Introduction

The Kamchatka arc is a long-lived active subduction system with high magma production rates and unusually mafic magmas. It is situated in a complex plate boundary system with a triple junction between the Pacific-, Bering Sea- and Okhotsk Plates, transform boundaries, and active and extinct convergent boundaries. It is further complicated by subduction of the Emperor seamounts and a slab tear (Yogodzinski et al. 2001). Petrological and geochemical studies of Kamchatka volcanic rocks have focused on the evolution of single long-lived stratovolcanoes at the active arc front and their regional comparison (Duggen et al. 2007; Ponomareva et al. 2007; Portnyagin et al. 2005, 2007a, b; Churikova et al. 2001, 2007; Ozerov 2000; Volynets 1994; etc.). Numerous mafic monogenetic cones, lava fields, and plateau-basalts, found along the arc front, in the Central Kamchatka graben depression, and in the back arc region, however, are not well studied. Their age, spatial distribution, compositional variations, and magma sources are still poorly understood. These are large volumes of mafic magmas erupted during the Pliocene–Holocene, a time when Kamchatka evolved to its present complex setting, and thus these rocks may be crucial in better understanding the complex tectonic setting of the arc.

Quaternary volcanism in Central Kamchatka occurs in three zones parallel to the trench (Fig. 1): (1) Eastern Volcanic Front (EVF), (2) Central Kamchatka Depression (CKD) as an intra-arc rift zone, and (3) the Sredinny Range (SR), the latter representing the back-arc of the presently active volcanic front in Kamchatka. The Southern Kamchatka arc, up to the latitude of the Malki–Petropavlovsk fault zone (Fig. 1), is considered as a continuation of Kurile island arc (Avdejko et al. 2002). The northern end of volcanic activity in Kamchatka is marked by Shiveluch volcano and has been related to the NW–SE Bering transform fault zone (e.g. Gorelchik et al. 1996; Yogodzinski et al. 2001; Davaille and Lees 2004; Portnyagin et al. 2005), which marks the termination of the actively subducting Pacific plate.

Fig. 1
figure 1

General plate tectonic setting of Kamchatka and area studied. The outline of volcanic zones are from http://kamchatka.ginras.ru/ (Sredinny Range active volcanism zone after (Pevzner 2006)). Plate boundaries after DeMets et al. (1990), Zonenshain and Savostin (1979) and Lander et al. (1994). White arrow show the Pacific-Okhotsk convergence after Scholl (2007). SR Sredinny Range, CKD Central Kamchatka Depression, EVF Eastern Volcanic Front, SK South Kamchatka. Location of the sampled sites in Sredinny Range are numbered: 1 Tobeltsen cone, 2 Spokojny volcano, 3 Tekletunup stratovolcano and cinder cones, 4 Ozernovsky lava field, Left and Right Ozernaya plateau, 5 Sedanka lava field and Titila volcano, 6 Gorny Institute volcano, 7 Dvuh’urtochnoe plateau, 8 Alnej-Chashakondzha massif, 9 Kekuknajsky lava field, 10 Kruki Ridge, 11 Esso monogenetic cones, 12 Ichinsky stratovolcano and monogenetic lava field, 13 Achtang stratovolcano and monogenetic cones, 14 Khangar volcano. Location of samples from other studies, dated with 40Ar-39Ar method in this study: 15 Daik at Khukhch Mount (Perepelov et al. 2006b); 16 Shield volcano under Bolshoj (Perepelov et al. 2006a); 17 Shield volcano under Belogolovsky (Churikova et al. 2001); 18 Bakening plateau (Dorendorf et al. 2000a). Black symbols—volcanoes studied in this paper, gray symbols—samples from other studies, dated with 40Ar-39Ar method in this study; white symbols—volcanoes not studied in this paper

Full size image

Variable magma sources, their spatial variations, and the composition of different slab fluids involved in magma genesis were characterized for a 220 km-wide transect across the arc at the latitude of the Kluchevskaya Group of volcanoes in (Churikova et al. 2001, 2007). According to these studies, the process of magma generation in Kamchatka has been described as the interaction of four components: a variably depleted N-MORB mantle wedge, OIB-type enriched mantle, slab fluids as well as slab melts at the northern termination of volcanism at Shiveluch (Yogodzinski et al. 2001). Melting degrees range from 10 to 20% and the presence of an OIB-type enriched source is required for back-arc magmatism.

This study focuses on the Kamchatka back arc region of the Sredinny Range (SR), a region with abundant Quaternary mafic volcanism for which the plate tectonic setting is basically unknown (Volynets 1994; Tatsumi et al. 1995; Churikova et al. 2001, 2007; Perepelov 2004, 2005; Trubitsin et al. 1998; Avdejko et al. 2002; Fedorov and Shapiro 1998; Pletchov 2008, etc.). We will present new geochemical, isotopical, and geochronological data for these volcanic rocks and their possible variations in space and time based on a representative collection that covers this region. We also discuss causes and mechanisms of melting and present a model for the origin and setting of back arc volcanism in Kamchatka.

Geological setting and sampling

The Eastern Volcanic Front terminates in the north with a group of late Pleistocene cinder cones on the eastern slope of the Kumroch Range almost as far north as the mouth of the Kamchatka River (Fig. 1). The Central Kamchatka Depression terminates at the Shiveluch volcano, but a few smaller monogenetic vents exist up to Nachikinsky volcano ~150 km NE of Shiveluch (Ponomareva et al. 2007). Today, the Sredinny Range is located 400 km behind the present trench, but represented the active volcanic front in the Eocene and switched to the present back-arc position after the accretion of the Kronotsky terrain and the formation of the recent active volcanic front (see also Appendix 1 in Electronic Supplementary Material). Some authors estimate the age of this collision between 40 and 20 Ma (e.g. Alexeiev et al. 2006), others date it from 7 to 2 Ma (Lander and Shapiro 2007). An active Benioff zone below the SR is observed only up to the latitude of the Holocene Khangar volcano at nearly 350 km above the slab (Gorbatov et al. 1997). Thus, back arc seismicity is absent further north even though there is evidence for abundant recent volcanism in this area.

In contrast to the main southern part of Kamchatka, where the Pacific Plate subducts below the Okhotsk Plate with a rate of ~8 cm/a (Scholl 2007), the Bering part of the North American Plate (north of Shiveluch volcano) is likely to have been stable relative to Kamchatka since Miocene times (Trubitsin et al. 1998; Baranov et al. 1991; Park et al. 2001; Baranov et al. 1991; Avdejko et al. 2002; Davaille and Lees, 2004). Therefore, there is no active subduction and we should not expect young volcanism in this back arc region. However, voluminous Holocene volcanic activity in SR back arc was documented by tephrochronological and radiocarbon methods at Khangar (Bazanova and Pevzner 2001), Ichinsky (Pevzner et al. 2000; Pevzner et al. 2002), and Alnej and Gorny Institute volcanoes (Pevzner 2006). Spokojny volcano is located in the northernmost part of the SR and was also active ~5400 14C years B.P. (Pevzner 2006). Moreover, voluminous Holocene monogenetic mafic volcanism is also observed throughout the northern part of the SR with erupted volumes similar to those behind the active arc (Pevzner and Volynets 2006). Large fields of Late Pleistocene–Holocene cinder cones with basaltic and basaltic andesite composition were also described in the Sedanka river area (Dirksen et al. 2004). Monogenetic lava flows Kireunskij and Left Belaja River of the volcanic massif Alnej–Chashakondzha formed 2600 14C years B.P. (Pevzner 2004); the Ozernovsky lava flow covers an area of about 100 km2 and erupted about 9000 14C years B.P. ago (Pevzner et al. 2002). The northernmost Holocene monogenetic center of all Kamchatka is the Tobeltsen cinder cone, located about 180 km to the north of Shiveluch volcano. Its eruption occurred about 3500 14C years B.P. (Pevzner 2006). It is therefore obvious that large volumes of Holocene mafic lavas erupted all along the Sredinny Range back arc.

