1 INTRODUCTION

Extensive and voluminous Cenozoic basalts are widely distributed in the Southeast Asian region (Fig. 1; Ho et al., 2000; Fedorov and Koloskov, 2005; Sun et al., 2009; Zou and Fan, 2010; Wang et al., 2012; Zhang et al., 2018), including the South China Sea (SCS) Basin (Yan et al., 2008; Zhang et al., 2017, 2018), Hainan Island, Leizhou Peninsula (Zou and Fan, 2010; Wang et al., 2012; Li et al., 2013; Liu et al., 2015), Fujian-Zhejiang coast (Ho et al., 2003; Huang et al., 2017), Taiwan Island (Tian et al., 2019), and parts of Vietnam (An et al., 2017; Hoang et al., 2018) and Thailand (Yan et al., 2018). The detailed geological setting of Southeast Asia is present in Supplementary 1. The SCS has shown the evidence of the diverse arrays of spatially and temporally complex tectonic processes, including continental rifting, seafloor spreading, subduction, and terrane collision, which make up a complete Wilson cycle (Zhou et al., 2009; Li et al., 2014; Yan et al., 2018). The SCS is a highly complex research area worthy of in-depth study. Several tectonic dynamic hypotheses have been proposed in previous studies to explain the Cenozoic volcanic activities in SCS and its surroundings on the basis of geochemical and geophysical evidence. These hypotheses include the upwelling of the mantle plume (Zhou et al., 2009; Yan et al., 2018), the retreat and withdrawal of the subducted Paleo-Pacific plate (Shi and Li, 2012), tectonic extrusion related to the India-Eurasia collision (Briais et al., 1993), seafloor tension as a result of the subduction of proto-SCS (Hall, 2002), and mantle-plume-spreading-ridge interaction model (Yan et al., 2018; Yu et al., 2018; Zhang et al., 2018). However, the origin of volcanic activities in SCS and its surrounding areas remains controversial or ambiguous. For example, the dynamic mechanism that triggers the expansion of SCS is yet unknown, and the connection between the Hainan mantle plume and the expansion of SCS is unclear. Moreover, the influence of the Hainan mantle plume on volcanic activities in Southeast Asia has not been studied in detail. Given these knowledge gaps, analysis of the geochemical data of Cenozoic volcanic rocks covering a wide range of areas surrounding the SCS is necessary. A geochemical dataset contains tens of indicators quantifying the contents of trace elements and isotopes in volcanic rock samples. Many of them exhibit strong correlations with each other because the generating mechanisms of volcanic rocks are few. Hence, it is necessary to extract principal components (PCs) from these multidimensional data, which reflect independent factors contributing to the rock contents. Herein, we employed a suitable research method, principal component analysis (PCA), first proposed by Jimenez-Espinosa et al. (1993), to explore the characteristics and properties of the mantle sources of volcanic activities in SCS and its surrounding regions.

Even in the era of big data, PCA remains an important tool to extract the hidden information from datasets in high-dimensional space owing to its effective dimension-reduction capacity (Zhou et al., 2018). It can easily identify the most "main" contributors and simplify the structures of a dataset, remove noise and redundancy, reduce the number of dimensions of the original data, and make the information hidden behind complex data of high-dimensional space more visible (Abdi and Williams, 2010). Therefore, it is widely used and regarded as one of the most valuable applications of linear algebra (Zhao, 2016). Improvements in theory and advancement in computer technologies have enabled the use of PCA to solve more geoscience problems. For example, PCA has been employed to visualize multiband remote sensing data (Shimizu et al., 1997), investigate the hydrochemical characteristics of groundwater systems (Peng et al., 2015), analyze atmospheric aerosols in Mexico City (Miranda et al., 2000), and describe the recurrent snowmelt pattern of multiyear remotely sensed snow cover (Woodruff and Qualls, 2019).

