Trace element tectonic discrimination of granitoids: inspiration from big data analytics

ABSTRACT The tectonic environment of granitoids has always been a concern of the academic community. The Nb vs. Y and Rb vs. Nb+Y diagrams have had a substantial impact. The present work uses more than 110,000 granitoid samples (SiO2 ≥ 56%) from the globally shared database to discuss the validity and also explain why these diagrams used for discrimination between the different tectonic settings of granitoid rocks. The amount of data from the spreading center is sparse and the data are highly scattered and so, the present study focuses mainly on granites from the ocean islands and convergent margins tectonic environments. On the TAS diagram most of the ocean island data are alkaline series and trachy-basalt series, and some are bimodal. In contrast, the granitoids on the convergent margin are mainly sub-alkaline. This work shows that the tectonic discrimination diagrams of granitoids remain valid, and only the boundaries need to be slightly adjusted.


Introduction
The tectonic environment of granitoids has been the topic of extensive study for a very long time.When the tectonic environment of basalt was introduced as a topic at the beginning of plate tectonic theory, the tectonic environment of granitoids was also raised (Pearce & Cann, 1973;Pearce & Norry, 1979;Wood, 1980;Shervais, 1982;Pearce et al., 1984aPearce et al., , 1984b;;Vermeesch, 2006).The introduction of big data has ushered in a new age of scientific research as we enter the twenty first century, and its most significant developments are as follows: (1) the replacement of deterministic thinking with uncertainty thinking; and (2) the introduction of big data methodologies.
People constantly seek to obtain a clear conclusion in their study.For example, we must always define the tectonic setting when discussing basalt research to present certainty in thought.When analyzing a hitherto unidentified set of basalts, we classify them as convergent margins, marking a transition from uncertainty to certainty.
Does the characteristic basalt of the convergent margin derive from the subduction zone?Obviously, the answer is negative, although it might also be related to the conditions within the plate, which can only be confirmed by additional examination.What is the subduction process if it is found that the basalt was produced in the subduction zone?What effects does subduction have on basalt?Are there more factors at play?Questions such as the origins of the other factors require additional inquiry.
Since almost all of our previous studies were derived by induction, by the process of inference from the individual to the general, research suggests that induction from the individual to the general is impossible (Popper, 1979).Tectonic discrimination diagrams are the most typical product of the induction approach.Basalts of unknown tectonic origin were allocated to a certain tectonic environment after a series of studies, and a definitive result was established as a result of induction.There are two problems: due to the limitations of induction, the outcomes of the research, which is described by specific areas or regions, may not be precise.The discrimination diagram is also a summary of information from certain locations; it lacks the meaning of universal quantification and cannot resolve the questions surrounding basalt across all regions.Thus, the conclusions of the majority of research involving discrimination diagrams are unreliable.They have not been falsified and do not have a universally quantified meaning.How can this problem be resolved?One solution involves the application of big data strategies, especially comprehensive data models.For example, global databases have been used to study the tectonic discrimination diagrams of basalts (Wang et al., 2016;Yang et al., 2016aYang et al., , 2016b;;Chen et al., 2017;Di et al., 2017;Wang et al., 2017aWang et al., , 2017b)).Research shows that the discrimination diagrams currently used in science have fundamental problems, and there are only a few discrimination diagrams that still in use and applicable.
Regarding the tectonic setting of basalt, we have conducted extensive research, and while some of the obtained results are encouraging, they require further investigation.The tectonic environment of granitoids is more complex than that of basalt, since basalt is derived from the mantle, while granitoids are mainly derived from the crust, and only a few granitoids can be derived from mantle sources.The mantle is relatively simple and uniform, while the crust is relatively complicated and heterogeneous in composition.The vast majority of prior theories and methods of tectonic discrimination diagrams for basalt and granitoid are only applicable to magmatic rocks in ocean basins and not continental crust.For this reason, this paper discusses only the tectonic environment of granitoids associated with oceanic crust.Basalt occurs in three main environments: spreading centers, oceanic islands, and convergent margins.Since mid ocean ridge granitoids (MORG), i.e. spreading center granitoids, are unique, most of them are characterized by initial melt and belong to M-type granitoids.This study focuses on distinguishing between the tectonic environments of convergent margin granitoids (CMG) and ocean island granitoids.However, the latter was not included in the study of Pearce et al. (1984b).We believe that, after decades, the amount of data should be substantial, but it remains very low.The total available data points for the global spreading center are 27,696, but there are fewer than 1,000 samples of granitoids from the spreading center, with even fewer samples accessible.Identifying the MORG tectonic environment is relatively complex, and is discussed below.Only the problem of distinguishing between granite in the Ocean Island granitoid (OIG) and convergent margin granitoid (CMG) environments (Figures 1 and 2) is investigated in this paper.

