Petrogenesis and an Evaluation of the Melting Conditions of the Late Permian ELIP Picrites, SW China: Constraints Due to Primary Magma and Olivine Composition

Abstract

The late Permian Emeishan large igneous province (ELIP) in SW China is a melting product of the Emeishan mantle plume. Recently, it has been debated whether peridotite or pyroxenite is the dominant lithology of the mantle source in the ELIP. To address this, systematic analyses of bulk-rock and coexisting spinel and olivine compositions were conducted on picrites from Lijiang–Yongsheng, Dali–Binchuan, Yumen, Muli, and Ertan. The ELIP picrites exhibit positive TiO₂–CaO and negative MgO–CaO correlations, as well as low FC3MS values (−0.24–0.1), supporting a peridotite-dominated mantle source. This lithology of the mantle source is also supported by the high 100 × Mn–Fe (1.43–1.73) and Mn–Zn (13.6–18.4) values but low 10,000 × Zn–Fe (8.0–12.7) ratios of the olivine phenocrysts. The estimated mantle potential temperature for Lijiang, Yongsheng, Yumen–Ertan, Muli, and Dali–Binchuan picrites decreased away from Lijiang and Yongsheng, suggesting that the Lijiang and Yongsheng areas were the center of the ELIP. The Lijiang–Yongsheng primary magma shows similar SiO₂ content but lower Al₂O₃ contents (average of 8.24 wt.%) and higher MgO contents (average of 21.42 wt.%) than those of Dali–Binchuan primary magma (Al₂O₃: 9.86 wt.%; MgO: 19.02 wt.%). Also considering the high Gd–Yb (average of 3.05) and La–Yb (average of 14.61) ratios and mantle potential temperature (average of 1599 °C), we proposed that Lijiang–Yongsheng lavas are produced via the melting of a garnet–peridotitic mantle. In contrast, the Dali–Binchuan lavas with low Gd–Yb (average of 1.91) and La–Yb (average of 5.88) ratios can be explained by their formation in the garnet–spinel transition zone of a peridotitic mantle. The Yumen–Ertan primary magma displays similar mantle potential temperature (average of 1600 °C), Al₂O₃ and FeO content, and Gd–Yb ratios to those of Lijiang–Yongsheng lavas, indicating that YumenvErtan primary magma may be attributed to the partial melting of garnet with minor peridotite. Therefore, heterogeneous plume-head mantle sources lead to the evaluation of melting conditions of the late Permian ELIP picrites.

1. Introduction

Due to the large volume of magma erupted, large igneous provinces (LIPs) commonly contain numerous world-class base metal deposits (e.g., Fe–Ti oxide deposits) [1,2] and serve as potential Au reservoirs under the crustal magmatic underplate [3]. For example, the Permian Emeishan large igneous province (ELIP), situated in the western part of the Yangtze craton, is a melting product of the Emeishan mantle plume [4,5,6,7,8]. To date, although previous authors have conducted a large amount of research on the genesis of the ELIP [9,10,11,12], there remain two contradictory views about the lithology of the mantle source of the ELIP basalts: peridotite vs. pyroxenite [7,13,14,15,16]. Understanding the mantle-source lithology of ELIP picritic–basaltic rocks is crucial for unraveling their petrogenesis.

When the nature of the mantle-source lithology of the ELIP is investigated, picrites provide a better comparison than ELIP basalts [4,17,18]. The correlation between CaO and TiO₂ [19] and the indicator parameter FC3MS (FeO/CaO − 3 × MgO/SiO₂, wt.%) [20,21] can help identify the melt characteristics of peridotite and pyroxenite. Moreover, experimental studies indicate that olivine grains in magmas derived from peridotite- and pyroxenite- dominated sources have significant differences in 100 × Mn–Fe, 10,000 × Zn–Fe, and Mn–Zn ratios [22]. Notably, a high-Mg picrite sample is only a mixture of olivine and melts at varying degrees, rather than primary liquids [14,17,23], resulting in the bulk-rock composition no longer containing complete information about the primitive magma composition [24]. Thus, this study used the PRIMELT3 software (GitHub, San Francisco, CA, USA) to re-evaluate the primary magma compositions and melting conditions (e.g., temperature and 22pressure) in generating ELIP picrite [24]. The PRIMELT3 model differs from other traditional methods of constraining the primary magma composition based on the olivine composition [25].

