Geochemistry and Texture of Clinopyroxene Phenocrysts from Paleoproterozoic Picrobasalts, Karelian Craton, Fennoscandian Shield: Records of Magma Mixing Processes

: This paper presents an integrated major and trace element data and crystal size distribution analysis for zoned clinopyroxene phenocrysts hosted in variolitic and massive picrobasalts of the Suisaari Formation, Karelian Craton, Eastern Fennoscandian Shield. Clinopyroxenes in variolitic and massive lavas occur as unzoned, reverse, and normally zoned crystal. Oscillatory-zoned clinopyroxenes are only observed in variolitic lavas. The obtained data were examined in order to evaluate the contribution of magmatic processes such as magma mixing, contamination and fractional crystallization to the formation of various zoning patterns of clinopyroxene phenocrysts. Clinopyroxene phenocrysts in both variolitic and massive lavas originate from similar primary melts from a single magmatic source. The obtained data on composition and texture of clinopyroxene phenocrysts together with the crystal size distribution (CSD) analysis suggest that crystallization of the massive lavas mainly involves fractionation in a closed magmatic system, whereas the crystallization of the variolitic lavas is determined by processes in an open magmatic system. The results provide novel information on the evolution of Paleoproterozoic magmatic systems in the Karelian Craton. phenocrysts. The results provide novel information on the evolution of Paleoproterozoic magmatic systems in the Karelian Craton.


Introduction
Pyroxenes are abundant rock-forming minerals in volcanic rocks. They exhibit diverse varieties of textural, zoning and compositional features that record the crystallization history of magma from early crystallization (where pyroxenes form large phenocrysts) to the final stage (in which pyroxene microcrysts crystallize in the groundmass) [1][2][3]. Therefore, the chemical composition and texture of pyroxenes from volcanic rocks are usually used as the markers of the evolution of magma composition during crystallization in open and closed magmatic systems. Additionally, compositional and textural characteristics of pyroxenes can provide information on the physicochemical conditions of crystallization including temperature, pressure, and oxygen fugacity variations during the evolution of magmatic systems [4][5][6]. Furthermore, pyroxenes are more resistant to metamorphic alterations under greenschist facies compared to olivine, which provides the possibility of their use as petrogenic indicators for a wide compositional range of volcanic rocks.
Pyroxenes have been intensively studied worldwide over the past 40 years. Numerous studies of pyroxenes from geochemically diverse volcanic rocks of various geodynamic settings [2,[7][8][9][10][11][12] have demonstrated that their composition and texture are reliable indicators by which to estimate the evolution and ascent histories of magmas. This approach has been successfully applied to decipher the evolution of modern volcanic systems [13][14][15] as well as Mesozoic-Paleozoic paleovolcanic series of volcanic rocks [16][17][18].
In the Paleoproterozoic picrobasalts of the Suisaari Fm, Ludikovian, Onega Basin of the Karelian Craton in the Eastern Fennoscandian Shield, clinopyroxene is an ubiquitous mineral. The most intriguing feature of Suisaari basalts is a wide occurrence of variolitic lavas described by F. Yu. Levinson-Lessing more than 100 years ago [19]. The origin of variolitic lavas has been an intensively debated problem over the last 40 years [20][21][22][23][24][25][26][27][28]. Yalguba variolitic lava origin models have included liquation [19][20][21][22][23], mixing of felsic and mafic magmas [24,25], quenching crystallization from a strongly overheated melt [26,27] and low-temperature secondary alterations of basic rocks [28]. Despite intensive study, several questions (including the crystallization mechanism and role of magma mixing and contamination by crustal material in variolitic texture formation) remain unresolved. Clinopyroxenes that are abundant in all Suisaari Fm lavas could give insight into the origin of mafic melts and their crustal evolution, including fractional crystallization, contamination by crustal materials and magma system evolution. Since clinopyroxenes from volcanic rocks of Suisaari Fm exhibit a wide range of compositional, zoning and textural patterns [29], the study of their major and trace element compositions and textures provides the opportunity to decipher the petrogenesis of host picrobasalts.
This paper presents the integrated major and trace element data and a crystal size distribution analysis for zoned clinopyroxene phenocrysts hosted in variolitic and massive picrobasalts of Suisaari Fm. The obtained data were examined in order to evaluate the contribution of magmatic processes such as magma mixing, contamination and fractional crystallization to the formation of various zoning patterns of clinopyroxene phenocrysts. The results provide novel information on the evolution of Paleoproterozoic magmatic systems in the Karelian Craton.

