DATA
Strontium-isotope mapping of the Kyushu islands, southwestern Japan, using stream sediments
2025 Volume 59 Issue 1 Pages 9-25
Details
2025 Volume 59 Issue 1 Pages 9-25
A geochemical mapping study was conducted to investigate the spatial distribution of 87Sr/86Sr ratios in Kyushu. The study aimed to understand the factors controlling the geochemical distribution of Sr isotopes in stream sediments more completely. Using a thermal ionization mass spectrometer, we analyzed the 87Sr/86Sr of 307 stream sediments collected for geochemical mapping in Japan. The results showed that systematic variations in 87Sr/86Sr ratios correlate with surface geology. Stream sediments derived from mafic volcanic and pyroclastic rocks displayed low 87Sr/86Sr ratios (0.704–0.708), whereas those derived from felsic volcanic, granitic, and metamorphic rocks showed intermediate Sr isotope ratios (0.705–0.711). In the northwestern and southeastern areas underlain with accretionary complex and non-accretionary sedimentary rock, high 87Sr/86Sr ratios (0.705–0.721) were observed in the stream sediments. Based on the 87Sr/86Sr data of stream sediments derived from these rock types, we found a predominant supply of clastic materials originating from the Asian continent to the Paleogene back-arc basin in northwest Kyushu and the Paleogene subduction zone in southwest Kyushu. This suggests that the Sr isotope map is useful for the provenance analysis of clastic particles in sedimentary rocks. Furthermore, we present a map of the 87Sr/86Sr ratio in the exchangeable fraction of stream sediments. This fraction, which dissolves easily in water and is accessible for plants, is important for provenance studies of agricultural products. The 87Sr/86Sr ratio in the exchangeable fraction was predominantly affected by the presence of plagioclase in stream sediments and exhibits a positive correlation with that in the bulk fraction.
Regional spatial distribution patterns of multi-elemental concentrations, known as “geochemical maps”, provide us with fundamental information on elemental behavior in nature. The maps are created using soil substances and stream sediments and are essential for environmental assessment and mineral exploration. Recently, sub-continental and cross-boundary geochemical mapping projects have been conducted in addition to nationwide geochemical mapping (Xie and Chen, 2001; Salminen et al., 2005; De Vos et al., 2006; De Caritat and Cooper, 2011; Reimann et al., 2014a, 2014b; Smith et al., 2014). In Japan, the Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology (AIST) has constructed a comprehensive nationwide geochemical map of elements within stream and marine sediments mainly for environmental assessment and to examine particle transfer processes from land to sea (Imai et al., 2004, 2010).
In recent years, geochemical maps have been found wide applications in provenance and migration analyses in archeology, identification of the geographical origin of agricultural products, and forensic investigations (Ariyama et al., 2006; Nakai et al., 2014). For these purposes, isotopic data would be more informative than elemental concentration data because isotopic ratio changes in various physicochemical processes are far less than those of chemical composition. Accordingly, Sr isotopic data have been widely used for this purpose (Hodell et al., 2004; Rummel et al., 2010; Voerkelius et al., 2010; Thornton, 2011; Hiraoka et al., 2016). This is because Sr in rocks and sediments on the Earth’s surface, released by weathering processes, is incorporated into plants and animals via the nutrient cycle and food chain (Crock et al., 1986).
In Japan, Asahara et al. (2006) constructed a local Sr isotope map within a 75 km × 35 km area using stream sediments and found that the Sr isotope ratios of stream sediments accurately reflected those of source rocks. Subsequently, Jomori et al. (2013) constructed a regional spatial 87Sr/86Sr distribution map in western Japan at a scale of 30,000 km2 using fine stream sediments sampled for Japanese nationwide geochemical mapping (Imai et al., 2004). They clarified that the large-scale Sr isotope distribution corresponded well even with that of the large tectonic zone. Despite their small range investigation, Jomori et al. (2017) and Minami et al. (2017) further confirmed three key findings: 1) 87Sr/86Sr ratio variation among different stream sediment grain sizes is much smaller than variations associated with distinct rock bodies; 2) the fine grain fractions (<180 μm) of stream sediments exhibit 87Sr/86Sr values that are closest to those of the source rock; 3) the 87Sr/86Sr values of fine stream sediments (<180 μm) can be simply expressed as a function of the exposed rocks in the catchment area. Minami and Suzuki (2018) compared the 87Sr/86Sr ratios of agricultural products with those of soil substances in production fields and stream sediments near rice fields. They found that the 87Sr/86Sr ratios of rice plants were comparable to those of the bulk and acetic acid-extracted fraction, which consists of Sr adsorbed onto the surfaces of sediment particles that dissolves easily in water and is readily adsorbed by plants from paddy fields, and also those of stream sediments and water near the rice field.
In this study, we constructed an Sr isotope map for Kyushu, the southernmost of the four main islands in Japan. The study area consists of various rock types including accretionary complex and sedimentary, granitic, volcanic, and metamorphic rocks formed mainly during the Jurassic, Cretaceous, Paleogene, Neogene, and Quaternary periods and divided by tectonic lines. The Kyushu region accounts for about 19% of the nation’s agricultural output. Moreover, there are a number of archeological materials in Kyushu that have been brought from East Asia. First, we discuss how accurately the Sr isotope distribution corresponds to the complex geological structure, even at the large scale of 40,000 km2. Second, we create a bioavailable Sr isotope-ratio map of the Kyushu region for provenance analyses of archeological samples and agricultural products.
Figure 1 shows the sampling location in Kyushu, which includes several isolated islands, including Iki, Hirado, Fukue, Amakusa, Tanegashima, and Yakushima islands. The Tsukushi and Miyazaki plains are flat plains that are known to be major rice-producing areas. Fukuoka, Kumamoto, and Kagoshima are the largest cities in mainland Kyushu. A total of 366 stream sediment samples were collected from various locations in Kyushu during 1999–2004 for a geochemical mapping project conducted by the Geological Survey of Japan (Imai et al., 2004) (Fig. 1). In addition, 25 stream sediments were collected from isolated islands around Kyushu from 2013 to 2014 (Ohta, 2018). The sampling density was approximately one point per 100 km2.
Sampling locations of stream sediments in Kyushu. The circle symbols represent the 391 stream sediment samples collected for geochemical mapping. 87Sr/86Sr values were determined using 307 samples (blue circles).
