Remagnetization of Permian Emeishan basalts: Constraints on the timing of native copper mineralization in northeast Yunnan Province, China

New paleomagnetic results from the Permian Emeishan basalts in the Zhaotong area, NE Yunnan province, China show four natural remanent magnetization components. Detailed stepwise thermal demagnetization of basaltic samples from 16 flows from the Dadi section, which represent basalt units III and IV, isolated two groups of characteristic remanent magnetizations. Samples in unit IV (five flows) record a southwest declination and a moderate downward inclination that is considered to be a partial remagnetized remanence. The bottom flows from unit III (11 flows) record a normal polarity direction, interpreted as a remagnetization, which yields a tilt-corrected mean direction of Ds/Is = 8.8°/31.6° (N = 9, ks = 39.7, α 95 = 8.3°), with a corresponding paleomagnetic pole at 77.1°N, 240.0°E (K = 49.2, A95 = 7.4°). The secondary directions have steeper inclinations than primary ones that have been successfully recovered from other studies in this area of the Emeishan basalts. By comparison with the Phanerozoic paleomagnetic poles of the South China Block, the preferred timing of remagnetization is the Lower-Middle Jurassic. Field relationships suggest that the remagnetization of the Emeishan basalts is coeval with the spatially related, but localized, copper mineralization. Thus the timing of the main copper mineralization hosted in the Emeishan basalts is hypothesized to occur in the Early-Middle Jurassic.


INTRODUCTION
The Emeishan flood basalts form a large igneous province (LIP) that covers an area of around 2.5 × 10 5 km 2 in southwestern China, and is considered to have originated from a mantle plume (Xu et al., 2001;He et al., 2003;He et al., 2007). The timing of the Emeishan volcanism has been dated to the Middle to Late Permian (e.g., 259-260 Ma), and is temporally related to the late Guadalupian mass extinction (e.g., Courtillot et al., 1999;Ali et al., 2002;Ali et al., 2005, Ali et al., 2010Hou et al., 2002;Shellnutt and Jahn, 2011;Shellnutt et al., 2012;Luo et al., 2013;Yan et al., 2013;Li et al., 2018). A series of paleomagnetic studies on the Emeishan basalts have been conducted for more than three decades (McElhinny et al., 1981;Lin et al., 1985;Zhao and Coe, 1987; Van der Voo et al., 1993;Huang and Opdyke, 1998;Zhu et al., 1998;Ali et al., 2002;Liu and Zhu, 2009;Zheng et al., 2010;Liu et al., 2012). It was noted in early studies that both north-northeast shallow-up (normal) and southwest down (reverse) directions are primary remanent magnetizations (Huang and Opdyke, 1998;Zheng et al., 2010;Liu et al., 2012). However, some researchers have observed that the normal and reverse polarity directions are not antipodal and suggested that some directions are contaminated by unresolved overprinting, with reverse directions more affected than the normal ones (Huang and Opdyke, 1998;Liu et al., 2012). The potential causes of remagnetization, however, have not been established.
There are many traces of copper mineralization and smallvolume copper deposits spatially associated with the Emeishan basalts (Zhu et al., 2003;Wang et al., 2006;Li et al., 2009). The Cu concentration of the Emeishan basalts is high (i.e., an average of ∼170 ppm) and the copper mineralization grades range from about 1.0 to 4.0 wt% Cu (Zhu et al., 2003;Wang et al., 2010). Since 1958, tens of copper mines have been established in Sichuan, Yunnan and Guizhou provinces ( Figure 1A). However, none is of significant economic value. A number of studies have suggested that the copper mineralization of the Emeishan basalts was the result of mid-to low-temperature (i.e., <350°C) hydrothermal activity (Wang et al., 2006;Li et al., 2009;Wang et al., 2011). However, the age of copper mineralization remains equivocal. Wang et al. (2006) suggested that either Triassic magmatism or hydrothermal activity at ∼230 Ma is associated with the copper deposits, whereas, Li et al. (2009) suggested that the main copper mineralization phase was during the Middle to Late Jurassic. Ar 39 -Ar 40 whole-rock dating on basalt sample from the Emeishan basalts yield a range of Mesozoic and Cenozoic overprint cooling ages (Hou et al., 2002;Ali et al., 2004). However, the relationship between copper mineralization, the Ar 39 -Ar 40 overprint ages and paleomagnetic doubts from the Emeishan basalts is unclear.
