CaO = 20.3–20.5 wt %, Al2O3 = 6.8–7.2 wt %, Na2O = 1.7–1.9 wt %, Ca/Al = 3.9–4.0, Na/Ti = 3.5–4.2), although both have similar Mg# values (89.5–92.0 and 89.7–90.9 for sponge-textured and nonspongy clinopyroxenes, respectively).
 Olivines and orthopyroxenes within xenoliths are compositionally homogenous with moderate Fo contents (89.3–90.0) and Mg# values (89.7–91.0), respectively. Within the margins, some orthopyroxenes are enclosed in the olivines (Figure 3b). Spinels within sample TS-X03 contain low concentrations of Cr2O3 (9.7–10.8 wt %) and total FeO (11.0–11.5 wt %), high concentrations of MgO (19.6–20.8 wt %) and Al2O3 (56.8–58.7 wt %), and have high Mg# (75.7–77.3) and low Cr# (10.0–11.2) values. Spinels throughout the xenoliths analyzed during this study are geochemically homogenous, and all are nonspongy in shape.
4 Geochemical Results
 The major and trace element and Sr-Nd-Hf isotope compositions of basalts from Panshishan and Tashan volcanoes are given in Table 1, and the analytical methods used are summarized in Appendix A.
Table 1. Major (wt %), Trace Element (10−6), Sr-Nd-Hf Isotopic Compositions of Panshishan and Tashan Basalts in Nanjing Volcanic Field
Fe2O3T, total iron as Fe2O3; mg#, molar Mg/(Mg + Fe).
Ti/Ti* = TiN/(NdN−0.055 × SmN0.333 × GdN0.722).
εNd = [(143Nd/144Nd)sample/(143Nd/144Nd)CHUR-1] × 104; where (143Nd/144Nd)CHUR = 0.512638.
εHf = [(176Hf/177Hf)sample/(176Hf/177Hf)CHUR-1] × 104; where (176Hf/177Hf)CHUR = 0.282772.
Na2O + K2O
Na2O + K2O
 The Panshishan and Tashan basalts contain medium to high SiO2 (46.5–52.3 wt %) concentrations, high total alkali (Na2O + K2O = 6.0–11.6 wt %) and Al2O3 (14.1–17.4 wt %) contents, low to medium concentrations of MgO (1.8–9.5 wt %), low TiO2 (0.8–1.9 wt %) and CaO (4.1–7.8 wt %) concentrations, and low Ca/Al (0.3–0.6) and high Na/Ti (2.8–11.2) ratios. These rocks are therefore classified as basanites to phonotephrites using the nomenclature of Le Bas et al. , with only one (sample 08PSS01) classified as a tephriphonolite (Figure 4). These basalts have obvious correlations between SiO2 and TiO2 (not shown), SiO2 and Fe2O3T (not shown), MgO and CaO (Figure 5a), MgO and Fe2O3T (not shown), MgO and K2O (not shown), and MgO and Al2O3 (not shown); these trends are not present in geochemical data for other alkaline basalts from this area (our unpublished data). No correlations are observed between MgO and other major elements. The Panshishan and Tashan basalts have higher total alkali contents and lower CaO and TiO2 contents (Figures 4 and 5a, TiO2 not shown) at a given MgO content than other alkaline basalts within the Nanjing basaltic field. For convenience, we refer to basalts from the Panshishan and Tashan volcanoes and other alkaline basalts from the Nanjing basaltic field as low-Ca and high-Ca alkaline basalts, respectively.
 Compatible element (Ni, Cr, and Sc) concentrations within low-Ca alkaline basalts positively correlate with MgO (Figure 5b, Ni and Cr versus MgO not shown), whereas incompatible elements (Rb and Hf) negatively correlate with MgO (Figure 5c, Rb versus MgO not shown). In comparison, barring Ni and Cr versus MgO, the high-Ca alkaline basalts do not have similar relationships between these elements and MgO. These high-Ca basalts have smooth chondrite-normalized rare earth element (REE) patterns without Eu or Ce anomalies and are light rare earth element (LREE)-enriched (La/Yb = 34.0–98.5; not shown). In addition, the low-Ca alkaline basalts have higher LREE/heavy rare earth element (HREE) and middle rare earth element (MREE)/HREE ratios than the high-Ca alkaline basalts. In a primitive-mantle-normalized incompatible element diagram (Figure 6), the low-Ca alkaline basalts resemble OIBs (Figure 6), with Nb and Ta enrichments, and Pb depletions relative to the LREE. The low-Ca and high-Ca alkaline basalts have similar patterns to each other, although the former have patterns with significant Rb-Ba fractionation (Rb/Ba = 0.09–0.18) and more significantly negative Ti anomalies (Ti/Ti* = 0.15–0.63).