Tectonic discrimination diagram of granitoids: the results of big data research
Numerous systems for establishing the tectonic discrimination diagrams of granitoids have been proposed in previous studies (Batchelor & Bowden, 1985;Frost & Frost, 2008;Frost et al., 2001;Harris et al., 1986;Laurent et al., 2014;Maniar & Piccoli, 1989;Pearce et al., 1984b;Schandl & Gorton, 2002), but the scheme of Pearce et al. (1984b) is the most effective and well known in academic circles; hence, this work focuses mostly on that scheme.
In this study, the total volume from three series of granitoid data given in the database is used, which is approximated as follows: 369 samples from spreading centers, 5,122 samples from ocean islands, and 45,786 samples from convergent margins (all data in this paper are from the PetDB and GEOROC databases).In Pearce's Y-Nb and (Y+Nb)-Rb discrimination diagrams, the convergent margin granitoids and the ocean island granitoids are sufficiently distinguished.When analyzing a substantial amount of data, the figure retains good discriminatory power, requiring only small adjustments (Figures 1 and 2).However, in the Y-Nb discrimination diagram, a portion of the convergent margin data plots on the ocean island region.More than 80% of these data come from the Apennine-Maghrebides Chain, which is characterized by its high Rb, Nb, and Y contents.These characteristics place it in the category of ocean island granitoids, but the high Rb content is indistinguishable from Ocean Island granitoids.However, Pearce's Y-Nb and (Y+Nb)-Rb discrimination diagrams show no identifiable pattern in the distribution of granitoids at the spreading center (Figure 3).
Figure 2 shows the data from different regions at the convergent margins.Based on the data from New Zealand, Izu, and Honshu, the earlier discrimination diagrams may be credible, but the boundary must be adjusted.The concentrations of large ion lithophile elements (LILEs), such as Rb, and high field strength elements (Nb, Y) in convergent margin granitoids are greater than those previously estimated, and the concentrations of the same elements in ocean island granitoids are likewise marginally higher.Therefore, a portion of the former OIG can be allocated to the CMG.
However, there are a few exceptional circumstances that must be considered.For instance, some Honshu data plot inside the ocean island region; however, it is hypothesized that these data originated from the rear arc basin and are not considered at this time.This section focuses on the Apenninic-Maghrebides Chain, and the majority of the data have typical OIG characteristics.
The spreading center granitoid (MORG) data acquired for this investigation are the least abundant and the most widely dispersed (Figure 3).Why? Pearce et al. (1984b) used data from several normal and anomalous ocean ridge granites to construct the ORG boundary of the discrimination diagrams, which indicated that MORG data are quite rare.

