petrological implications of strontium isotope
TRANSCRIPT
Geochemical Journal, Vol. 22, pp. 237 to 248, 1988
Petrological implications of strontium isotope compositions of the Kinpo volcanic rocks in Southwest Japan:
Ascent of the magma chamber by assimilating the lower crust
TAKERU YANAGI, HIROSHI ARIKAWA*, REIKO HAMAMOTO and
ISAMU HIRANO**
Department of Geology, Faculty of Science, Kyushu University, Fukuoka 812, Japan
(Received December 7, 1988; Accepted February 3, 1989)
In order to define the contaminant and the contamination process, strontium isotopic compositions were determined on thirty-six rock samples from lava flows of the Quaternary Kinpo volcano in the Bep
pu-Shimabara Graben, Southwest Japan. Forty-two rock samples were also analyzed for major chemical compositions. The rocks are all subalkalic. All but a few tholeiitic basalts are calc-alkalic with abundant phenocrysts. The rocks are characterized by rather high Sr contents, ranging from 387 ppm to 1106 ppm. 87Sr/86Sr ratio ranges from 0.70377 to 0.70537. Marked linear relationships are found between 87Sr/86Sr and 1/Sr of the rocks. Repetitive occurence of basalts and basaltic andesites in the stratigraphic succession and lines of
petrographic evidence for magma mixing suggest the operation of a periodically refilled magma chamber under the Kinpo volcano. The linear relationships between 87Sr / 86Sr and 1 / Sr, and also between some major oxide components are interpreted as the manifestation of eruption of lavas during the repeated magma mixing. The increase of 87Sr / 86Sr ratio is postulated to be due to the contamination by melts derived from the granulitic basic lower crust during the upward migration of the chamber to attain the heat balance.
INTRODUCTION
A considerable variation of 87Sr/86Sr ratios exists in volcanic rocks from island arcs (Faure, 1986). Their values are low and their range is narrow in rocks from immature arcs with thin crust
(Stern, 1982; Notsu, et al., 1983; von Drach et al., 1986; Bailey et al., 1987). But 87Sr/86Sr ratios are rather high and their range is wide in rocks from mature arcs with abundant granitic plutons
(Ishizaka et al., 1977; Matsuhisa and Kurasawa, 1983; Notsu, 1983; Kurasawa, 1985). This indicates that the magmas react in varing degrees with thick crust on the way to the surface. Accordingly, in order to compare the magmas sup
plied to mature and immature arcs, a reliable
geochemical method should be devised to eliminate samples contaminated by crustal
materials. Then with this method we can know
pristine strontium isotope compositions of the magmas supplied to the mature arcs from the
mantle. For this object, what we should do is to know in what situation the reaction takes place and how the reaction proceeds. On the other hand our knowledge concerning the genesis of arc volcanics has much improved in the last decade. Evidence for magma mixing
(Eichelberger, 1975; Sakuyama, 1981; Koyaguchi, 1986) is commonly found in all types of volcanic rocks, and now the magma mixing is
generally accepted as one of the key processes to form the variety of volcanic rocks in island arcs
*Present address: Idemitsu Oil Exploration Co., Ltd., Marunouchi, Chiyoda-ku, Tokyo 100, Japan **Present address: Tateno Dam Construction Office, Kyushu Regional Construction Bureau, Ministry of Construction, 561-3 Shimonabe, Kumamoto 862, Japan
237
238 T. Yanagi et al.
(Gill, 1981). The concepts of magma mixing and a periodically refilled magma chamber have become important ideas for understanding the
genesis of arc volcanics (Yanagi and Ishizaka, 1978). Therefore our interest is to know how the contamination, which causes variation in 87Sr/86Sr ratio , is incorporated into the batch fractionation scheme in the refilled magma
chamber. A time sequence of events happened in a
magma chamber is well expected to be reflected on the volcanic succession. How the contamination has proceeded in the chamber may be
projected on the volcanic succession as the stratigraphic variation of 87Sr/86Sr ratio. The associated variation of chemical composition of volcanic rocks may give constraints on the differentiation proceeding in the chamber.
With these in mind we will report the variations of 87Sr/86Sr ratios and chemical compositions in the volcanic succession of the Kinpo volcano situated in the western end of the Beppu-Shimabara Graben (Matsumoto, 1979), Southwest Japan, and discuss how the replenishment of magma into the chamber is related to the contamination.
GEOLOGICAL SETTING
The Quaternary Kinpo volcano lies in the western end of the Beppu-Shimabara Graben extending from Beppu to Shimabara cities to about 150 km long (Fig. 1). The volcano, 15 km x 10 km in size, is a composite stratovolcano comprising four closely associated volcanic cones and some parasitic andesite domes. The cones in the south are deeply dissected by erosion, indicating that these cones are older. Portions of the cones with abundant pyroclastics are much eroded and hence changed to valleys and depressions. The
portions covered with thick lava flows remain as ridges. The internal structure of the cones is accessible in this part. Cones in the north, Ninotake and Sannotake, are relatively young. They preserve their original slopes. There, it is difficult to examine their internal structure.