Geological Setting
The Yalguba Ridge is located in the central part of the Paleoproterozoic Onega Basin, in the southeastern part of the Archaean Karelian craton, Fennoscandian Shield [30,31] (Figure 1). The Yalguba Ridge locality belongs to the Suisaari Fm of the Ludicovian (approx. 2.0 Ga) volcanicsedimentary succession of the Middle Paleoproterozoic. Suisaari Fm occurs mainly in the western part of the Onega Basin within an area of approximately 2000 km 2 and comprises mostly maficultramafic (picritic and picrobasaltic) volcanic rocks [31]. The total thickness of Suisaari Fm does not exceed 420 m and is dominated by lavas (75 vol.%) compared with volcano-sedimentary rocks with widespread pyroclastic products (25 vol.%) [31,32]. A depositional age of the Suisaari Fm is constrained by the Sm-Nd mineral isochron age of 1975 ± 24 Ma obtained for the Konchezero gabbroperidotite sill that is coeval to Suisaarian mafic lavas [33].
Subvolcanic rocks that are co-magmatic with the lavas from Suisaari Fm are represented by dolerite and peridotite dykes and sills. The Yalguba Ridge section represents the upper part of the Suisaari Formation and is represented by massive, pillowed and variolitic mafic lavas interbedded with tuffs with thickness up to 3 m. The rocks are altered from the prehnite-pumpellyite to greenschist facies. Detailed petrography has been described in [22,31,32,34].

Sampling and Methods
Twenty samples of variolitic lavas and 10 samples of massive lavas were collected from outcrops of the Yalguba Ridge. Sampling was undertaken across variolitic lava sequences with a thickness of 16 m and an underlaying massive lava flow with a thickness of 10 m. The major element composition of clinopyroxene from 20 thin sections was studied by energy-dispersive spectroscopic (EDS) analyses, and trace element compositions of clinopyroxene phenocrysts from four thin sections were analyzed by secondary ion mass spectrometrometry.
Clinopyroxene phenocrysts were detected in the variolitic lava both in globules and matrices, as well as in the massive lavas. The phenocryst content was low and clinopyroxene phenocrysts were more abundant in the variolitic than in the massive lavas. All samples of the variolitic lavas contained clinopyroxene phenocrysts, whereas these were observed in only two out of 10 samples of the massive lavas. All clinopyroxene phenocrysts were well-preserved and showed no secondary alteration, with the exception of phenocrysts from the matrices of variolitic lavas, which exhibited signs of chloritization.
Major and minor element compositions and textural features of clinopyroxenes were analyzed using a VEGA II LSH (Tescan, Brno, Czech Republic) scanning electron microscope coupled with an INCA Energy 350 energy dispersive detector at the Centre for Collective Usage, Karelian Research Centre, RAS (Petrozavodsk, Russia). The analyses were performed on polished thin sections under the following analytical conditions: W cathode, 20 kV accelerating voltage, 20 nA beam current, 2 μm beam diameter, counting time of 90 s. The accuracy of analyses was better than 1% for the major elements and 5% for the minor elements.
Based on the backscattering electron (BSE) imaging and EDS analysis, a set of representative clinopyroxenes with various zoning patterns were selected for trace element analysis using secondary ion mass spectrometry techniques. Concentrations of minor elements for these samples were determined using a Cameca IMS 4f secondary ion mass spectrometer (Cameca, Gennevilliers, France) at the Yaroslavl Branch of the Physicotechnical Institute of RAS following the analytical procedure in [17]. The concentration uncertainties were determined to be 5-10% at concentrations >1 ppm, and up to 20% at concentrations of 0.1-1 ppm.
Crystal size distribution (CSD) analysis was carried out on the thin section BSE images of the variolitic and massive lavas using the ImageJ software. Clinopyroxene grains were isolated by threshold filters from felspar and groundmass and manually vectorized to define the grain boundaries. Grains smaller than 10 pixels and grain fragments in the image edges were eliminated from the analysis. CSD statistics were calculated for 698 clinopyroxene grains from the globules, 298 grains from the variolitic lava matrix and 434 grains in the massive lavas. Crystal parameters such as length, width, area, centroid, and fit ellipse were determined using the Image J software in order to characterize the clinopyroxene morphology. CSD correction software [35] was used to calculate 3D parameters from 2D section data. CSD data are listed in Supplementary Materials Table S2.

Results
Major element composition data obtained for 350 clinopyroxenes, including phenocrysts and microcrysts from the variolitic and massive lavas of the Yalguba Ridge, indicated that they all corresponded to augite [36], with En content ranging from 31% to 52%.

Major Element Contenst and Zoning Patterns
"Homogenous" clinopyroxene phenocrysts (HP type) occurred both in variolitic and massive lavas as euhedral to subhedral grains with a size of 400-700 m. They had no signs of zoning both in polarizing light and BSE (Figure 3a,f). The amount of "homogenous" clinopyroxene phenocrysts was less than 30% in the variolitic lavas, and 70% in the massive lavas. In the variolitic lavas, HP were characterized by the gradual decreasing of the Mg# (Mg# = 100 × Mg/(Mg + Fe) in mole considering all iron as Fe 2+ ) values from 83-88 in the central parts of the grains to 78-83 in their margins. In the massive lavas, Mg# of HP gradually decreased from 90-92 in the central parts of the grains to 80-82 in their margins ( Table 1). The decreasing of Mg# was coupled with increasing TiO2 content and decreasing Cr2O3 in both types of lavas. Chemical trends within center-to-margin profiles of HP clinopyroxenes were similar to that of normal-zoned clinopyroxenes.