This paper presents new paleomagnetic results from the Emeishan basalts from the Dadi section in the Zhaotong area, northeast Yunnan province. This section is associated with native copper deposits and data from this region allows us to obtain the timing of copper mineralization.

GEOLOGIC SETTING AND SAMPLING
The Emeishan basalts are exposed across large parts of Yunnan, Sichuan and Guizhou provinces in southwest China ( Figure 1A). The main body of the Emeishan LIP (ELIP) is bounded by the Longmenshan-Xiaoqinghe faults to the north-northwest and the Ainaoshan-Honghe faults to the southwest (He et al., 2003;Liu and Zhu, 2009; Figure 1A). The basalt sequences range in thickness from 1.0 to 5.0 km in the west to several hundred meters in the east (Ali et al., 2005). A mantle plume head located in the western part of Binchuan-Miyi area, northwest of Kunming has been suggested to explain these thickness variations He et al., 2007). The Emeishan basalt sequences in northeast Yunnan province are usually over 1,000 m thick. In this area, the basalts unconformably overlie the Middle Permian Maokou Formation and are unconformably overlain by the Upper Permian Xuanwei Formation. The Maokou Formation is mainly comprised of fossiliferous marine limestone . The Xuanwei Formation is composed of continental deposits of sandstones, mudstones, thin coal seams with basal conglomerates in the northern parts (Yunnan Bureau, 1989). The Emeishan basalt sequence in this area is divided into four stratigraphic units, I-IV up section (Wang et al., 2006;Liu et al., 2012). After the eruption of the Emeishan basalt, the most significant magmatic activity in this area was in the Late Triassic (Yunnan Bureau, 1989;Li et al., 2009). Regional faults mostly with a northeast strike ( Figure 1B), are part of the Mesozoic shortening belt system within the Yangtze China Block that resulted from the Yanshanian Movement (Yang et al., 1986;Yan et al., 2003;Li et al., 2009;Dong et al., 2015). Copper mineralization is largely concentrated in unit IV of the Emeishan basalts, which are generally 50-100 m thick and composed of 3-5 distinct basalt flows. The copper ore bodies and mineralization bodies, tens of centimeters to 2 m thick, mainly are located in schistose amygdaloidal basalt, amygdaloidal basaltic breccias and carbonaceous sedimentary rocks between of different basalt flows. The footwall and hanging wall of the ore bodies are generally massive basalt (Li et al., 2009). The native copper deposits in the northeast Yunnan province are mainly distributed along the limbs of the Xiaozhai and Maolin synclines (Wang et al., 2006;Li et al., 2009; Figure 1B). The northwest limb of the Xiaozhai syncline dips at 30-48°S and the southeast limb dips at 20-37°N (Wang et al., 2006). All of the copper mines in northeast Yunnan province are of limited scale or only have small Cu mineralization. For instance, the estimated reserve of one occurrence on the southeast limb is 16,000 tons of Cu mineralization at an average grade of 1.5 wt% Cu with a strike length about 500 m and an average thickness of 10 m (Zhu et al., 2003).