The basement of the volcano is covered with
thick Quaternary deposits and Aso welded tuffs. Accordingly it is difficult to directly know the
geology of the basement. Fragments of granites and schists are often found as xenoliths in lavas. These types of rocks may be the main constituents of the geologic units around the valcano. Cretaceous granitic batholiths are extensively ex
posed both to the north and to the southeast of the volcano. Schists of intermediate metamor
phic grades are associated with these granitic batholiths. The volcanic rocks are divided into six stratigraphic units as shown in Fig. 1. Each unit comprises many layers of lava flows and
pyroclastics. These stratigraphic units are called Matsuo (Mt), Kokinpo (Kk), Ishigamiyama (ly), Sannotake (St), Ninotake (Nt), and Ichinotake
(It) in the chronological order. Lavas, excluding a few basalts, are calc
alkalic andesites with 53-63%wt SiO2. They are
porphyritic with abundant phenocrysts. The phenocrysts consist of, in the order of abundance, plagioclase, clinopyroxene, orthopyroxene, hornblende, magnetite, and olivine. Their total amount ranges from 27% to 66%, generally decreasing with increasing Si02 content. Horn
blendes are much abundant in lavas of early stages (Mt, Kk, ly). The common association of olivine with hornblende in the same rock and the abundance of plagioclase phenocrysts with many minutes glass inclusions are petrographic characteristics of mixed magmas (Tsuchiyama, 1985). With the change of rock type from acidic andesite to basalt, hornblende phenocrysts are reduced in abundance and become much decom
posed into compact aggregates of minute minerals such as plagioclase, clinopyroxene, orthopyroxene, and magnetite. Basic rocks increase in the amount of plagioclase phenocrysts having sieve textures with many glass inclusions, or having zones clouded by abundant dust of minute glass grains. Detailed petrographic examination verified that all lava flows bear disequilibrium petrographic characters suggesting the role of magma mixing in their origin.
Sr isotopes of Kinpo volcanic rocks, Japan 239
I.
Kinpo ~.4
Beppu-Shimabara Graben Aso welded tuff
® Ichinotake volcanics ® Yoshio Formation Ninotake volcanics
Sannotake volcanics
® Ishigamiyama volcanics Kokinpo volcanics
® Matsuo volcanicsFault
Fig. 1. Geological sketch map of the Kinpo volcano.
ANALYTICAL METHODS AND RESULTS
Major chemical compositions of volcanic
rocks were determined with a X-ray fluorescence spectrometer RIGAKU CR11009. Details on the sample preparation and the calibration of standards are shown in Nakada et al. (1985). All standard glass disks were synthesized from chemical
reagents. Nakada et al. (1985) discussed the re
producibility and the precision in detail. The relative error for Si02 determination is about 0.5%. Analytical results on a dry basis are listed in Table 1, in the stratigraphic order with the
youngest at the top. For determination of strontium content and
strontium isotope composition, the conventional
isotope dilution method was employed. Ba(N03)2
coprecipitation method (Baadsgaard and Lerbekmo, 1983) was adopted here for stron
tium separation. Strontium was purified from
barium with cation exchange resin Dowex 50W
X8. Isotopic analyses were done on spiked samples with a mass spectrometer JEOL JMS05RB. All the data were normallized to 87Sr/86Sr=0.1194. Measurements of isotope com
position of spiked E and A SrCO3 samples during the course of this study show the average 87Sr / 86Sr ratio of 0.70799±0.00006(1a). The
average of standard deviations of 87Sr/86Sr ratios in the samples is 0.0001(0.014%). Rb, content was also determined by the isotope dilution method. Analytical errors are less than 0.5% for Sr and less than 1 % for Rb. Analytical results are listed in Table 2. The chemical compositions show that all the lavas are subalkalic. When plotted on the Si02 vs. Total FeO / MgO (T.FeO / MgO) discrimination diagram (Miyashiro, 1974) most of the data lie in the field of calc-alkalic volcanic rocks (Fig. 2). Lavas of the early stages (Mt and Kk) lie more closely to the discrimination line or in the field of tholeiitic volcanic rocks (Fig. 2). Lavas
240 T. Yanagi et al.
Table 1. Chemical compositions of lavas from the Kinpo volcano
Sample No.
Unit Si02 Ti02 A1203 Fe203 MnO MgO CaO Na20 K20 P205 Total
Ichinotake
82042519
Ninotake 82091101
82091102
82091104 82091113
82091116 82050416
82091115
Sannotake
82091112
82091111
82090901
82091109 82091108
82091107 82091106
82091105
Ishigamiyama
82042501 Kokinpo
82030715
84032307 82042805
82042321
82043002
84032304
82042927
82050511
82042907
82030816
82042330
82050413
82042513
82090510
82090502
82031609
Matsuo
82042202
82031009B
82031012
82030806
82030906
82032107
82042904
82032403
82030726
It
Nt
Nt
Nt
Nt
Nt
Nt
Nt
St
St
St
St
St
St
St
St
ly
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Mt
Mt
Mt
Mt
Mt
Mt
Mt
Mt.