Figure 2 shows a geological map of the study area simplified from the seamless geological map of Japan at a 1:200,000 scale (Geological Survey of Japan, AIST, 2015). Details of the geology of Kyushu were obtained from Moreno et al. (2016). Five large tectonic lines are found in Kyushu: the Usuki-Yatsushiro Tectonic Line (UYTL), dividing Kyushu into north and south regions; the Butsuzo Tectonic Line (BTL); the Nobeoka Trust (NT), bending into a dogleg shape in the southern region; the Kokura-Tagawa Fault (KTF), running north to south in the northeast region; and the Hatashima-Ariake Fault (HAF) located in the northwest region.
Geological map of Kyushu, Japan. The map was simplified from the seamless digital geological map of Japan at a scale of 1:200,000 (Geological Survey of Japan AIST, 2015). UYTL, Usuki-Yatsushiro Tectonic Line; BTL, Butsuzo Tectonic Line; NT, Nobeoka Thrust; KTF, Kokura-Tagawa Fault; HAF, Hatashima-Ariake Fault. (A) Sedimentary rock (Neogene–Quaternary); (B) Sedimentary rock (pre-Paleogene); (C) Sedimentary rock of the accretionary complex associated with metagabbro, metabasalt, and ultramafic rock; (D) Alkaline volcanic rock (mostly mafic volcanic rock); (E) Non-alkaline mafic volcanic rock; (F) Non-alkaline felsic volcanic rock; (G) Pyroclastic flow deposit, debris, and tephra; (H) Granitic rock; (I) Metamorphic rock (mostly high-pressure type).
The Permian accretionary complex (Fig. 2; C) is located near the KTF in northeastern Kyushu. The Jurassic accretionary complex (Fig. 2; C) crops out in a narrow zone between the UYTL and BTL. Cretaceous and Paleogene accretionary complexes (Fig. 2; C) are located south of BTL and south of NT, respectively.
Most metamorphic rocks (Fig. 2; I) in the Kyushu area are the high-pressure type. The Triassic–Jurassic high-pressure metamorphic rock (Suo metamorphic rock) is distributed around the Tsukushi Plain. The Cretaceous high-pressure metamorphic rocks (Nagasaki and Sanbagawa metamorphic rock) are located at the western and eastern ends of northern Kyushu, respectively. They predominantly consist of pelitic, mafic, and psammitic schists.
Cretaceous granitic rocks (Fig. 2; H) widely intrude into the northern region. Neogene biotite-granodiorite and biotite-monzogranite are distributed in the Osumi Peninsula in the southern part of Kyushu. Coarse-grained biotite granite is intruded in the central part of Yakushima Island, and hornblende-biotite bearing granite and granodiorite are exposed sporadically to accretionary complexes.
Neogene–Quaternary alkaline volcanic rocks are distributed within a limited area (Fig. 2; D). The distribution of Neogene–Quaternary non-alkaline felsic rock is also restricted to the Fukue and Nakadori islands and the Osuzu and Hokusatu regions (Fig. 2; F). In contrast, Neogene–Quaternary non-alkaline mafic volcanic rocks are widely distributed in both the northern and southern regions of Kyushu (Fig. 2; E). Volcanic rocks are associated with large pyroclastic flow deposits, debris, and tephra that erupted during the formation of large-scale calderas (Fig. 2; G). In particular, the Aso pyroclastic rock (Aso-4 event: mainly c.a. 90 ka, Seki et al., 2016) that erupted from the Aso Caldera and Ito pyroclastic rocks (c.a. 29 ka, Okuno, 2002) that erupted from the Aira Caldera are distributed widely in the central and southern parts of Kyushu, respectively.
Pre-Cretaceous non-accretionary sedimentary rock (Fig. 2; B) occurs in narrow zones directly above the UYTL, between the UYTL and BTL, and in the Amakusa Islands. Paleogene non-accretionary sedimentary rock (Fig. 2; B) is distributed west of the KTF, west of the HAF, and on the Amakusa Islands. The distribution of Neogene sedimentary rocks (Fig. 2; A) is restricted to the Fukue and Nakadori islands, Tanegashima Island, and the Miyazaki Plain. Quaternary sedimentary rock (Fig. 2; A) is widely distributed in the Tsukushi and Miyazaki plains, although it is restricted to a small area elsewhere.
Geochemical analysisThe collected stream sediments were dried in air and sieved through a 180-μm screen. Magnetic minerals contained in the samples were removed using a magnet to minimize the effects of magnetic mineral accumulation (Imai et al., 2004). The geochemical data of 53 elements in 366 stream sediments from Kyushu and 25 samples from isolated islands have been reported by Imai et al. (2004) and Ohta (2018), respectively.
87Sr/86Sr was determined for 307 of the 391 samples that were analyzed for 53 elements. The procedure for 87Sr/86Sr measurement is described in detail by Jomori et al. (2013). Samples (50–70 mg) were heated at 950°C for 5 h to remove organic matter and then digested with a HF/HClO4 mixture. The degraded product was evaporated to dryness, and the residue was dissolved in 2.4 mol L–1 HCl. The Sr in the sample solutions was separated from the other elements using a quartz column filled with a cation-exchange resin (BioRad AG 50WX8, 200–400 mesh; BioRad Laboratories, Hercules, CA, USA). The Sr fraction separated using the resin was loaded onto a tantalum filament, and the Sr isotope ratios were measured using a thermal ionization mass spectrometer [TIMS; VG Sector 54 (GV Instruments Ltd., Wythenshawe, UK) at Nagoya University and Thermo TRITON (Thermo Fisher Scientific, Waltham, MA, USA) at Kochi Core Center]. 87Sr/86Sr ratios were normalized to 86Sr/88Sr = 0.1194. The value of repeated analyses of standard NIST-SRM987 during this study was 0.710238 ± 0.000011 (n = 5: mean ± 2s). The 87Sr/86Sr values of the stream sediments determined here are listed in Table S1 and are associated with the Rb and Sr concentrations (Imai et al., 2004; Ohta, 2018) and 87Rb/86Sr ratios calculated from the Rb and Sr concentrations.