The basalt section in the Zhaotong area, northeast Yunnan province, was sampled near the village of Dadi (DD; 27.22°N, 103.54°E; Figure 1B). The total sampled sequence at Dadi section is ∼140 m, containing a thick conglomerate bed (∼5 m) of the Upper Permian Xuanwei Formation and at least 21 flows, each ranging from 1 to 12 m ( Figure 1C). Unit III of the Emeishan basalts forms the base of this section and a tuff layer separates units III and IV of the basalts ( Figure 1C). The upper part of the section is overlain by conglomerate bed of the Xuanwei Formation ( Figure 1C). The contact between unit IV of the basalts and the conglomerate is a parallel, unconformable boundary. The paleomagnetic directions recorded in the basaltic cobbles from the Xuanwei Formation conglomerate pass a conglomerate test, suggesting that the magnetization in the underlying Emeishan basalts is primary (Liu et al., 2012). The upper parts of the unit IV basalts (sites DD2-6; ∼50 m; Figure 1C) have been studied by Liu et al. (2012) and record a southwest shallow downward (reverse) direction, which is consistent with the reverse polarity magnetization elsewhere in the ELIP. The focus in this study is on the underlying sixteen flows, including the lower eleven basalt flows (sites DD12-22, each 1-5 m thick), mid-basalt tuffs (∼8 m) and five upper basalt flows (sites DD7-11, each 5-10 m thick) sampled in this study. The ∼8 m purple-red tuff marks the boundary between basalt unit III and unit IV (Yunnan Bureau, 1989; Figure 1C). Sites DD7-11 are part of unit IV and were separated by a ∼3 m rivulet from sites DD2-6, that is labeled as a fault in Figure 1C. Sites DD12-22 belong to unit III. The basalt unit III in adjacent Shuimo (SM) section in Zhaotong area yields primary northeast and shallow up (normal) directions (Liu et al., 2012). The individual basalt flows grade from massive, to vesiculated, to brecciated from bottom to top. Four abandoned mining adits are visible, two are within the DD9 and DD10 flows (Unit IV) near the sample sites and two others were ∼1 km from the sampled section in the flows DD14 and DD17 (Unit III).
A total of 176 oriented samples were taken from the sixteen flows using a handheld petrol drill with water cooling. Both magnetic and sun compasses were used for in-situ core orientation, and less than 5°difference was found between the two orientations. Samples were further processed in the laboratory into specimens (generally 0.5 cm height cylinder to reduce the intensity with the diameter of 2.54 cm) suitable for paleomagnetic measurements. Fresh off-cuts were used for rock magnetic and microscope investigations.

METHODS
All specimens were subjected to progressive thermal demagnetization in 13-15 steps up to 550 or 585°C using an ASC TD-48 oven in a magnetically shielded room (residual field <300 nT). The remanence was measured using a 2G Enterprises 755R SQUID magnetometer. The magnetic component directions were determined using principal component analysis (Zijderveld, 1967;Kirschvink, 1980) as implemented by the PaleoMag software using linear fitting (Jones, 2002) or great circle fitting by using PMGSC software. All data fits were made using at least four points. To compare with the thermal demagnetization behavior, 24 sister samples were subjected to stepwise alternating field (AF) demagnetization with eleven steps up to 80 mT using a 2G-760R long-core SQUID magnetometer with an inline AF demagnetization system. In most cases, however, AF demagnetization does not fully resolve the paleomagnetic directions, therefore only the thermal demagnetization results are used for further analysis. Fisher statistics (Fisher, 1953) were used to compute the site mean directions and the equal area projections of the directions were analyzed and plotted with the PmagPy software (Tauxe, 2010). The same attitude of the stratum (strike/dip 357°/20°) for the upper sites DD2-6, which was measured from the overlying Xuanwei Formation sediments (Liu et al., 2012), is used for all tilt corrections.
To characterize the composition and domain state of the magnetic carriers, 27 hysteresis loops were measured on a Princeton Measurement MicroMag 3,900 Vibrating Sample Magnetometer (VSM3900) at room temperature with a field range of ±1 T. Isothermal remanent magnetization (IRM) acquisition up to 1 T and backfield demagnetization to −0.5 T of representative specimens were also conducted on the VSM3900 and the coercivity components of each specimen were analyzed (Kruiver et al., 2001). The hysteresis parameters are plotted on a Day-plot (Day et al., 1977;Dunlop, 2002). The temperature dependence of susceptibility was measured in air from room temperature to 700°C using an AGICO MFK1 Kapabridge equipped with a CS-3 high-temperature furnace in a field of 200 A/m at a frequency of 976 Hz. To investigate the microtexture of the basaltic samples directly, polished thin sections were studied by reflected light microscope. All paleomagnetic and magnetic measurements were performed in the Paleomagnetism and Geochronology Laboratory of the Institute of Geology and Geophysics, Chinese Academy of Sciences.