In this study, we employed PCA to mine geochemical data (e.g., trace elements and isotopes) and determine mantle sources in SCS and its surrounding areas. It is challenging to determine independent potential sources due to the complexity and diversity of chemical-isotopic indicators and sources (Chen et al., 2020). PCA transforms the original tens of variables (i.e., the contents of trace elements or isotopes) into several independent PCs, which are easier to explain and are linear combinations of the original variables. Each PC represents one independent geochemical or isotopic source for the samples (Chen et al., 2020). The success of our analyses proves that PCA is an effective, feasible, and applicable technique for dissecting geochemical data and understanding the nature of mantle sources of volcanic activities. More importantly, we introduced a suitable study method for the analyses of geochemical data.

2 ANALYTICAL METHOD

PCA is a spatial projection process. The space formed by the original indicators is usually nonorthogonal, indicating a correlation between the original indicators, whereas the space formed by the derived indicators is orthogonal, indicating that the derived indicators are independent of each other (Jimenez-Espinosa et al., 1993).

Herein, PCA for geochemical data was conducted as follows (Zhou et al., 2018): (1) we obtained a correlation matrix for the original indicators, assessed the inner dependence of these indicators was, and preliminarily determined potential mantle sources reflected by the indicators; (2) we performed PCA calculations, including the standardization of the data, determination of the linear transformation between the PCs and original indicators, evaluation of the proportion of each PC to explain the total data variance, and computation of the PC values for each sample; (3) we selected only the PCs explaining over 5% of the data variance to achieve dimensionality reduction; (4) we determined which magma source from those determined in (1) is most sensitive to each selected PC according to the transforming relationship between PC and the original indicators; and (5) we evaluated the spatial influence range and degree of each magma source according to the spatial distribution of the PC values.

Assuming w_ij is the content of j^th trace element measured from i^th rock sample, the corresponding standardized indicator x_ij can be expressed as

where w_j and s_j are the mean and standard deviation of the content of j^th trace element of all samples, respectively. Then, the correlation matrix C=(c_jk)_N×N is computed, which is an (N×N) square and symmetrical matrix whose jk entry is the correlation between the j^th and k^th columns of X= (x_ij)_M×N. The eigenvalues (λ) and corresponding eigenvectors (u) are then determined from the following equation:

For an N-dimensional dataset, N combinations of eigenvalues and eigenvectors can be determined. We only retain several of them with a significant eigenvalue (> 5% total eigenvalue). The k^th new indicators (i.e., PC values) for the i^th sample can be calculated as follows:

where u_jk is the j^th element of the k^th eigenvector determined from Eq.2. Substituting Eq.1 into Eq.3, we obtain a linear relationship between PCs and the original indicators.

3 RESULT 3.1 PCA of trace elements

Fifteen trace element indicators (Rb, Nb, Ba, Hf, Th, U, La, Ce, Nd, Sm, Eu, Tb, Yb, Lu, and Y) are selected (Supplementary 1) for volcanic rock samples (< 33 Ma) collected from China (SCS, Hainan Island, Fujian-Zhejiang coast, Taiwan Island), and Vietnam and Thailand. (Supplementary Table S1).

3.1.1 Correlation analysis of trace elements as original indicators

Based on the correlation matrix of the trace elements obtained from the present study (Supplementary Table S2), it is revealed that 15 trace elements can be divided into three significant groups. The first group includes 11 trace elements (Rb, Nb, Hf, Th, U, La, Ce, Nd, Sm, Eu, and Tb) with correlation coefficients exceeding 0.6; in fact, most of the correlation coefficients obtained reached 0.9. The 11 elements are strongly positively correlated and generally enriched or depleted simultaneously, which are quite consistent to oceanic island basalt (OIB)-type magmatic source (Sun and McDonough, 1989). Thus, we can preliminarily speculate that this group represents an OIB-type mantle source. The second group includes three trace elements, including Yb, Lu, and Y, whose correlation coefficients exceed 0.8. This group may represent enriched mid-ocean ridge basalt (E-MORB)-type mantle sources because E-MORB can generally cause the enrichment of Yb, Lu, and Y contents (Sun and McDonough, 1989). The third group contains only Ba element, which is not significantly correlated with any of the 14 other elements (correlation coefficients < 0.21). This result reveals that Ba element is a good independent indicator. Thus, Ba element may reflect a subduction-related mantle source (e.g., subduction-related fluids or sediments) (Leeman et al., 1994; Jicha et al., 2009; Hanyu et al., 2012).