Clinopyroxene phenocrysts with normal zoning (NZ) also occurred in the variolitic and massive lavas. These were characterized by the presence of a high-Mg core in the central part of the grain. The core was mantled by a rim that could be well-recognized in BSE (Figure 3b,g). Clinopyroxene phenocrysts with normal zoning can be divided into two subgroups. Type I normal zoning clinopyroxene phenocrysts (NZ Type I) were represented by relatively large (400-1000 m) euhedral to subhedral clinopyroxenes that occurred both in variolitic and massive lavas. The NZ Type I phenocrysts had a euhedral core surrounded by a thin rim with sharp interfaces between zones. The core/rim thickness ratio was 10:1 ( Figure 3b,g). The zoning is defined by the contrast differences in major element contents. In the variolitic lavas, the Mg# decreased from 88-93 in the core to 79-86 in the rims (Table 1). In the massive lavas, Mg# values decreased from 85-92 in the cores to 80-86 in the rims. The variation of Mg# with TiO2 and Cr2O3 was also observed in the variolitic and massive lavas such that Cr2O3 slightly decreased and TiO2 slightly increased within declining Mg# ( Figure 4, Table  S3).
Type II of normal-zoning phenocrysts (NZ Type II) was represented by clinopyroxenes with relatively small (200-300 m) grain sizes. This sub-type of normal-zoning phenocryst is characterized by a presence of rounded cores enriched in MgO (Figure 3c,h) and thick rims with a core-to-rim thickness ratio of 1:1. The trends of core-to-rim compositional changes were similar to those of NZ Type I clinopyroxenes (Table 1). It should be noted that cores of NZ Type II phenocrysts in the massive lavas were characterized by high Cr2O3 contents up to 1.4 wt.% ( Figure 4). The major element chemistry of the high-Cr cores was indistinguishable from the other high-Mg pyroxenes. The amount of normal-zoned clinopyroxene phenocrysts was higher in the variolitic lavas (40%) than in the massive lavas (20%). Moreover, in the variolitic lavas, NZ Type I was more abundant, forming up to 30% of phenocrysts compared to NZ Type II (10%).
Reverse-zoned clinopyroxenes (RZ type) formed subhedral 200-250 m grains. They were less abundant in the variolitic (20%) and massive (10%) lavas compared with the NZ phenocrysts. The RZ phenocrysts usually have rounded cores surrounded by a rim that can be well recognized in the BSE images (Figure 3d,i). The core-to-rim thickness ratios varied from 10:1 in the massive lavas to 1:1 in the variolitic lavas. The RZ phenocrysts in the variolitic and massive lavas had Mg-poor cores but different Mg# values. In the massive lavas the cores of the RZ type phenocrysts had an Mg# = 82-84, and in the variolitic lavas Mg# values were 68-75 (Table 1). The rims in RZ grains were more magnesian than the cores both in the massive (Mg# = 86-89) and variolitic (Mg# = 79-84) lavas. The Mg# values in the RZ clinopyroxenes coincided with Mg# values in the rims of normal-zoned clinopyroxene phenocrysts (Figure 4). In contrast with the normal-zoned clinopyroxenes, the cores of RZ clinopyroxenes had lower Cr2O3 contents and have higher TiO2 contents compared to the rims. The Al2O3 and Na2O contents showed narrow core-to-rim variations with no relation to the variations in Mg# (Figure 4). Oscillatory-zoned clinopyroxenes (OZ type) were identified only in the variolitic lavas and were the most rare phenocryst type, forming less than 10% of phenocrysts. They commonly had a subhedral morphology with rounded cores (~20-  Table 1). The cores had high TiO2 and low Cr2O3 contents compared with the mantles and rims. The TiO2 content decreased in the mantles and slightly increased in the rims, whereas the Cr2O3 content was relatively constant in the mantles and rims.

Trace Element Composition of Clinopyroxene Phenocrysts
Cores and rims of zoned clinopyroxenes had various rare earth element (REE) contents depending on the zoning type, but were all characterized by similar convex upward REE distribution patterns with minor negative Eu anomalies. Compositional similarity was also pronounced in the trace element distribution characterized by the depletion in high field strength elements (HFSEs) ( Figure 5).
HP type phenocrysts were characterized by convex upward REE distribution patterns with REE = 13-30 ppm, depletion in both LREE (light rare earth elements), heavy rare earth elements (HREE) ([La/Sm]n = 0.33-0.37, [Gd/Yb]n = 2.05-2.30), and minor negative Eu anomalies (Eu/Eu* = 0.76-0.78) (Eu/Eu* = EuN/(SmN + GdN) 1/2 ) (Figure 5a). The REE content showed no significant variations along the crystal profile. HP type phenocrysts were characterized by a depletion of HFSEs with negative Nb and Zr anomalies. Nb/Nb* varied from 0.23 in the variolitic lavas to 0.51 in the massive lavas. The central parts of HP phenocrysts showed no significant Pb anomalies, whereas the margins of the grains showed slightly negative Pb anomalies (Figure 5b).