Rock Magnetism
The room temperature hysteresis loops of specimens from the flows DD7-22 are closed by ≤ 0.2 T and have low coercivity (H c ) values range from 3.8 to 11.7 mT with a mean of 6.4 ± 2.2 mT (n 27; Figure 2). The saturation magnetizations (M s ; acquired at 1 T) range from 2.4 to 6.0 Am 2 kg −1 , with saturation remanence (M rs ) values (0.2-0.6 Am 2 kg −1 ).
The IRM acquisition curves indicate that the specimens reach >95% saturation in fields of ∼0.2 T ( Figure 3). The back-field demagnetization indicates that remanence coercivity (H cr ) is generally less than 30 mT, with average of 20.9 ± 4.8 mT (n 27). There is single coercivity component with coercivity midpoint (B 1/2 ) range from 22.4 to 50.1 mT, with a mean of 33.5 ± 6.5 mT of conducted samples ( Figure 3; Table 1).
Temperature dependent susceptibility of representative samples from unit IV and III sharply drops at ∼585°C (Figure 4). All together the rock magnetism strongly suggests that (low-Ti) magnetite is the dominant magnetic carrier for the flows from Units III and IV on this section.
The M rs /M s vs. H cr /H c value on the Day-plot show that the bulk magnetic properties of the DD7-22 samples fall into the pseudo-single-domain region and generally follow the single domain (SD) and multidomain (MD) mixing curves, suggesting mixtures of SD and MD magnetite grains (Day et al., 1977;Dunlop, 2002; Supplementary Figure S1).

Paleomagnetic Directions
The lava flows are classified into two groups according to their stratigraphic units sampled. Unit III is comprised of flows DD12 to DD22 (106 samples) and unit IV is comprised of flows DD7 to DD11 (70 samples). The demagnetization diagrams for representative samples are shown in Figures 5, 6 and the paleodirections are summarized in Figure 7 and listed in Tables 2, 3.
For unit III specimens, one or two components can be isolated from the demagnetization trajectories ( Figure 5). A lowtemperature component (referred to as III-1) is separated below ∼300°C in 85 out of 106 unit III specimens with maximum angular deviation (MAD) less than 15°. Eleven outliers were further rejected for inclusion in the means, including ten directions with geographic declination (D g ) range from 90°to 270°, with positive inclinations and one with negative inclinations. The final calculated mean directions of III-1 components is D g /I g 10.6°/45.8°(n 74, k g 11.1, α 95 4.9°) and 13.7°/46.0°(N 9, n 74, k g 63.3, α 95 5.9°) on specimen and site level, respectively. These two mean directions are similar at the 95% CI. The latter one corresponds to a geomagnetic pole of 77.8°N, 187.8°E (N 9, K g 56.7, A 95 6.2°) and is used for further discussion (Table 2; Figure 7A). After tilt correction, the mean direction of III-1 component is D s /I s 352.0°/48.1°(N 9,  After removal of the low-temperature component, the characteristic remanent magnetization (ChRM) component of unit III specimens (III-2) can be isolated between ∼350-420°C and 585°C by a well-defined linear segment decaying to the origin on the orthogonal vector diagrams with MAD less than 10°( Figure 5). The site mean III-2 direction is D g /I g 19.4°/25.6°( N 9, n 81; k g 39.7, α 95 8.3°) before and D s /I s 8.8°/31.6°( N 9, n 81; k s 39.7, α 95 8.3°) after tilt correction. This corresponding to a geomagnetic pole of 67.2°N, 226.0°E (K g 42.2, A 95 8.0°) and 77.1°N, 240.0°E (K s 49.2, A 95 7.4°), respectively (Table 3; Figure 7D).