Rare picrites are found at several levels of the volcanic successions in the Lijiang–Yongsheng, Dali–Binchuan, Yumen, Muli, Ertan, and Song Da regions [13,14,25,26,27,28]. The ELIP’s magmatism lacks a simple model due to the lithological and geochemical complexity of erupted magmas at the local (e.g., cross-section), regional (single and different LIPs), and global scales [10]. Therefore, this study systematically collected the geochemical (major and trace elements) and mineralogical (coexisting olivine and Cr-spinel) compositions from all ~260 Ma ELIP picrites. Furthermore, we used the PRIMELT3 software and an Al-in-olivine thermometer to recalculate the primary magma composition, the mantle potential temperature (the temperature that the solid adiabatically convecting asthenospheric mantle would have if it could reach the surface metastably without melting [29]), and the olivine crystallization temperature (T_{Al-in-olivine} and T_Sc-Y). Identifying the mantle-source messages and mantle melting behaviors can contribute to deciphering the deep magmatic processes of ELIP picrite formation.

2. Geological Setting

The late Permian ELIP (~260 Ma; [2]) lies in the western part of the Yangtze Craton, and its western boundaries are defined by the Ailaoshan–Red River fault [4,5,30]. There has been wide speculation that the ELIP is associated with a mantle plume at the base of the continental lithosphere [2,31,32,33]. Since it is one of the world’s largest continental LIPs, large Emeishan continental flood basalt and mafic–ultramafic intrusions of at least 0.25 × 10⁶ km² have been discovered there (Figure 1) [11].

The Emeishan Permian flood basalt is divided into western, middle, and eastern parts. The thickness of the Permian flood basalt sequence in the ELIP varies from ~5000 m in the west to several hundred meters in the east [4]. The eastern part of the ELIP is mainly basaltic lava. The middle part comprises basaltic lava with a small amount of alkali–acid volcanic rocks. The western side of the Emeisham basalt province consists primarily of tholeiitic and subordinate minor rhyolite and picrate [34].

The coeval picrite and picritic basalt lie in several levels of the volcanic successions that have been reported in only a few locations, including the Lijiang–Yongsheng, Dali–Bingchuan, Ertan, Yumen, and Song Da areas [13,35,36]. Notably, using the term ‘picrite’ to describe the Permian high-Mg volcanic rocks in the Song Da district is still controversial. Glotov et al. (2001) [32] found pyroxene spinifex textures of Song Da rocks. In contrast, Anh et al. (2011) [26] regarded olivine in the Song Da high-Mg volcanic rocks as either elongated or equant, rather than spinifex-textured.

3. Picrite Samples and Methodology

3.1. Permian Emeishan Picrite

The primary textures and modal compositions of the Permian Emeishan picrites from the Dali–Binchuan, Lijiang–Yongsheng, Yumen, Ertan, and Muli areas [11,12,13,14,28,35,37] are generally similar. Detailed petrographic characteristics are given below.

The Dali, Binchuan, Wuguijing, and Wulongba picrites occurred in the Dali–Binchuan area. The Dali picrites occur as massive and pillowed lavas, indicating a subaqueous environment of eruption [35]. At Binchuan, the studied picritic lavas occur among basalts and andesites in the upper part of an exceptionally thick (45-km) succession [30]. The Wuguijing and Wulongba picritic lava flows typically occur in the lower parts of their volcanic sequences [12]. These picrite samples have a porphyritic texture, containing 20 vol.% to 35 vol.% euhedral olivine phenocrysts ranging in diameter from 0.2 to 4 mm [12,16,38]. Serpentine replacement along micro-fractures in olivine is more common in Wuguijing, indicating that the Wuguijing picrites are more altered than those from Dali, Binchuan, and Wulongba. Small-melt and Cr-spinel inclusions are found in some large olivine phenocrysts [10]. The groundmass includes anhedral clinopyroxene, plagioclase microlites, and minor Fe–Ti oxide minerals [36,37].

Representative picrate samples from Lijiang–Yongsheng are highly porphyritic, and there are olivine phenocrysts of up to 40 vol.% in Lijiang, Tanglanghe, Maoniuping, and Yongsheng picrites. The Lijiang picrites, collected near Lijiang [33], Daju, and Shiman City [18], occur as 3–50 m-thick intercalations with pyroxene–phyric high-Ti basalts [18]. There is also evidence of picrite rocks in the Yongsheng area, about 10 km north of Binchuan [10]. At Yongsheng, the picrite samples were collected from different levels of a 50 m-thick lava flow within a 1–2 km-thick volcanic sequence [4]. Notably, the picrite samples from Tanglanghe and Maoniuping are more altered than those from Wulongba [12]. Melt inclusions and Cr-spinel inclusions are enclosed in some fresh olivine cores. The fine-grained groundmass comprises plagioclase, clinopyroxene, phlogopite, apatite, and Fe–Ti oxide minerals [10,35].