Mt
61.28 0.63 18.62
56.80
58.12
58.86
61.45
58.50
58.36
56.81
57.40
55.23
52.95
53.73
52.97
58.14
60.57
58.11
0.68
0.66
0.70
0.59
0.64
0.66
0.75
0.73
0.84
0.88
0.85
0.88
0.71
0.63
0.70
17.13
17.39
17.55
17.34
17.03
16.97
17.82
17.37
18.06
16.67
16.53
16.40
17.89
17.71
17.44
58.57 0.66 17.80
51.27
56.46
58.31
59.33
56.87
57.93
52.92
58.18
57.60
57.86
59.98
59.70
58.92
56.96
59.04
58.39
57.14
54.81
59.68
50.75
56.19
57.46
57.48
62.81
53.19
1.02
0.86
0.72
0.70
0.86
0.79
1.00
0.65
0.79
0.82
0.67
0.86
0.77
0.88
0.78
0.72
0.85
0.91
0.69
1.10
0.87
0.89
0.79
0.62
0.97
17.90
17.79
18.11
17.97
17.51
17.84
17.67
17.08
17.55
18.09
18.16
17.81
17.95
18.22
17.68
18.33
18.14
18.16
19.13
18.77
18.22
18.09
17.11
17.87
17.74
6.53
7.64
7.37
7.21
6.02
7.09
7.09
7.72
7.63
7.96
8.79
8.42
8.65
7.36
6.44
7.57
6.96
9.79
7.78
7.23
6.89
7.60
7.28
9.07
7.10
7.51
7.16
6.54
7.23
7.31
7.71
7.37
7.08
7.32
8.03
6.48
9.16
7.89
8.12
7.69
5.74
9.14
0.16
0.16
0.15
0.15
0.13
0.14
0.14
0.15
0.16
0.16
0.17
0.16
0.16
0.15
0.14
0.15
0.15
0.16
0.17
0.16
0.16
0.16
0.15
0.17
0.16
0.17
0.14
0.16
0.15
0.16
0.15
0.16
0.16
0.15
0.19
0.17
0.18
0.18
0.16
0.16
0.13
0.15
2.13
5.44
4.77
4.51
3.43
4.86
4.78
4.59
5.06
5.05
6.51
5.66
6.44
4.76
3.71
4.67
3.21
4.90
3.44
2.69
2.73
3.13
3.51
4.30
4.43
3.76
3.46
2.73
3.18
2.44
2.95
2.91
3.09
3.44
3.96
2.72
4.74
3.77
3.19
3.90
2.46
4.91
6.43
8.09
7.13
6.93
6.62
7.23
7.63
7.86
7.66
8.26
9.28
8.96
9.35
7.04
6.59
7.04
7.82
9.48
7.68
7.31
6.98
8.07
7.42
9.64
7.86
7.55
7.30
6.74
6.41
7.16
7.41
6.90
7.46
7.69
8.76
6.72
10.64
8.09
7.31
7.72
5.95
8.32
2.69
2.84
2.77
2.65
3.20
3.04
3.08
2.81
2.62
2.65
2.19
2.58
2.16
2.84
3.20
2.81
3.17
3.14
3.07
3.20
3.41
3.31
3.17
3.18
2.94
3.10
3.22
2.88
3.01
3.40
3.58
2.97
3.23
2.92
3.12
3.15
2.91
3.27
3.32
3.25
3.39
2.97
1.54
1.29
1.43
1.35
1.57
1.43
1.42
1.37
1.47
1.40
1.87
2.09
1.88
1.23
1.51
1.23
1.44
1.36
1.74
1.58
1.65
1.69
1.56
1.46
1.41
1.41
1.41
1.41
1.76
1.79
1.72
1.87
1.33
1.36
1.19
1.14
0.97
1.31
1.74
1.53
1.32
1.54
0.08
0.15
0.14
0.12
0.11
0.13
0.13
0.16
0.19
0.19
0.34
0.42
0.36
0.12
0.13
0.12
0.12
0.20
0.23
0.19
0.15
0.21
0.21
0.20
0.19
0.17
0.17
0.14
0.18
0.28
0.22
0.20
0.20
0.16
0.22
0.19
0.17
0.22
0.20
0.20
0.19
0.21
100.09
100.22
99.90
100.03
100.46
100.09
100.26
100.04
100.29
99.80
99.65
99.40
99.25
100.24
100.63
99.84
99.90
99.22
99.22
99.50
99.97
99.41
99.86
99.61
100.00
99.61
99.63
99.41
100.29
100.18
99.80
99.83
99.99
99.17
99.35
100.07
99.39
100.01
100.48
99.83
100.48
99.14
of the later stages (St and Nt) are plotted much apart from the discrimination line with low and almost constant T.FeO/MgO ratios, indicating that calc-alkalic nature is much developed in the
rocks of later stages.
Marked linear relationships are found among
selected components such as Si02, CaO and T.Fe203 (Fig. 3). The linear relationship is also
Sr isotopes of Kinpo volcanic rocks, Japan 241
Table 2. Analytical results of Rb and Sr concentrations
Sample No.
