Samples used for grain size and mineral analyses were treated to remove the organic matter and carbonate, separating the coarse and fine fractions. The procedure was carried out as follows: suitable amounts of sediment sample were placed in a beaker, oxidized and disaggregated by 30% H₂O₂. The above steps were repeated every 12 h until no air bubbles remained to completely remove the organic matter. Then, the solution was desalinated with deionized water and centrifuged a total of 6 times. Afterward, the sample was placed in a water bath box at a constant temperature of 35°C after adding twice sufficient 30% acetic acid, continuously shaking the sample, and repeating the centrifuge procedure. Finally, the carbonate-free fraction was wet-sieved through a 63 μm (240 meshes) sieve. Oversize and undersize subsamples were collected separately to dry them for later use.

The grain size distribution of the carbonate-free coarse fraction (>63 μm) was analyzed by a standard sieve shaker test, with a size interval of 1Φ. The relative content of each grain size after weighing was calculated. For the carbonate-free fine fraction (<63 μm), prior to the grain size measurements, subsamples were disaggregated with 3 mL of 0.05 mol/L (NaPO₃)₆ and sonicated for 30 min. Grain size distributions were determined using a laser particle size analyzer (Mastersizer 2000, Malvern Company) with a measurement range from 0.02 to 2 000 μm. The grain size detection resolution was 0.01Φ with a relative error of less than 3% during repeated tests. Grain size data were exported based on the detection interval of (1/4)Φ. Finally, the relative proportion for each grain size fraction was computed by normalization. The grading standard was the Udden-Wentworth grade (Udden, 1914; Wentworth, 1922), using the formulas of McManus to calculate grain size parameters (McManus, 1988).

From the coarse-grained fractions, the microscopic observation and identification were carried out using an OLYMPUS SZ61 stereo microscope. In addition, the morphology of the biogenic component was examined by scanning electron microscopy (SEM, Quanta 200, FEI Company) and energy dispersive spectroscopy (EDS, GENESIS 2000, EDAX Company). The beam spot and work distance of SEM are 5.0 and 10 mm (25 kV, high vacuum mode), respectively, and the magnification varies from 150 to 600. From the fine-grained fractions, the mineral composition was found using the X-ray powder diffraction method with a Bruker D8 ADVANCE diffraction meter produced by Bruker (Germany). Diffraction patterns in the 3°–65° range were obtained using a 0.02° (2θ) step scan and 4°/min scanning speed. Operating conditions were 40 kV and 100 mA using Cu-target radiation. Peak areas were calculated after manual baseline correction using Jade 6.5 software, following the semiquantitative k-value method (Chung, 1974).

Among the carbonate-free fraction of each sample, distinct differences in terms of grain size characteristics are apparent. For the grain size parameters of this fraction, see Table 1. The mean grain size ranges between 1.96Φ and 8.19Φ with a mean of 5.53Φ. The sorting coefficient is 2.07–4.04 with an average of 2.64, indicating its poor sorting generally or even particularly poor sorting for very few samples. The skewness shifts from very positive to very negative, with the value of –2.34 to 3.65. Forty percent of all samples show a majority of fine-grained particles due to the negatively skewed distribution, while others have the opposite grain size composition. The kurtosis is over 2 and the range is 2.47–4.48, which is leptokurtic. Figure 2 illustrates the variation trend in the mean grain size distribution. The finest grain size occurs at SW12 from the north margin of this area, whereas SW10 from the southeast of the ridge axial region shows the coarsest grain size. Generally, the particle sizes in the carbonate-free fraction become slightly smaller from southeast to northwest.

The grain size distribution of the carbonate-free fraction can be summarized in three basic types: unimodal, bimodal and irregular bimodal patterns. Four samples are unimodal with a mode of approximately 1.95–3.91 μm (Fig. 3a). They barely contain the fraction coarser than 500 μm, suggesting an obvious negative skewed distribution. Most samples show a bimodal distribution pattern in which the coarse and fine modal sizes are at 125–250 μm and 1.95–3.91 μm, respectively (Fig. 3b). Some of these samples have a significant positive skewed tendency. Further, the size of the two modes in a very few samples is reversed. In the irregular bimodal pattern, there is little difference in the content of two modes, with a modal size of approximately 500 μm and 125 μm, respectively (Fig. 3c). In particular, gravel-grained particles, which have different contents from 11.68% to 68.34%, occur in the >2 000 μm fraction.