The representative Yumen, Ertan, and Muli picrite samples are massive rocks with a porphyritic texture, containing about 20 vol.% olivine phenocrysts [11,12]. A picritic lava flow with a thickness of ~5 m is interbedded within a basaltic sequence in Yumen [11]. Ertan picrites occur near the bottom and in the middle part of a flood basalt pile with a thickness of ca. 1 km [30]. Spherical weathering is common in the Ertan picrite outcrop [12], and no fresh olivine but numerous pseudomorphs after olivine have been found in Muli picrite [39], indicating that these picrites are severely altered. Micron-sized Cr-spinel crystals with a diameter of up to 0.4 mm are enclosed in olivine phenocrysts or occur as isolated grains within the groundmass [28]. Secondary minerals are composed of serpentine, talc, and chlorite [39].

3.2. Methodology

This study used the PRIMELT3 model and major element compositions of bulk-rock to calculate the primary magma compositions of the Emeishan picrites. Based on mass balance, PRIMELT3 reconstructed the primary magma composition by adding or subtracting olivine to/from the lava composition [24]. Then, the primary magma MgO contents (weight%) were used to obtain the mantle potential temperature (T_P): T_P (°C) = 1025 + 28.6 × MgO − 0.084 × MgO₂. In the formula, the systematic errors for T_P were within ±42 °C [24].

According to the Al concentrations between coexisting olivine and Cr-spinel, the olivine crystallization temperatures for ELIP picrites could be calculated with an Al-in olivine thermometer (T[K] = 10,000/(0.575 + 0.884 × Cr^# − 0.897 × Ln(Al₂O₃^olivine/Al₂O₃^spinel); Cr^# (=Cr/[Cr + Al]) < 0.69) [40].

We also reused these olivine compositions and major compositions of bulk-rock to estimate the magma crystallization temperatures using an Sc-Y geothermometer (T[K] = [−3230 − 100 × P(Ga) − 1402 × Mg^#ol + 1933 × ${\mathrm{X}}_{\mathrm{CaO}}^{\mathrm{melt}}$ + 2612 × (${\mathrm{X}}_{{\mathrm{NaO}}_{0.5}}^{\mathrm{melt}}$ + ${\mathrm{X}}_{{\mathrm{KO}}_{0.5}}^{\mathrm{melt}}$) − 569 × ${\mathrm{X}}_{{\mathrm{SiO}}_2}^{\mathrm{melt}}$]/[−1.471 − log₁₀(${\mathrm{D}}_{\mathrm{Sc}}^{\mathrm{ol}/\mathrm{melt}}$ − ${\mathrm{D}}_{\mathrm{V}}^{\mathrm{ol}/\mathrm{melt}}$)) [41].

4. Geochemical Characteristics

4.1. Bulk-Rock Compositions for Picrite Samples

Major- and trace-element data for the ninety-nine Emeishan picrite samples from the Dali–Binchuan, Lijiang–Yongsheng, and Yumen–Muli areas (Table S1) were compiled from the published literature (e.g., [1,33,42,43]). We filtered the bulk-rock data for the ELIP picrites using the following criteria: (1) an ablation loss-on-ignition (LOI) of less than 6 wt.%, (2) low SiO₂ contents (<47 wt.%) [28] and total alkalis (K₂O + Na₂O < 3 wt.%), and (3) high MgO (>16 wt %) contents [37]. After correction for the LOI, all of the picrite samples displayed low SiO₂ contents of 42.41 wt.%–48.33 wt.%, total alkalis (K₂O + Na₂O = 0.19 wt.% − 2.24 wt.%), and Al₂O₃–TiO₂ ratios (3.26–9.52), but they displayed high MgO contents (17.81 wt.%–29.65 wt.%) (Table S1). Based on the classification criteria proposed by Li et al. (2019) [44], the Songda picrites belonged to the komatiite group (Figure 2a).

Previous researchers have classified the Emeishan picrites as high-Ti and low-Ti series according to their bulk TiO₂ contents and Ti–Y ratios (high-Ti magma has TiO₂ > 2.5 wt.% and Ti–Y > 500) [30] or using the 0.08 × MgO + TiO₂ − 2.91 parameter [45]. However, we did not distinguish the ELIP picrites into low- and high-Ti groups. The Emeishan picrites contain Fe–Ti oxide minerals [10,16] that may alter the rock composition [17], demonstrating the ineffectiveness of classifying according to the TiO₂ concentration and Ti–Y ratios. On the other hand, when the classification scheme proposed by He et al. was used (2010) [45], the Muli picrite samples spanned a continuous range from −0.14 to +0.23 (Figure 2b, Table 1), demonstrating the inadequacy of separating rocks based on the 0.08 × MgO + TiO₂ − 2.91 parameter.