Characteristics of spreading center granitoids
The spreading center granitoids belong to the M-type, which is the most complex (including the initial melts and felsic dikes) among the granitoids with small numbers, as shown in Figure 3.The basaltic magma of the mid ocean ridge was produced by the partial melting of depleted peridotite and pyroxenite in the upper mantle.During magmatic ascent, this material normally undergoes a modest amount of fractional crystallization (Langmuir et al., 1992).Less than 4% of the data from various marine environments (including mid ocean ridges) have more than 56% SiO 2 (Byerly et al., 1976;Geist et al., 1995;Gunnarsson et al., 1998;Haase et al., 2005;Perfit et al., 1999;Regelous et al., 1999;Smith et al., 2003;Wanless et al., 2010Wanless et al., , 2011)).
The evolution of felsic magma in the oceanic crust is important because it reveals how the initial continental crust growth occurred.These felsic crusts must have developed from mafic magma or mafic rocks, either through distinct crystallization, partial melting, assimilation, or a mixture of these processes (Foley et al., 2002;Rollinson, 2008;Smithies et al., 2009).Some felsic magma was produced when hydrothermally altered oceanic crustal rocks partially melted (France et al., 2010;Koepke et al., 2007;O'Nions & Grönvold, 1973).However, several geochemical and experimental studies indicate that the fractional crystallization of basalt fluids appears to be the predominant process (Berndt et al., 2004;Grove & Juster, 1989;Haase et al., 2005;Wanless et al., 2010).The heat generated by magmatic fractional crystallization can cause amphibole to dissolve hydrothermally, releasing water and volatiles (Wanless et al., 2010(Wanless et al., , 2011)).Whether by partial melting or assimilation, variations in the composition of the initial melt and the course of dissociative crystallization have led to the formation of numerous rock types (Haase et al., 2005).Thus, understanding the assimilation-fractional crystallization mechanism in the oceanic crust is crucial to our overall comprehension of the genesis of evolutionary melts (Freund et al., 2013).
Figure 4 shows the distribution of spreading center granitoids.The spreading center is where the majority of mid ocean ridge basalts are generated, and it produces some granitoids, such as diorite, quartz diorite, granite, and oceanic plagioclase granite (Aumento, 1969;Casey, 1997;Dick et al., 2000;Expedition 305 Scientists, 2005;Silantyev, 1998).These granitoids are also found in the plutonic rocks of the majority of ophiolites (Coleman, 1977;Koepke et al., 2007;Zhang et al., 1992).The majority of these granitoids are found as intrusions in gabbro sequences, and the patterns and sizes of these intrusions vary substantially.In deep sea drill cores, some of these rocks are as small as a few millimeters to a few centimeters, resembling "felsic veins".In contrast, other felsic material emerges in gabbros and rock walls as meter-scale plutons or reticules as xenoliths (Juteau et al., 1988;Zhang et al., 1992).
In Oman's ophiolites, felsic intrusions of up to 100 meters in length are described as "late intrusions" (Lippard, 1986).According to petrographic investigations, the distinctive felsic plutonic rocks in the oceanic crust are formed of quartz, Na-or Ca-enriched plagioclase, amphibole, and apatite, as well as numerous accessory minerals, including iron and titanium oxides (Koepke et al., 2007).
In the ISMA classification of granitoids, the spreading center granitoid belongs to a class of M-type granites.M-type granite is the most complex of the granitoids; in a previous study on granitoid discrimination diagrams, this type was the most dispersed (Figure 3).
Compared to the convergent margin granitoids, the amount of spreading center granitoids is minuscule.In this study, there are 39,589 samples from convergent margin granitoids, 4,351 samples from ocean island granitoids, and 270 samples from spreading center granitoids from the three primary databases.In various situations, the number of granitoids (SiO 2 ≥ 56%) declines exponentially.The data for the spreading center granite are filtered according to the general range of the content of the normal granitoids' major elements to ensure that there is a certain amount of plagioclase, quartz, K-feldspar, and a small number of dark minerals (amphibole, biotite, etc.) in the rock.However, after such filtering, only approximately 190 samples of granite remain.This is 70% of the total spreading center data, which indicates that nearly one third of the data in the database may be incorrect.The initial melt was in the process of transitioning the rocks from partial melt to granitic magma, and thus are, not considered granitoids.Therefore, it is usual for the spreading center data on the tectonic discrimination diagram to be quite dispersed.
The relations between SiO 2 , K 2 O, and MgO for the major granitoid elements at the spreading center are shown in Figures 5 to 8. Figure 5 shows the relation between SiO 2 and the other major elements.(1) The MORG can be divided into two types: low silicon and high silicon with SiO 2 = 65%.Both types of granitoids are enriched in TiO 2 , particularly low-silicon granitoids, and the average TiO 2 content is 2%, which is higher than the average TiO 2 content of mid ocean ridge basalt (MORB).As shown in Figure 5, the high TiO 2 granitoids are also enriched in FeO, MgO, CaO, P 2 O 5 , MnO, etc. (2) The K 2 O content of the granitoids in the spreading center is similarly unusual, and the data are strangely dispersed (SiO 2 -K 2 O discrimination diagrams in Figure 5), and they may be roughly classified into two categories.If SiO 2 = 65% is still taken as the threshold, the low-silicon group is poor in K 2 O.With the increase in SiO 2 content, K 2 O increases slightly, and the two are positively correlated (Figure 5).The high-silicon group can be divided into two groups with K 2 O contents, and the low K 2 O content is approximately 1.0-1.5%.The high K 2 O content is approximately as high as 4%.There is no connection between the two groups, which obviously come from two different source areas.The basalt and granite in the ocean basin are sodium-enriched and potassium-poor, and there are few instances of exceptionally high K 2 O. (3) Al 2 O 3 at the spreading center is also separated into two groups, namely, low Al 2 O 3 , at approximately 13%, and high Al 2 O 3 , at greater than 15% (15-17%).
Figure 6 shows the relation between K 2 O and other major elements, and can also be divided into two categories based on the potassium content.Most of the other major elements displayed correlations with K 2 O in the low-potassium group, but the high- potassium group had no relations with other elements.This shows that the properties of the low-potassium and high-potassium groups are completely different.Low-potassium magma has the characteristics of typical granitic magma, whereas high-potassium magma has no regularity to follow.The aforementioned circumstances demonstrate that the source of granitoid melting in the spreading center is more complex than previously believed.In Figure 3, the spreading center data also plot within the ocean island and convergent margin areas, indicating that granitic melt may not be the exclusive source.In addition to the residues in the melt, it may be necessary to consider the degree of partial melting, the contamination of OIB-like source rocks, and some potassiumenriched mantle-derived magmas.
Because of the correlation between potassium and the other major elements in the low potassium group, we independently screened the low-potassium MORG (K 2 O ≤ 2.0%) to create a discrimination diagram (Figure 7).Therefore, we believe that the K 2 O ≤ 2.0% granitoid data in the spreading center are reliable, as they reflect the actual features of the spreading center granitoids, but the high-K type may not sufficiently determine the tectonic environment.
Figure 8 displays the relationship between SiO 2 and TiO 2 for MORG, OIG, and CMG.Although the number of MORG data points is very small, is the data are clearly divided into two groups: high silicon and low titanium, and low silicon and high titanium (Figure 6).Ti should be enriched in the residual phase during partial melting because it is a high field strength element.Why does the melt have such a high content?The anomalous Ti enrichment of MORG could be caused by the following reasons.(1) It may be related to the impurity of the sample, which may be caused by substances containing some residual phases in the felsic veinlets.We researched some ophiolites in Yunnan, and we have seen the remains of pyroxene and plagioclase (calcium enriched) in the granitic melt, which obviously cannot be the constituent minerals of granitic melt.(2) The granitoid is the result of basalt fractional crystallization.(3) The granitoid produced by partial melting may be able to extract ilmenite and leftover clinopyroxene from basaltic magma.The TiO 2 content of granitoids at the convergent margin is generally lower than 1%.Although the ocean island granitoid is relatively enriched in TiO 2 , it is generally less than 1.5%, while the TiO 2 content of the low silicon group in the spreading center is approximately 2%, which is higher than that of the ocean island granite.
Figure 9 shows the relation between Sr and Y for MORG, CMG and OIG.At the convergent margin, the granitoids are enriched in Sr but poor in Y, with Y often being lower than 30.The Sr values of the ocean island granitoids are slightly lower, but the  Y values are slightly higher, and the Y values are mostly lower than 60.The spreading center is extremely poor in Sr and enriched in Y; Sr is lower than 150, and Y is mostly concentrated between 50 and 150, suggesting that the spreading center granitoids formed with the lowest pressure and the highest temperature.