We also measured 87Sr/86Sr in the exchangeable fraction of the sediment samples in 69 of the total 391 samples to extract elements in carbonates or those weakly adsorbed onto the materials. Approximately 200 mg of sample before heating was mixed with 10 mL of 1 mol L–1 ammonium acetate (pH = 6.6–7.1) and left overnight in a PFA tube at room temperature. The supernatant was separated from the residue by centrifugation at 3,500 rpm for 10 min. The solution, with a few drops of HNO3 added, was evaporated to dryness, and the residue was dissolved in 3 mL of 2.4 mol L–1 HCl. The subsequent procedure was the same as that described above for the sediment samples (Jomori et al., 2017).
Preparation of geochemical mappingThe 2,000-m grid maps covering Kyushu and the surrounding isolated islands were constructed using geographic information system software (ArcGIS 10.8; Environmental Systems Research Institute Japan Corporation, Tokyo, Japan) following the method described by Ohta et al. (2004). Stream sediment is a composite of the products resulting from weathering and erosion of soil and rocks in the watershed area upstream of the sampling site (Howarth and Thornton, 1983). The sampling site was presumed to express the average chemical concentrations and isotopic ratios in the drainage basin. The watershed area for each sample was calculated using ArcGIS software based on a digital elevation model (50-m mesh data) obtained from the Geospatial Information Authority of Japan. Figure 3 shows the geochemical maps of the watersheds for Rb and Sr concentrations and 87Rb/86Sr and 87Sr/86Sr ratios. Percentile ranges are used to select the elemental concentration and isotopic ratio intervals in the color image maps: 0 ≤ x ≤ 5, 5 < x ≤ 10, 10 < x ≤ 25, 25 < x ≤ 50, 50 < x ≤ 75, 75 < x ≤ 90, 90 < x ≤ 95, and 95 < x ≤ 100%, where x represents the elemental concentration, according to Reimann (2005). The constructed Sr isotope maps are provided in tiff and kml formats and are used to display the data in GIS and mapping software such as Google Earth and Google Maps (see Supplementary Material).
Spatial distributions of Rb and Sr concentrations (a, b), and 87Rb/86Sr and 87Sr/86Sr ratios (c, d) in Kyushu.
We assumed that when a specific rock underlies more than half the area of a river basin, it is representative of the geology of the watershed and mainly controls the elemental abundances and Sr isotopes (Ohta et al., 2004; Jomori et al., 2017). A total of 391 samples collected from Kyushu and the isolated islands were classified into 8 different types according to the dominant lithology, as proposed by Ohta et al. (2021). In the study area, sedimentary rocks (Sed); accretionary complexes, comprising mainly sandstone, mudstone, and mélange matrix (Acc); granitic rock (Gr); non-alkaline felsic and mafic volcanic rock (Fv and Mv); alkaline mafic volcanic rock (aMv); pyroclastic rock (Py); and high-pressure type metamorphic rock (Mp) were considered typical lithologies. The median elemental concentrations in the stream sediments classified by the dominant lithology in the watershed are summarized in Table S2.
Because the purpose of this study was to determine the correspondence between geology and Sr isotopic ratios, the key features of the geochemical maps are briefly summarized below. Ohta et al. (2022) studied the factors that determine the spatial distribution patterns of elemental concentrations in stream sediments. They found that Li, K2O, Rb, Cs, and Tl were more abundant in the northern area and on the south side of the UYTL, where accretionary complexes, granitic rock, and felsic volcanic rock are distributed, while these elements were less abundant on the north side of the UYTL, where mafic volcanic, pyroclastic, and metamorphic rocks are distributed (see Rb in Fig. 3a). The spatial distributions of MgO, CaO, TiO2, MnO, T-Fe2O3, Sc, V, Co, and Sr were opposite to those of Li, K2O, Rb, Cs, and Tl (see Sr in Fig. 3b). Their high-concentration areas mainly correspond to the areas covered by mafic volcanic and pyroclastic rocks, especially around the Aso Volcano.
The spatial distributions of 87Rb/86Sr and 87Sr/86Sr also appeared to correspond to the geological map (Fig. 3c, d). The 87Rb/86Sr and 87Sr/86Sr ratios were high in the northern part of the KTF, southwest of the HAF, Amakusa Islands, and southeast of the UYTL, which are covered by pre-Paleogene sedimentary rocks, accretionary complexes, and metamorphic rocks. Both ratios were low in the eastern to central parts of the northern regions and the center to the western part of the southern regions of Kyushu Island, which are covered by granitic, volcanic, and pyroclastic rocks.
Comparison of 87Sr/86Sr ratios in the exchangeable and bulk fractions of stream sedimentsFigure 4 shows the relationship of 87Sr/86Sr ratios in the exchangeable and bulk fractions of stream sediments. The exchangeable fraction of stream sediments consists of Sr weakly adsorbed to the surface of sediment particles from stream water and the surface of weathered minerals. Therefore, metals in the fraction dissolve easily in water and are readily available to plants. Stream sediment samples derived from granitic (mainly Cretaceous), mafic volcanic, and pyroclastic rocks, which have low to medium 87Sr/86Sr ratios in the bulk fraction, plotted roughly on the 1:1 line. In contrast, stream sediments derived from sedimentary rocks (mostly Paleogene) and accretionary complexes (mostly Paleogene) had lower 87Sr/86Sr ratios in the exchangeable fraction than in the bulk fraction. This result is consistent with the report of Jomori et al. (2017), who found that 87Sr/86Sr values in the exchangeable fraction of stream sediments corresponded closely to those of stream water collected at the same sites but were systematically lower than those in the bulk fraction.
Relationship of 87Sr/86Sr values between bulk and exchangeable Sr in stream sediments classified by dominant lithology. The abbreviations are explained in the main text. The short blue dashed line indicates the least square fitting results using a cubic function. One parenthesized data set was excluded from the calculation because it was clearly an outlier.
Figure 5 shows the relationship between 87Rb/86Sr and 87Sr/86Sr in samples derived from the felsic (Fv) and mafic (Mv) volcanic rocks. Notably, two samples (6016 and 6018) collected from the felsic rock exhibited the highest 87Rb/86Sr (2.82–3.54) and 87Sr/86Sr (0.71001–0.71046) values (Fig. 5a), which are comparable to those of the Osuzu welded tuff (Terakado et al., 1988). Stream sediments derived from Fv in Fukue Island (Fke 10) also had high 87Rb/86Sr (1.78) and 87Sr/86Sr (0.70830) values. The other samples derived from Fv in southwestern Kyushu had much lower isotopic ratios, comparable to those of their source rocks (Hosono et al., 2003, 2008).