To elucidate the origin and source(s) of volcanism in this area we studied a >200 km profile of monogenetic centers, plateau lavas, and stratovolcanoes along the Sredinny Range from the southern Achtang lava field through Ichinsky, Esso, Kekuknajsky, Kruki Ridge, Alnej, Dvuh’urtochnoe plateau, to Sedanka area, Gorny Institute volcano, Ozernovsky lava field and Tekletunup volcano in the north (Fig. 1; Appendix 1 in ESM). Particular attention was paid to young volcanoes, whose Holocene age was established by radiocarbon and tephrochronological dating (Dirksen et al. 2004; Pevzner et al. 2002; Pevzner 2004). Some older (but mostly undated) Late Miocene–Pliocene plateau basalts of Ozernovsky area, Kruki Ridge, and Dvuh’urtochnoe plateau were also sampled for comparison. The detailed description of the localities is found in Appendix 1 (ESM).

Analytical techniques

Analytical work for this study has been done at the Geowissenschaftliches Zentrum at the Georg-August Universität, Göttingen (Abteilung Geochemie). Concentrations of major and minor elements (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, P) as well as selected trace elements (Nb, Zr, Y, Sr, Rb, Ga, Zn, Cu, Ni, Co, Cr, V, Ba, Sc) were analyzed in 117 samples by X-ray fluorescence spectroscopy (XRF) using a Philips PW 1480 automated sequential spectrometer. For calibration of major and trace elements, about 100 reference materials were used, a wide variety of synthetic and international geochemical reference samples from the US Geological Survey, the International Working Group “Analytical standards of minerals, ores and rocks”, the National Research Council of Canada, the Geological Survey of Japan, the South African Bureau of Standards, the National Institute of Standards and Technology, etc. The glass disks, used for both major and trace element determination, were prepared by mixing prefused lithium tetraborate, lithium metaborate, and LiF (4,200 mg) with the sample powder (700 mg) in platinum–gold crucibles and fused for 15 min at 1,100°C after which the melt was poured into pre-heated, polished 32 mm-diameter molds. Elements were analyzed using a 3-kW rhodium target X-ray tube for sample excitation. The detailed operating conditions and the specific details of analytical precision and accuracy are given in Hartmann (1994) and Hartmann and Wedepohl (1993). Analytical errors for major elements are better than 1% and for trace elements around 5%.

Additional trace elements were analyzed on 72 samples by inductively coupled plasma mass spectrometry (ICPMS). Whole-rock powders (~100 mg) were dissolved (as described in Yokoyama et al. 1999) in Teflon beakers with a 4 ml mixture of HF/HNO3 (1:1) and evaporated, then taken up in 2 ml HF/HClO4 (1:1) and refluxed two times with decreasing amounts of HClO4 at temperatures up to 190°C. After evaporation, 0.5 ml of 6 N HCl was added and evaporated. For the measurements, samples were redissolved in HNO3. Rock standards JA-2 and BHVO-1 were analyzed continuously together with samples from which we estimate an external error about 10–15% and higher for Ta. Some samples (293/1, 820904-1, 820904-22, SR0411, SR0424, SR0426, SR0428, SR0412, ALN0401) were repeatedly dissolved with different digestion schemes, but problems with unrealistically low Nb/Ta ratios either due to incomplete acid digestion or loss of HFSE to the container walls remained. For these samples, Nb, Ta, Hf, and Zr concentrations were determined by laser ablation technique on the glass disks with 1:3 lithium tetraborate/sample proportions. With this method, all samples, as well as standard sample BB gave close to chondritic ratios for HFS elements. From this, we recalculated concentrations based on XRF Zr values and the Zr/Nb, Zr/Hf and Nb/Ta ratios from laser ablation data.

Isotope ratios for Sr, Nd, and Pb were measured with a Finnigan MAT 262 RPQ II + mass spectrometer at Göttingen. Whole-rock powders (~100 mg for Sr and Nd and ~150 mg for Pb) were dissolved in HF/HNO3. The Sr and Nd-isotope ratios were corrected for mass fractionation to 86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219 and normalized to values for NBS987 (0.710245), and La Jolla (0.511847), respectively. Measured values of these standards over the last 3 years in Göttingen University were 0.710239 ± 0.000004 for Sr and 0.511844 ± 0.000003 for Nd. External errors (2σ) are estimated at <0.0004% for Sr and Nd isotopes. Lead isotopes were corrected to NBS981 standard (Todt et al. 1984). Normalization of our data to recommended values was done using variable mass fractionation factors (0.046%; 0.177%; 0.013%, and 0.004%) for different measurement periods. This was preferred over using an average fractionation factor of 0.075% amu because one set of five samples had to be normalized with a relatively high fractionation correction of 0.177% amu. Procedural blanks for Sr, Nd and Pb (261, 135 and 91 pg, respectively) were insignificant.

40Ar/39Ar step heat analysis for 11 samples was performed in the geochronology laboratory at the University of Alaska Fairbanks (UAF) on 250–500 micron-sized whole rock chips. The monitor mineral MMhb-1 (Samson and Alexander 1987) with an age of 513.9 Ma (Lanphere and Dalrymple 2000) was used to monitor neutron flux (and calculate the irradiation parameter, J). The samples and standards were wrapped in aluminum foil and were loaded into aluminum cans of 2.5 cm diameter and 6 cm height. The samples were irradiated in the research reactor of McMaster University in Hamilton, ON, Canada for 20 megawatt-hours.

Upon their return from the reactor, the samples and monitors were loaded into 2 mm diameter holes in a copper tray that was then loaded in an ultra-high vacuum extraction line. The monitors were fused, and samples heated, using a 6-watt argon-ion laser following the technique described in York et al. (1981), Layer et al. (1987), and Layer (2000). Argon purification was achieved using a liquid nitrogen cold trap and a SAES Zr-Al getter at 400°C. The samples were analyzed in a VG-3600 mass spectrometer at the Geophysical Institute, University of Alaska Fairbanks. Argon isotope ratios were corrected for system blank and mass discrimination, as well as calcium, potassium and chlorine interference reactions following McDougall and Harrison (1999). System blanks generally were 2 × 10−16 mol40Ar and 2 × 10−18 mol36Ar, which are 10–50 times smaller than fraction volumes. Mass discrimination was monitored by running both calibrated air shots and a zero-age glass sample. These measurements were made on a weekly to monthly basis to check for changes in mass discrimination.