 The low-Ca alkaline basalts have limited ranges in Sr, Nd, and Hf isotopic compositions (87Sr/86Sr = 0.70320–0.70329, εNd = 6.8–7.4, εHf = 9.5–13.0); these compositions are the most depleted of all Cenozoic alkaline basalts within the Nanjing basaltic field (Figure 7) and are more depleted than the Quaternary Nushan basalts, which are the most depleted basalts among the high-Ca alkaline basalts (Figure 7). However, the low-Ca alkaline basalts are still less depleted than the depleted mid-ocean ridge basalt (MORB) mantle (DMM; 87Sr/86Sr = 0.70263, εNd = 9.6, εHf = 17.3) [Workman and Hart, 2005]. In addition, small-scale but systematic changes in Hf isotopic compositions associated with changes in elemental ratios are also present within the Panshishan basalts. The most isotopically depleted end-member basalt also has lower Ca/Al, Ti/Ti*, and Lu/Hf ratios (Figure 8).
 The low MgO concentrations (1.8–9.5 wt %) and positive correlations between each of CaO, Ni, Cr, and Sc with MgO (Figure 5b, Ni and Cr not shown) of these low-Ca alkaline basalts may initially be thought to relate to processes such as fractional crystallization (FC) or assimilation and fractional crystallization (AFC) that occurred during the formation of these basalts. However, the presence of abundant mantle xenoliths (5–80 cm diameter) in these basalts (Figure 2a) suggests that these alkaline magmas ascended rapidly, without time to undergo FC or AFC. Moreover, these low-Ca rocks are the most isotopically depleted basalts within the Nanjing volcanic field (Figure 7), suggesting that these basalts are not crustally contaminated. Instead, we propose a two-component mixing model, primarily as the low-Ca alkaline basalts have systematic correlations between Hf isotope compositions and elemental ratios, including Ca/Al, Lu/Hf, and Ti/Ti* (Figure 8). The high-Ca end-member has higher Ca/Al, Ti/Ti*, and Lu/Hf ratios than the low-Ca end-member but similar ratios to high-Ca alkaline basalts in the Nanjing basaltic field, suggesting that the high-Ca end-member of the low-Ca alkaline basalts is related to these high-Ca alkaline basalts. These high-Ca alkaline basalts have OIB-like features (e.g., positive Nb and Ta anomalies and large ion lithophile element (LILE) enrichments; Figure 6b), similar to the majority of asthenospheric-mantle-derived Cenozoic alkaline basalts in eastern China [Basu et al., 1991; Liu et al., 1994; Peng et al., 1986; Song et al., 1990; Tang et al., 2006; Xu et al., 2005; Zeng et al., 2010; Zhi et al., 1990; Zhou and Armstrong, 1982; Zou et al., 2000]. The broad and relatively enriched Sr-Nd isotopic compositions (Figure 7a) and low Ce/Pb ratios (13.7–21.1) of these basalts also suggest that the magmas that formed these basalts were derived from a region of the mantle either that contained a crustal component or that assimilated crustal material during ascent.
 The low-Ca end-member of the low-Ca alkaline basalts contains low concentrations of CaO, MgO (Figure 5a), and compatible elements (e.g., Sc; Figure 5b); has low Ca/Al, Ti/Ti*, and Lu/Hf ratios (Figure 8); contains high concentrations of incompatible elements (e.g., Hf; Figure 5c); and has high εHf values (Figure 8). The following sections focus on the potential origins of this low-Ca component.
5.1 Melts from Carbonated Peridotite?
 Previous research on strongly alkaline Cenozoic basalts elsewhere in eastern China suggested that melting of carbonated peridotite can produce melts characterized by incompatible element enrichments, negative K and Ti anomalies, and superchondritic Zr/Hf ratios [Zeng et al., 2010]. Similar characteristics are also present in the low-Ca alkaline basalts analyzed here (Zr/Hf = 49.8–52.9; Figure 6a), suggesting a similar origin. However, alkaline basalts derived from a region of carbonated mantle have elevated Ca/Al ratios [Zeng et al., 2010], whereas the low-Ca component considered here has very low Ca/Al ratios (Figure 9a), meaning that a model whereby magmas were sourced from melting of carbonated peridotite cannot explain the genesis of the low-Ca component within these low-Ca alkaline basalts.