Stratigraphic Rock unit type
Rb (ppm)
Sr (ppm)
Rb / Sr 87Sr / 86Sr
Ninotake
82091101
82091102
82091104
82091113
82091116
82050416
82091115
Sannotake
82091112
82091111
82090901
82091109
82091108
82091107
82091106
82091105
Ishigamiyama
82042501 Kokinpo
82030715
84032304
82043002
84032101
82030816
82042907 82042330
82050413
82042513
82090510
Matsuo
84032201W
82031609
82031009B
82031012
82030806
82030906
82032107
82042904
82032403
82030726
Nt
Nt
Nt
Nt
Nt
Nt
Nt
St
St
St
St
St
St
St
St
ly
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Kk
Mt
Mt
Mt
Mt
Mt
Mt
Mt
Mt
Mt
Mt
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Basalt
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Andesite Andesite
Andesite
Andesite
Andesite
Andesite
Andesite
Basalt
Andesite
Andesite
Andesite
Andesite
Andesite
39.3
44.2
38.6
49.0
44.1
44.0
45.3
41.2
36.9
68.9
58.1
50.0
35.8
48.9
37.7
42.3
17.37
45.2
49.7
34.2
36.2
37.3
38.1
50.9
57.5
41.5
35.1
32.6
29.1
17.36
10.14
30.4
53.2
40.5
35.8
27.2
488
443
422
424
450
464
510
577
709
965
974
976
387
412
387
490
1106
610
613
868
567
587
476
485
532
805
682
699
632
662
700
591
516
579
565
1050
0.0806
0.0997
0.0915
0.1156
0.0978
0.0948
0.0887
0.0714
0.0520
0.0715
0.0596
0.0512
0.0925
0.1185
0.0973
0.0863
0.0157 0.0742 0.0811 0.0394 0.0638 0.0635 0.0800 0.1049 0.0910 0.0514
0.0515
0.0466
0.0461
0.0262
0.0145
0.0515
0.1031
0.0700
0.0634
0.0259
0.70463 ±12
0.70463+ 9
0.70487±19
0.70469±12
0.70456±10
0.70451± 8
0.70463 ± 6
0.70424± 7
0.70410±13
0.70377± 8
0.70383+ 7
0.70380± 6
0.70477.h 8
0.70491.t 9
0.70477± 6
0.70537±20
0.70392±18
0.70447±17
0.70432±18
0.70411± 7
0.70461±11
0.70455 ±24
0.70474± 7
0.70480± 6
0.70442±16
0.70420± 6
0.70473± 7
0.70474±12
0.70462±12
0.70495±14
0.70427±10
0.70461± 6
0.70464± 6
0.70462 ± 11
0.70512± 3
0.70435± 9
found between MgO and A1203 (Fig. 4), though
the variation of A1203 is very slight. In Fig. 5 87Sr /86 Sr ratio is plotted on the ver
tical axis and 1000 / Sr is plotted on the horizontal axis. In this diagram the lavas from the same unit show marked linear arrangements with
positive slopes. Linearity is excellent. Some of the lavas from Unit Mt is higher in 87Sr / 86Sr ratio than others, although the rocks from this unit
split into two different linear arrangements. The lavas of the Units St and Nt are low in 87Sr/86Sr ratio and lie on the almost same correlation line. The correlation lines shift their positions pro
gressively downward in Fig. 5 with eruption time. At the same time, the slopes progressively become gentler, resulting in a radial variation of the correlation lines.
If two isotopically distinct magmas are mixed
242 T. Yanagi et al.
65
60
Si02
55
50
Calc-alkalic
• D
0
•
A
A
000
00
6P
~ 0
0
0
•, 0 0
0
a
% 10.0
9.0
8.0
7.0
6.0
0
0
CO k
A
0Ib 0
.pe
T.Fe203
Tholeiitic
1.0 2.0 3.0
T.Fe0/MgO
Fig. 2. SiO2 vs T.FeO/MgO discrimination diagram
for the Kinpo lavas. The line of discrimination between thollitic and calc-alkalic volcanic rocks is taken
from Miyashiro (1974). T.FeO means the total iron content in rocks, expressed as FeO. Solid circle, Matsuo volcanics(Mt); open circle, Kokinpo volcanics(K); open circle with a virtical bar, Ishigamiyama volcanics: open triangle, Sannotake volcanics(St); solid triangle, Ninotake volcanics(Nt); open circle with a horizontal bar, Ichinotake volcanics(It).
11.0
10.0
9.0
8.0
7.0
6.0
5.0
0
0
0 8
.
A
n
o (1°001x48
CaO
pp
O
.
50 55 60
Si02 (°/o)65
Fig. 3. T.Fe2O3 vs SiO2 and CaO vs SiO2 diagrams
for the Kinpo lavas. Both T.Fe2O3 and CaO decrease linearly with the increase of SiO2. The total iron content in rocks is expressed as T.Fe2O3. The symbol annotations are the same as in Fig. 2.
A1203
20.0
15.0
10.0
.o
od~vio0 ZA
2.0 3.0 4.0 5.0 6.0 7.0
MgO
Fig. 4. A1203 vs MgO diagram for the Kinpo lavas . The variation of A1203 is very limited. The symbol annotations are the same as in Fig. 2.