Demagnetization behavior of the unit IV specimens is more complicated. In general, a component with low unblocking temperature (IV-1), up to ∼300°C, can be isolated from most samples ( Figures 6A,B), generating similar mean directions of D g /I g 5.8°/41.6°(n 60; k g 14.1, α 95 4.8°) and 6.5°/41.0°(N 4; n 60; k g 61.7, α 95 11.8°) on specimen and site level, respectively. The latter one is used for further discussion and corresponds to a geomagnetic pole of 85.9°N, 181.7°E (N 4, K g 63.9, A 95 8.8°; Table 2). In some cases, the unblocking temperature of IV-1 component is as high as ∼500°C and overlaps with the ChRM component (IV-2, Figures 6C,D). The IV-2 component, which has unblocking temperatures of ∼500°C-580°C, can only be isolated from few specimens (five out of 70) from site DD10 using linear fit method ( Figures 6A,B) and in most cases overlaps with the IV-1 component. This behavior is revealed in vector component diagrams as curved demagnetization trajectories and on equal-area diagrams as great circles (Figures 6C,D). A combined analysis of remagnetization great circles and directions using the method of McFadden and McElhinny (1988) Figure 7C).

Mineralogy Inspection
Two representative specimens, DD10-10 from unit IV and DD21-03 from unit III, were investigated using optical microscopy. The two samples both have a fine matrix with a predominance of plagioclase microphenocrysts (Figures 8A-F). Interspersed pyroxene and other opaque minerals with a small amount of olivine are also observed ( Figures 8C,F). The grain size of the plagioclase laths ranges from 0.05 to 0.2 mm and the pyroxene has variably altered to chlorite ( Figures 8B,C,E,F). The amygdales found with the vesicles of DD21-03 are composed of chlorite and zeolite ( Figures 8E,F), which are commonly the result of epithermal alteration (Wei and Powell, 2003). Compared with the specimens in the upper flows, which record a primary remanence (sample DD4-03; Figures 8G-I), a higher amount of chlorite is observed in specimens DD10-10 and DD21-03 in this study. The higher degree of chlorite formation suggests that flows DD7-22 may have experienced a higher degree of hydrothermal alteration.

FIGURE 6 | (A-D)
Representative orthogonal plots (Zijderveld, 1967) of stepwise thermal demagnetization behavior of Emeishan basalt unit IV. Symbols are the same as in Figure 6.
Frontiers in Earth Science | www.frontiersin.org January 2021 | Volume 8 | Article 590939 8 the folding occurred after the Upper Cretaceous, which is contradicts current theories (Yang et al., 1986;Yan et al., 2003;Li et al., 2009;Dong et al., 2015). Furthermore, the Fisher statistic precision parameter of the mean direction before (k g ) and after (k s ) tilt-correction is equal that indicates the tilt correction is invalid (Van der Voo, 1990). Therefore, the in-situ III-1 directions are preferred, which suggests that the III-1 represent viscous remanent magnetization acquired during Paleogene to Quaternary period.
The mean direction of IV-2 component has a southwest declination and a moderate down inclination (D g /I g 195.1°/ 31.2°, n 22, k g 32.9, α 95 11.9°; D s /I s 196.7°/27.9°, n 22, k s 50.1, α 95 9.6°; Table 3; Figure 7C). The k s /k g > 1.5, indicating the tilt-corrected directions were better clustered that suggest the IV-2 components were obtained before folding. The tilt-corrected IV-2 mean directions are roughly equal to the primary remanence obtained from our previous work in this area where a mean reverse polarity direction of D s /I s 205.9°/10.0°(N 5, k s 78.4, α 95 8.7°; Figure 9A) was obtained from the upper flows from this section (Liu et al., 2012). Comparison of the mean directions of IV-2 to the primary ones reveals an angular difference of ∼19°, although this bias is significant, it is likely due to contamination by unresolved overprints in the new data, as evidenced by the need to use great circles. The VGP generated from IV-2 components is close to late Permain poles and far away from any younger poles (Huang et al., 2008;Huang et al., 2018 and references therein; Figure 9B). From this we infer that unit IV basalts have experienced a significant partial remagnetization event.