The RZ type clinopyroxenes from the variolitic lavas had higher REE content compared to that of the massive lavas, which decreased from REE = 117-137 ppm in the cores to REE = 64-66 ppm the rims. The reverse-zoned clinopyroxenes from the massive lavas had lower REE content but   g,h). The normalization values are from [37]. The whole-rock composition of the massive lavas (yellow colored line) is shown in e and f. The whole-rock composition of the variolitic lavas (dark grey color for matrix Type I, light grey color for matrix Type II, and rose color for the globule composition) is shown in (g) and (h) (modified from [34]).

Crystal Size Distribution of Clinopyroxenes
The analyses of crystal size distributions (CSDs) are becoming increasingly used to study the crystallization kinetics of magmatic systems [12,38,39]. The CSD results obtained for clinopyroxenes from the variolitic and massive lavas of the Yalguba Ridge are plotted on a population density (ln(n), m −4 ) vs crystal size (L, m) diagram. Figure 6 shows that clinopyroxene from the variolitic lavas had inverse relationships between grain length and population density. CSD curves of clinopyroxenes in the globules and matrices of the variolitic lavas were characterized by kinked linear trends marking two size crystal populations. The CSD curves of these clinopyroxenes showed linear trends with a slope of −0.06 for a crystal size smaller than 120 m, and -0.03 for a crystal size larger than 120 m. In contrast, the CSD curves of clinopyroxenes from the massive lavas exhibited a concave-downward CSD at sizes smaller than 20 m ( Figure 6) and a linear trend with a slope of -0.05 at a crystal size larger than 20 m.
To estimate the grain size, an equivalent circle diameter (ECD) was calculated from the measured surface area that corresponded to the diameter of a circle with an area equivalent to the crystal area. Clinopyroxenes both from the variolitic and massive lavas were characterized by an ECD lower than 100 m with maximum size values between 25-40 m. The aspect ratio (major-axis length/minor-axis length of the best fit ellipse) of clinopyroxenes from both variolitic and massive lavas (AR, defined as AR = major axis length/minor axis length of the best fit ellipse) were also comparable, varying from 1 to 3. Moreover, there was similarity in the 2D orientation, evidenced by the similar values of alignment factors (AFs) for the samples from the variolitic (AF = 0.06) and massive (AF = 0.10) lavas. The low values of AFs reflected a lack of specific orientation. It should be mentioned that the studied samples were not oriented, so the measured 2D orientations were relative.

Discussion
The variolitic and massive lavas of the Yalguba Ridge belong to the Suasaari Fm of the Onega Basin, which represents a continental flood basalt province in the Fennoscandian Shield. Previous lithostratigraphic, chemical, and isotope studies [33,34] have revealed that the volcanic rocks of the Onega Basin originated from a single mantle plume. The uppermost Onega Basin volcanic series (including the Suisaari Fm) have compositional features comparable with the modern oceanic plumederived magmas, including high MgO contents, enrichment in Ti and other HFSEs, and differentiated REE patterns. These features indicate the origin of Suisaarian primary melts may have been due to a deep-seated mantle melting caused by the mantle plume rising [33]. In contrast to high-Mg Suisaarian basalts, low-Mg ones have shown signs of contamination and fractional crystallization characterized by enrichment in SiO2 and depletion of MgO and HFSE. The available geochemical, isotopic, and geochronological data indicate that generation of approximately 1.96-1.97 Ga occurred within several short pulses of melting caused by the mantle plume rising [33,[40][41][42]. These pulses contaminated a variety of high-and low-Mg mafic rocks in various proportions.
The origin of the variolitic texture in volcanic rocks has been studied for a long time, but is still debated. Since clinopyroxene is a ubiquitous mineral in the lavas of the Yalguba Ridge, we used it as a petrogenic indicator to decipher the key stages in the magma system evolution in order to clarify the differences in crystallization history of the variolitic and massive lavas.
The evolution of magmatic systems involves complicated processes such as fractional crystallization, magma mixing, and contamination by crustal materials that might be recorded in the compositional and textural features of pyroxenes. Three general groups of pyroxene populations are commonly classified in magmatic systems [39,43] including autocrysts, antecrysts, and xenocrysts. Autocrysts crystallized from the final melt, which corresponded to the whole-rock compositions of the host rocks. Antecrysts are introduced in the magma, in which they are now hosted, through different magma replenishment events or via magma convection, whereas xenocrysts are incorporated as a result of contamination. Since autocrysts are the only crystal in equilibrium with their host's whole-rock composition, they can be distinguished from antecrysts or xenocrysts by evaluating for chemical equilibrium.