3.1.2 PCA calculations and dimensionality reduction

Fifteen PCs with linear expressions for the trace element contents is obtained after data normalization and spatial projection calculations (see the Method part for detail). The proportions of data information explained by each PC are respectively 65.9%, 19.3%, 6.5%, 3.2%, and 1.8%, etc. PCs accounting for over 5% of the data information are selected for analysis. In fact, only the first three PCs explain 91.7% of the data information. Table 1 shows the trace element coefficients of these three PCs, which are designated PC1, PC2, and PC3.

Corresponding to PC1, the Rb, Nb, Hf, Th, U, La, Ce, Nd, Sm, Eu, and Tb coefficients have largest positive values compared with those corresponding to PC2 and PC3. The Yb, Lu, and Y coefficients corresponding to PC2 have larger positive values compared with those corresponding to PC1 and PC3. The Ba coefficient corresponding to PC3 has a large positive value (Table 1). Thus, the trace elements affecting PC1 are Rb, Nb, Hf, Th, U, La, Ce, Nd, Sm, Eu, and Tb, those mainly affecting PC2 are Yb, Lu, and Y, and that affecting PC3 is Ba. This analysis is consistent with the findings described in Section 3.1.1. Thus, PC1, PC2, and PC3 may represent an enriched OIB-type mantle source, a depleted MORB-type mantle source, and the involvement of subduction-related fluid/sediment, respectively.

3.1.3 Comparison of the combined and standard samples

We merge similar samples with proximate sampling sites and identical lithological features and ages into one combine sample, by taking the average value of their PCs as the corresponding PC value of the combined sample. As a result, 110 combined samples are obtained. Five standard samples, including C1 chondrite, primary mantle, normal MORB (N-MORB), enriched MORB (E-MORB), and OIB, are also selected for comparison. Supplementary Table S3 lists the PC values calculated from the trace elements of the five standard and 110 combined samples. The PC1 value of OIB (1.89) is much higher than its PC2 and PC3 values (0.03 and -0.51). In addition, the PC2 values of N-MORB and E-MORB (2.24 and 0.87) are higher than their PC1 and PC3 values (-2.93 and -0.19; -3.01 and -0.22) (Supplementary Table S3). This result proves that PC1 and PC2 are closely related to OIB-type and MORB-type mantle sources, respectively, and that PC3 is not strongly correlated with the five standard samples. The results of this analysis are consistent with the results discussed earlier in Sections 3.1.1 and 3.1.2. In addition, the PC2 values of 1, 19–23, and 25 combined samples (1.25–2.41) are higher than their PC1 and PC3 values (< 0; Supplementary Table S3), which indicates that these samples are significantly affected by MORB-type mantle sources. 2–11, 33–55, 57–59, 66–72, 76–87, and 89–110 combined samples are characterized by higher PC1 values compared with their PC2 and PC3 values (Supplementary Table S3), thus suggesting that these samples are mainly influenced by OIB-type mantle sources.

We use cluster analysis to understand the similarity between the combined and standard samples. In the cluster maps obtained (Fig. 2), samples with similar geochemical features are clustered on the same branch. Most volcanic rocks from the SCS and its surrounding regions are clustered on the same branch as the OIB standard sample (Fig. 2), which indicates that deep enriched OIB-type magmatic sources widely affect the volcanic activities of the SCS and its surrounding regions. 1, 19–23, and 25 combined samples are clustered on the same branch as the N-MORB and E-MORB standard samples (Fig. 2), thus revealing that these volcanic samples are mainly affected by MORB-type spreading ridge mantle sources.