The  Table 3). When compared to the APWP of the South China Block, the poles generated from III-2 components lies distinct from Permian poles ( Figure 9B). The III-2 geomagnetic poles before and after tilt correction fall within the 95% confidence ellipsoids of Late Jurassic (J 3 ) and Middle-Early (J 1-2 ) poles, respectively (Huang et al., 2008;Huang et al., 2018 and references therein; Figure 9B). This observation is unlikely to be the result of paleosecular variation (PSV). Langereis et al. (2012) undertook a PSV analysis of the Permo-Carboniferous Reverse Superchron (PCRS), which ended at ∼263 Ma, shortly before the emplacement of the ELIP. They found that PSV during the PCRS was comparably low to that found during the Cretaceous Normal Superchron (Biggin et al., 2008), but distinctly lower than more recent time periods. From our eleven lava flows we calculate a VGP scatter value of 11.3°and that from 30 flows at the SM section and upper five flows at the DD section is 10.8°and 9.2°(at an equatorial paleolatitude), respectively (Liu et al., 2012), which is comparable to the equatorial values from both superchrons. This suggests that PSV immediately after the PCRS remained low and is unlikely to explain the observed difference in pole positions from our data. We conclude that our III-2 directions correspond to an overprint.
The F-test (Butler, 1992) confirms that the geomagnetic poles before and after tilt correction of III-2 components are consistency with J 3 (67.4°N, 218.8°E; K 78.0, A 95 7.6°; N 6; calculated from data listed in Huang et al., 2018   Notes: n/N, numbers of the chosen/conducted samples or site; D g (D s ), I g (I s ), declination and inclination before (after) tilt correction; k/K, Fisher statistic precision parameter of the mean direction/poles; α 95 , radius of the 95% confidence circle about the mean direction; Lat/Long, the longitude/latitude of the paleomagnetic pole or the virtual geomagnetic pole in stratigraphic coordinates; A 95 , the 95% confidence circle about the paleomagnetic pole. a Indicates the site-mean direction is combined by adjacent flows.
Frontiers in Earth Science | www.frontiersin.org January 2021 | Volume 8 | Article 590939 Besse, 2001), respectively. This indicates that the tilt-corrected paleodirections recorded in III-2 components were acquired during Early to Middle Jurassic while the III-2 in-situ directions were contemporary to Late Jurassic. If this is true, it confines the folding occurred during Early to Middle Jurassic. The most important folding event affecting the Yangtze platform during the Jurassic to Cretaceous periods was the Yanshanian Movement, which has a fold axis trending generally toward NNE- Equal-area projections of the paleomagnetic poles generated in this study (blue square) and those of the Yangtze Block since Late Permian (Huang et al., 2008; and references therein). P 2 , T 1 /T 2 , J 1 /J 2 /J 3 , K 1 / K 2 , E, Q indicates Late Permian, Early/Late Triassic, Early/Middle/Late Jurassic, Early/Late Cretaceous, Eocene and Quaternary, respectively. Frontiers in Earth Science | www.frontiersin.org January 2021 | Volume 8 | Article 590939 11 NE (Yang et al., 1986). The flow dips at our sample locality have a north strike (eg, the strike of Xuanwei Formation sediments is 357°) and that fitting was suggested to have occurred during the Yanshanian Movement (Yang et al., 1986;Yan et al., 2003;Li et al., 2009;Dong et al., 2015). However, it is difficult to determine the timing of folding accurately. If the folding is earlier than the acquisition of the III-2 components, then the in-situ directions suggest Upper Jurassic remagnetization. Given these considerations, it is tentatively interpreted that the tiltcorrected directions were acquired prefolding (i.e., Early to Middle Jurassic), suggesting the Emeishan basalts of Unit III on this section experienced significant overprinting.

Mechanism of the Remagnetization
Native copper mineralization within the Emeishan flood basalts is present at many locations, but this is a highly localized and spatially discontinuous feature, being mainly restricted to unit IV. However, the occurrence of copper mineralization is not controlled by the distribution of the Emeishan basalts themselves, but mainly localized by faults and/or folds within basalt (Li et al., 2009; Figure 1B). At the studied section, two mining tunnels are in basalts flows at sites DD9 and DD10 in unit IV. Combined with the contamination of the unit IV paleomagnetic directions by a notable overprint, this suggests the lower unit IV basalts have experienced a remagnetization event associated with copper mineralization. However, the overlying basalt flows (DD2-6) as well as conglomerates, and other sections in this area preserve primary magnetizations (cf. Liu and Zhu, 2009;Liu et al., 2012). We speculate that the remagnetization of the Emeishan basalts is spatially related to copper mineralization and is a local phenomenon.