Sr isotopes of Kinpo volcanic rocks, Japan 243
87Sr/86Sr
0.706
0.705
0.704
0.703
0
0 0 0 0
.
VC 0
.
a
00
oft c
A
.A 16
0.5 1.0 1.5 2.0 2.5 3.0
10001Sr
Fig. 5. 87Sr/ 86Sr vs 1000/ Sr diagram for the Kinpo lavas. Linear relationships are found between 87Sr/ 86Sr and I/ Sr for individual stratigraphic units.
with each other at different proportions, the mixed magmas may trace exactly a straight line with a certain slope on this diagram. The line connects the two end members. If a basic magma differentiates by settling plagioclase, the residual magma may trace a horizontal line with a constant 87Sr/86Sr ratio. The linear arrangement of data with a positive slope, therefore, is indicative of the mixing origin of the lava flows; mixing of isotopically distinct magmas. For each stratigraphic unit the mixing is almost approximated by a binary system with a basaltic high-Sr magma having low 87Sr / 86Sr ratio and a dacitic low-Sr magma having high 87Sr/86Sr ratio, so far as Sr contents and S7Sr / 86Sr ratios are concerned. The radial variation of correlation lines suggests that the dacitic low-Sr magma lowered its 87Sr / 86Sr ratio successively with time .
Ishizaka et al. (1977) analyzed volcanic rocks from the Myoko volcano group and reported the same fact that the 87Sr/86Sr ratio is high at about 0.7053 in rocks of early stages but decreases successively with time to 0.7042.
ESTIMATE OF CONTAMINANT
The variation of 87Sr/86Sr ratio in the Kinpo volcanic rocks shows that the magmatic system is open. Therefore the problem is what kind of materials are involved in the origin of the magmas. Since the lower crust at high temperatures is the first material for the magma to meet with during the ascent from the mantle to the surface, it is the most promising candidate as the
contaminant. But the lower crust is not accessible for direct observation. Granulitic and gabbroic xenoliths in basalts in Uta-jima, a small islet off the Japan Sea coast of Southwest Japan,
are believed to represent a part of the lower crust because of their occurrence as xenoliths, basic compositions, and high-P high-T metamorphic characters (Murakami, 1975). The same
granulitic and gabbroic xenoliths are found in basalts in northern Kyushu (Ishibashi, 1970). Though they are a member of the upper crust, Cretaceous granities are another candidate, since they are rocks making a large part of the basement around the volcano, and they are often contained as small xenoliths in the lava flows.
In connection with the selection of these
244 T. Yanagi et al.
0.712
0.709
0.706
0.703
87Sr/
86 Sr
A •r
M
rJ
0 0
Cr~~ o °
©°°° o0 0
°0
B 0 00 0
iDUtajima glanulites
a
0 a
0 5.0 10.0
1000/Sr
Fig. 6. 87Sr/86Sr vs 1000/Sr diagram for granites in North Kyushu, and granulitic and gabbroic xenoliths in basalts from Utajima, southwest Japan. Lines AB and CD limit the area within which the Kinpo lavas lie. The
granites (squares) and the xenoliths (circles) lie on the extension of the trend shown by the volcanic rocks. M indicates the possible composition of the primary magmas.
87Sr/86Sr
0.709
0.706
0.703
B
Utajima granulites
o .' Melt derived from
B: •• granulites I
A
C
0oFlk
a
a
D'
0
o° 11
oalik
Granites in North Kyushu
0
0 0.1 0.2 0.3 0.4
Rb/Sr
Fig. 7. 87Sr/86Sr vs Rb/Sr diagram for the Kinpo lavas, granites in North kyushu, and granulitic and gabbroic xenoliths from Utajima. Solid circle, Kinpo lavas; open circle, granulitic and gabbroic xenoliths from Utajima; square, granites in North Kyushu. Lines AB and CD limit the area within which the compositions of the Kinpo lavas are distributed. Area BB'D'D is the field in which the contaminant should lie. The rectangle shows the composition range of melts derived by partial melting of the granulitic and gabbroic xenoliths.
Sr isotopes of Kinpo volcanic rocks, Japan 245
representative contaminants, it should be
noticed that the conspicuous increase of seismic P-wave velocity at a depth of about 15 km is ob
served in the northern and central part of
Kyushu (Mitsunami, 1974). The increase of the
P-wave velocity in the depth interval between
about 16 km and 30 km is very small and
smooth, suggesting a homogeneous constitution of the lower crust.