Scatter diagrams of the 110 combined and five standard samples (Fig. 3) are constructed based on the PC values (Supplementary Table S3) and clustering analysis results (Fig. 2). These diagrams intuitively show that most volcanic rocks in the SCS and its surrounding regions (black crosses in Fig. 3) are clustered together, which is consistent with the PC values of OIB. This finding confirms that volcanic activities around the SCS are closely related to an OIB-type mantle source.

3.2 PCA of Sr-Nd-Pb isotopic ratios

Five isotopic indicators (⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd, ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁸Pb/²⁰⁴Pb) were examined from 623 volcanic rock samples (< 33 Ma) obtained from China (SCS, Hainan Island, Fujian-Zhejiang coast, Taiwan Island), Vietnam and Thailand. (Supplementary Table S1).

3.2.1 Correlation analyses of the original indicators (isotopic ratios)

Analysis of the correlation matrix of five isotopic ratios of 623 volcanic samples (Supplementary Table S4) reveals that these ratios could be divided into two groups. The first group includes the isotopic ratios of ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb, whose correlation coefficients exceed 0.8. This result indicates that these Pb isotopic ratios are strongly positively correlated. The second group includes the isotopic ratios of ⁸⁷Sr/⁸⁶Sr and ¹⁴³Nd/¹⁴⁴Nd, whose correlation coefficients are negative in value (Supplementary Table S4).

3.2.2 PCA calculations and dimensionality reduction

After data normalization and spatial projection calculation, we obtain five PCs with linear expressions for the isotopic ratios. The proportions of data information explained by each PC are respectively 79.9%, 13.6%, 3.7%, and 2.4%, etc. PCs accounting for over 5% of the data information are selected for further analysis; thus, only the first two PCs, which could explain 93.5% of the data information, are considered. Table 2 shows the Sr-Nd-Pb isotope coefficients corresponding to these two PCs recorded as PC1 and PC2.

PC2 is characterized with relatively low ⁸⁷Sr/⁸⁶Sr (-0.665), ²⁰⁷Pb/²⁰⁴Pb (0.195), and ²⁰⁸Pb/²⁰⁴Pb isotopic coefficients (0.348), a relatively high ¹⁴³Nd/¹⁴⁴Nd coefficient (0.446), and a similar ²⁰⁶Pb/²⁰⁴Pb isotopic coefficient (0.447) in comparison with those of PC1 (Table 2; ⁸⁷Sr/⁸⁶Sr=0.397, ¹⁴³Nd/¹⁴⁴Nd=-0.439, ²⁰⁶Pb/²⁰⁴Pb= 0.450, ²⁰⁷Pb/²⁰⁴Pb=0.470, ²⁰⁸Pb/²⁰⁴Pb=0.476), thus suggesting that PC2 represents a relatively depleted MORB-type mantle source (Zindler and Hart, 1986; Sun and McDonough, 1989). PC1 may reflect a slightly enriched OIB-type mantle plume influenced by an enriched mantle 1 (EM1) -type mantle source (Zindler and Hart, 1986; Sun and McDonough, 1989).