Current understanding of the apenninic-maghrebides chain
The majority of data from the convergent margin of the Apenninic-Maghrebides Chain plot in the ocean island area in tectonic discrimination diagrams (Figure 2).Based on Figure 10, the volcanic rocks are mostly of the alkaline series, followed by the trachy-basalt series, and the least common rocks are the subalkaline series.This founding indicates that the tectonic environment of the Apenninic-Maghrebides Chain is relatively complex.
Figure 12 shows the data of Cenozoic "orogenic" and "nonorogenic" volcanic rocks from Sardinia, Spain, Corsica-Estérel, Languedoc, the northern Apennines and the Alps (Lustrino et al., 2017).Iovine et al. (2018) also agreed that the volcanic rocks in this area belong to the trachy-basalt series and alkaline series, which are obvious intraplate products.
Based on their investigation of volcanic rocks in the Sardinian Trough, Lustrino et al. (2013) determined that the igneous rocks in this region can be divided into two groups.The first group, which is related to plate subduction (38-15 Ma), records related metasomatism, varying degrees of crustal contamination in shallow layers, the separation of crystallization, and partial melting of basic rocks.The second group has major and trace element characteristics with intraplate magmatic rocks (Iovine et al., 2017;Lustrino, 2010).
Some scientists believe that the rift is related to subduction, and subduction is responsible for the rift's formation (Iovine et al., 2017;Lustrino et al., 2017;Lustrino, 2010;Turco et al., 2006), which may possibly be the reason why the subduction environment is believed to dominate the region.Another theory holds that rifting is not related to subduction but rather to the mantle plume beneath the Mediterranean Sea (Bell et al., 2013) or the extension of the lithospheric mantle (Ferrari et al., 2012).Figure 13 shows two models of continental extension (Turco et al., 2006).
In Figure 14, we have plotted the Apenninic-Maghrebides chain data onto a geological map, using the classification principle determined by the distribution of the two types of volcanic rocks shown in Figure 10 ( Turco et al., 2006).The data with Na 2 O + K 2 O >10% belong to the alkaline series and trachy-basalt series, while data with Na 2 O + K 2 O ≤ 10% and SiO 2 > 60% belong to the subalkaline series.The North Apenninic Arc (NAA) and South Apenninic Arc (SAA) are separated into two segments by the WTB rift, as shown in Figure 14, suggesting that the arc system and the rift system may be two independent systems that were formed almost simultaneously by the tectonic mechanism of lower  2017)).This diagram shows the data pertaining to Cenozoic "orogenic" and "nonorogenic" volcanic rocks from Sardinia, Spain, Corsica-Estérel, Languedoc, northern Apennines and Alps.mantle activity, rather than being induced by the subduction of the arc system.Figure 14 reveals that the volcanic rocks of the arc system are dispersed throughout the NAA and SAA regions, which are located more than 300 kilometers from the subduction zone.Rift volcanic rocks are distributed within the arc system volcanoes and superimposed on the arc system volcanoes from 3.5 Ma to the present day, according to Turco et al. (2006).