Clinopyroxene-Melt Equilibrium
The equilibrium between clinopyroxene and melt represented by whole-rock composition of the variolitic and massive lavas was tested using the partition coefficient of Fe-Mg between clinopyroxene and melt, which has been estimated to be KdFe-Mg = 0.275 ± 0.067 [44]. Figure 7 shows that in massive lavas, the margins of several HP clinopyroxenes (KdFe-Mg = 0.19-0.29) and rims of the NZ (KdFe-Mg = 0.21-0.24) and OZ (KdFe-Mg = 0.20-0.25) clinopyroxenes were in equilibrium with their host rock compositions at Mg# = 0.52. The cores of OZ clinopyroxenes slightly deviated from the equilibrium (KdFe-Mg = 0.30-0.32) and appeared to crystallize from the calculated equilibrium melt, with Mg# = 0.48-0.49. The NZ cores showed more distinct deviations (KdFe-Mg = 0.14-0.16) from the equilibrium, and should have crystallized from the melt at Mg# = 0.60-0.68, which is higher than the Mg# of host massive lavas. In general, the compositions of clinopyroxene phenocrysts from the massive lavas appeared to be close to equilibrium compared with the clinopyroxene phenocrysts from the variolitic lavas. Similar to the massive lavas, the margins of several HP clinopyroxenes from the variolitic lavas also exhibited the equilibrium features (KdFe-Mg = 0.28-0.44) with their host rock composition at Mg# = 0.59. In contrast with the massive lavas, the rims of NZ, RZ, and OZ clinopyroxenes from the variolitic lavas showed disequilibrium features (KdFe-Mg = 0.35-0.42) and were calculated to be in equilibrium with a melt at Mg# = 0.48-0.53. Additionally, several cores of NZ clinopyroxenes were in equilibrium with the whole-rock composition (KdFe-Mg = 0.17-0.28), whereas the cores of RZ and OZ clinopyroxenes exhibited strong deviations from the equilibrium (KdFe-Mg = 0.61-0.80). The Mg# of melts, which should have been in equilibrium with the cores of RZ and OZ clinopyroxenes from the variolitic lavas, were calculated to be Mg# = 0.35-0.38 and Mg# = 0.40-0.50, respectively. Since the KdFe-Mg values of the cores from RZ and OZ clinopyroxenes were the furthest from the equilibrium value, they could be interpreted as the antecrysts crystallized from the early batch of magmas or xenocrysts, which are foreign to the magma system. A clinopyroxene-only pressure independent thermometer [45] was used to calculate the temperatures of crystallization for the clinopyroxenes, which were determined to be in equilibrium with the host's whole-rock composition. The temperatures of the clinopyroxenes from the variolitic lavas were determined to be 1192  26 С, which is similar to the crystallization temperature calculated for clinopyroxenes from the massive lavas (1161  34 С) (Table S5).

Origin of Clinopyroxene Zoning
Clinopyroxene phenocrysts of variolitic and massive lavas exhibited similar zoning patterns, but were abundant to different extents. The massive lavas contained the lowest abundances of clinopyroxene phenocrysts, and the most abundant zoning pattern was HP clinopyroxenes, whereas the zoned clinopyroxenes were rarely recognized. In contrast, the variolitic lavas contained greater proportions of clinopyroxenes dominated by NZ and HP zoning patterns. Moreover, the OZ phenocrysts were recognized only in the variolitic lavas and not detected in the massive lavas. The variable proportions of the clinopyroxenes with different zoning patterns in the massive and variolitic lavas suggest that the melts that formed these rocks experienced different crystallization histories.
HP clinopyroxenes in the massive lavas are characterized by high Mg# and Cr2O3 contents gradually decreasing from the center to the margins (Figures 3a,f and 4), with low REE content. The central parts of HP clinopyroxenes in the massive lavas are likely to crystallize from the most primitive mafic melt. Moreover, the margins of HP clinopyroxenes in the massive lavas were determined to be in equilibrium with their host whole-rock composition (Figure 7), and could be interpreted as autocrysts crystallized from a melt. Since the HP clinopyroxene phenocrysts represented the most abundant zoning pattern in the massive lavas, it can be assumed that the crystallization of the massive lavas mainly involved crystallization in a closed system or in a system with suppressed magma mixing. HP clinopyroxenes were also recognized in the variolitic lavas, where they exhibited similar compositional variations but lower Mg# and Cr2O3 contents.