Age spectra are shown in Fig. 2. A summary of all the 40Ar/39Ar results is given in Table 1, with all ages quoted to the ±1 sigma level and calculated using the constants of Steiger and Jaeger (1977). Analytical details can be found in Appendix 2 (ESM). The integrated age is the age given by the total gas measured and is equivalent to a potassium–argon (K–Ar) age. The spectrum provides a plateau age if three or more consecutive gas fractions represent at least 50% of the total gas release and are within two standard deviations of each other (Mean Square Weighted Deviation less than ~2.7). For some samples, multiple runs were done and the average age reflects a weighted average of the plateau or weighted mean ages for the runs.

Fig. 2
figure 2

40Ar/39Ar age spectra and isochron diagrams for basaltic rocks from Kamchatka. Ages are reported at the 1-sigma level (see Table 1). Location units correspond to the sample sites numbers in Fig. 1

Full size image
Table 1 Summary of 40Ar/39Ar step heat analyses for Kamchatka samples
Full size table

Results and discussion

The age of plateau and back arc volcanism in Kamchatka

Previously, most age information on pre-Holocene magmatic complexes of Kamchatka was derived from topographic comparison and palinological data from sedimentary deposits (Ponomareva et al. 2007). First 40Ar/39Ar dates for volcanic rocks from the Central Kamchatka Depression (Calkins 2004) show that the lava shield that forms the base to the Holocene Kluchevskaya group was generated 260–280 ka ago. We dated by 40Ar/39Ar several Late Miocene–Pliocene rocks from the Sredinny Range including plateau lavas and shield volcanoes underlying the Late Pleistocene–Holocene edifices (Table 1). Age spectra with well-defined plateaus (Table 1) are shown in Fig. 2 and range from ~6 Ma down to ~20 ka.

  1. 1.

    Plateau lavas in the Sredinny Range back arc of Kamchatka are Late Miocene (6 Ma) to Pliocene (3 Ma) in age. Given this age range and the volume and area estimated from exposed sections (Ponomareva et al. 2007), these plateau lavas have an average eruption rate of 10 × 106 tons/year, which is about twice the “normal” arc magma production rate but still considerably less that Late Pleistocene to Holocene eruption rates in the Central Kamchatka Depression (100–160 × 106 tons/year; Ponomareva et al. 2007). These plateau lavas, which presently are in a back arc position with respect to the active volcanic front, all have a well-expressed arc geochemical signature (see next section).

  2. 2.

    At 3.66 to 3.78 Ma different types of rocks—arc type plateau and intra-plate basalts as dikes (Perepelov et al. 2006b)—erupted simultaneously suggesting the existence of different or transitional tectonic regimes at this time.

  3. 3.

    The younger rocks at the Sredinny Range (<1 Ma to recent) are less depleted in HFSE and combine the subduction and within-plate-type signatures (see discussion below). These rocks are similar to present back arc rocks at Ichinsky (Churikova et al. 2001; Pearce 2005).

  4. 4.

    The formation of extensive plateau lavas and large shield volcanoes is a typical feature of arc volcanism in Kamchatka. These plateaus formed in different regions at different times: (a) Present back arc volcanic rocks (<1 Ma) are underlain by the Sredinny Plateau lavas (6–3 Ma), (b) Late Pleistocene–Holocene volcanoes of the Central Kamchatka Depression are built on an extensive plateau of basalts and mafic andesites which range in age from 260 to 280 ka and (c) the Eastern Volcanic Front also is constructed on extensive shield volcanoes which have not yet been dated but are likely of Late Pleistocene age.

Geochemistry of Kamchatka back arc lavas

The majority of studied rocks are olivine–clinopyroxene–plagioclase basalts with only a few more evolved compositions (andesites and dacites). In the former, olivine and plagioclase are always present as main phenocryst phases. Clinopyroxene is rare and often even completely absent as phenocrysts, with exception of Dvuh’urtochnoe plateau basalts, where it is very abundant. Rocks crystallized to a different degree, from almost aphyric to serial-porphyric, and are rich in Fe and Ti oxide phases (titanomagnetite and spinel) in the groundmass but rarely as phenocrysts.

Most Late Pleistocene–Holocene rocks belong to the calc-alkaline medium-K series and fall into the field of previously studied rocks of the Ichinsky volcano and surrounding monogenetic cones (Volynets et al. 1991; Churikova et al. 2001; Volynets and Churikova 2004; Volynets 2006; Bychkov et al. 2003). Some samples from Sedanka, Kekuknajsky and Tekletunup monogenetic fields belong to high-K series and fall into the field of monogenetic Ichinsky basalts with within-plate affinities.

The composition of the Late Miocene–Pliocene plateau basalts is different from any previously studied SR samples in major and trace elements (Figs. 3, 4, 5), and in comparison with Late Pleistocene–Holocene rocks, show systematic enrichments in CaO and depletion in TiO2, Na2O, and P2O5. While all plateau lavas have low K2O (Fig. 3a), samples from Left Ozernaya plateau belong to high-K series and have the highest K2O values (Fig. 3a). The Left Ozernaya plateau is also characterized by low silica and high FeO and MnO concentrations and by high TiO2 and P2O5 compared to other analyzed plateau basalts. It is interesting that the Right- and Left-Ozernaya plateaus (ROP and LOP) have rather different major element compositions, although geographically they are located very close to each other.

Fig. 3
figure 3

K2O vs. SiO2, Na2O vs. SiO2, TiO2 vs. SiO2, Al2O3 vs. SiO2 and Al2O3 vs. MgO diagrams for samples of the Sredinny Range. Light gray field corresponds to reference data from Ichinsky WPT (within-plate type basalts); dark gray field—SR IAB (island-arc-type basalts) from Churikova et al. (2001). The overall trends are typical for Kamchatka arc volcanoes. Rocks from Alnej-Chashakondzha and Gorny Institute volcano have most evolved compositions; majority of other sampled sites are represented by basalts and basaltic andesites. Effect of plagioclase fractionation is negligible below SiO2 55% and MgO 4%. Classification lines on K2O (wt.%) versus SiO2 (wt.%) diagram after Le Maitre et al. (1989)

Full size image
Fig. 4
figure 4

N-MORB-normalized trace elements patterns for rocks from Sredinny Range. Gray reference field includes the entire data set for SR Late Pleistocene–Holocene lavas, presented in this paper. The order of incompatible elements is after Hofmann (1988) with Cs and all REE added. N-MORB and OIB values after Sun and McDonough (1989). OIB composition: black dots. The arc signature is well pronounced even on the very northern part of the SR (a, c, e, g), where there is no subducting plate expected in Quaternary times. Late Pleistocene-Holocene volcanic rocks (a–e, g) are less enriched in LILE and less depleted in HFSE—compared to older plateau lavas (f, h)