5.2 Melts from Recycled Crust?
 Melting of eclogite or pyroxenite within the mantle can produce magmas with low Ca/Al ratios if clinopyroxene is left as a primarily residual phase [Herzberg and Asimow, 2008], a process that has been observed during the experimental melting of eclogite [Spandler et al., 2008]. Eclogite and/ or garnet pyroxenite within the mantle forms either directly from recycled crustal material or is produced by reaction between melts derived from crustal material and SiO2-undersaturated peridotite nearby [Rapp et al., 1999; Sobolev et al., 2005, 2007; Yaxley, 2000]. If eclogite or pyroxenite within the mantle is directly derived from recycled continental materials, melting of this material would generate basaltic magmas with a more pronounced garnet signature than entirely peridotite-derived melts. A simple calculated batch-melting curve for carbonated garnet peridotite compositions in the plots of La/Yb versus Sm/Yb indicates that high-Ca alkaline basalts of the Nanjing basaltic field, in addition to alkaline basalts from Shandong and Nushan, plot on the melting curve but at different degrees of melting and with the Anfengshan basalts having higher Sm/Yb ratios than other basalts at a given La/Yb ratio (Figure 9b). This “garnet signature” suggests that the magmas that formed these basalts were derived from a region of the mantle that contained eclogite [Chen et al., 2009; Zeng et al., 2010]. Low-Ca alkaline basalts plot below this melting curve, indicating an eclogite-free source for these basalts. In addition, the low Ni and Cr concentrations within the low-Ca alkaline basalts suggest that these magmas were probably not formed during partial melting of a reaction pyroxenite [Sobolev et al., 2007]. Therefore, it seems clear that the low-Ca component within these basalts was not derived from melting of recycled crustal material.
5.3 Products of Asthenosphere-Lithosphere Interaction?
 Previous geochemical studies on Cenozoic basalts from the Datong [Xu et al., 2005] and Taihang Mountain [Tang et al., 2006] areas suggest that asthenosphere-derived melts may react with the lithospheric mantle at the boundary between these two regions of the mantle, a process that may modify the composition of basaltic melts, leading to increases in SiO2 and Cr concentrations, decreases in Al2O3 and CaO concentrations, and the development of a kink in chondrite-normalized REE patterns at Gd by the replacement of orthopyroxene and the precipitation of olivine [Xu et al., 2005]. However, this asthenosphere-lithosphere interaction does not change the Sc concentration of basaltic magmas. In addition, the smooth REE patterns of the low-Ca alkaline basalts discussed here do not support this model. This indicates that asthenosphere-lithosphere interaction cannot explain the chemical characteristics of the low-Ca alkaline basalts.
5.4 Products of Peridotite Xenolith-Host Magma Interaction?
 The presence of core-mantle textures within peridotite xenoliths hosted by low-Ca alkaline basalts is indicative of interaction between the xenoliths and host basaltic magmas (Figures 2b and 2c). This means that the mineralogical and geochemical differences between the core and margin of xenolith sample TS-X03 may provide important evidence of a genetic relationship between peridotite xenoliths and the host basaltic magmas.
5.4.1 Reaction or Melting?
 The distinct core-mantle textures observed in peridotite xenoliths during this study may be indicative of a reaction between the xenoliths and the host magma (Figures 2b and 2c); however, orthopyroxenes and spinels within the xenoliths do not show reaction textures. In addition, olivines, orthopyroxenes, and spinels within xenolith cores are chemically near-identical to the same minerals within xenolith margins. These observations are inconsistent with the occurrence of a reaction between peridotite xenoliths and the host basaltic melts.