Normal zoning in clinopyroxenes from the variolitic and massive lavas is defined by a decrease in Mg# and Cr2O3 contents coupled with an increase in TiO2 and REE contents from the cores to rims (Figure 4, Table 1). It is noteworthy that the NZ cores were characterized by positive Nb anomalies in contrast to the NZ rims, which exhibited negative Nb anomalies. Positive Nb anomalies are typical for the most primitive picrobasalts of Suisaari Fm and Zaonega Fm [33], which is interpreted [33] to be an indicator of the uncontaminated character of melts produced these rocks. High Mg# and positive Nb anomalies of the NZ cores suggest crystallization from the primitive melt with a composition similar to the high-Mg picrobasalts from the Suisaari Fm. Based on their textural features, two types of NZ clinopyroxenes characterized by similar compositional variations were recognized in the variolitic and massive lavas. Notably, NZ Type I clinopyroxenes were abundant and NZ Type II was rarely detected in the variolitic and massive lavas. The cores of NZ Type I clinopyroxenes with high Mg# and low REE content were assumed to have crystallized from a less evolved melt. The cores exhibited no distinct resorption texture (Figure 3b,g) and the rim overgrowth was likely formed as a result of fractional crystallization-the process responsible for driving the geochemical evolution of the melt causes in subsequent rim growth, resulting in a distinct geochemically character compared to the rim/core. In contrast, the NZ Type II clinopyroxenes had a separating resorption surface between the cores and rims (Figure 3c,h), suggesting it is unlikely that the overgrowth rims originated due to fractional crystallization. The plausible explanation for this zoning pattern is that the core was a fragment of clinopyroxene crystallized in the more mafic magma. NZ Type II cores might be resorbed and subsequently overgrown by the rim with low Mg# during the mixing of evolved magma or convection from high-Mg to Mg-low zones of the magmatic system. NZ Type II pyroxenes exhibited relatively slight variations in core-rim composition, and NZ Type II cores had slightly low Mg# compared to NZ type I cores (Table 1), which originated from fractionation. Additionally, the compositions of NZ Type II cores were in equilibrium or close to equilibrium with the host rock composition. Based on this data, we assume that convection is favored to produce NZ Type II clinopyroxenes, though we cannot exclude that NZ Type II formation involves magma mixing.
Reverse zoning and resorption textures are common in pyroxenes from various rocks [18,43,46], and are considered to record the open system processes, including magma mixing and contamination by host rocks [4,43]. Some of the clinopyroxenes in the variolitic lavas and a single crystal in the massive lavas exhibited reverse zoning that manifested as increasing Mg# from the core to rims, instead of Mg# decreasing (as is the case in normal zoning). Additionally, RZ clinopyroxenes are characterized by distinct disequilibrium features such as an overgrowth of rims on resorbed cores with dissolution surfaces. It is noteworthy that the RZ clinopyroxenes from the massive lavas were characterized by lower rim thicknesses and low resorption extents compared to those from the variolitic lavas (Figure 3d,i). However, reverse zoning can record not only variations of magma composition but also the variations of crystallization conditions (e.g., temperature or oxygen fugacity). If the reverse zoning reflects the variations of magma composition, there will be a correlation between Mg# and the other element contents [43].
The RZ clinopyroxenes from the variolitic and massive lavas displayed a positive correlation between Mg# and Cr2O3, and a negative correlation between Mg# and TiO2, although this was better in the clinopyroxenes from the massive lavas (R = 0.75 and R = −0.86) than from the variolitic lavas (R = 0.67 and R = −0.60). These data provide additional evidence that reverse zoning leaves a record open magma system processes. The Mg-poor clinopyroxenes were assumed to crystallize from the more evolved melts, whereas the Mg-rich clinopyroxenes corresponded to the more primitive magmas. The similarity in REE and trace element distribution patterns between the cores and rims of the reverse-zoned clinopyroxenes (Figure 5e,f) precludes an origin involving crustal contamination in an open system. Therefore, the cores of RZ clinopyroxenes from the variolitic and massive lavas of the Yalguba Ridge were interpreted as antecrysts, suggesting that they can originate from magma mixing processes or convection within the magmatic system. In the case of RZ clinopyroxenes from variolitic lavas, magma mixing processes are favored, since convection is unlikely to produce such strong variation in the core-rim composition. Therefore, the cores of RZ clinopyroxenes are assumed to have crystallized from an early batch of magma during the magma mixing processes, revealing an evolved magma composition with low Mg# and Cr2O3 contents and high REE content, while the rims corresponded to a more primitive melt. In contrast, the RZ clinopyroxenes from the massive lavas exhibited less pronounced differences in core-rim composition and were determined to be close to equilibrium with the host's whole-rock composition (Figure 7), meaning they are likely to have been produced by convection in a closed system triggered by injection of high-Mg magma.