(a–d) 87Rb/86Sr–87Sr/86Sr plot for stream sediments derived from felsic volcanic rock (Fv) and mafic volcanic rocks in the northeastern [Mv (NE)], northwestern [Mv (NW)], and southwestern [Mv (SW)] regions (Fig. S1). Source rock data were obtained from the following sources: Terakado et al. (1988) for Fv (Osuzu)a; Hosono et al. (2003, 2008) for Fv-Mv (Hokusatsu)b; Miyoshi et al. (2011) for Mv (Aso)c; Notsu et al. (1990), Kita et al. (2001), and Sugimoto et al. (2006) for Mv (NE)d; Kurasawa (1985) for Mv (NW)e; Nakamura et al. (1990) and Notsu et al. (1990) for aMvf; Kurasawa et al. (1984), Kurasawa (1986), and Hosono et al. (2003, 2008) for Mv (SW)g.
The mafic volcanic rocks were divided further into those of the northeastern area [Mv (NE)], northwestern area [Mv (NW)], and southwestern area [Mv (SW)] (Fig. S1). Most stream sediments derived from Mv (NE) exhibited low 87Rb/86Sr (0.174–0.497) and 87Sr/86Sr (0.70409–0.70494) (Fig. 5b). These values are comparable to the low values observed in the northwestern volcanic rocks of 87Rb/86Sr (0.101–0.340) and 87Sr/86Sr (0.70389–0.70488), although rhyolite and dacite in Aso Volcano exhibit exceptionally high 87Rb/86Sr (0.50–4.55) (Notsu et al., 1990; Kita et al., 2001; Sugimoto et al., 2006; Furukawa et al., 2009; Miyoshi et al., 2011). The 87Rb/86Sr and 87Sr/86Sr ratios of stream sediments derived from Mv (NE) were mostly comparable to those of the mafic components.
Figure 5c shows that half of the stream sediment samples derived from Mv (NW) have 87Rb/86Sr (0.346–1.01) and 87Sr/86Sr (0.70473–0.70550) ratios similar to those of source rocks, while the remaining samples have much higher 87Rb/86Sr (1.08–3.66) and 87Sr/86Sr (0.70808–0.71720) ratios than those of source rocks. The host rocks [Mv (NW)] have low 87Rb/86Sr (0.024–0.076) and 87Sr/86Sr (0.70363–0.70509) ratios (Kurasawa, 1985; Nakamura et al., 1990; Kita et al., 2001).
In Figure 5d, the 87Rb/86Sr (0.479–2.53) and 87Sr/86Sr (0.70523–0.70666) ratios of sediments derived from Mv (SW) are plotted in the range of those of source rocks (Hosono et al., 2003, 2008; Kurasawa, 1986; Kurasawa et al., 1984).
Systematic variations in 87Rb/86Sr and 87Sr/86Sr in stream sediments derived from pyroclastic rocksThe pyroclastic flow deposits were further subgrouped into three volcanoes (calderas) and eruption ages: Py (Ito), Py (Aso), and Py (Futago). Three samples collected from Unzen-dake and Hirado Islands were subgrouped into Py (Oth) (Fig. 6).
87Rb/86Sr–87Sr/86Sr plot for stream sediments derived from Aso, Futago-dake, and Ito pyroclastic rocks. Source rock isotopic data were obtained from Hunter (1998) for Py (Aso-4)a and from Kurasawa et al. (1984) for Py (Ito)b. The dashed lines of Py (Aso)a and Py (Ito)b represent mean 87Sr/86Sr ratios of Aso-1, -2, -3, and -4 pyroclastic flow deposits (n = 41) (Hunter, 1998) and Ito pyroclastic flow deposits (n = 4) (Kurasawa et al., 1984), respectively.
The 87Rb/86Sr ratios of Py (Aso) vary widely from 0.156 to 1.77, whereas their 87Sr/86Sr ratios are almost constant at 0.70413 ± 0.00021 (n = 41: mean ± 2s) (Hunter, 1998). Many stream sediment samples derived from Py (Aso) have very low 87Rb/86Sr (0.252–0.484) and Sr isotopic ratios (0.70418–0.70471), suggesting that they originated from basaltic pyroclastic rocks. Some samples collected from the margin of Py (Aso), which adjoins the UYTL, had systematically high 87Rb/86Sr (0.547–1.10) and 87Sr/86Sr ratios (0.70498–0.70629).
The 87Sr/86Sr ratios of stream sediments derived from Py (Ito) were 0.70629 ± 0.00104 (n = 17: mean ± 2s), which is almost close to 0.70610 ± 0.00009 (n = 4: mean ± 2s, shown as the dotted line in Fig. 6) for Ito pyroclastic flow deposits (Kurasawa et al., 1984). The deviation of the 87Rb/86Sr data of the stream sediments derived from Py (Ito) was much larger than for those of Py (Aso) and Py (Futago) (Fig. 6).
Systematic variations in 87Rb/86Sr and 87Sr/86Sr in stream sediments derived from granitic rocksCretaceous granitic rocks on the western and eastern sides of the KTF (Fig. 2) are hereafter referred to as K-Gr (west) and K-Gr (east), respectively. The 87Rb/86Sr (0.352–1.76) and 87Sr/86Sr (0.70568–0.70750) values for stream sediments derived from K-Gr (west) correspond to those of granitic rocks collected from the northeastern part of Kyushu (Owada et al., 1999) (Fig. 7; open diamonds). Stream sediments derived from K-Gr (east) have apparently higher 87Rb/86Sr (2.10–3.45) and Sr isotopic ratios (0.70821–0.70834) than those of stream sediments derived from K-Gr (west). K-Gr (west) reportedly has higher Sr concentrations and a lower initial Sr isotope ratio than those of granitic rocks distributed in southwestern Japan, excluding the Kyushu region (Owada et al., 1999; Moreno et al., 2016). Stream sediments derived from K-Gr (west) had higher Sr concentrations (mean of 239 mg/kg) than those from K-Gr (east) (mean of 134 mg/kg).