3.2.3 Comparison of combined and standard samples

By taking the average value of their PCs as the corresponding PC value of the combined sample, we merged similar samples into 61 combined samples. Supplementary Table S5 gives the PC values calculated from the Sr, Nd, and Pb isotopic ratios of the 61 combined and five standard samples. OIB and EM1 have relatively high PC1 values (-1.968–5.547; -0.189–2.260; Supplementary Table S5), while MORB has a high PC2 value (-2.536–1.308; Supplementary Table S5). Therefore, we suppose that PC1 likely represents an enriched OIB- and EM1-type mantle source and that PC2 represents a depleted MORB-type spreading ridge. This conclusion is consistent with the results described in Section 3.2.2. 1–4, 6–7, 9–12, 16–27, 32–43, 45–46, and 49–61 combined samples have relatively significantly higher PC1 values relative to PC2 values, which indicates that these samples are affected by the OIB- and EM1-type mantle source (Supplementary Table S5; Fig. 4). Scatter diagrams and cluster analysis directly illustrate that these samples are clustered together, consistent with the PC values of OIB and EM1 (black cross in the Fig. 5). These results indicate the samples are related to OIB- and EM1-type mantle sources.

4 DISCUSSION

Trace elements were analyzed via PCA, and three PCs that explain 65.9%, 19.3%, and 6.5% of the variance found were extracted. They include an enriched OIB-type mantle plume source with fairly high trace element content (PC1), a depleted MORB-type mantle source featuring spreading ridges with enriched Yb, Y, and Lu contents (PC2), and subduction-related fluids/sediments with large variations in Ba contents (PC3; Table 1). PCA was employed to analyze the Sr, Nd, and Pb isotopic ratios, and two PCs could explain 79.9% and 13.6% of the observed variance. The two PCs include a relatively enriched OIB-type mantle plume source containing large amounts of EM1-type components with high ⁸⁷Sr/⁸⁶Sr, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb isotopic coefficients but a low ¹⁴³Nd/¹⁴⁴Nd isotopic coefficient (PC1), and a typical depleted MORB-type mantle source with relatively low Pb and Sr isotopic coefficients and a high Nd isotopic coefficient (PC2; Table 2).

The OIB-type mantle plume revealed by PC1 values obtained from PCA calculations of trace elements and isotopic ratios is the most important magma source dominating volcanic activities in SCS and its surrounding areas. OIB-type volcanic activities around SCS are most likely due to the Hainan mantle plume (Hoang et al., 1996; Yan et al., 2006, 2018; Zhou et al., 2009; Zou and Fan, 2010; Wang et al., 2012, 2013; Huang et al., 2013; Liu et al., 2017). We compared the PC1 values of volcanic rocks in different periods to understand the degree of influence of the Hainan mantle plume on volcanic activity in different periods in SCS and its surrounding areas (Supplementary Table S6). The PC1 values of nine 0–12.6-Ma Hainan Island combined samples are consistently positive and range from 0.38 to 4.62; In contrast, the PC1 values of fourteen 12–33-Ma Hainan Island samples are negative (-2.66–-0.84) (Supplementary Table S6; Wang et al., 2012; Liu et al., 2015). This result indicates that an enriched OIB-type mantle plume plays a more significant role in Hainan volcanic activities of 0–12.6 Ma than in during the period of 12–33 Ma (Supplementary Table S6). The above observations reveal that younger combined samples collected in the same area have higher PC1 values than older samples. Similar phenomena have been observed in SCS and its surrounding areas, including the seamounts of SCS, expansion center of SCS, Zhejiang-Fujian coast, Thailand, and Vietnam (Supplementary Tables S6 & S7). In detail: (1) in the seamounts of SCS, eight younger combined volcanic rocks (3– 8 Ma) have high positive PC1 values ranging from 1.38 to 7.63, whereas three older volcanic rocks (16.5 Ma) have relatively low positive PC1 values (0.75–1.27) (Supplementary Table S6; Yan et al., 2008, 2015); (2) in the expansion center of SCS, the 3.8–12.8 Ma combined volcanic samples have high PC1 values (1.29–7.63), whereas the three older volcanic rocks (15–17 Ma) have relatively low negative PC1 values (-2.90–-2.01) (Supplementary Table S6; Zhang et al., 2017, 2018); (3) in the Zhejiang-Fujian costal region, younger 0–16.2 Ma volcanic rocks have significantly higher PC1 values (1.27–7.08) than the three older volcanic rocks (16– 18 Ma), which have low positive or even negative values (-2.86–-2.60) (Supplementary Table S6; Huang et al., 2017); (4) in Thailand and its surrounding areas, older volcanic rocks (8.8–11 Ma) have negative PC1 values (-0.18), whereas younger volcanic rocks (0.6– 3.3 Ma) have high positive PC1 values ranging from 0.44 to 4.62 (Supplementary Table S6; Yan et al., 2018); (5) in Vietnam, some older volcanic rocks (7.0–16.5 Ma) have low PC1 values ranging from -1.81 to -0.51, whereas some younger volcanic rocks (0.2–9.6 Ma) have significantly high positive PC1 values (1.07–4.93) (Supplementary Table S6; Hoang et al., 2018). Thus, the observed phenomenon is not accidental, instead, quite common in our study area. We speculate that the influence of the Hainan mantle plume on the Cenozoic volcanic activity in this area gradually increases. Specifically, the influence of the Hainan mantle plume on nearby young volcanic activities less than 13 Ma is much stronger than that of volcanic activities greater than 13 Ma (Supplementary Tables S6 & S7; Supplementary 1).