Previous studies [12,43,47] have shown that oscillatory zoning of minerals might be caused by the kinetics of crystal growth or magma processes such as magma mixing or convection. If the banding of oscillatory zoning is fine (15 m), it is thought to be kinetically controlled, and if the banding is coarse then oscillatory zoning reflects the compositional variations in the magma due to dynamic magmatic processes [43]. Oscillatory zoning was detected only for the clinopyroxenes from the variolitic lavas of the Yalguba Ridge, and was defined by a Mg-poor core, a relatively Mg-rich mantle, and a Mg-poor rim (Figure 4). Oscillatory zoning is characterized by the band widths ranging from 15 to 30 μm, suggesting magma composition variations. In general, there are two possibilities of oscillatory zoning in open magmatic systems, including repeated chamber replenishment events and crystal convection. OZ clinopyroxenes from the variolitic lavas had rounded cores with resorption features (Figure 3e,j) surrounded by euhedral mantles and rims with sharp boundaries. Moreover, the cores exhibited a significant chemical disequilibrium with the whole-rock host composition, whereas the mantles and rims were closer to equilibrium with some rim compositions with chemical equilibriums (Figure 7). The cores of OZ clinopyroxenes were assumed to be antecrysts, with the core-mantle zone mainly reflecting changes in magma composition due to magma mixing, whereas the rims were likely produced by overgrowth on the mantle during fractionation processes. Similar to the RZ clinopyroxenes, Mg-poor cores of the OZ clinopyroxenes were attributed to crystallization from evolved magma, whereas the Mg-rich bands corresponded to crystallization from primitive magma.
Zoned clinopyroxenes in the variolitic and massive lavas exhibited similarity in REE and trace element patterns ( Figure 5). The cores and rims of zoned and unzoned clinopyroxenes exhibited similar compositional features such as convex upward REE distribution patterns with minor negative Eu anomalies and depletion in HFSE elements. The data suggest the clinopyroxenes in both variolitic and massive lavas originated from similar primary melts, probably a single magmatic source. However, the variations in texture, zoning, and composition of clinopyroxene phenocrysts between the variolitic and massive lavas indicate that the differentiation processes operated to different extents in the magma system. Though at the early stage clinopyroxenes from the variolitic and massive lavas crystallized from similar primary melts, their subsequent magmatic evolution significantly differed between the variolitic and massive lavas. The obtained data suggest that the crystallization of clinopyroxenes, which caused the different zonation patterns in the massive lavas, mainly involved fractionation in a closed magmatic system, whereas the crystallization of the variolitic lavas was determined by processes in an open magmatic system. Additional support for this conclusion comes from the CSD analysis of the clinopyroxenes in the variolitic and massive lavas.
It has been demonstrated [38,39] that CSD curves would be linear if the crystal growth were defined by homogeneous nucleation increasing exponentially with time in an open system corresponding to fractional crystallization. Kinked CSD curves likely reveal more complex crystallization resulting from various processes such as mixing, mechanical sorting, or Ostwald ripening [38,39]. CSD curves of clinopyroxenes from variolitic lavas are characterized by kinked CSD curves with a concave-upward curvature at sizes larger than 120 m, which indicate the accumulation of a population of larger clinopyroxenes due to magma mixing in an open system ( Figure 6). Clinopyroxene from the massive lavas exhibited mainly linear CSD curves, indicating that fractional crystallization was the dominant process. The concave-downward CSD curve observed for sizes smaller than 20 m might have corresponded to Ostwald ripening, which involves the growth of large grains at the expense of small grains. The obtained CSD data for the variolitic and massive lavas coincide with the results of texture and zoning studies of the clinopyroxene phenocrysts, which suggest different crystallization histories for the variolitic and massive lavas.

Petrological Model and Conclusions
The present study of geochemistry and texture for the clinopyroxene phenocrysts from the Yalguba Ridge reveals that the variolitic and massive lavas are likely to have originated from a melt with a Mg# = 0.60-0.68, which could be close in composition to the most primitive picrobasalts of Suisaari Fm [33]. The supposed primary melt experienced variable extents of differentiation, including both closed and open system processes. The low amount of zoned clinopyroxene phenocrysts and the linear CSD curve suggest that crystallization of the massive lavas is likely to occur in a closed system or magmatic system with a suppressed accumulation of clinopyroxene antecrysts due to the magma recharge. The variolitic lavas of the Yalguba Ridge have a more complicated evolution history. A wide variety of clinopyroxene zoning patterns and kinked CSD curves prove that crystallization of the variolitic lavas took place in an open system and involved magma mixing and crystal convection in addition to magma fractionation. The magmatic processes for the variolitic and massive lavas can be generalized into the stages illustrated in Figure 8. Stage I is common for both variolitic and massive lavas, and it involves the crystallization of high-Mg (Mg# > 88) cores of NZ clinopyroxenes from both of those lavas, as well as homogeneous clinopyroxenes from the massive lava and mantles of OZ clinopyroxenes from the variolitic lava, which crystallized from the most primitive melt. Chemical characteristics of these clinopyroxenes, including high Cr2O3 content, low REE content, depletion of HREE, and positive Nb anomalies, correspond to the specific features of picrobasalts from Suisaari Fm [33]. However, although stageclinopyroxenes from variolitic and massive lavas crystallized from a similar primary magma, their subsequent crystallization evolutions differed significantly.