Table 1. Bulk-rock compositions, mantle potential temperature (T_P), and olivine crystallization temperature (T_{Al-in-olivine} and T_Sc-Y) of the Emeishan picrites.

Location	Bulk-Rock Compositions						Olivine Compositions
	MgO	0.08 × MgO + TiO2 − 2.91	Gd–Yb	La–Yb	TP	Data Source	Fo	10,000 × Zn–Fe	TSc-Y	Data Source	TAl-in-olivine	Data Source
	wt.%				°C		mol.%		°C		°C
Lijiang	19.04–27.53	0.21–0.62	2.31–3.32	9.14–20.94	1573–1660	[16,18,33,35]	86–92	7.59–10.38	1455 ± 49	[16]	1283 ± 75	[16,33,46]
Tanglanghe	17.90–20.77	0.48–0.56	2.81–3.00	11.35–11.58	1550–1552	[25]	86–90	7.00–9.43	1358 ± 22	[28]	1247 ± 41	[12]
Maoniuping	20.40–24.07	0.38–0.63	2.85–3.12	8.43–19.36	1555–1582	[25]	86–91	6.91–9.27	1366 ± 31	[28]	1260 ± 45	[12]
Yongsheng	21.95–23.76	0.61–0.67	3.22–3.62	16.67–18.4	1614–1649	[43]	/	/	/		/
Dali	17.00–23.95	(−0.52)–(−0.1)	1.97–2.39	5.39–7.62	1506–1588	[35,37,38,42]	82–94	5.43–12.69	1379 ± 52	[16,46]	1305 ± 74	[13,16,33,44,46]
Binchuan	20.59–21.08	0.09–0.37	2.68–3.15	11.47–17.92	1581–1592	[47]	/	/	/		1282 ± 30	[13,43]
Wuguijing	17.30–18.90	0.09–0.37	2.68–3.15	11.47–17.92	1523–1574	[25]	87–90	7.10–9.37	1352 ± 41	[28]	1301 ± 55	[12]
Wulongba	19.40–21.03	(−0.22)–(−0.16)	2.07–2.09	3.01–3.09	1516–1539	[25]	86–92	7.18–9.77	1376 ± 36	[28]	1328 ± 88	[12]
Yumen	17.80–20.30	0.38–0.49	3.11–3.77	11.95–14.02	1592–1637	[11]	86–92	8.42–11.84	1375 ± 49	[28]	/
Ertan	18.30–20.90	0.26–0.72	3.12–4.57	13.71–24.63	1565–1635	[12]	/	/	/		1245 ± 48	[12]
Muli	16.64–21.16	(−0.14)–0.23	2.35–3.01	1.18–7.87	1555–1603	[39]	/	/	/		/
Songsa	13.86-22.48	(−1.07)–(−0.63)	1.07–1.61	0.47–6.06	1522–1631	[12,17]	/	/	/		/

Notes: 1. The mantle potential temperatures (T_p) were recalculated using the primary magma compositions. 2. Olivine grains coexisting with Cr-spinel with low Cr^# (<0.69) were taken into account in the calculation of the continuous range of T_{Al-in-olivine}.

Table 1. Bulk-rock compositions, mantle potential temperature (T_P), and olivine crystallization temperature (T_{Al-in-olivine} and T_Sc-Y) of the Emeishan picrites.