PCA is a powerful tool that can reveal the potential mantle sources of Southeast Asian volcanic activities based on geochemical or isotopic indicators. PCA of the trace elements of volcanic samples from Hainan Island, seamounts in SCS, and the expansion center of SCS yielded two PCs that explain over 85.2% of the observed variance; these PCs are represented by the red (PC1; enriched OIB-type mantle plume) and blue (PC2; depleted MORB-type spreading ridge) lines in Fig. 6, respectively. The ordinate of Fig. 6 represents the relative degree of influence of the two PCs on the trace element compositions of the volcanic samples. < 13-Ma volcanic samples from Hainan Island (0–12.6 Ma), seamounts of the SCS (3–8 Ma), and the expansion center of the SCS (7.4–12.8 Ma) have similar and high PC1 values (0.38–4.62, 1.38–7.63, 5.14), indicating that these samples are affected to the same degree by the OIB-type mantle plume (red lines in the Fig. 6a). The < 13-Ma samples from the expansion center of SCS (7.4–12.8 Ma) have high PC1 (5.14) and PC2 (2.25) values, thus indicating that these samples are simultaneously influenced by the OIB-type mantle plume (red lines in the Fig. 6a) and MORB-type spreading ridge (blue line in the Fig. 6a). Therefore, OIB-type mantle plumes play a significant role in the formation of volcanic activities less than 13 Ma in the expansion center of SCS (Fig. 6a). Volcanic activities in the expansion center of SCS are affected by MORB-type spreading ridges and OIB-type mantle plumes, validating the mantle-plume-ridge interaction model in the expansion center of SCS. In addition, volcanic activities in the expansion center of SCS in the periods of 7.4–12.8 and 15–17 Ma showed similar and relatively high PC2 values (2.25; 1.54–2.13; Fig. 6a–b), indicating that MORB-type spreading ridges play an equally important role in volcanic activities in the expansion center of SCS during these periods. However, the low PC1 values of the 15–17-Ma samples (-2.90–-2.01) but a much higher PC1 value of the 7.4–12.8-Ma samples (5.14; Fig. 6a–b) indicate that an OIB-type mantle plume begins to appear at the expansion center of SCS after 13 Ma, which strongly proves that the extension of SCS is not triggered by an OIB-type mantle plume. The extension of SCS appears to occur before the appearance of the mantle plume, which invalidates the notion that the Hainan mantle plume causes the expansion of SCS.