Full size image
Fig. 5
figure 5

Incompatible element ratio plots for SR rocks. Symbols are same as in Fig 3. a Th/Yb versus Ta/Yb (Pearce 1983) distinguishes between primitive island arc rocks from depleted and enriched mantle sources. The displacement from the mantle field to higher Th/Yb ratios is caused by slab fluid/melt enrichment in Th over Yb, whereas variations in the mantle source are shown by a change in both ratios. Groups 3 and 4 rocks have a depleted mantle source even compared to NMORB; Group 1 has a variable addition of an enriched mantle source component. Group 2 takes the intermediate position. b, d, e, f low fluid-mobile/immobile element ratios are typical for Group 1; highest values are observed in Group 4. Rocks from Group 2 (Alnej-Chashakondzha) are similar in U/Nb (f) to Group 3 and 4 (plateau basalts), but in Nb/Ta, Zr/Sm and Zr/Y (c, d) to Quaternary monogenetic basalts

Full size image

All trace elements patterns have typical arc signatures with variably depleted HFSE and HREE and are enriched in LILE and LREE (Fig. 4). Most (with the exception of single analysis of Kruki Ridge basalt) have HREE concentration lower than NMORB expressed as YbNMORB/Ybsample by a factor of 1.05–2.5. LOP, ROP and some Holocene and Late Pleistocene basalts of Alnej–Chashakondzha massif and Tekletunup area show a maximum degree of depletion (YbNMORB/Ybsample > 1.65), i.e. this factor is not correlated with the age of the studied rocks.

Most Late Pleistocene–Holocene lavas have Nb and Ta concentrations (compared to LREE) higher than in typical arc-front lavas in Kamchatka, reflecting variable enrichment of the source in these elements (Group 1; Fig. 4a–e) compatible with previously published analyses for the SR (Churikova et al. 2001). They form a rather coherent group in all trace element ratio diagrams and have relatively low (but still arc-like) fluid mobile/HFSE elements ratios (Ba/Nb, Th/Ta, U/Nb).

Rocks from the Alnej–Chashakondzha volcanic massif (Group 2) have an intermediate composition between Late Pleistocene–Holocene and Late Miocene–Pliocene plateaus (Fig. 5a, b, e). These rocks have trace element patterns similar to lavas from Achtang volcano in southern part of SR.

Older rocks from the underlying Late Miocene–Pliocene plateau basalts (Group 3; 3.05 ± 0.14 Ma–6.19 ± 0.05 Ma, Table 1) show arc-type trace element patterns similar to the rocks of the Eastern Volcanic Front, some are even more depleted in HREE. Group 3 includes basalts from the Right Ozernaya plateau, Dvuh’urtochnoe plateau basalts, and Kruki Ridge basalts, as well as a single analysis of a Pleistocene cinder cone from the Kekuknajsky area. All these older plateau rocks are much more depleted in HFSE and REE compared to younger SR rocks, while fluid-mobile elements (e.g. Ba, Pb, Sr) have similar or higher concentrations. This group is thus enriched in U/Nb, Ba/Nb, Th/Ta, Sb/Ce, and La/Yb compared to the younger Late Pleistocene–Holocene monogenetic basalts (Fig. 5).

An extreme case in all diagrams is represented by Left Ozernaya plateau basalts (Group 4), the oldest studied samples (6.19 ± 0.05 to 6.09 ± 0.05 Ma). These are distinguishable from other plateau lavas in many trace element ratios (Fig. 5), but are particularly enriched in Ba, Sr, and LREE, depleted in HFSE, and they have higher La/Yb, Th/Ta, Th/Yb and lower Zr/Y, Zr/Sm ratios (Fig. 5; Table 2). These rocks have the highest Ba/Nb ratio observed (Fig. 5d)—5 times higher than in other plateau lavas and up to 80 times higher than in any other samples from the area. These large variations in Ba/Nb ratio are due to both variable Nb and Ba contents, although influence of low Nb concentrations to this ratio is more pronounced.

Table 2 Major and trace element concentrations in Sredinny Range volcanic rocks. Major elements concentrations are given in wt.%, trace elements in ppm
Full size table

Portnyagin et al. (2005) report substantial variations in composition of volcanic rocks from the Central Kamchatka Depression with respect to the location of volcanoes relative to the Bering zone of transform faults: Nb/Y, Dy/Yb are higher and Ba/Nb is lower in northern CKD volcanic rocks (volcanoes Khajlula and Nachikinsky, north of the Bering transform fault) in comparison with Kluchevskoy and Shiveluch, located above the actively subducting Pacific Plate, south of the Bering transform fault. In contrast, according to our data, volcanic rocks from Sredinny Range do not show any regular spatial changes in composition but rather vary with time (Fig. 6).

Fig. 6
figure 6

a Nb/Y, Ba/Nb and Dy/Yb variations in SR back arc rocks from the south to the north. b Northern CKD rocks (after Portnyagin et al. 2005) for reference.An increase in Nb/Y, Dy/Yb and the Dy/Yb decrease in CKD rocks at crossing the Bering transform fault marks the edge of the actively subducting Pacific plate. Note that in Sredinny Range back arc rocks the observed compositional variations are not connected with the location of volcanic centers relative to the Pacific plate edge (Bering transform fault). At the same time, SR Late Miocene–Pliocene volcanic rocks (plateau) differ from Late Pleistocene–Holocene volcanic rocks (stratovolcanoes and monogenetic centers) by lower Nb/Y, Dy/Yb and higher Ba/Nb ratios

Full size image

Sr, Nd, and Pb isotopes

Sr-, Nd-, and Pb-isotope data are presented in Table 3. All analyzed samples have “primitive” Sr and Nd isotopic compositions, as is typical for Kamchatka rocks from the EVF and CKD (Fig. 7) and fall in the range of 87Sr/86Sr 0.703074 to 0.703359 and 143Nd/144Nd 0.513018 to 0.513106. The exceptions are samples AB0221 (Holocene cinder cone from Sedanka area), which has a very low, N-MORB-like Sr isotope ratio and also a slightly higher Nd isotope ratio (0.513108), and TT-1019/1 (Pleistocene cinder cone from Tekletunup area), which shows relatively high Sr (0.703557) and low Nd (0.512916) isotope ratios, close to the enriched source of Churikova et al. (2001). Lead isotopes are within the field of previously reported SR compositions (Fig. 8). Minor variations in lead isotopes exist but are relatively insignificant with respect to the overall variations in Kamchatka (see inset in Fig. 8a). A noteworthy feature is the complete absence of correlations between radiogenic isotopes and fluid-mobile/immobile element ratios (Fig. 8d). Another surprising result is the isotopic similarity for most samples, despite large geographical, geological, and compositional differences given by variable HFSE/HREE ratios and HFSE contents, which indicate a heterogeneous mantle source. Within the small isotopic compositional range, there is no systematic spatial variation. However, with respect to age, minor isotopic differences occur, which we will exploit in the reconstruction of magma sources with respect to their tectonic setting.