The tectonics of the apenninic-maghrebides chain
The above information indicates that in the Apenninic-Maghrebides Chain, the recent chain activities are mainly from rift valley activities, and rift volcanism is more developed than arc volcanism (Turco et al., 2006).What is the tectonic environment of this chain?To solve this problem, we collected all volcanic rock data from multiple tectonic environments (including Archean) around the world (Figure 15).As shown in Figure 15, the ocean islands have extremely distinctive characteristics, including the highest alkalinity (Na 2 O + K 2 O) and bimodal distribution, and are relatively complex, which may be classified into three series: (1) the transitional series between the trachy-basalt series and the subalkaline series; (2) the transitional series between the trachy-basalt series and alkaline series; and (3) the alkaline series (Figure 15).The oceanic island rocks have the highest alkalinity of all volcanic rocks, indicating that they are mainly derived from the enriched mantle.Seamounts also have three series, with the subalkaline series predominating, while the trachy-basalt series and alkaline series dominate in oceanic island rocks.This indicates that seamounts do not always belong to a single tectonic environment but  rather to numerous environments.Seamounts are a geographical feature that can originate from convergent margins, back arcs, rear arcs, ocean islands, and oceanic plateaus.The oceanic plateau is mainly composed of basalt, followed by trachy-basalt series.The intraplate volcanic rocks are mainly transitional types that belong to the sub-alkaline series and trachy-basalt series, followed by the alkaline series.Continental flood volcanic rocks plot mainly within the subalkaline series, followed by the trachy-basalt series and then the alkaline series.The complex volcanic rocks are mainly trachy-basalt series.Rift volcanic rocks are divided into two categories: continental rifts and rift volcanoes.They are all bimodal volcanism and subalkaline series, in which continental rifts contain little alkaline series, and the rift volcano's alkaline series content is relatively high.The spreading center and the back arc basin mainly belong to the subalkaline series, and the Archean also plots within the subalkaline series.Among these, the convergent margin is special, which is why we constructed Figure 15.
Convergent margin volcanic rocks include island arcs, continental arcs, and forearc volcanic rocks, mainly of the subalkaline series.However, why are there a large number of rocks that belong to the alkaline series?We know that the convergent margin includes a compressional subduction zone and an extensional zone with rift characteristics in front of the subduction zone.The Apenninic-Maghrebides Chain consists of two series of rocks: the predominantly alkaline series to trachy-basalt series and the less common subalkaline series.Similar to the Apenninic-Maghrebides Chain, the Aeolian arc is located in the Mediterranean Sea, and it has been demonstrated that the majority of the volcanic materials sampled from these two arcs are not from the subduction zone but rather from extensional contexts (Lustrino et al., 2017;Mazzeo et al., 2014;Turco et al., 2006).If the data of the Apenninic-Maghrebides Chain and Aeolian arcs are removed from Figure 15, the convergent margins are primarily a subalkaline series (Figure 16), which is consistent with the earlier understanding of the academic community and conforms to the characteristics of typical arc regions (Izu, New Zealand, Andes, Honshu arc, etc.).Comparing the data of the Apenninic-Maghrebides Chain to the volcanic rocks of other tectonic settings in the world reveals that while volcanic rocks from other tectonic settings are predominantly basalt, the Apenninic-Maghrebides Chain has predominantly intermediate rocks (trachyte phonolite) with few basic rocks.The Aegean arc, which is also located in the Mediterranean, exhibits similar characteristics to the Apenninic-Maghrebides Chain.Does this imply that crust-mantle mixing may play a significant role in magma formation?Is this connected to the fact that the Apenninic-Maghrebides Chain exhibits both convergent margin and ocean island characteristics in Figure 2?
The magmatic activity from the two types of tectonic settings in the Apenninic-Maghrebides Chain was generally concomitant during the Cenozoic, and arc magmatism was relatively developed in the early stage, as shown in Figure 17.From the Miocene to the Quaternary, the rifting increased significantly.There are a large number of samples without age data in the database, and we estimate that most of them may be modern samples.The arc system in the Apenninic-Maghrebides Chain may not be directly related to the rift.The arc has its own evolutionary history.The rift is the result of crustal expansion rather than arc magmatism.The overlap in age and geographical distribution further indicates that the origin of the rift magmatism may have nothing to do with subduction.The Apenninic-Maghrebides Chain is not a single tectonic setting but includes both rift and convergent margin characteristics.Therefore, it may be necessary to subdivide the Apenninic-Maghrebides Chain.