Location	Bulk-Rock Compositions						Olivine Compositions
	MgO	0.08 × MgO + TiO2 − 2.91	Gd–Yb	La–Yb	TP	Data Source	Fo	10,000 × Zn–Fe	TSc-Y	Data Source	TAl-in-olivine	Data Source
	wt.%				°C		mol.%		°C		°C
Lijiang	19.04–27.53	0.21–0.62	2.31–3.32	9.14–20.94	1573–1660	[16,18,33,35]	86–92	7.59–10.38	1455 ± 49	[16]	1283 ± 75	[16,33,46]
Tanglanghe	17.90–20.77	0.48–0.56	2.81–3.00	11.35–11.58	1550–1552	[25]	86–90	7.00–9.43	1358 ± 22	[28]	1247 ± 41	[12]
Maoniuping	20.40–24.07	0.38–0.63	2.85–3.12	8.43–19.36	1555–1582	[25]	86–91	6.91–9.27	1366 ± 31	[28]	1260 ± 45	[12]
Yongsheng	21.95–23.76	0.61–0.67	3.22–3.62	16.67–18.4	1614–1649	[43]	/	/	/		/
Dali	17.00–23.95	(−0.52)–(−0.1)	1.97–2.39	5.39–7.62	1506–1588	[35,37,38,42]	82–94	5.43–12.69	1379 ± 52	[16,46]	1305 ± 74	[13,16,33,44,46]
Binchuan	20.59–21.08	0.09–0.37	2.68–3.15	11.47–17.92	1581–1592	[47]	/	/	/		1282 ± 30	[13,43]
Wuguijing	17.30–18.90	0.09–0.37	2.68–3.15	11.47–17.92	1523–1574	[25]	87–90	7.10–9.37	1352 ± 41	[28]	1301 ± 55	[12]
Wulongba	19.40–21.03	(−0.22)–(−0.16)	2.07–2.09	3.01–3.09	1516–1539	[25]	86–92	7.18–9.77	1376 ± 36	[28]	1328 ± 88	[12]
Yumen	17.80–20.30	0.38–0.49	3.11–3.77	11.95–14.02	1592–1637	[11]	86–92	8.42–11.84	1375 ± 49	[28]	/
Ertan	18.30–20.90	0.26–0.72	3.12–4.57	13.71–24.63	1565–1635	[12]	/	/	/		1245 ± 48	[12]
Muli	16.64–21.16	(−0.14)–0.23	2.35–3.01	1.18–7.87	1555–1603	[39]	/	/	/		/
Songsa	13.86-22.48	(−1.07)–(−0.63)	1.07–1.61	0.47–6.06	1522–1631	[12,17]	/	/	/		/

Notes: 1. The mantle potential temperatures (T_p) were recalculated using the primary magma compositions. 2. Olivine grains coexisting with Cr-spinel with low Cr^# (<0.69) were taken into account in the calculation of the continuous range of T_{Al-in-olivine}.

Figure 2. Geochemical characteristics of picrite samples from the Lijiang–Yongsheng, Dali–Binchuan, Yumen, Muli, and Ertan areas. All data are presented in Table S1. (a) Discriminant diagram for ELIP picrite samples, adapted from Li et al. (2019) [44]; all of the MgO, Al₂O₃, and TiO₂ content plotted here was recalculated to be 100% volatile-free. (b) Classification based on the parameter (0.08 × MgO + TiO₂ − 2.91) proposed by He et al. (2010) [45]. (c) The relationships between the TiO₂ content and CaO–Al₂O₃ ratio for the bulk-rock compositions. (d) Diagram of La–Yb versus Gd–Yb ratios. Data sources: Lijiang: Zhang et al. (2006) [18], Hanski et al. (2010) [35], Zhang et al. (2021) [33], and Wu et al. (2022) [16]; Tanglanghe, Maoniuping, Wuguijing, and Wulongba: Yu et al. (2020) [44]; Yongsheng: Liu et al. (2022) [44]; Dali: Hanski et al. (2010) [35], Liu et al. (2017) [44], Wu et al. (2018) [44], and Li et al. (2014) [44]; Binchuan: Yu et al. (2019) [47]; Yumen: Yao et al. (2021) [11]; Ertan: Yu et al. (2024) [12]; Muli: Li et al. (2010) [39].

Figure 2. Geochemical characteristics of picrite samples from the Lijiang–Yongsheng, Dali–Binchuan, Yumen, Muli, and Ertan areas. All data are presented in Table S1. (a) Discriminant diagram for ELIP picrite samples, adapted from Li et al. (2019) [44]; all of the MgO, Al₂O₃, and TiO₂ content plotted here was recalculated to be 100% volatile-free. (b) Classification based on the parameter (0.08 × MgO + TiO₂ − 2.91) proposed by He et al. (2010) [45]. (c) The relationships between the TiO₂ content and CaO–Al₂O₃ ratio for the bulk-rock compositions. (d) Diagram of La–Yb versus Gd–Yb ratios. Data sources: Lijiang: Zhang et al. (2006) [18], Hanski et al. (2010) [35], Zhang et al. (2021) [33], and Wu et al. (2022) [16]; Tanglanghe, Maoniuping, Wuguijing, and Wulongba: Yu et al. (2020) [44]; Yongsheng: Liu et al. (2022) [44]; Dali: Hanski et al. (2010) [35], Liu et al. (2017) [44], Wu et al. (2018) [44], and Li et al. (2014) [44]; Binchuan: Yu et al. (2019) [47]; Yumen: Yao et al. (2021) [11]; Ertan: Yu et al. (2024) [12]; Muli: Li et al. (2010) [39].