 Whereas some of the clinopyroxenes in the core of mantle xenolith sample TS-X03 have a distinct core-mantle texture (Figures 3a and 3c), all of the clinopyroxenes within the margin of this xenolith have a sponge texture (Figures 3b and 3d) and are generally smaller than the clinopyroxenes in the core. Several possible explanations can account for these textures, including the following: (1) mantle metasomatism prior to entrainment of xenoliths [Franz and Wirth, 1997; Ionov et al., 1995, 2005; Liang and Elthon, 1990]; (2) reaction between clinopyroxene and the host magma during transport [Brearley et al., 1984; Shaw and Klügel, 2002; Shaw et al., 2006]; (3) incongruent partial melting triggered by fluid penetration [Carpenter et al., 2002; Guzmics et al., 2008]; and (4) decompression-induced partial melting within the mantle [Carswell, 1975; Su et al., 2010]. We suggest that low-pressure melting, rather than metasomatism or reaction, caused the development of sponge texture within peridotite xenoliths hosted by low-Ca alkaline basalts in the study area, and this hypothesis is supported by the following points. Firstly, sponge-textured minerals are almost euhedral and have distinct boundaries with adjacent grains (Figure 3), which is inconsistent with the strong infiltration and destruction processes that occur during reactions between xenoliths and host magma. Secondly, metasomatized melts contain elevated concentrations of highly mobile incompatible elements (e.g., K, Na) and are deficient in compatible elements (e.g., Mg, Cr), whereas element exchange generally leads to enrichment in Na and reduced Mg# of peridotites during peridotite-melt reactions [e.g., Frey and Prinz, 1978; Kepezhinskas et al., 1995, 1996; Su et al., 2010; Xu et al., 1998; Zheng et al., 2005]. The sponge-textured clinopyroxenes in sample TS-X03 have lower Na2O concentrations (0.5–0.7 wt %) than nonspongy clinopyroxenes, although both have similar Mg# values; these geochemical features do not support a metasomatic origin for these textures. Thirdly, sample TS-X03 is representative of fertile peridotite, as evidenced by the high modal abundance of clinopyroxene (17%) and orthopyroxene (30%) in the core of this xenolith, and therefore is likely to have undergone low-pressure melting during ascent.
5.4.2 Low-Pressure Melting of Peridotite Xenoliths
 The evidence presented above indicates that low-pressure partial melting is the most likely cause of the sponge-textured clinopyroxenes in the mantle xenoliths. This means that the spongy-textured clinopyroxenes are a melting residue, and the chemical differences between the nonspongy clinopyroxenes and the spongy-textured clinopyroxenes can provide indirect evidence of the geochemistry of the melts released during low-pressure partial melting of the xenoliths. Sponge-textured clinopyroxenes generally contain higher concentrations of CaO (22.4–23.6 wt %), lower concentrations of Al2O3 (3.6–5.2 wt %) and Na2O (0.5–0.7 wt %), and have higher Ca/Al ratios (5.9–8.9) and lower Na/Ti ratios (0.8–1.6) than nonspongy clinopyroxenes, indicating low Ca/Al ratios and high Na/Ti ratios for released low-degree melts from clinopyroxenes. Mantle xenoliths are fragments of the lithospheric mantle that underwent melting before becoming a part of the lithosphere; in addition, the fact that clinopyroxene is the most fusible phase within peridotite means that it is more susceptible than the rest of the peridotite assemblage during such melting event(s). If we take clinopyroxenes from the Nushan peridotite xenoliths as an example, the concentrations of Na2O and Na/Ti ratios increase with increasing degree of partial melting (Figure 10), as evidenced by the marked increase in Na and decrease in Ti compatibility in clinopyroxene with increasing pressure [Blundy et al., 1995; Kinzler, 1997]. The sponge-textured clinopyroxenes within these xenoliths have much lower Na2O concentrations and Na/Ti ratios than clinopyroxenes within Nushan peridotite, indicating that the former most likely were formed during melting under very low-pressure conditions.
 In addition, the margin of xenolith sample TS-X03 contains more olivine (70% modal abundance) and less orthopyroxene (14%) than the core of the xenolith (ol: 52%; opx: 30%). In its margin, some orthopyroxenes are enclosed in the olivines (Figure 3b), a relationship that is the opposite of a peritectic olivine texture. During melting, orthopyroxene may break down either congruently to a melt of orthopyroxene composition or incongruently to olivine plus a SiO2-enriched melt. Incongruent melting is limited to pressures below 0.5 GPa in a simple Fo-SiO2 system and results in the formation of a small amount of olivine [Kushiro et al., 1968]. The melting of peridotite at higher pressures causes the production, rather than the consumption, of orthopyroxene [Kinzler, 1997]. Melting of orthopyroxene over a wide range of pressure conditions would also produce melts with different geochemical compositions, as melting experiments at 0.4 and 1 GPa formed SiO2-rich alkaline melts, whereas melting at 2 GPa produced melts that only had slightly elevated SiO2 concentrations and did not contain high alkali concentrations [Shaw, 1999; Shaw et al., 1998]. The crystallization of a large amount of olivine is most likely to be related to a melt with a high concentration of alkalis, primarily as alkalis increase the phase volume of olivine [Kushiro, 1975], and the mixing of this melt can explain the high alkali concentrations present within the low-Ca alkaline basalts analyzed during this study.