87Rb/86Sr–87Sr/86Sr plot for stream sediments derived from Cretaceous granitic rocks located on the east and west sides of KTF [K-Gr (east) and K-Gr (west)] and those from Neogene granitic rock in Osumi Peninsula [N-Gr (Osumi)] and Yakushima Island [N-Gr (Yakushima)]. Source rock 87Rb/86Sr and 87Sr/86Sr data were obtained from Shibata and Ishihara (1979) for N-Gr (Yakushima)a; Yanagi et al. (1971) and Terakado et al. (1988) for N-Gr (Osumi)b; Owada et al. (1999) for K-Gr (west)c; Kawano and Yuhara (2008) for K-Gr (west)d.
Stream sediments derived from Neogene granitic rocks (N-Gr) in the Osumi Peninsula and Yakushima Island have systematically higher 87Rb/86Sr and 87Sr/86Sr ratios than those derived from K-Gr (west) and lower ratios than those derived from K-Gr (east) (Fig. 7). Unlike K-Gr, the 87Rb/86Sr and 87Sr/86Sr ratios of stream sediments derived from N-Gr are systematically lower than those of their source rocks (Yanagi et al., 1971; Shibata and Ishihara, 1979; Terakado et al., 1988).
Systematic variations in 87Rb/86Sr and 87Sr/86Sr in stream sediments derived from metamorphic rocksFigure 8 shows the relationship between 87Rb/86Sr and 87Sr/86Sr in stream sediment samples derived from high-pressure metamorphic rocks, which were subgrouped into Triassic–Jurassic Suo metamorphic rock (Tr–J-Mp) and Cretaceous Nagasaki-Sanbagawa metamorphic rocks (K-Mp). Five stream sediment samples derived from K-Mp are plotted near the dotted line drawn between the two data points of the Sanbagawa pelitic schists (Utsunomiya et al., 2011). The 87Rb/86Sr (1.20–3.61) and 87Sr/86Sr (0.70931–0.71082) ratios of samples derived from K-Mp are well comparable to those of stream sediments derived from K-Mp in the other regions (Fig. 8; dashed circles, Jomori et al., 2013). The remaining three samples were predominantly originated from mafic schist and serpentinite. These samples plot near the data of mafic schists with low 87Rb/86Sr (0.025–0.596) and 87Sr/86Sr (0.70330–0.70483) ratios (Utsunomiya et al., 2011; Uno et al., 2014).
87Rb/86Sr–87Sr/86Sr plot for stream sediments derived from Suo metamorphic rocks (Tr–J-Mp) and from Nagasaki-Sanbagawa metamorphic rocks (K-Mp). The letters a, b, c, d, e, and f indicate the reference data obtained from Jomori et al. (2013), Utsunomiya et al. (2011), Uno et al. (2014), Shibata and Nishimura (1989), Miyamoto and Yanagi (1996), and Osanai et al. (1993), respectively. The areas enclosed by dashed lines represent the data of stream sediments derived from Sanbagawa metamorphic rocks (Jomori et al., 2013). The solid and short-dashed lines are the isochron lines for Suo pelitic schists (Shibata and Nishimura, 1989) and Haki, Kusano, and Tama granitic rocks contacting Suo metamorphic rocks (Osanai et al., 1993), respectively.
In contrast, stream sediments derived from the Tr–J-Mp plot are considerably below the 87Rb/86Sr–87Sr/86Sr isochron of their source rocks (Shibata and Nishimura, 1989). Their systematically low 87Sr/86Sr ratios cannot be explained by the simple mixing of their pelitic and mafic schists in the host rock. Instead, their 87Rb/86Sr (1.15–2.69) and 87Sr/86Sr (0.70735–0.70995) ratios were more consistent with those of Sanbagawa metamorphic rocks (K-Mp).
Systematic variations in 87Rb/86Sr and 87Sr/86Sr in stream sediments derived from non-accretionary and accretionary sedimentary rocksFigure 9a shows the relationship between 87Rb/86Sr and 87Sr/86Sr in stream sediment samples derived from accretionary sedimentary rocks (Acc). The host rocks of these samples were further divided into three subgroups based on their accretion ages: Paleogene, Cretaceous, and pre-Jurassic (mostly Jurassic), hereafter designated Pg-Acc, K-Acc, and J-Acc, respectively. When the sum of J-Acc and K-Acc or that of Pg-Acc and K-Acc had an exposed area exceeding 50% of the catchment area, it was expressed as J–K-Acc and K–Pg-Acc, respectively.
(a) 87Rb/86Sr–87Sr/86Sr plot for stream sediments derived from Jurassic (J-Acc), Cretaceous (K-Acc), and Paleogene (Pg-Acc) accretionary complexes. (b) Relationship between 87Sr/86Sr and exposed area of mafic rocks in the watershed. (c) 87Rb/86Sr–87Sr/86Sr plot for stream sediments derived from sedimentary rocks. The abbreviations preK-Sed, Pg-Sed, N-Sed, Q-Sed, and Sed (Oth) are explained in the main text. The area enclosed by the dotted line represents the 87Rb/86Sr–87Sr/86Sr region of stream sediments derived from Pg-Acc for comparison. Source rock isotopic data were obtained from Hosono et al. (2003), Jahn (2010), Shibata and Ishihara (1979), and Terakado et al. (1988).
Pg-Acc samples have the highest 87Rb/86Sr (0.496–6.47) and 87Sr/86Sr (0.70806–0.72094) ratios among the three subgroups. The 87Rb/86Sr (0.822–3.12) and 87Sr/86Sr (0.70560–0.71191) ratios of the K-Acc samples overlap with those of J-Acc (0.848–4.83 for 87Rb/86Sr and 0.70629–0.71533 for 87Sr/86Sr). The 87Rb/86Sr and 87Sr/86Sr ratios of the stream sediments derived from Pg-Acc and K-Acc exhibit values similar to those of source rocks (Hosono et al., 2003; Jahn, 2010; Shibata and Ishihara, 1979; Terakado et al., 1988) (Fig. 9a). The relatively low 87Sr/86Sr values of the sediments derived from J-Acc and K-Acc may have been caused by the coexisting igneous rocks of Acc or mafic volcanic rocks adjoining these rocks. Indeed, a negative correlation was observed between the 87Sr/86Sr ratio and the exposed area of these mafic rocks in the watershed of the stream sediment samples derived from Acc (Fig. 9b). However, even after excluding samples with more than 10% exposed area of mafic rocks to calculate the mean 87Sr/86Sr ratio, stream sediments derived from Pg-Acc still had a higher mean 87Sr/86Sr ratio (0.71415) than those from J-Acc and K-Acc (0.71325 and 0.71074, respectively).