Another example of copper mineralization in massive volcanic rocks is in the Keweenaw Peninsula, United States, which contains large (∼50 million tons) copper deposits hosted by ca. ∼1.1 Ga mafic lavas and conglomerates Zhu et al., 2003). Geochemical studies reveal the copper associated within both Keweenawan rocks (Keweenaw Peninsula, Michigan, United States) and Emeishan basalts are related to low-temperature hydrothermal activity having assemblages of low-grade, secondary hydrothermal phases (Zhu et al., 2003;Li et al., 2005;Li et al., 2009;Wang et al., 2011). Remagnetization within the Keweenawan rocks has been demonstrated to be associated with the formation of native copper formation, with increasing degrees of overprinting moving west along the lava flow strikes (Halls and Pesonen, 1982). White (1978) proposed a model for the Keweenawan copper deposits where metamorphic fluids migrated up-dip along the strike and leached the copper from the host volcanic rocks. This model is consistent with the observation that the overprint of the host basalts was found to increase along strike coincident with an increase in burial metamorphism in the underlying Portage Lake Volcanic Group Palmer et al., 1981).
The remagnetization characteristics of the Emeishan basalts associated with copper mineralization are, however, different from those of Keweenawan rocks in at least two ways. First, the remagnetization is in rocks of an area of weak copper mineralization associated with the faults; and second, the location of remagnetization is stratabound by the copper mineralized flows. Additionally, the copper mineralization in the Emeishan basalts is mostly associated with bitumen and is dispersed Palmer et al., 1981;Zhu et al., 2003;Li et al., 2009). These differences are interpreted that the copper bearing hydrothermal fluids of the Emeishan basalts have a different origin than metamorphic fluids and that they probably migrated along the fault systems. Geochemical data suggest that native copper mineralization in the Emeishan basalts was a result of epigenetic hydrothermal activity associated with strong organic-inorganic interactions (Li et al., 2004;Wang et al., 2006;Li et al., 2009). The hydrothermal fluids are most likely due to basinal fluids that originally formed by meteoric waters (Li et al., 2009). These basin fluids migrated along the faults, leaching the Emeishan basalt flows and deposited copper in localized areas associated with fault systems. Compared to the extensive metamorphic fluids that concentrated copper in Keweenawan rocks, the basin fluids that leached the Emeishan basalts was probably inefficient over a much smaller scale.
Zeolite, bitumen and chlorite are present in the rocks of flow DD7. Microscopic investigation confirms the chlorite and zeolite within the DD7-22 samples. The higher chlorite content of flows DD7-22 when compared with flows DD2-6, suggests that DD7-22 experienced a higher degree of hydrothermal activity. This suggests that the remagnetization of the Emeishan basalt flows DD7-22 is due to the hydrothermal activity associated with copper mineralization. We suggest that the hydrothermal fluids migrated through the Emeishan basalts using the large tuff unit separating units III and IV of the basalts and the fault systems as pathways diverting laterally, reducing, but not stopping, the upward flow into unit IV. This would explain the reduction of the remagnetization signal moving up section. The main magnetic carrier of the DD7-22 basalt samples is (low-Ti) magnetite, which is suggested by the typical unblocking temperature. Thus, the magnetite within the DD7-22 samples is most probably autogenic and carries a chemical remanence magnetization due to the hydrothermal activity.
Many paleomagnetic studies in southwest China have demonstrated the effects of widespread remagnetizations (Kent et al., 1987;Dobson and Heller, 1992;Liu et al., 2011). Some 39 Ar-40 Ar data from whole-rock of basalt samples dating on the Emeishan basalts has yielded anomalously young ages at ∼175, ∼142, ∼98 and ∼42 Ma (Hou et al., 2002;Ali et al., 2004). The Longmenshan fault system to the north of the ELIP ( Figure 1A), has been interpreted responsible for these overprint dates (Ali et al., 2004). The collision of the North and South China blocks, initially at the eastern end of Qinling Orogen in Late Permian to Early Triassic and suturing westward until the final suture occurred in the Early to Middle Jurassic (Zhao and Coe, 1987;Zhu et al., 1998;Yang and Besse, 2001), has been suggested as a potential cause of the older overprinting events (Ali et al., 2004). The 39 Ar-40 Ar ages of secondary basaltic minerals (eg, laumontite, actinolite) in the copper ores in NE Yunnan yield two age groups (226-228 Ma and 134-149.1 Ma; Zhu et al., 2005). Our paleomagnetic data suggest a significant remagnetization occurred during the Early to Middle Jurassic, which is consistent with previous age estimates for mineralization. Considering that copper mineralization is widespread in Yunnan, Sichuan and Guizhou provinces ( Figure 1A) and is associated with faults, the timing of formation of fault systems is probably coeval with mineralization. The faulting associated with the Yanshanian Movement in the opened up pathways for fluid migration through the Emeishan basalts, where copper was provided. This interpretation of paleomagnetic data is consistent with the suggestion that the main copper mineralization in this region is Jurassic (Li et al., 2009).