Table 3 Isotopic composition of 87Sr/86Sr, 143Nd/144Nd, 206Pb/204Pb, 207Pb/204Pb and 208Pb/204Pb in the representative analysis of Sredinny Range volcanic rocks
Full size table
Fig. 7
figure 7

143Nd/144Nd vs. 87Sr/86Sr fields show previously reported compositions for Kamchatka rocks (Churikova et al. 2001; Dorendorf et al. 2000a, b) and new data from the SR back arc. Arrows indicate three components: slab fluid, MORB-type source and enriched mantle source. Other symbols like in Fig. 3

Full size image

Interpretation

The rocks sampled along the SR, are divided into two main types of different age: groups 1 and 2: overlying Late Pleistocene–Holocene monogenetic cones and stratovolcanoes and groups 3 and 4: underlying Late Miocene–Pliocene plateau. These two types of different age are compositionally uniform but significant differences exist between them (highly depleted N-MORB mantle source with low Ta/Yb, Zr/Sm, Zr/Y and high Ba/Nb, U/Nb, Th/Ta in groups 3 and 4 vs. variably enriched mantle source with high Ta/Yb, Zr/Sm, Zr/Y and low Ba/Nb, U/Nb, Th/Ta in groups 1 and 2). The SR rocks of different ages thus have rather variable geochemistry and significant variation in HFSE and fluid-mobile element contents at quite similar isotopic composition. They also differ in their morphology (plateau vs. monogenetic cones or stratovolcanoes) and erupted volumes. Sampling of rocks of preferably basaltic composition and the use of incompatible element ratios allows us to exclude possible effects of fractionation processes, and our choice of samples with most diverse compositions for isotope analysis should cover all variations in the potential magma sources: (a) mantle, depleted to a different degree by previous melting events; (b) enriched mantle; (c) subducted sediments; (d) assimilated crustal material; (e) partial slab melts (so-called “adakite component”), and (f) slab fluids generated by dehydration of the subducted oceanic crust (McCulloch and Gamble 1991; Sun and McDonough 1989; Hofmann 1988; Pearce and Parkinson 1993; Ayers 1998; Plank and Langmuir 1988; Poli and Schmidt 2002; Münker et al. 2004). We will discuss these potential components below.

Mantle source compositions

In determining the composition of the mantle source of our magmas, we followed the idea of McCulloch and Gamble (1991) that in a steady-state melting regime, the slab is required to supply the dominant proportion of LILE and lesser amounts of LREE, while HFSE and HREE are derived predominantly from the mantle, depleted to a different degree by previous melting events and partly replenished by asthenospheric flow. The method of calculating the mantle composition before its enrichment by a subduction component is based on the assumption that transport of HFSE and HREE in the slab fluid is limited (Pearce 1983; Pearce and Parkinson 1993). Such conservative behavior of HFSE during subduction in Kamchatka rocks was confirmed by high precision HFSE and Hf–Nd isotope measurements (Münker et al. 2004). However, it has been argued (Kessel et al. 2005) that under high-P/high-T conditions in deep slabs, a composite supercritical fluid may be responsible for elevated HFSE in (back-) arc settings. Here, however, low (MORB-like) Nd- and Sr-isotopic ratios in SR rocks and the absence of any correlation between radiogenic isotopes and enriched slab signatures prevent significant sediment involvement in magma genesis. Moreover, Nb and Ta concentrations in most enriched lavas from our collection are so high (26 and 1.6 ppm, respectively) that—in absence of an OIB-type mantle component—we should assume that about 90% of these elements (i.e. 23 ppm Nb and 1.4 ppm Ta) would have to be transported with such a supercritical component, which we consider unrealistic. Therefore, a mantle source composition can be constructed by interpolating between HFSE and HREE on a “spider-diagram” (Fig. 9, dotted line). Based on existing data on mineral-melt distribution coefficients (Rollinson 1993; White 2000; Green et al. 2000; Zack et al. 1997; Gregorie et al. 2000; Kennedy et al. 1993; Ionov et al. 1997; LaTourrette et al. 1995; Foley et al. 1996; Johnson 1998; Hauri et al. 1994; Hart and Dunn 1993; many others) and mantle mineralogy (Koloskov 1999; Jagoutz et al. 1979; etc., see Table 4), we calculated the composition of the mantle wedge before fluid addition, so as to get a best fit of approximated and modeled trace element distribution patterns. The degree of melting was estimated using Na2O6.0–CaO6.0 systematic (Plank and Langmuir 1988) and HREE content. Our data, when plotted on a (Na2O)6.0 versus (CaO)6.0 diagram, fall on the Plank and Langmuir (1988) trend of decreasing CaO and increasing Na2O at decreasing melting degree. Most old plateau lavas have CaO from 9.07 to 10.77 wt.% at Na2O 3.01–2.10 wt.% (melting degree about 17–20%), while younger rocks tend to group at the lower right part of the diagram with CaO 7.41–9.10 wt.% and Na2O 4.23–3.42 wt.% (melting degrees 10–15% and lower). These rough estimates will be refined by modeling of HREE behavior at different melting extents.

Fig. 8
figure 8

a, b Pb isotope systematic in comparison with previously studied Kamchatka (Churikova et al. 2001; Dorendorf et al. 2000a, b), Aleutians rocks (Miller et al. 1994), N-MORB (Miller et al. 1994) and pelagic Pacific sediments (Miller et al. 1994; Kersting and Arculus 1995). c, d208Pb/204Pb vs. 87Sr/86Sr and 87Sr/86Sr vs. Ce/Pb for SR rocks in comparison with previously reported Kamchatka and Aleutians rocks (Miller et al. 1994). Aleutian samples are marked with crosses, other symbols as in Fig. 3, fields as in Fig. 7

Full size image

According to our modeling, the mantle source of Late Pleistocene–Holocene rocks of SR can be described as a mixture of variably depleted N-MORB-type mantle (spinel lherzolite), with different amounts of enriched OIB-type mantle (30–55% garnet lherzolite, Fig. 9a, b) with degrees of melting between 8 and 10%. All Late Miocene–Pliocene plateau basalts are most likely produced by relatively high degrees of melting (about 20%) of only a strongly depleted N-MORB-type mantle for which we assume an amphibole harzburgite modal composition (Fig. 9d). In the Alnej–Chashakondzha area there were only few samples with MgO >4%. Therefore, we chose the sample with the lowest SiO2 content as representative of the most primitive melt. The composition of the mantle source was calculated as a mixture of 86% depleted N-MORB-type mantle (spinel lherzolite) and 14% enriched OIB-type mantle at 10% melting (Fig. 9c). Our model implies an amphibole-bearing residue after melting. Preliminary temperature estimates limited to a few samples (Volynets unpublished) are relatively high (>1,100°C for some rocks) but inconclusive with respect to differences between the magmatic temperatures for different age groups. If confirmed, these high temperatures would place the mantle source at the uppermost limit of amphibole stability in the mantle wedge (Foley 1991; Fumagalli and Poli 2005; Niida and Green 1999; Schmidt and Poli 1998; Wallace and Green 1991).

Our interpretation supports an OIB-type mantle component (Volynets 1994; Churikova et al. 2001, 2007; Münker et al. 2004); however, different mantle sources are evident in slightly variable isotopic compositions for only a few samples (Fig. 7). The HREE-depletion in Kamchatka arc rocks relative to N-MORB source was proposed to be a result of formation of older voluminous plateau basalt magmatism (Churikova et al. 2001; Münker et al. 2004). However, we observe that mantle sources of these Late Miocene–Pliocene plateau basalts (LOP and ROP) already had extremely depleted HFSE and HREE (Ybsample/YbNMORB between 0.46 and 0.78). This leads us to the conclusion that the melting event that caused a regional depletion of the mantle wedge below the Kamchatka back arc likely occurred prior to formation of the Late Miocene plateau basalts.