Figure 9c shows the relationship between 87Rb/86Sr and 87Sr/86Sr in stream sediment samples derived from non-accretionary sedimentary rocks, which were further grouped into Q-Sed, N-Sed, Pg-Sed, and preK-Sed according to their Quaternary, Neogene, Paleogene, and pre-Cretaceous formation ages, respectively. When the sum of Q-Sed, N-Sed, and Pg-Sed had an exposed area exceeding 50%, it was expressed as Sed (Oth). The 87Rb/86Sr and 87Sr/86Sr values of the samples derived from Pg-Sed and N-Sed mostly overlapped those of samples derived from Pg-Acc (Fig. 9c). The 87Sr/86Sr values of stream sediments derived from Q-Sed and Sed (Oth) changed considerably from 0.7046 to 0.7168, resembling those of samples from Pg-Sed and N-Sed, but their 87Rb/86Sr values were lower than 2.56. 87Rb/86Sr and 87Sr/86Sr ratios in samples derived from preK-Sed were 0.7092–0.7111 and 1.55–1.60, respectively.
Stream sediments generally reflect the isotopic characteristics of source rocks. However, some samples exhibited different isotopic data from their source rocks. To use the Sr isotope map to identify the origin of agricultural products and for forensic investigations, it is crucial to elucidate why the isotopic data of stream sediments differ from those of their source rocks.
A small deviation of approximately 0.001 in the 87Sr/86Sr ratio between the stream sediments and their source rocks can be explained by the exposure of coexisting rocks. The 87Sr/86Sr ratios of stream sediments derived from Py (Ito) and Py (Aso) were systematically higher than those of the source rocks (Fig. 6). The supply of stream sediments is assumed to be directly proportional to the lithological area (Rose et al., 1970). Watershed analysis suggested that K-Acc was exposed in 10–32% of the watershed of samples derived from Py (Ito). Considering the mixtures of 10, 20, and 30% K-Acc with 90, 80, and 70% Py (Ito), the 87Sr/86Sr ratios increased in the order of 0.70610, 0.70637, and 0.70669 (Fig. S2). Samples derived from the marginal part of Py (Aso), which exhibit systematically high 87Rb/86Sr (0.547–1.10) and 87Sr/86Sr ratios (0.70498–0.70629), can be similarly explained. These mostly coexist with Cretaceous high-temperature metamorphic rock, and Neogene Fv and J-Acc, which are exposed in 10–26% of the watershed. The supply of clastics from these rocks increased the 87Sr/86Sr ratio to 0.70586 (Fig. S3). Furthermore, the 87Sr/86Sr ratios of stream sediments derived from Py (Ito) and Py (Aso) increased with decrease of the CaO/K2O and Yb/La ratios, which can be attributed to the contribution of materials from Acc because stream sediments derived from Py have more abundant MgO, CaO, TiO2, T-Fe2O3, and heavy rare earth elements (Gd–Lu) and less abundant K2O, Rb, and Cs than those from Acc (Ohta et al., 2022) (Figs. S4). Thus, the deviation in the 87Sr/86Sr data of stream sediments derived from Py (Ito) and Py (Aso) are explained well by this mixing model.
However, the higher 87Rb/86Sr and 87Sr/86Sr ratios in samples 7003, 8026, 8027, 12001, and Iki 01, which were derived from Mv (NW), than those of their source rocks cannot be explained by the same reason stated above (Fig. 5c). The locations of these samples were underlain by N-Sed and Pg-Sed, while Mv was distributed throughout the upper river basin. The exposed areas of coexisting N-Sed and Pg-Sed in the watershed ranged from 1 to 18%, which is too small to explain the high 87Sr/86Sr ratios obtained by the simple mixing model (Rose et al., 1970). Watanabe et al. (1981) reported that Pg-Sed and N-Sed are the basement rocks of Tara-dake and are exposed on the river floor because large amounts of precipitation actively erode the volcano via radial river valleys. This suggests that the contribution of material (Pg-Sed and N-Sed) exposed on the river floor or nearby sites to the sampling points was considerable. In fact, the 87Sr/86Sr and 87Rb/86Sr ratios of these samples completely overlapped those of samples derived from Pg-Sed and N-Sed (Fig. 5c). The K2O/CaO and La/Yb ratios of the above five samples were similar to those of samples derived from Pg-Sed and N-Sed, whereas the other samples derived from Mv (NW) had low K2O/CaO and La/Yb ratios, comparable to those of the Hokusatsu mafic volcanic rocks (Fig. S5).
The pseudo-isochron lines obtained for the stream sediments derived from N-Gr had a gentler slope and higher initial ratio compared to sediment samples derived from K-Gr (Fig. 7). The N-Gr in Kyushu had high K2O/Na2O and low Fe(III)/Fe(II) ratios, suggesting its formation by re-melting of older sediments (Nakada and Takahashi, 1979). In fact, the mean K2O/Na2O ratios of stream sediments derived from N-Gr on Yakushima Island and the Osumi Peninsulas are 0.91 (n = 5) and 0.89 (n = 6), respectively, which are higher than those of sediments derived from K-Gr (0.70: n = 15). The higher initial 87Sr/86Sr ratios of the stream sediments derived from N-Gr reflect that their source rocks were formed by the melting of sediments from Paleogene accretionary complexes, which have high Sr isotopic ratios (Nakada and Takahashi, 1979). However, the 87Rb/86Sr and 87Sr/86Sr ratios of the sediments derived from N-Gr were much lower than those of their source rocks by 1.0–3.0 and 0.001–0.005, respectively (Fig. 7). This result indicated that stream sediments were less abundant in K-feldspar than that in the source rocks. K-feldspar elevates 87Rb/86Sr and 87Sr/86Sr in a sample, but it generally does not become a small grain because of its resistance to physical weathering processes, similar to quartz (e.g., Goldich, 1938). Minami et al. (2017) confirmed that the relative abundance ratio of K-feldspar to plagioclase decreased with a decrease in stream sediment grain size. Consequently, the 87Rb/86Sr and 87Sr/86Sr of stream sediments in the <180-μm fraction are systematically lower than those of their source rocks.