Implications to Future Study
Paleomagnetic studies on the Emeishan basalts report the prevalence of normal magnetizations over the presumably younger reversed polarity remanence (Huang and Opdyke, 1998;Ali et al., 2002;Liu et al., 2012). This study provides one explanation for this phenomenon in that parts of the Emeishan basalt unit IV of reverse polarity have been remagnetized during Early to Middle Jurassic. The steeper inclinations that characterize remagnetized sites suggest that caution should be exercised if similar steep inclinations are observed at other Emeishan flow sites, as these may be the result of remagnetization by younger hydrothermal activity. This observation may explain some previous results from the Emeishan basalts, for example, the directions observed in the study of Huang et al. (1986). The native copper deposits hosted in the Emeishan basalts are spatially associated with the remagnetization of the neighboring basalts flows. This process likely occurred in the Early to Middle Jurassic, which constrains the timing of copper mineralization to this period. The copper mineralization is generally small scale in volume and associated with local structure and lithology features. Paleomagnetic studies aiming to obtain primary remanent magnetizations should avoid such areas of the affected flood basalts. The high chlorite content of remagnetized basalts (DD21-03 and 10-10; Figure 8) identified in the microscopic investigation aided in the identification of remagnetized samples and future studies should include microscopic work. Future paleomagnetic studies in this region may provide further useful insights into the timing of copper mineralization, particularly in association with studies focusing on the Yanshanian Orogeny.

CONCLUSION
Stepwise thermal demagnetization on 176 basalt samples from 16 sites in the Dadi section from units III and IV of the Permian Emeishan basalt Formation yielded two distinct ChRMs. The samples in the upper unit IV have complicated magnetizations, with demagnetization behavior typical of extensive, but not complete remagnetization. These data yield a mean tilt-corrected direction of D s /I s 196.7°/27.9°(n 22, k s 50.1, α 95 9.6°) that most probably represents a primary remanence direction, but is likely contaminated by unresolved overprints that resulted from hydrothermal activity. The lower flows from unit III have been completely remagnetized, and yield a mean tilt corrected direction of D s /I s 8.8°/31.6°(N 9, k s 39.7, α 95 8.3°), which is of normal polarity and corresponds to a geomagnetic pole at 77.1°N, 240.0°E (K s 49.2, A 95 7.4°). The overprint directions have steeper inclinations (∼19°) than the primary directions that have been successfully recovered from other sites.
By comparison with the Phanerozoic paleomagnetic poles of the South China Block, the inferred age of remagnetization time is likely the Early to Middle Jurassic. Field relationships reveal that the remagnetization of the Emeishan basalts is spatially related to copper mineralization and is likely to be associated with fluid migration through the fault systems. The remagnetization event related to copper mineralization is nevertheless a local phenomenon in the Emeishan basalts.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
CY performed the study and measurements, SL assisted with fieldwork and measurements. All authors contributed to data analysis, interpretations and writing of the manuscript.

ACKNOWLEDGMENTS
We thank Lei Wang and Jiyun Yin for the help with sampling in the field, to Mingjian Cao for the help with the microscope investigation, to Randy Enkin for the PMGSC program. We are also grateful for the detailed review by Henry Halls, Jason Ali, John Geissman and two anonymous reviewers, and suggestions from James Tyburczy that help us improve the manuscript.