The granulitic and gabbroic xenoliths
(Ishizaka et al., 1984), and granites from northern Kyushu (Yanagi, 1975) are plotted in the s7Sr / 86Sr vs 1000 / Sr plot of Fig . 6. Lines AB and CD limit the area within which the Kinpo volcanic rocks are plotted. It is evident that the xenoliths and the granites lie on the extension of the trend shown by the Kinpo volcanic rocks. Therefore both of these seems to be good can
didates as the contaminant. Figure 7 is a 87Sr / 86Sr vs Rb / Sr diagram on
which the granulitic and gabbroic xenoliths and the granites are plotted again. On this diagram, too, the magma produced by binary mixing traces a straight line connecting the two end member magmas. The Kinpo volcanic rocks lie within the area bounded by the lines AB and CD. In this diagram the contaminant should lie in a wide area limited by the lines BB' and DD'. However, none of the granites nor the xenoliths lie in this area. Therefore it is not likely that these granulitic and gabbroic rocks or the
granitic rocks were directly incorporated into the magmas under the Kinpo volcano. Then, in the next place melts produced from these rocks will be examined. To discuss the partial melting of rocks of
granitic compositions, distribution coefficients should be known for rubidium and strontium.
The distribution coefficient of strontium between
plagioclase and coexisting melt is dependent on
plagioclase composition and is about 2 to 3 for sodic plagioclase (Yanagi, 1975; Cox et al., 1979),
while about 3 to 5 for K-feldspar (Yanagi, 1975;
Cox et al., 1979). Bulk distribution coefficient of strontium between a melt and coexisting granitic
solid is dependent mostly on the abundances of
plagioclase and K-feldspar in the solid. Taking
the abundances of plagioclase and K-feldspar in the granites as about 50% and 20%, respectively, the bulk distribution coefficient of strontium is estimated to be about 2. The bulk distribution coefficient of rubidium
is mostly determined by the abundances of biotite and K-feldspar. The distribution coefficient of rubidium for biotite and Kfeldspar are about 2 and 0.7, respectively
(Yanagi, 1975; Cox et al., 1979). Since the amount of biotite is less than 10% in the
granites, the bulk distribution coefficient of rubidium is less than 0.4. Therefore, in partial melting of granitic rocks the produced melt decreases in strontium, while increases in rubidium. It is evident that the melts so formed have Rb/Sr ratios larger than those of the parental granites. Such melts are plotted far apart from the area BB'D'D toward the right hand side of the parental granites in Fig. 7. Therefore both the granites themselves and the melts derived from them are not the candidate. On melting of the basic granulitic xenolith, the strontium distribution coefficient depends mostly on the abundance of plagioclase in the solid, which is about 60% (Murakami, 1975). Since the strontium partition coefficient for calcic plagioclase is about 1.5 to 2 (Yanagi, 1975; Cox et al., 1979), the bulk partition coefficient between granulites and their melts is estimated to be close to unity. The average Sr content of the xenoliths is about 408 ppm (Ishizaka et al., 1984). Since rubidium is not incorporated into
crystal structures of the constituent minerals of the gabbroic granulites; plagioclase, ortho
pyroxene and clinopyroxene, the distribution coefficient of rubidium between the granulitic rocks and the derived melts may probably be less than 0.1, and possibly close to about 0.04
(Yanagi, 1975; Cox et al., 1979). Therefore rubidium is concentrated in melts on melting of the granulites. Accordingly the Rb/Sr ratio of the melts is magnified by about 25 times at maximum, depending on melting degrees. Since Rb / Sr ratio of granulites is about 0.004
(Ishizaka et al., 1984), it is possible to produce melts with Rb / Sr ratios in a range of 0.004 to
246 T. Yanagi et al.
0.1. This range is in accordance with the probable range of Rb / Sr ratio (0.03 to 0.13) of the contaminant (Fig. 7). 87Sr/86Sr ratio of the
granulitic and gabbroic xenoliths ranges from 0.705 to 0.706 (Ishizaka et al., 1984). This means that the melts produced by partial melting of the
granulites may lie within the rectangle of Fig. 7. The melts lying in this area is most suitable as the contaminant. Accordingly it is concluded that magmas which produced the Kinpo volcanic rocks were contaminated by melts derived by partial melting of the basic granulites which probably make the lower crust. The increase in A1203 of the dacitic andesites (Fig. 4) may be inter
preted in this context, since A1203 is high in the granulitic and gabbroic xenoliths (Murakami,. 1975).
DISCUSSION
Fractionation model
The isotopic evidence for repeated magma mixing and repetitive occurrence of basalts and basaltic andesites in the volcanic succession are indicative of a refilled magma chamber under the Kinpo volcano. Batch fractionation in an arctype refilled magma chamber was first proposed and discussed by Yanagi and Ishizaka (1978) as the main magmatic process which can transform the tholeiitic primary magma into calc-alkalic volcanic rocks and eventually into a magma with the chemical composition of the average continental crust. Furthermore they stated that the heat balance in the refilled magma chamber is necessary to keep the batch fractionation working. The concept of heat balance is the important clue to solve the problem how magma settles its chamber in the crust (Yanagi, 1981). The chamber model lying at a shallow depth in the crust was discussed by Yanagi et al. (1988) in connection to repeated volcanic eruptions of the Sakurajima volcano in the historic time. These
chamber models are summarized schematically in Fig. 8 to show how the contamination
proceeds. If the chamber lies at a depth where the influx
of heat associated with the replenishment of
magmas exceeds the loss to the surface through
thick crust, the chamber may grow in volume
with time. The rocks around the chamber is
heated and the accumulated energy may bring
about melting of the ceiling of the chamber,
resulting in the migration of the chamber toward
the shallow depths. During the migration from
the M-discontinuity, the magma digests the
crustal material. This migration occurs in
association with the repetitive supply of primary
magmas from the mantle. The rate of heat loss
to the surface increases while the migration
proceeds. The migration continues until the chamber rises to a depth where the heat loss is
balanced with the heat supply by the primary
magmas.