How can the tectonic environments of MORG, CMG and OIG be distinguished?
The basalts of convergent margins, ocean islands and spreading centers can be clearly distinguished.The geochemical properties of volcanic arc basalt (VAB) involve LILE enrichment and the depletion of high field strength elements (HFSEs); ocean island basalt (OIB) involves both LILE and HFSE enrichment; and spreading center basalt (MORB) involves LILE depletion and HFSE enrichment (Pearce et al., 1984a).
Nb and Y are HFSEs, and Rb is a LILE. Figure 1 shows that CMG is depleted in Nb, whereas OIG is enriched in Nb, indicating that the two are clearly distinct.Y has a secondary role, although overall, Y in the convergent margin is low and in ocean islands it is high (Figure 1).Why is it possible to distinguish the spreading center from convergent margins and ocean islands?According to the data, the K 2 O content in the spreading center is less than 2%, the average Y content surpasses 100, the average Nb content is approximately 20, and the average Nb content in the ocean island is approximately 100.Both the ocean island and the spreading center are enriched in HFSEs.Ocean island material, however, originates from the lower mantle, whereas spreading center material originates from the upper mantle.Ocean island material contains more HFSEs than the spreading center material.
Because the thickness of the oceanic crust is quite thin (ranging from 7-8 km at the spreading center to approximately 30 km at the subduction zone), the composition is relatively simple, consisting primarily of basic rocks.Granitoids in the ocean basin are formed mostly by the partial melting of basic rocks with hydrothermal fluid.Therefore, the majority of these rocks inherit the geochemical properties of their source rocks (e.g.gabbro, diabase, basalt).The CMG is generated by the partial melting of the volcanic arc basement, which is predominantly composed of volcanic arc basalt; therefore, the CMG inherits the geochemical properties of convergent margin basalt (CMB), and the geochemical properties of CMB are characterized by an LILE abundance and HFSE deficit.The granite resulting from the partial melting of this parent rock is enriched in LILEs and depleted in HFSEs.Ocean island basalt (OIB) contains elevated levels of LILEs and HFSEs.Ocean island basalt is derived from oceanic mantle plumes, which results in elevated levels of LILEs and HFSEs, and the basement of oceanic islands is primarily composed of OIB.Granite (i.e.OIG) formed from OIB inherits the same features as OIB, and it is also enriched in LILEs and HFSEs.Similarly, the spreading center granite (MORG) shares the same characteristics (Pearce et al., 1984a).CMG, OIG, and MORG (K 2 O ≤ 2%) have similar geochemical characteristics as CMB, OIB, and MORB, respectively, so the aforementioned three granite tectonic settings can therefore be identified.Y diagram, Rb vs. Nb+Y diagram) given by Pearce et al. (1984b) is still effective for distinguishing different types of granitoids with little adjustment and can be adapted to global data.Global data studies show that the Y and Nb contents in OIG are higher than those in CMG.These two types of granitoids can be clearly distinguished within the tectonic discrimination diagram of Pearce et al. (1984b).