PCA of the isotopic ratios of volcanic samples from Hainan Island, seamounts in SCS, and the expansion center of SCS revealed two PCs that explain over 93.5% of the observed variance. These PCs are represented by the red (PC1; an enriched OIB-type mantle plume containing large amounts of EM1-type components) and blue (PC2; a depleted MORB-type mantle source) lines in Fig. 7, respectively. The ordinate of Fig. 7 represents the relative degree of the influence of the two PCs on the isotopic ratios of the Hainan Island and SCS volcanic samples. Isotopic PCA calculations showed the same results for the trace elements. The volcanic samples less than 13 Ma obtained from the expansion center of SCS (7.4–12.8 Ma) are influenced by PC1 (1.342) and PC2 (0.799) (Fig. 7), thus confirming the mantle-plume–ridge interaction at the expansion center of SCS. The PC2 values of the 7.4–12.8-and 15–17-Ma volcanic samples from the expansion center of SCS are 0.799 and in the range of -0.449 to 1.001, respectively. The similar PC2 values suggest that MORB-type spreading ridges have almost similar effects on the production of 7.4–12.8-Ma and 15–17-Ma volcanic activity. The PC1 values of the 7.4–12.8-Ma samples obtained from the expansion center of SCS (1.342) are significantly higher than those of the 15–17-Ma samples (-3.875–-1.224), thus suggesting that OIB-type mantle plumes have more significant effects on the 7.4–12.8-Ma volcanic activities than on the 15–17-Ma volcanic activities. This result shows that the expansion of SCS is not caused by the Hainan mantle plume.

By the method of the PCA and cluster analyses of trace elements and isotopic ratios, volcanic activities in these regions, which extend from Zhejiang-Fujian in the north to Vietnam-SCS in the south and from Thailand in the west to Southern Taiwan Island in the east, may be affected by the Hainan mantle plume (Supplementary 1). The successful analyses reveal that PCA is effective and feasible for analyzing the geochemical data of volcanic samples and discovering geochemical indicators and mantle sources of volcanic activities. Thus, PCA is a suitable statistical method for analyzing geochemical data.

5 CONCLUSION

PCA was employed to analyze geochemical data and interpret the characteristics and types of mantle sources of Cenozoic volcanic activities in SCS and its surrounding regions. We analyzed 15 trace elements and 5 isotopic indicators of 623 volcanic rock samples obtained from China (SCS, Hainan Island, Fujian-Zhejiang coast, Taiwan Island), Vietnam and Thailand. to characterize the geochemical properties of the volcanic rocks, determine the types of mantle sources, and evaluate the degree of influence of each mantle source.

PCA revealed three PCs that explain 65.9%, 19.3%, and 6.5% of the total data variance based on the geochemical contents of the 15 trace elements. In addition, two PCs that explain 79.9% and 13.6% of the data variance were extracted based on the Sr-Nd-Pb isotopic ratios. The three main PCs identified are enriched OIB-type mantle plume, depleted MORB-type spreading ridge, and subduction-related fluid/sediment.

In the Southeast Asian region, the effect of the mantle plume on relatively younger volcanic activities (< 13 Ma) is greater than that on older volcanic activities (> 13 Ma) at the same location. Volcanic activities in these regions, which extend from Zhejiang-Fujian in the north to Vietnam-SCS in the south and from Thailand in the west to Southern Taiwan Island in the east, may be affected by the Hainan mantle plume. PCA was employed to verify the mantle-plume–ridge interaction of volcanic activities beneath the expansion center of SCS, and the results refute the hypothesis that the tension in the SCS is triggered by the Hainan plume.

The results of this study demonstrate that PCA is effective and feasible for analyzing the geochemical data of volcanic samples and discovering the geochemical indicators and mantle sources of volcanic activities.

6 DATA AVAILABILITY STATEMENT

All data generated and/or analyzed during this study are available in the figshare repository (https://doi.org/10.6084/m9.figshare.21780764).

Electronic supplementary material

Supplementary material (Supplementary 1 and Supplementary Tables S1–S7) is available in the online version of this article at https://doi.org/10.1007/s00343-022-1407-8.