The ELIP picrites from the Dali–Binchuan, Lijiang–Yongsheng, and Yumen–Muli areas shared high CaO concentrations (5.76 wt.%–10.75 wt.%), exceptionally low TiO₂ (<2.5 wt.%; Figure 2c) and Al₂O₃ (5.28 wt.%–10.62 wt.%) content, and a strong enrichment of incompatible trace elements (e.g., La–Yb = 10.39 ± 5.592, Gd–Yb = 2. 74 ± 0.55; Figure 2d). The characteristics of the relative loss of large-ion lithophile elements (Rb, Sr, and P; Figure 3a) and high-field strength elements (Nb, Th, and Hf), as well as the enrichments in light rare earth elements, proved their mantle-plume activity [28]. These Permian picrites displayed ocean island basalt (OIB)-like patterns (Figure 3b) and almost linear REE patterns with enrichments in light REEs relative to heavy REEs. Most picrites were depleted in most trace elements relative to ocean island basalts (OIBs), except for the Muli picrite (Figure 3b).

4.2. Olivine Compositions

The published olivine compositions of the Dali–Binchuan, Lijiang–Yongsheng, and Yumen picrite samples [16,28,46] are presented in Table S2. Olivine phenocrysts were Fo-rich [100 × Mg/(Mg + Fe²⁺), molar] and ranged from 82.5 mol.% to 93.7 mol.%. They had high CaO contents (0.20 wt.%–0.54 wt.%; Figure 4a) that differed from those of mantle-derived olivine xenocrysts (CaO < 0.1 wt.%) [49], indicating that the olivine phenocrysts crystallized from magma, rather than mantle xenocrysts. The variable Fo values (94 to 82; Table 1) of the olivine crystals in the Dali–Binchuan, Lijiang–Yongsheng, and Yumen picrite samples were also suggestive of crystallization from magmas because the olivine of mantle xenocrysts is commonly less variable in composition [50].

The olivine phenocrysts in the ELIP picrites had much higher Ni concentrations (up to c. 3897 ppm; Table S2) than those from the primitive mantle and depleted peridotite (1960 ppm) [48] or mantle peridotite (2800 to 3100 ppm) [51]. Moreover, the Ni content of the olivine phenocrysts showed a positive correlation with the Fo values (Figure 4b).

The coexisting spinel grains of the ELIP picrites had Cr₂O₃ values varying from 31.39 wt.% to 54.46 wt.% and Al₂O₃ values varying from 7.56 wt.% to 28.66 wt.%, with Cr^# [100 × Cr/(Cr + Al), molar] ranging from 28 to 82 and Mg^# [100 × Mg/(Mg + Fe²⁺), molar] ranging from 24 to 68 (Table S3).

5. Discussion

5.1. Mantle Source for Picrites: Peridotite vs. Pyroxenite

The CaO concentration (5.76 wt.%–10.75 wt.%) of the ELIP picrites was relatively high, with a positive correlation with the TiO₂ content (Figure 5a), excluding the necessity of a metasomatic pyroxenite contribution (e.g., [19,52]). Moreover, the FC3MS values of the ELIP picrites ranged from −0.24 to 0.1, and all of them were within the range of peridotite (FC3MS < 0.65; Table 1) [21], further indicating that the source region of the ELIP picrites should mainly be mantle peridotite (Figure 5b). Furthermore, the MgO, Al₂O₃, and CaO contents of the ELIP picrites were plotted in the field of experimental melts of peridotite (Figure 5c,d).

Olivine phenocrysts are the most important and earliest crystalline silicate minerals in picritic porphyry, and this study also used the olivine composition to evaluate the residual mineralogy of the source. Although rare olivine phenocrysts were altered [28], there was no evidence of negative Nb or Ti anomalies in the primitive mantle-normalized spider diagram (Figure 3a), indicating that negligible crustal contamination could not affect the olivine composition. The negligible crustal contamination was consistent with the Th–Ta and La–Nb ratios, which were similar to those of the primary mantle (Figure 6a,b), and the lower γOs(t) values of the Lijiang picrites relative to global plume-related picrites [7].