 The breakdown of orthopyroxene discussed above means that low-pressure melting residues of peridotite have higher clinopyroxene/orthopyroxene ratios than residues left after high-pressure melting. In other words, melting of peridotite under low-pressure conditions is characterized by residues with excess clinopyroxene, whereas melting under high-pressure conditions is characterized by residues with excess orthopyroxene. This indicates that low-pressure melts should retain more clinopyroxene-related characteristics, such as lower Ca/Al ratios, than high-pressure melts (i.e., primitive basaltic magmas). Peridotite partial melting experiments also suggest that melts with low Ca/Al and high Na/Ti ratio are produced under low-pressure, rather than high-pressure, conditions [Hirose and Kushiro, 1993] (Figure 11), consistent with the model described above. Negative correlations between Hf isotope compositions and Ca/Al ratios (Figure 8) also strongly support a model involving mixing between xenolith-derived melts and primitive basaltic magmas. The low-Ca alkaline basalts with lower Ca/Al ratios are more isotopically depleted, indicating greater contributions of xenolith-derived melts in these basalts.
 However, it is unclear why the xenoliths are zoned if the low-Ca component was derived from low-pressure melting of these peridotite xenoliths. This zoning can be explained by heat diffusing from the host magma, which is another important factor during peridotite melting. Xenoliths with diameters of 5–80 cm can take a few hours or more to thermally equilibrate with the host magma, depending on the temperature difference between the xenoliths and the magma during entrainment. There is less melting in the core as it takes time for the heat to diffuse through the xenolith.
5.5 Genesis of Low-Ca Alkaline Basalts
 As discussed above, interactions between peridotite xenoliths and basaltic magmas modified the compositions of both components. We consider that the Nushan basalts represent the primary melt compositions of low-Ca alkaline basalts from the Panshishan and Tashan volcanoes, primarily as the Nushan basalts also have similar Sr-Nd isotopic compositions (Figure 7a) but have negligible evidence of interaction between xenoliths and basaltic magma. Interaction between xenolith-derived melts and the primary basaltic melt caused an increase in SiO2 (Figure 4) and incompatible element concentrations (e.g., Hf; Figure 5c), Na/Ti ratios (Figure 11), and εHf values in addition to decreases in CaO and compatible element concentrations (e.g., Sc; Figure 5b) and Lu/Hf and Ca/Al ratios (Figures 8 and 11). We also modeled the melting of peridotite xenoliths to assess whether the low Ca/Al and high Na/Ti ratios of low-Ca alkaline basalts in the Nanjing basalt field were inherited from peridotite-xenolith-derived melts. This modeling indicates that a low degree of batch melting of peridotite xenolith material can release melts with high Na/Ti and low Ca/Al ratios (Figure 11). Mixing between primary basaltic melts and low-degree, xenolith-derived melts can reproduce the chemical characteristics of the low-Ca alkaline basalts discussed above (Figure 9) and provides a viable mechanism for the genesis of these basalts; we estimate that the proportion by weight of xenoliths contaminated in these magmas may be more than 10%.
 Globally, intraplate alkaline basalts are geochemically variable, and previous studies that attempted to explain the genesis of these basalts focused on changes in mantle sources. These studies proposed a number of genetic models for the basalts, including derivation from the asthenospheric mantle [e.g., Chen et al., 2009; Han et al., 1999] and from the lithospheric mantle [e.g., Chung et al., 1995; Humphreys and Niu, 2009; Pilet et al., 2008] and formation during lithosphere-asthenosphere interaction at the boundary between these two regions of the mantle [e.g., Gorring et al., 2003; Perry et al., 1987; Tang et al., 2006]. However, our understanding of the genesis of these alkaline basalts and the mantle source of the magmas that formed these basalts also needs to consider potential interactions between mantle xenoliths and host basaltic magmas during ascent. Previous studies focused on the influence of peridotite xenoliths [Klügel, 1998] but ignored the effects of the xenoliths on the composition of the host basaltic magmas. Cenozoic basalts in eastern China have highly variable compositions (Figure 11), with most having low Na/Ti and high-Ca/Al ratios that indicate negligible influence by mantle xenoliths. However, the low-Ca alkaline basalts in the Nanjing basaltic field have clearly been modified by mixing with melts generated during low-pressure melting of peridotite xenoliths, leading to basalts that have negative correlations between Na/Ti and Ca/Al ratios. The high Na/Ti (>3.0) and low Ca/Al ratios (<0.7) that are present in the majority of basalts analyzed during this study suggest that these ratios may be a useful tool for assessing the effect of interaction between peridotite xenoliths and host basaltic magmas.