The outcome of our model is to some extend dependent on the choice of end-members and modeling parameters (composition, partition coefficients, and modal mineralogy of the source). We selected mantle source compositions and distribution coefficients, which allowed us to get the results closest to the observed values. For example, the OIB-like mantle source cannot be replaced by a less depleted mantle as proposed for Karymsky lavas (Portnyagin et al. 2007a), because HFSE contents would remain too low. Other partition coefficients (e.g. Workman and Hart 2005) for modeling the most enriched sample AV0292 resulted in slightly lower concentrations for most of elements in the modeled mantle source (difference is <4.2%), a higher OIB contribution (65% instead of 55%) and higher degrees of melting (11% instead of 8%). For the most depleted samples, the difference between two models in element concentrations would be higher and to achieve the better match with the real rock composition it would require to apply lower degrees of mantle melting. In any case, the inferred temporal variation in source composition is robust and does not depend critically on the modeling parameters.

Sediments and crust contamination

Pelagic sediments and their melts have high Pb- and Sr- concentrations and isotopic ratios (average 206Pb/204Pb ~ 18.8; 87Sr/86Sr ~ 0.711, Plank and Langmuir 1998). The addition of <1% amount of sediments to the mantle source, therefore, will cause a strong increase of isotope ratios in comparison with “pure” MORB mantle (Fig. 10). Since all our samples are very low in Sr- and Pb- and high in Nd-isotopes (i.e. close to MORB mantle), there is no evidence for a significant sediment involvement in magma generation in the Kamchatka back arc.

Fig. 9
figure 9

Modeled source compositions before slab fluid addition in comparison with the actual rock compositions. Dotted line: composition of mantle before fluid addition (after Pearce 1983; Pearce and Parkinson 1993) based on the relative compatibility of the trace elements. Bold black line: modeled mantle source composition. gray field: reference for all volcanic rocks from the studied area; gray line—composition chosen for geochemical modeling. All data are normalized to N-MORB (Sun and McDonough 1989). The amounts of each component and degree of melting from the model is given in insets

Full size image

This conclusion is in agreement with previously published data based on Pb and Be isotope data (Kersting and Arculus 1995; Churikova et al. 2001; Tsvetkov et al. 1989, etc.). Unradiogenic Sr isotope ratios also preclude significant assimilation of old continental crust in Kamchatka magmas, in particular for the most mafic rocks (Churikova et al. 2001).

An adakite component in SR magmas?

Participation of slab melts in magma generation would result in high Sr/Y (Sr/Yb) and La/Yb ratios and increased silica content at higher Mg concentrations. Due to the strong preference for HREE by garnet in the residue after melting of the subducted oceanic basalts, a significant LREE/HREE enrichment would be expected and be coupled to depletion in HREE (Fig. 11), which is not observed. Thus, an adakite component (i.e. slab melts) can be excluded from the discussion.

Fig. 10
figure 10

Schemes of mixing between N-MORB source and sediment, defined by strontium and lead isotopic compositions and concentrations of these elements in the end-members; SR volcanic rocks data are plotted for comparison. Black star: average Pacific sediment composition. White star: average N-MORB composition. a Black line indicates mixing curve between average N-MORB and average Pacific sediment. Gray star on the background and gray curve correspond to N-MORB with an elevated Sr isotope ratio of 87Sr/86Sr = 0.703, representing N-MORB + seawater, and relevant mixing curve between this MORB and sediment. Our data plot near this second mixing curve. b Mixing curve, calculated for Sr isotopes and Sr concentrations in end-members, and modeled partial melting line for the mix (gray line with indication of melt amount) and modeled partial melting line for the mix of N-MORB and subduction fluid (black line with indication of melt amount). Composition of the subduction fluid is after (Stolper and Newman 1994). c modeled mixing curve, calculated for Pb isotopes and Pb concentrations in the end-members, modeled partial melting of the mix (gray line) and modeled partial melting for the mix of N-MORB and subduction fluid (black line). d La/Sm vs. Pb (ppm) for the samples, analyzed for isotopic composition and modeled mixing curves for the partial melting of the mix of NMORB-source and sediment melt (gray line) and mix of NMORB-source and subduction fluid (black line). For further discussion see text

Full size image
Fig. 11
figure 11

Comparison of trace elements and REE in Aleutian adakites, Komandor Mg-andesites and SR rocks, studied here (gray field). a REE distribution patterns for compared rocks. Concentrations are normalized to C1 chondrite (Sun and McDonough 1989). Black thin line TT-1019/2 sample (Tekletunup area), black dotted line 293/1 sample (Chashakondzha area). b Sr/Yb vs. La/Yb in compared rocks. Typical adakites, as well as Mg andesites from Komandor islands with slab melt in their source, have much higher Sr/Yb and La/Yb then all studied SR rocks. Composition of Aleutian adakites and Komandor andesites after Yogodzinski et al. (1995)

Full size image

Composition and amounts of slab fluids

According to the recent data on mineral-melt-fluid distribution coefficients (Ayers 1998; Brenan et al. 1995; Keppler 1996; Stalder et al. 2001; Kessel et al. 2005) slab fluids should be enriched in LILE (K, Cs, Rb, Ba, Pb), to a lesser extent—LREE, and depleted in HFSE (Nb, Ta, Zr, Hf), Th and HREE (Gill 1981; Tatsumi et al. 1995; Churikova et al. 2001). U, Ba, Th, Pb are easily transported in fluids, while Nb, Ta, and Ce are much less soluble. Both element groups, however, should behave similarly during crystal-melt equilibration. Therefore, Ba/Nb, Th/Ta, U/Nb, Pb/Ce ratios do not vary during partial melting but should increase with minor fluid involvement. In the more depleted mantle source, less fluid is needed to elevate these ratios. To estimate the fluid composition and to distinguish the relative contribution of each element by mantle source and subduction component, we first added variable amounts of a fluid component to all modeled mantle compositions. The fluid composition was mostly based on Stolper and Newman (1994), only Ba, Nb, and Pb concentrations are taken from Brenan et al. (1995). For a detailed description of the calculation method, see Churikova et al. (2001). This fluid-modified source was then batch-melted to different degrees, using an amphibole-bearing mantle. The amount of fluid and the melting degree was adjusted individually in each case to get maximal agreement with observed whole rock-concentrations for most fluid-mobile elements, as Rb, K, Ba, Sr, and Pb (Fig. 13). An example of this step of the modeling is represented in Fig. 12, for model parameters, see Table 5.

Fig. 12
figure 12

Calculated mixtures of the modeled mantle source (bold line) and slab fluid after (Stolper and Newman 1994) for Sedanka area (black line with circles). Relative amounts of each component are shown in the right upper part of the diagram. Elevated Nb, Ta, Zr, Y concentrations in the fluid are shown by white circles; low Ba, Pb, Sr concentrations—black circles; gray line-modeled rock composition

Full size image
Table 4 Partition coefficients, used for modeling in this work
Full size table

Calculated compositions are usually over-enriched in Y, Zr, Nb, and Ta compared to observed compositions. This difference is less pronounced in models for compositions that include an OIB component, due to the lesser amount of the fluid involved in their genesis and higher HFSE concentrations in the OIB component. The composition of fluids in all rocks is rather similar. However, in detail, we see that HFSE and LREE elements are elevated in all Late Pleistocene–Holocene lavas compared to Late Miocene–Pliocene plateau basalts. At the same time, LILE are more enriched in the depleted rocks of Late Miocene–Pliocene age and show similarity to the lavas from the recent arc front.