Since the magma in the chamber digests the crustal material, the magma is contaminated and increased in its 87Sr/86Sr ratio (O'Hara, 1977). Lavas erupted during the magma mixing have 87Sr/86Sr ratios varing between those of the
primary magma and the contaminated magma in the chamber. Once the chamber has migrated up to a level where the heat balance is realized, the magma in the chamber has no excess energy to melt the wall rock. Resultantly, the 87Sr / 86Sr ratio decreases with the successive replenishment of the primary magma from the mantle. The variation in 87Sr /16 Sr ratio observed in the Kinpo volcanic rocks accords well with this model. The same but much more marked case has been observed in the Myoko volcano group (Ishizaka et al., 1977), where the 87Sr/86Sr ratio increases first in the rocks of very early stages, passes the maximum, then decreases in middle stages and becomes constant in the rocks of later stages.
Lava eruption during the mixing The linear~relationships between 87Sr/86Sr
and 1 / Sr (Fig. 5) are realized only when lava eruptions occur during the magma mixing. Since the crystallization of plagioclase makes a large change in strontium content of the residual magma, the linear relationships would be obscured by the crystallization associated with
the mixing. The fact that tight linear relationships are found in all stratigraphic units (Fig. 5)
Sr isotopes of Kinpo volcanic rocks, Japan 247
(a) (b)
V
Upper Crust
Lower Crust
CS
V
iC
Mantle
0 P 0 PFig. 8. Schematic illustration of a periodically refilled magma chamber ascending in the crust . (a) The magma chamber at the initial stage. The magma chamber forms at first at the crust-mantle boundary by the accumulation of the primary magma successively ascending from the depth in the mantle . The rocks above the chamber is heated, then partially melted, and eventually incorporated into the magma. The crystallization always continues and the crystals formed in the chamber gravitationally settle down on the floor. This results in the upward migration .of the chamber. (b) The magma chamber ascending in the crust (modified from Yanagi et al., 1988). The chamber rises until it reaches a depth where the heat balance is realized. The magma replenishment to the chamber goes in the following way. When the primary magma accumulates at the crust-mantle boundary , the plug starts subsiding and the accumulated magma simultaneously starts moving through an opening between the plug and the cylinder into the chamber in the crust. The periodical replenishment continues so far as the primary magma is supplied from the mantle. The plug is composed of cumulate formed through batch fractionation and the assimilation in the chamber. The cumulate is represented by dots. V, volcano; C, magma chamber; P, primary magma periodically ascen
ding from the mantle.
indicates that the lava eruption occurred during the repeated magma mixing, and that their chemical variation is not due to the crystallization differentiation, but to the magma mixing. The same phenomenon is found in the Sakura
jima volcano (Yanagi et al., 1988). Petrographic and petrochemical characters of the historic lava flows in the Sakurajima volcano are fully explained by a model in which the continuous inflow of basaltic magma occurred for last 520 years into the chamber originally filled with a dacitic magma. No effect of crystallization is discernible on the chemical compositions of the lava flows.
CONCLUDING REMARKS
The existence of a periodically refilled magma chamber such as that depicted by Yanagi and Ishizaka (1978) is suggested under the Kinpo volcano. All lavas so far analyzed were erupted
during the repeated magma mixing. As exemplified by Yanagi and Ishizaka (1978), the mixed magmas evolved in the periods between repeated magma replenishments. On each replenishment the magma mixing occurred between the primary and the evolved dacitic magma. The 87Sr/86Sr ratios, low in basic and high in acid lavas, suggest the contamination by
crustal materials during the crystallization differentiation. Rather suppressed increase in Rb / Sr ratio during the contamination prefers the involvement of the melt produced by partial melting of the heated granulitic lower crust. The successive decrease in 87Sr/86Sr ratio of the dacitic magma in later stages indicates a pro
gressive decrease in the amount of melt incorporated into the evolving magma during the growth of the volcano. The reduction of crustal involvement probably means the decrease of the
upward-migration rate of the magma chamber
248 T. Yanagi et al.
with time.
Acknowledgments-We are very thankful to S.
Nakada and T. Nishiyama for their discussion and
supports during the course of this work. Mrs. E. Abe
assisted us in drawing the figures. The manuscript was
improved by a critical reading by Y. Matsthisa and
two anonymous reviewers.
REFERENCES
Baadsgaard, H. and Lerbekmo, J. F. (1983) Rb-Srand U-Pb dating of bentonites. Can. J. Earth Sci.
20, 1282-1290.
Bailey, J. C., Larsen, O. and Frolova, T. I. (1987) Strontium isotope variations in lower Tertiary Quaternary volcanic rocks from the Kurile island arc. Contrib. Mineral. Petrol. 95, 155-165.