The olivines in the ELIP picrites had 10,000 × Zn–Fe values of 8.0–12.7 (Table S2), showing a similar trend to that of olivines in Edenteka and Baffin Island picrites derived from a peridotite source [56]. Moreover, the 100 × Mn–Fe (1.43 to 1.73; Table S2) and Mn–Zn (13.6 to 18.4) ratios in the ELIP picrites were consistent with those in olivines crystallized from melts derived from peridotite (Figure 7a,b).

5.2. Primary Magma Composition and Mantle Potential Temperature

As demonstrated above, melts derived from peridotite-dominated sources could be identified using the covariation between MgO and CaO in the bulk-rock composition, as well as the high Mn–Zn and 10,000 × Zn–Fe ratios of primitive olivine phenocrysts (Figure 6 and Figure 7). Moreover, Emeishan picrite samples exhibiting early pyroxenite crystallization could be further identified and filtered using the PRIMELT3 software. Finally, forty-nine Emeishan picrite samples from the Dali–Binchuan, Lijiang–Yongsheng, and Muli areas were used here to recalculate the primary magma compositions and mantle potential temperature (Table S4).

Here, our estimated SiO₂ content for the Dali–Binchuan primary melt (46.94 ± 0.72 wt.%) was similar to that of the Lijiang–Yongsheng primary melt (46.10 ± 0.43 wt.%; Figure 8a). However, the Lijiang–Yongsheng primary melt exhibited lower Al₂O₃ content (7.09 wt.%–9.19 wt.%) and higher MgO content (19.49 wt.%–23.87 wt.%) than those of the Dali–Binchuan primary melt (Al₂O₃: 8.13 wt.%–10.67 wt.%; MgO: 17.75 wt.%–21.15 wt.%) (Figure 8b). The MgO content was positively correlated with the melting temperature [57]; thus, the high MgO content of the Lijiang–Yongsheng primary melt indicated a high mantle melting temperature (T_p).

When the equation of Herzberg and Asimow (2015) [24] was employed, the T_P value for the Dali–Binchuan primary melts ranged from 1506 to 1592 °C (averaging 1538 ± 24 °C), which was consistent with the T_P values (1590 °C; 17) of Daying picrite located ~12 km southwest of Binchuan Town [12]. Those of the Lijiang–Yongsheng and Yumen–Muli primary melts were estimated to be 1550–1660 °C and 1555–1637 °C, slightly higher than those for the Dali–Binchuan primary melts (Figure 9). Here, our re-estimated T_P values of Lijiang–Yongsheng, Dali–Binchuan, and Yumen–Muli primary magma surpassed the average ambient mantle temperature (1318 ± 32 °C) by about 200 °C [59], indicating the melting of the anomalously hot mantle. Notably, the estimated T_P for the Lijiang, Yongsheng, Yumen, Ertan, Muli, and Dali–Binchuan picrites decreased away from Lijiang and Yongsheng, suggesting that the Lijiang and Yongsheng areas were the center of the ELIP.

5.3. Olivine Crystallization Temperature: T_{Al-in-olivine} and T_Sc-Y

Olivine phenocrysts have several Cr-spinel inclusions [10,16], implying a simultaneous crystallization of olivine and Cr-spinel [33]. According to the Al concentrations between coexisting olivine and Cr-spinel, the olivine crystallization temperatures for the ELIP picrites could be calculated using an Al-in-olivine thermometer. The spinels in this study exhibited Cr no. values (Cr/(Cr + Al), mol) in the range of 0.570–0.685 (Table S3), and they were all within the calibration range of the Al-in-olivine thermometer (0–0.69; 40).

The calculated crystallization temperatures for olivine phenocrysts from different ELIP picrites generally increased with the increase in Fo (Figure 10a–h), which was consistent with the expected cooling during magma evolution. Notably, the Dali, Binchuan, Wulongba, Wuguijing, Tanglanghe, Maoniuping, Lijiang, and Ertan picrite samples yielded similar crystallization temperatures (T_{Al-in-olivine}) for the olivine phenocrysts (Figure 10i), with averages of 1305 ± 74 °C, 1282 ± 30 °C, 1328 ± 88 °C, 1301 ± 55 °C, 1247 ± 41 °C, 1260 ± 45 °C, 1283 ± 75 °C, and 1245 ± 48 °C, respectively (Table S3). This study’s T_{Al-in-olivine} for ELIP picritic-magma was similar to that (1378 ± 55 °C) of Daying picrite located ~12 km southwest of Binchuan Town, indicating that our estimates of T_{Al-in-olivine} are valid.