Our approach assumes that the difference between the modeled mantle wedge base line and observed whole-rock composition is a result of fluid addition. Given that little is known about HFSE and HREE behavior in the fluid phase at high pressures and temperatures, the concentrations of HFSE elements in our model are somewhat arbitrary. Recent investigations of trace element behavior at high pressures show that at pressures equivalent to 180 km depth a supercritical liquid with melt-like solubility is observed, mobilizing most of the key trace elements (Kessel et al. 2005). On the basis of high-precision measurements of HFSE in Kamchatka volcanic rocks, (Münker et al. 2004) concluded that while Lu, Zr, Hf behave conservatively, Nb and Ta may be transported by subduction fluids, but the amount added is insignificant with respect to the overall mass balance. Probably, HREE and HFSE can be partitioned to the fluid to some (although not substantial) amounts under Kamchatka back-arc conditions, where this deep fluid is possibly silica-rich and more readily dissolving these elements (Keppler 1996; Stalder et al. 2001; Churikova et al. 2007). Th levels are relatively uniform for rocks of highly variable Nb and Ta. If a “dense” fluid were significantly involved rather than an enriched (OIB-) mantle component, it would be hard to explain that only Nb and Ta are carried by this fluid and Th is not. In general, Sredinny Range fluids demonstrate not only internal conformity but also are also similar to fluids that were calculated for the EVF and CKD Kamchatka lavas by Churikova et al. (2001). Substantial differences, as originally seen in distinct whole-rock compositions, are observed only for the most fluid-mobile element concentrations and their ratios—K/La, Cs/Rb, Ba/Th (Fig. 14). Cs concentrations in EVF fluids are much higher due to its particularly high mobility in fluids and the proximity of the EVF to the trench. Fluids in Late Miocene–Pliocene plateau magma sources of Group 3 are equally enriched in all these ratios, lower in Ce/Pb compared to Late Pleistocene–Holocene monogenetic lavas, and thus are similar to fluids involved in EVF magma genesis. Samples from Left Ozernaya River rocks of Group 4 show very similar composition to Group 3 but are distinct only for some of the most fluid-mobile elements (e.g. Cs, Mo). Noll et al. (1996), Hattori and Guillot (2003) and Wörner et al. (2001) used the observation that chalcophile and siderophile elements are even more readily partitioned into a fluid and therefore can even better constrain fluid involvements. Rocks of the first group are rather low in Sb/Ce ratio, probably due to the low amount of fluids in their source. Left Ozernaya River plateau again show lower values of Sb/Ce at higher Ba/Nb than in lavas of Group 3. These data suggest that all Late Miocene–Pliocene plateau lavas were formed from the same depleted N-MORB mantle source but were influenced by fluids of slightly different composition.

Fig. 13
figure 13

Modeled fluid compositions for SR rocks in comparison with EVF, CKD and Ichinsky WPT fluids, from this study and from Churikova et al. (2001). Gray reference field corresponds to the entire modeled fluid compositions for all SR rocks from this study. Concentrations are normalized to N-MORB (Sun and McDonough 1989)

Full size image
Table 5 Modeling parameters
Full size table

The amounts of fluid and OIB components in the source are correlated with the age of the rocks studied. Late Pleistocene and Holocene volcanic rocks (Group 1 and 2) have higher amounts of enriched mantle component in their source (14–55%); they also have relatively small estimated fluid additions to their sources (0.4–2%). Late Miocene–Pliocene lava flows and plateau basalts (Groups 3 and 4) do not have an OIB-type component in their source and have higher apparent fluid additions to their sources (1.5–4%; on average more than 2%). These rocks are also produced by higher degrees of partial melting (up to 20% in comparison with 8–10% for Late Pleistocene–Holocene rocks). In addition, they have a different mantle source mineralogy (harzburgite for LOP instead of lherzolite for the majority of SR rocks).

These observations lead us to two main conclusions with respect to mantle sources and fluids. First, the subducted slab plays an important role in Kamchatka back-arc magmatism, also at its northern end, where no active subduction can be traced by geophysical methods (Davaille and Lees 2004). In addition, we should not underestimate the importance of the enriched mantle component in producing back-arc magmas, without an OIB component many geochemical features of Late Pleistocene–Holocene volcanic rocks would be difficult to explain.

A second important observation is the similarity of slab fluid composition in rocks for all age groups for the whole Sredinny Range. Even though we found subtle differences (see above), the modeling results are quite robust: slab fluids in all Late Pleistocene–Holocene volcanic rocks are relatively enriched (i.e. less depleted) in HFSE and LREE and have similar overall patterns compared to the fluids in Late Miocene–Pliocene plateau lavas. This suggests that surprisingly similar fluids segregated from the same Pacific Plate below the arc front during the Late Miocene–Pliocene and the back-arc in the Quaternary.

The origin of plateau lavas and recent volcanism at the Sredinny Range

As known from geological and geophysical studies (Legler 1977; Lander and Shapiro 2007; Avdejko et al. 2002; Alexeiev et al. 2006), the active volcanic front of the Proto-Kamchatka subduction zone was located in Eocene–Miocene times in the present position of the Sredinny Range. Later, the Kamchatka arc system has been modified by stepwise accretion from south to north of the Kronotsky terrains (Fig. 16 in ESM) between 40 and 2 Ma (Alexeiev et al. 2006; Lander and Shapiro 2007).

Our data constrain the style of magmatism with respect to these tectonic events. The systematic change in SR rock geochemistry with time is interpreted as the result of this arc shift and a massive slab rollback as suggested by tectonic reconstructions. In this scenario, the SR represented the active volcanic front until ~3 Ma with voluminous plateau basalts derived from an extremely depleted N-MORB-like mantle source and high degrees of melting. The geochemical uniformity of this first group, also in comparison to the composition of modern arc-front magmas, suggests that the same processes and sources were involved.

Since the arc-front shifted trench-ward, the SR represents the modern back arc region. This is confirmed by the continuous geochemical zonation from arc front to back arc at the latitude of Ichinsky volcano, i.e. strong and gradual increase in LILE, LREE, and HFSE (Churikova et al. 2001, 2007). The younger group (Late Pleistocene–Holocene) is characterized by low-degree melting of a partly enriched (OIB-) mantle. The geochemical uniformity of all Late Pleistocene–Holocene volcanic rocks of the SR again suggests that these rocks were essentially formed from the same sources. Comparing the two groups, we found systematic differences in the composition of the mantle wedge; however, relative to larger variations in other arcs, these differences are comparatively small. To explain this relative homogeneity, we argue that volcanism in the northern part of the SR was mainly caused by large degrees of fluid-induced melting with slab fluids dominating the element budged while a significant sediment component is absent.