The 87Rb/86Sr and 87Sr/86Sr ratios of the stream sediments derived from Tr–J-Mp were much lower than those of their host rocks (Fig. 8). Nishimura (1998) reported that Tr–J-Mp underwent contact metamorphism by the K-Gr intrusion in Kyushu. Figure 8 shows the stream sediments derived from the Tr–J-Mp plot near the 87Rb/86Sr–87Sr/86Sr isochron of K-Gr in contact with Tr–J-Mp (Osanai et al., 1993). It seems reasonable that the gentle slope in the 87Sr/86Sr–87Rb/86Sr plot obtained from the stream sediments derived from Tr–J-Mp reflects thermal metamorphism. However, the Rb-Sr ages of muscovite sampled from Tr–J-Mp, which underwent thermal metamorphism were 214–207 Ma and did not become younger because the metamorphic temperature was likely too low to re-equilibrate Sr isotopes (Shibata and Nishimura, 1989). Furthermore, the pseudo-isochron obtained from the stream sediments derived from Tr–J-Mp showed a much lower initial Sr isotopic ratio (ca. 0.7059) than those obtained from the source rocks (0.7068). Unfortunately, explaining the low 87Rb/86Sr and 87Sr/86Sr values of stream sediments derived from Tr–J-Mp using the currently available datasets is difficult.
Application to provenance study of sediments in accretionary complexes of the Shimanto Belt and non-accretionary sedimentsThe study of Ohta and Minami (2019) suggests that stream sediments can be used as a substitute for rock samples in analyses of the origin of sedimentary rocks in Japan, where the ground surface is widely covered with vegetation and finding suitable rock outcrops is challenging.
Stream sediments derived from Q-Sed showed a wide range of 87Rb/86Sr (0.41–2.56) and 87Sr/86Sr (0.7046–0.7168) ratios (Fig. 9c). Four samples with 87Sr/86Sr (<0.709) ratios were collected from the Tsukushi and Kumamoto plains. These samples also had high CaO/K2O (0.60–5.07) and low La/Yb (6.0–13.7) ratios, which are comparable to those of samples derived from Py (Aso) (Fig. S6). These results are consistent with the report that the Q-Sed in the Tsukushi Plain is composed of pyroclastic flow deposits and basalt-andesite materials distributed in the upper river basin (Shimoyama et al., 2010). In contrast, the remaining samples derived from Q-Sed, with high 87Sr/86Sr (>0.710) ratios, were less influenced by mafic components, instead being more influenced by coexisting Pg-Sed and N-Sed (Fig. S6). In fact, the CaO/K2O (0.37–0.74) and La/Yb (12.8–23.4) ratios of these samples are comparable to those of samples derived from Pg-Sed and N-Sed (Fig. S6). Thus, we can apply the Sr isotope map to provenance analysis of the clastic particles of sedimentary rocks in the study area.
The Shimanto Belt, comprising K-Acc and Pg-Acc, is thought to have formed entirely under the same tectonic setting (e.g., Moreno et al., 2016). Nevertheless, stream sediments derived from Pg-Acc in Kyushu had higher 87Rb/86Sr and 87Sr/86Sr ratios than those derived from K-Acc (Fig. 9a). Jomori et al. (2013) reported that the sedimentary rocks of accretionary complexes formed through the mixing of ocean-derived mafic materials with high Ca/Ti, low La/Yb, and low 87Sr/86Sr and continental material with low Ca/Ti, high La/Yb, and higher 87Sr/86Sr ratios. In fact, stream sediments derived from J-Acc included more abundant CaO, TiO2, Cr, Co, and Ni than those from K-Acc and Pg-Acc because J-Acc is composed of a mélange unit and oceanic plate stratigraphy-associated serpentinite and ultramafic rock (Fig. S7a). In contrast, sediments derived from Pg-Acc in Kyushu had systematically lower CaO/TiO2 ratios and higher La/Yb ratios than those from J-Acc and K-Acc (Fig. S7a). These facts suggest that the higher Sr isotope ratios in stream sediments derived from Pg-Acc in Kyushu were caused by an increased input of detrital materials originating from the Asian continent.
87Rb/86Sr (1.30–4.87) and 87Sr/86Sr (0.7071–0.7192) ratios of stream sediments derived from Pg-Sed were comparable to those of samples derived from Pg-Acc, despite the difference in their depositional environments (Fig. 9c). The chemical compositions of stream sediments derived from the Pg-Sed samples were also similar to those of the Pg-Acc samples, which were poor in MgO, CaO, Sr, and heavy rare earth elements and abundant in K2O, Rb, and light rare earth elements (Fig. S7b). The Pg-Sed in Kyushu was deposited in the forearc shallow basin during the Middle Eocene and was assumed to be a broadly coeval Paleogene accretionary complex, which was deposited on the trench in the Middle Eocene to Early Oligocene (Saito, 2008). Teraoka et al. (1998) pointed out that Paleogene and Cretaceous non-accretionary rocks have a common origin of clastic with accretionary complexes of the same age—felsic to intermediate volcanic and granitic rocks derived from the Asian continent. Similar to the case of Pg-Acc, clastics originating from the Asian continent have been concluded to elevate 87Rb/86Sr and 87Sr/86Sr ratios in stream sediments derived from Pg-Sed.
Variation of 87Sr/86Sr in the bulk and exchangeable fractions of stream sedimentsThe 87Sr/86Sr ratios in the bulk fraction were poorly correlated with those in the exchangeable fraction of stream sediment derived from Sed (mostly Pg-Sed) and Acc (mostly Pg-Acc) (Fig. 4). During chemical weathering of granitic rocks, Takagi et al. (2001) suggested that during chemical weathering of granitic rocks, plagioclase, with a lower 87Sr/86Sr than the whole rock, is more readily weathered than K-feldspar, biotite, and clay minerals, in which 87Sr/86Sr is relatively high. Minami and Suzuki (2018) and Jomori et al. (2017) also reported that the Sr in the exchangeable fraction is attributable to the large contribution of Sr from plagioclase because of its low resistance to chemical weathering and the ease with which Sr leaches out of host rocks and stream sediments. In fact, the 87Sr/86Sr ratios in the exchangeable fraction in stream sediments derived from Mv and Py with little K-feldspar, biotite, and clay minerals correlate with those in their bulk fractions. Teraoka et al. (1998, 1999) reported that sandstones of Cretaceous and Paleogene accretionary and non-accretionary sedimentary rocks contain approximately 22–34% quartz, 2–14% K-feldspar, 13–35% plagioclase, and 30–60% lithic fragments. The high 87Sr/86Sr ratios in the bulk fraction of stream sediments reflect the presence of K-feldspar and acidic rock fragments in their host rocks. Evidently, stream sediments derived from Acc and Sed had high K2O/Na2O ratios, which serve as a proxy for the mineralogical ratio of K-rich minerals (K-feldspar and micas) to plagioclase (Fig. 10). Furthermore, the K2O/Na2O ratios of the stream sediments correlated highly with the 87Sr/86Sr ratio in the bulk fraction, but not at all with the 87Sr/86Sr ratio in the exchangeable fraction (Fig. 10). The 87Sr/86Sr ratio in the exchangeable fraction in stream sediments derived from Gr increased with the K2O/Na2O ratio. The fact does not mean that Sr in K-rich minerals was not extracted at all by using ammonium acetate solution but that the change attributable to adding this extract was small compared to the overall characteristics of the Sr ratio of the bulk fraction. For these reasons, it can be inferred that the 87Sr/86Sr in the exchangeable fraction was derived mainly from plagioclase and that less Sr was extracted in the exchange fraction from K-rich minerals and felsic rock fragments, which have a high 87Sr/86Sr ratio and high resistance to chemical weathering.