Stage II is specific to the crystallization of variolitic lavas in an open system, and involves the recharging of a primitive melt with a high Mg# and low REE content into a magma mush and/or transport of crystal cumulates, which could have caused a resorption of early clinopyroxene grains and produced the RZ and OZ clinopyroxenes in variolitic lavas (Figure 8). The cores of RZ and OZ clinopyroxenes with low Mg# = 0.70, high TiO2 and REE content, and depletion of HFSE with negative Nb anomalies reflect the crystallization from the low-Mg evolved melt. Similar convex upward REE distribution patterns with minor negative Eu anomalies detected for the clinopyroxenes suggest that the low-Mg evolved melt originated from the fractionation of the primary magma. Depletion in HFSE with negative Nb anomalies might be assumed as an indicator of crustal contamination. The high-Mg composition of the mantle of the OZ clinopyroxenes shows the subsequent injection of primitive magma in an open system, which might additionally have produced the homogenous clinopyroxenes with high Mg#.
Stage III is similar for both variolitic and massive lavas, and involves crystal convection. Crystal convection in a recharged system in variolitic and massive lavas produces NZ Type II clinopyroxenes. Additionally, convection in a closed system is assumed to be responsible for the origin of the RZ clinopyroxenes in the massive lavas.
Stage IV involves fractionation of primary magma and produces homogenous and NZ Type I clinopyroxenes, as well as the RZ and OZ rims of both variolitic and massive lavas.
The obtained data reveal the different trends of the primary magma evolution in an open system for the variolitic lavas and a closed system for the massive lavas of the Yalguba Ridge, though they are likely to originate from a single primary magma.
The mechanism behind variolitic texture formation in volcanic rocks is a widely debated topic. The most widespread models proposed to explain this phenomenon include the liquation of basalt magma [19][20][21][22][23], or mixing of felsic and mafic magmas [24,25]. The obtained data for clinopyroxene phenocrysts in the variolitic lava of the Yalguba Ridge suggest that they crystallized from primary and evolved basaltic magmas derived from a single source. As such, it seems unlikely that the variolitic textures originated from the mixing of felsic and mafic magmas [24,25]. The liquation model [34] is based on whole-rock composition, which shows the differences in the REE and trace element distribution of globules and matrices in the variolitic lavas of the Yalguba Ridge ( Figure 6). Moreover, two types of matrix composition were distinguished whereby one of the matrix types is characterized by a high content of the large-ion lithophile elements Rb, Ba and Sr and a depletion of HFSE elements, suggesting crustal contamination. Clinopyroxenes from variolitic lavas are characterized by a depletion of HFSE with negative Nb anomalies that might be assumed to be an indicator of crustal contamination. However, the obtained data are insufficient to provide arguments for and against the liquation model, and can only indicate potential time constraints on clinopyroxene phenocryst crystallization and liquation events. It might be assumed that clinopyroxene phenocrysts crystallized before the variolitc texture formation, since the similar textures and zoning patterns of clinopyroxene phenocrysts were recognized in both globules and matrices of the variolitic lavas. Therefore, the formation of variolitic textures on the lavas of the Yalguba Ridge might have occurred at the final stages of open system evolution and not have been recorded in the clinopyroxene chemistry.
The major and trace elements and the crystal size distribution of clinopyroxene phenocrysts hosted in the variolitic and massive lavas of the Yalguba Ridge (Suisaari Formation Onega Basin, Russia) led to the following conclusions: 1. Clinopyroxene phenocrysts in both variolitic and massive lavas originate from similar primary melts from a single magmatic source.
2. Crystallization of the massive lavas of the Yalguba Ridge mainly involves fractionation in a closed system, whereas the crystallization of the variolitic lava occurs in an open magmatic system and involves magma mixing and crystal convection.
3. The formation of variolitic textures in the lavas of the Yalguba Ridge might occur at the final stages of magmatic crystallization.
Supplementary Materials: The following are available online at www.mdpi.com/2075-163X/10/5/434/s1, Table  S1: Whole-rock major (wt. %) and trace (ppm) element compositions of the variolitic lava [34] and massive lavas, Table S2: CSD data of clinopyroxene from the variolitic and massive lavas of the Yalguba Ridge, Table S3: Concentrations of major elements in representative clinopyroxene phenocrysts from the variolic and massive lavas of the Yalguba Ridge, Table S4: SIM analysis data for representative clinopyroxene phenocrysts from the variolic and massive lavas of the Yalguba Ridge, Table S5: KD (Fe-Mg) and crystallization temperatures calculated for clinopyroxenes from the variolitic and massive lavas using the clinopyroxene-only pressure independent thermometer [45].
Author Contributions: Conceptualization, S.A.S. and A.V.S.; investigation, S.A.S. and S.Y.C.; data curation, S.Y.C.; writing-original draft, S.Y.C.; writing-review and editing, S.A.S. and A.V.S.; visualization, S.Y.C. and A.V.S.; supervision, S.A.S. and A.V.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by state assignment to Institute of Geology Karelian Research Centre RAS АААА-А18-118020290085-4.