Compared to the volcanic front, mantle upwelling from the deep asthenospheric wedge must have aided melting in the back arc and the incorporation of the enriched OIB mantle components. Therefore, we suggest a general trend from fluid-dominated melting in the Eastern Volcanic front, to upwelling of a strongly fluid-fluxed mantle below the Central Kamchatka Depression (Dorendorf et al. 2000b) to melting of a fluid-enriched mantle aided by strong upwelling and decompression in the back arc region.

The nature of the northern termination of arc and back arc magmatism in Kamchatka

Geophysical research of the Kurile–Kamchatka triple-junction and Aleutian island arcs shows that the maximum depth of seismicity decreases from south to north below Kamchatka and seismic data do not show a clear signal for a down going oceanic plate below the northern part of the SR despite the fact that voluminous recent magmatism is observed (Pevzner and Volynets 2006). It has been shown by Kirby et al. (1996), however, that seismicity is lost at slab temperatures higher than 600–700°C due to its increasingly ductile behavior. Davaille and Lees (2004) also argued that the seismicity of the subducted Pacific slab is gradually decreasing to the north as result of slab edge heating in the mantle wedge and mostly due to a decreasing plate thickness. Moreover, the rate of subduction under Kamchatka across the Aleutian–Bering–Alfa fault zone from the active Pacific Plate to the stable Bering Sea Plate decreases from 8 cm/year to nearly nil, causing the increased heating of the lower velocity slab. Yogodzinski et al. (2001) argued that the edge of the Pacific Plate is traced below Shiveluch volcano while Portnyagin et al. (2005) show that slab edge may be further N and is marked by the Shisheisky complex (about 80 km N of Shiveluch volcano, Fig. 15).

Fig. 14
figure 14

Cs/Rb versus K/La and Ba/Th versus Ce/Pb for volcanic rocks from Kamchatka and for calculated fluids (insets in the upper right part of diagrams). Symbols are as in Fig. 3. Fields correspond to Kamchatka rock compositions for reference (Churikova et al. 2001; Dorendorf et al. 2000a, b). For discussion see text

Full size image
Fig. 15
figure 15

Tectonic configuration of the region, where three plates are joined on the base of data received by R/V “Volcanolog” (Seliverstov 1998). Pacific Plate subducts under Okhotsk Plate with the velocity 8 cm/year, and forms Kamchatka Trench on the boundary of these two plates. This trench is represented on the figure by solid line with triangles showing the subduction direction. The trench of the Proto-Kamchatka subduction zone is situated between the Bering Sea- and Okhotsk Plates and is shown by dotted line with triangles. Relative motion of Pacific Plate and Bering Sea Plate produces the system of the transform fault zones. It is possible to pick three main zones of this system: Aleutian trench, Bering transform fault, Alpha transform fault. These zones are represented by solid lines. Dotted lines show projections of these transform zones below the Okhotsk Plate. The data for Late Pleistocene–Holocene volcanism on the SR and on the north edge of the subducting Pacific Plate are represented by: upturned triangle—Shiveluch volcano (Yogodzinski et al. 2001), triangles—Shisheisky complex (Portnyagin et al. 2005), squares—data from (Pevzner and Volynets 2006), and diamonds—data from Churikova et al. (2001). New data from this paper are marked by circles

Full size image

Some information on the northern edge of the Pacific Plate and its border with the Bering Sea Plate is gained by the morphology of the down going oceanic plate (Baranov et al. 1991; Seliverstov 1998; Fig. 15) in the region of the triple junction, where this border is represented by a 150 km wide zone from the Aleutian- through Bering- to Alfa-transform faults. Based on the projected continuation of these faults under Kamchatka, we argue that the northern termination of the subducted Pacific Plate probably is not a single abrupt plate edge but rather is represented by a wide zone with segments separated by transform faults. Magmas with somewhat higher Sr/Y ratios were found at Shiveluch volcano and on the Shisheisky complex (Yogodzinski et al. 2001; Portnyagin et al. 2005) and were interpreted to result from a slab melt with a garnet residue derived from the heated slab edge at the northern slab termination. However, Holocene volcanism at this northern termination in the back arc does not show such an effect and this slab edge effect is weak at best in the volcanic front.

The en-echelon offset of the northern termination of Holocene volcanism to the N from the Eastern Volcanic Front to the Central Kamchatka Depression and the Sredinny Range probably better traces the slab edge. As no volcanoes of Holocene age are found north of the projection of the Alfa-fault, there is probably no active subduction to the N of the on-land projection of this transform.

With increasing depth, the melting regime may be increasingly modified by segmentation of the slab and mantle movement across the direction of the main vector of subduction (Davaille and Lees 2004) and thus magmatism will provide only a diffuse image of the slab edge in the back arc.

Conclusions

  1. 1.

    The rocks of the Sredinny Range are distinguished in terms of their morphology, geochronology, and geochemistry in two groups. The first group of Late Miocene–Pliocene age (3–6 Ma) represents voluminous plateau lavas derived from a highly depleted mantle source with substantially low HFSE and HREE and high fluid-mobile enrichments. The second group of Late Pleistocene–Holocene age (less than 1 Ma) are mostly monogenetic cones and stratovolcanoes that combine a subduction- with a within-plate-type signature with variable enrichment in all incompatible elements. Despite obvious differences between the two groups, all rocks belong to medium–high-K calc-alkaline series and are characterized by rather similar subduction trace elements signatures and primitive isotopic composition.

  2. 2.

    Trace element patterns and isotope composition of volcanic rocks allow us to identify three components that contributed to magma generation in the Sredinny Range of Kamchatka: (1) N-MORB-type mantle, depleted to a different extent, (2) enriched OIB-type mantle, and (3) a slab fluid. While the Late Miocene–Pliocene plateaus were produced by high-percentage melting (20%) of extremely depleted NMORB mantle, induced by large fluid addition, the Late Pleistocene–Holocene volcanic rocks were formed by lower degrees of melting (8–10%) from an enriched mantle source (average 70% variably depleted MORB + 30% OIB) and a smaller contribution from a slab fluid. Significant amounts of subducted sediments or their melts are notably absent in the magma source.

  3. 3.

    We argue that the systematic change in rock geochemistry at the same location is the result of the modification of the subduction setting of SR due to the shift of the subduction zone to the east during terrain accretion. The plateau lavas are in fact similar to the Kamchatka arc front lavas and represented the arc front at Miocene times. The overlying Quaternary rocks represent the modern back arc lavas to the presently active subduction zone.

  4. 4.

    Both, the uniform geochemistry of young volcanic rocks along the SR and the systematic across arc geochemical zonation from contemporary arc front to back arc indicates a trend from fluid-dominated melting in the Eastern Volcanic front, through the upwelling of a strongly fluid-fluxed mantle below the Central Kamchatka Depression to melting of a fluid-enriched mantle aided by strong upwelling and decompression in the present back arc region.

  5. 5.

    The northern edge of the subducting Pacific Plate is marked by the termination of Holocene volcanoes and is represented by a wide plate boundary transform fault zone. The absence of young volcanism in the Sredinny Range to the north of the on-land projection of the Alpha fault marks the plate boundary at depth.