Relationships between K2O/Na2O and 87Sr/86Sr ratios of stream sediments in the bulk fraction (a) and the exchangeable fraction (b). The abbreviations are explained in the main text.
However, applying the data to the estimate agricultural production areas is hindered by small numbers of 87Sr/86Sr ratios in the exchangeable fraction (n = 64). Thus, regression analysis of the relationship between 87Sr/86Sr ratios in the bulk and exchangeable fractions was applied to establish missing values (Fig. 4). A polynomial regression model was fitted to the 87Sr/86Sr ratios in the bulk and exchangeable fractions using the least squares method. When the degree of the polynomial function increased from 2 to 4, the R-score no longer changed for values above 3. Therefore, a cubic regression function was applied to the dataset (blue dotted line in Fig. 4). As a result, the p-values for the coefficients of the cubic regression model were all less than the significance level of 0.01 (Table 1). In the present study, the predicted values and 95% prediction intervals using this regression model were provided for the missing 87Sr/86Sr ratios in the exchangeable fraction (Table S1). The map of 87Sr/86Sr in the exchangeable fraction using the dataset integrating measurement and complementary data (Fig. 11b) was constructed at a higher resolution than the map based solely on measurement data (Fig. 11a). The higher resolution map was approximately mirrored by the map of the bulk Sr ratios from stream sediments.
Multiple regression analysis of 87Sr/86Sr in the bulk and exchangeable fractions using cubic regression function*
| Multiple R | 0.8075 |
| R squared | 0.6521 |
| Adjusted R square | 0.7984 |
| Residual standard error | 9.055 × 10–4 |
| Coefficient | Estimate | Standard Error | t value | P-value |
|---|---|---|---|---|
| Intercept (a) | –1315.5 | 472.9 | –2.782 | 0.007 |
| b | 5517.9 | 1992.6 | 2.769 | 0.007 |
| c | –7710.6 | 2798.5 | –2.755 | 0.008 |
| d | 3591.5 | 1310.1 | 2.741 | 0.008 |
* 87Sr/86Sr(ex) = a + b*[87Sr/86Sr(bulk)] + c*[87Sr/86Sr(bulk)]2 + d*[87Sr/86Sr(bulk)]3
Spatial distributions of 87Sr/86Sr in the exchangeable fractions using measurement data (a) and the dataset integrating measurement data and calculated values (b).
87Sr/86Sr ratios were determined using fine stream sediments (<180 μm), which were used for the geochemical mapping of 53 elements in Kyushu and isolated islands around the Kyushu mainland. The 87Sr/86Sr ratios in the bulk fraction of the stream sediments changed systematically according to surface geology. The 87Sr/86Sr ratios were 0.704–0.708 in stream sediments derived from Neogene and Quaternary mafic volcanic rocks and Quaternary pyroclastic rocks, although stream sediments derived from Neogene and Quaternary felsic volcanic rocks had high 87Sr/86Sr ratios (0.705–0.711). Furthermore, the 87Sr/86Sr ratios of stream sediments derived from Cretaceous and Neogene granitic rock and those from Triassic, Jurassic, and Cretaceous high-pressure type metamorphic rocks were 0.705–0.708 and 0.705–0.710, respectively. Stream sediments derived from Palaeogene–Neogene non-accretionary and Jurassic-Paleogene accretionary sedimentary rocks exhibited very high 87Sr/86Sr ratios (0.705–0.721), whereas those from Quaternary sedimentary rocks ranged widely (0.705–0.717).
The 87Sr/86Sr ratios of the fine stream sediments were largely comparable to those of their source rocks. However, certain samples exhibited a difference of 0.001–0.005 in 87Sr/86Sr from their source rocks. Such differences can be explained by the influence of different lithologies coexisting in their watershed areas, changes in the composition ratio of K-feldspar and plagioclase during weathering processes, and the exposure of basement rocks covered by volcanic rocks due to active erosion. With these exceptions, the 87Sr/86Sr map is a sensitive indicator of source rock origin. In fact, we successfully conducted a provenance analysis of clastic particles in non-accretionary and accretionary sedimentary rocks using the 87Sr/86Sr ratios of stream sediments. Furthermore, as for applying the Sr isotope map to identify agricultural product origin, a map of 87Sr/86Sr ratios in the exchangeable fraction of stream sediments is also provided. 87Sr/86Sr in the exchangeable fraction correlates positively with that in the bulk fraction for stream sediments derived from igneous and metamorphic rocks; however, we demonstrated a significant shift to a lower value in stream sediments derived from non-accretionary and accretionary sedimentary rocks relative to the bulk fraction. The results suggest that 87Sr/86Sr in the exchangeable fraction is derived largely from plagioclase in the stream sediment, whereas Sr is not extracted in the exchangeable fraction of K-feldspar, which is highly resistant to chemical weathering.
The authors thank Dr. Y. Asahara for the Sr analysis using TIMS. This work was supported by a Grant-in-Aid for Scientific Research (B) (Nos. 19300301, 22300308, and 18H00756) and (A) (No. 21H04359). We are grateful to Dr. T. Iizuka and one anonymous reviewer for their constructive comments.