We also estimated the magma crystallization temperatures using an Sc-Y geothermometer [41], assuming that an olivine-melt equilibrium pressure of 0.1 GPa, the T_Sc-Y for olivine phenocrysts from different ELIP picrites also correlated positively with Fo (Figure 11a–g). The calculated crystallization temperatures (T_Sc-Y) for the Lijiang, Maoniuping, Dali, Wulongba, Wuguijing, Tanglanghe, and Yumen picrites were 1337–1540 °C (1455 ± 49 °C), 1300–1416 °C (1366 ± 31 °C), 1249–1442 °C (1379 ± 52 °C), 1310–1442 °C (1376 ± 36 °C), 1291–1419 °C (1352 ± 41 °C), 1318–1389 °C (1358 ± 22 °C), and 1301–14,670 °C (1375 ± 49 °C), respectively (Table S2). Furthermore, we observed that the mean T_Sc-Y for primitive olivine (Fo ≥ 90) from the Dali–Binchuan picrites (1413 ± 46 °C) was slightly lower than that for the Lijiang–Yongsheng (1468 ± 40 °C) and Yumen (1444 ± 23 °C) picrites (Figure 11h), indicating heterogeneous plume-head mantle sources.

Above all, the olivine crystallization temperature (T_{Al-in-olivine} and T_Sc-Y) was lower than the T_P value (1550–1660 °C) from different ELIP picrites in this study, but it was higher than that of MORB (1270 °C; 40), which was consistent with the role of a mantle thermal plume for the ELIP.

5.4. Geodynamic Scenario for the Generation of the ELIP Picrites

The Dali–Binchuan primary magma composition had a high Al₂O₃ content due to partial melting at low pressure (e.g., [19,25,58]), while the Lijiang–Yongsheng primary melt had low Al₂O₃ content (Figure 8b) due to the high-pressure melting of peridotite [59]. In addition, lavas from Lijiang–Yongsheng were characterized by high Gd–Yb and La–Yb ratios (Figure 2d), which may be attributed to the partial melting of garnet–peridotite [9,25,30]. On the contrary, the Dali–Binchuan lavas contained lower Gd–Yb and La–Yb ratios, as they could have been produced in the garnet–spinel transition zone of a peridotitic mantle (Figure 2d).

Specifically, as shown in Figure 12, Dali–Binchuan lavas began melting at lower pressures (~4 GPa) than the Lijiang–Yongsheng lavas did (~5 GPa, Figure 12a). The shallower melting depth for the Dali–Binchuan lavas agreed with the Dali lavas’ slightly lower mantle potential temperatures (Figure 9) because melting began at a shallower point when the source was cooler [60]. The final melting pressures for the Dali–Binchuan lavas (ranging from 1 to 2 Gpa) were lower than those for the Lijiang–Yongsheng lavas (>2 GPa, Figure 12b), further suggesting that the Dali–Binchuan lavas were generated in the garnet–spinel transition zone of peridotitic mantle. Above all, the lithosphere of the Dali–Binchuan area was thinner than that beneath the Lijiang–Yongsheng areas if the estimated final melting pressure was representative of the depth of the lithosphere–asthenosphere boundary.

6. Conclusions

(1) The positive correlation between the TiO₂ and CaO content, the negative correlation between the MgO and CaO content, and the low FC3MS values of different ELIP picrites indicate a derivation from a peridotite-dominated mantle source. The peridotite mantle source is consistent with the high 100 × Mn–Fe and Mn–Zn ratios and low 10,000 × Zn–Fe values in the olivine phenocrysts.

(2) Based on the primitive magma compositions, the Lijiang–Yongsheng, Dali–Binchuan, and Yumen–Ertan melts in the ELIP could have been produced via mantle melting at 1550–1660 °C, 1506–1592 °C, and 1555–1669 °C, respectively. Those melts started to crystallize olivine at ~1400 °C, based on the Sc-Y geothermometer.

(3) The Lijiang–Yongsheng primary magma exhibited high MgO content (19.49–23.87 wt.%), Gd–Yb (3.05 ± 0.33) and La–Yb (14.61 ± 3.77) ratios, and mantle melting temperatures (1599 ± 34 °C), but it exhibited low Al₂O₃ content (8.24 ± 0.54 wt.%), which may be attributed to the partial melting of garnet–peridotite.

(4) The Dali–Binchuan lavas with high Al₂O₃ content (9.86 ± 0.67 wt.%), and low Gd–Yb (1.91 ± 0.81) and La–Yb (5.88 ± 2.88) ratios, may have been produced in the garnet–spinel transition zone of a peridotitic mantle.