Conclusion
It should be noted that this study only analyzes the granitoids of two different tectonic backgrounds, oceanic islands and convergent margins.Therefore, the understanding of this work is limited to the above two types of granitoids and excludes granitoids in other tectonic settings.
(2) The granitoids from spreading centers are more complex than previously thought.Based on the K 2 O content, they can be divided into high and low K types (Figure 5).However, many high K series data are widely scattered, suggesting that some of these rocks may represent the initial melt of partial melting (such as thin felsic veins) or include residual mineral inclusions from partial melting (such as elongate anorthite and clinopyroxene residues that are visible in thin sections or abnormally high Ti, Mg, Ca, and Fe contents in some samples), indicating that some melts have not yet reached the composition of granitoid magma.Most granitoids in spreading centers are potassium deficient, and the aforementioned discrimination diagrams can still be used to identify the tectonic environment of granitoids that are potassium deficient (K 2 O ≤ 2.0%) (Figure 7).(3) The problem of the Apenninic-Maghrebides Chain was discovered unintentionally in this study.The data collected from the Apenninic-Maghrebides Chain indicate that the chain is obviously incompatible with other areas at the convergent margins.The study of the Apenninic-Maghrebides Chain shows that magmatic activity from the rift is more developed than magmatism caused by subduction.Moreover, this rift is not caused by the extensional tectonics of the front edge of the plate subduction zone, although they originate from the same period and overlap in geographical distribution.The rift is superimposed on the arc system and not at the far end of the arc system.This shows that there is no connection between the arc and the rift.The Apennine rift is more developed than the arc system, and it mainly shows the characteristics of the extensional background rather than the characteristics of the subduction zone.This is the uniqueness of the Apenninic-Maghrebides Chain.

Figure 1 .
Figure 1.Nb vs. Y and Rb vs. Nb+Y diagrams of convergent margin granitoids and ocean island granitoids (form Pearce et al. (1984b)).The orange straight line is the original dividing line of the discrimination diagram, and the black curve is redefined based on the global ocean island and convergent margin data.The area within the color circle represents the 80% confidence interval of its distribution.

Figure 2 .
Figure 2. Nb vs. Y and Rb vs. Nb+Y diagrams of different regions of convergent margin granitoids.

Figure 3 .
Figure 3. Nb vs. Y and Rb vs. Nb+Y diagrams of spreading center granitoids.

Figure 5 .
Figure 5. SiO 2 and other major elements of spreading center granitoids.The red area in the figure is the point density area.We divide all the data density into five levels, and only the first three levels with the densest density are retained, which is the same as that below.

Figure 6 .
Figure 6.K 2 O vs. other major element diagrams of spreading center granitoids.

Figure 7 .
Figure 7. Nb vs. Y and Rb vs. Nb+Y diagrams of spreading center granitoids with K 2 O ≤ 2.0%.

Figure 11 .
Figure 11.Schematic geological and structural map of the Tyrrhenian margin of the Campania region (from Mazzeo et al. (2014)).

Figure 14 .
Figure 14.Distribution map of two types of volcanic rocks in the Apenninic-Maghrebides Chain (from Turco et al. (2006)).The hollow circle is a subalkaline series rocks; and the red squares are alkaline series and trachy-basalt series rocks.NAA: North Apenninic Arc; SAA: South Apenninic Arc; WTB: Western Tyrrhenian Basin.

Figure 15 .
Figure 15.TAS classification diagram of different tectonic environments.

Figure 16 .
Figure 16.TAS classification diagram of convergent margin volcanic rocks (excluding data from the Apenninic-Maghrebides Chain and the Aeolian arc).

Figure 17 .
Figure 17.Age distribution of two types of granitoids in the Apenninic-Maghrebides Chain.

( 1 )
It has been almost 40 years, and the granite discrimination diagram (Nb vs.