Cox, K. G., Bell, J. D. and Pankhurst, R. J. (1979) The interpretation of igneous rocks. George Allen & Uniwin, London, 450 p.
Eichelberger, J. C. (1975) Origin of andesite and dacite: evidence of mixing at Glass Mountain in California and at other circum-Pacific volcanoes. Geol. Soc. Amer. Bull. 86, 1381-1391.Faure, G. (1986) Principles of Isotope Geology. John Wiley & Sons, New York, 589 p.Gill, J. B. (1981) Orogenic andesites and plate tec
tonics. Springer-Verlag. Berlin Heidelberg, 390 p.Ishibashi, K. (1970) Petrochemical study of basic and
ultrabasic inclusions in basaltic rocks from nor thern Kyushu, Japan. Mem. Fac. Sci. Kyushu Univ. [D] 20, 85-146.
Ishizaka, K., Yanagi; T. and Hayatsu, K. (1977) A strontium isotopic study of the volcanic rocks of the
Myoko volcano group, central Japan. Contrib. Mineral. Petrol 63, 295-307.Ishizaka, K., Yanagi, T. and Murakami, N. (1984)
Isotopic study of mafic inclusions from Uta-jima, southwest Japan. Geochem. J. 18, 203-208.
Koyaguchi, T. (1986) Textural and compositional evidence for magma mixing and its mechanism;
Abu volcano group, southwestern Japan. Contrib. Mineral. Petrol. 93, 33-45.Kurasawa, H. (1985) Strontium isotopic composition
of the volcanic rocks from Unzen, Tara and Kumamoto-Kimpo areas in Kyushu, Southwest Japan. Bull. Volcanol. Soc. Japan 30, 273-286.
Matsuhisa, Y. and Kurasawa, H. (1983) Oxygen and strontium isotopic characteristics of calc-alkaline volcanic rocks from the central and western Japan arcs: Evaluation of contribution of crustal com
ponents to the magmas. J. Volcanol. Geotherm. Res. 18, 483-510.
Matsumoto, Y. (1979) Some problems on volcanic ac
tivities and depression structures in Kyushu, Japan.
Mem. Geol. Soc. Japan 16, 127-139.
Mitsunami, T. (1974) On the seismological observa tion in northern kyushu (4). Bull. Fukuoka Univ.
Education 24, 119-131.Miyashiro, A. (1974) Volcanic rock series in island arcs and active continental margins. Amer. J. Sci. 274, 321-355.
Murakami, N. (1975) High-grade metamorphis inclu sions in Cenozoic volcanic rocks from west San-in, southwest Japan. J. Japan. Assoc. Min. Petrol.
Econ. Geol. 70, 424-436.Nakada, S., Yanagi, T., Maeda, S., Fang, D. and Yamaguchi, M. (1985) X-ray fluorescence analysis of major elements in silicate rocks. Sci. Rept. Dept. Geol. Kyushu Univ. 14, 103-115.
Notsu, K. (1983) Strontium isotope composition in volcanic rocks from the Northeast Japan arc. J.
Volcanol. Geotherm. Res. 18, 531-548.Notsu, K., Issiki, N. and Hirano, M. (1983) Com prehensive strontium isotope study of Quaternary volcanic rocks from Izu-Ogasawara arc. Geochem. J. 17, 289-302.
O'Hara, M. J. (1977) Geochemical evolution during fractional crystallization of a periodically refilled
magma chamber. Nature 266, 503.Sakuyama, M. (1981) Petrological study of the Myoko and Kuroshime volcanoes, Japan: Crystallization sequence and evidence for magma mixing. J. Petrol. 22, 553-583.Stern, R. J. (1982) Strontium isotopes from circum Pacific intra-oceanic island arcs and marginal basins: Regional variations and implications for
magmagenesis. Geol. Soc. Am. Bull. 93, 477-486.Tsuchiyama, A. (1985) Dissolution kinetics of
plagioclase in the melt of the system diopside-albite anorthite, and origin of dustry plagioclase in andesites. Contrib. Mineral. Petrol. 89, 1-16.
Von Drach, V., Marsh, B. D. and Wasserburg, G. J.
(1986) Nd and Sr isotopes in the Aleutians: Multicomponent paranthood of island-arc magmas. Contrib. Mineral. Petrol. 92, 13-34.Yanagi, T. (1975) Rubidium-strontium model of for mation of the continental crust and the granite ate
the island arc. Mem. Fac. Sci. Kyushu Univ. [D] 22, 37-98.
Yanagi, T. (1981) Alternative magmatic processes of continental growth in an island arc. Mem. Fac. Sci.
Kyushu Univ [D] 189-206.Yanagi, T., Ichimaru, U. and Hirahara, S. (1988) Coupled chamber under the Sakurajima volcano. Proceedings of Kagosima International Conference on Volcanoes (in print).
Yanagi, T. and Ishizaka, K. (1978) Batch fractiona tion model for the evolution of volcanic rocks in an island arc: An example from central Japan. Earth
Planet. Sci. Lett. 40, 252-262.