 Cenozoic low-Ca alkaline basalts in the Nanjing basalt field show systematic correlations between Hf isotope compositions and incompatible element ratios, indicating mixing between two distinct end-member components. The high-Ca end-member has chemical characteristics similar to high-Ca alkaline basalts of the Nanjing basaltic field that are derived from the asthenospheric mantle. In comparison, the low-Ca end-member has low Ca/Al, Ti/Ti*, and Lu/Hf ratios, low concentrations of compatible elements (e.g., Sc and Cr), and high Na/Ti ratios, suggesting derivation from melting of peridotite xenoliths. The mineralogical and geochemical differences between the cores and margins of peridotite xenoliths hosted by low-Ca alkaline basalts are strongly suggestive of low-pressure melting of the xenoliths during ascent. The subsequent mixing of the melts released by this process modified the host basalts and produced the low-Ca alkaline basalts of the Panshishan and Tashan volcanoes.
 We are grateful to Ye Liu, Yue-Heng Yang, Wen-Lan Zhang, and Wei Pu for their technical support. Greg A. Valentine, Xun Yu, Xia-Yu Chen, and Lei Liu attended the field investigations of this study. We appreciate the thoughtful and constructive reviews provided by Editor Joel Baker and an anonymous reviewer. This study was supported by the NSFC (Grant 41202039), China Geological Survey (Grant 1212011220038), and the Fundamental Research Funds for the Central Universities (Grant 1127020608).
 Measurements of whole-rock major elements and trace elements were made at the Department of Geology, Northwest University, China. We used a RIX-2100 X-ray fluorescence spectrometer (XRF) to measure major elements. According to the measured values of standards (GSR-1 and GSR-3), the uncertainties are about ±1% for elements with concentrations >1.0 wt% and about ±10% for elements with concentrations <1.0 wt %. Trace elements, including rare earth elements (REEs), were determined using an ELAN 6100DRC inductively coupled plasma mass spectrometer (ICP-MS) after acid digestion (HF + HNO3) of samples in Teflon bombs. Analyses of USGS rock standards (BHVO-2, AGV-1, BCR-2, and G-2) indicate precision and accuracy better than 5% for Sc, V, Cr, Co, Ni, Rb, Sr, Y, Zr, Nb, Cs, Ba, Pb, U, and REEs and 10% for Hf, Ta, and Th. The results of the analyses of these standards can be seen in Table S2.
 Sr-Nd isotopic analyses were performed using a Finnigan Triton TI thermal ionization mass spectrometer at the State Key Laboratory for Mineral Deposits Research, Nanjing University. Detailed analytical procedures for the Sr and Nd isotopic measurements are given by [Pu et al., 2005]. Sr and Nd isotopic compositions were normalized to 86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219. Measured values for the NBS987 Sr standard and JNdi-1 Nd standard were 0.710266 ± 0.000005 for 87Sr/86Sr and 0.512121 ± 0.000003 for 143Nd/144Nd, respectively.
 Hf isotopic data were obtained using a Neptune multicollector mass spectrometer at the Institute of Geology and Geophysics, Chinese Academy of Sciences. A detailed analytical procedure for the Hf isotopic measurement is given by [Yang et al., 2010]. Hf isotopic compositions were normalized to 179Hf/177Hf = 0.7325. The measured value for the GSR-3 and BHVO-2 Hf standard was 0.282971 ± 0.000005 and 0.283106 ± 0.000006 for 176Hf/177Hf, respectively.
 Quantitative analyses (major elements) and back-scattered electron (BSE) images for minerals were carried out at the State Key Laboratory for Mineral Deposits Research, Nanjing University, using a JEOL JXA-8100 M electron-microprobe (EMP). The operating conditions were as follows: accelerating voltage of 15 kV and a probe current of 2 × 10−8 A. The diameter of the electron beam was 1 µm. The counting times at the peaks are 20 s for major elements. All data were corrected with standard ZAF correction procedures. Natural minerals and synthetic glasses were used as standards
All supporting information may be found in the online version of this article.