Based on the grain size distribution of the noncarbonate fraction, we selected three major coarse-grained fractions for microscopic identification, including >2 000 μm, 500–1 000 μm and 125–250 μm. In the fraction of >2 000 μm, only SW1 approximately 1 km from the 49.6°E hydrothermal field are dominated by massive or fibrous serpentine (Fig. 4a), whereas the remaining gravel-bearing samples, such as SW10, SW13, SW15 and SW18, mainly consist of volcanic breccia and volcanic glass (Figs 4b, c). The volcanic debris grains have an angular shape and develop a vesicular structure, with red-brown partial surface coloration due to iron oxidation.

In the 500–1 000 μm fraction, SW1 still mainly contains serpentine, while individual samples are rich in volcanic glass and volcanic ash, as well as plagioclase and little volcanic matter in certain other samples (Fig. 4d). Of these, plagioclase grains are in angular or subangular shapes and have shallow pits and dark impurities on parts of the surface, with the red-brown color that may be related to oxidation of iron.

In the fraction of 125–250 μm, plagioclase and siliceous bio-clasts are generally predominant, with the exception of SW1 and individual samples that are full of volcanic matter. These biogenic components are mainly radiolarian skeletons, followed by siliceous sponge spicules, with intact morphology (Fig. 4e). Through comparison of different size fractions, we discover that siliceous bio-clasts are obviously enriched in this fraction such that five samples comprise it almost exclusively: SW4, SW9, SW11, SW12 and SW14.

The XRD results show that 16 mineral phases are detected in the carbonate-free fine fraction (Table 2). The major mineral phases are clay minerals (18%–78%), including chlorite, illite, smectite, and kaolinite, quartz (12%–70%) and plagioclase (3%–32%), which are found at all of the stations, followed by pyroxene. Nontronite, pyrite, serpentine, talc, magnetite and hornblende are discovered at several stations near the hydrothermal field. Sphalerite and epidote are also distributed in areas distant from the field. Barite only appears at SW20, the farthest from the hydrothermal field.

The probability distribution of the cumulative grain size can directly and clearly reflect the characteristics of sediment transport and hydrodynamic environment. EM1 is the typical suspended load transport of one segment pattern (Fig. 6a). The total amount of suspended load section is approximately 95%, which corresponds to its relatively wider unimodal distribution. Therefore, it is shown that a large amount of suspended materials falling out from the seawater form the carbonate-free fine fraction in the SWIR surface sediments.

This finer end member consists mainly of clay minerals and quartz, with a small amount of silicates, sulfides and sulfate. Clay minerals other than nontronite in the seafloor have three main sources of terrigenous, submarine weathering and hydrothermal origin, which can be transported by wind, bottom current, ice-rafting, and so on (Weaver, 1989). Recent comparative analyses of clay minerals assemblages among different regions have indicated that the major source of clay minerals in the area is aeolian dust from the Namib and Kalahari deserts of southern Africa, as well as some from sediments of Antarctica offshore carried by the Antarctic bottom current^①

① Zhang Wenqiang, Fan Dejiang, Sun Xiaoxia, et al. The mineral composition and sources of the fine-grainedl sediments from 49.6°E hydrothermal field of the SWIR. Journal of Ocean University of China, https://doi.org/10.1007/s11802-019-3797-6

. In addition, clay minerals in sediments near the vent may have hydrothermal origins. Nontronite is found here and it is an indicator mineral (Herzig and Plüger, 1988; Severmann et al., 2004; Dias and Barriga, 2006) related to hydrothermal activity. Quartz is stable at ocean-bottom conditions and does not form authigenically in recent sediments. Most grains in pelagic sediments are in the 5–10 μm range, although the grain size varies with grain origin and transport mechanism (Rex and Goldberg, 1958). The high quartz abundance in the Southwest Indian Ocean is in partially derived from continental material of South Africa and Madagascar and partially derived from the Antarctic continent, as discussed in detail by Kolla et al. (1976b). Thus, the fine-grained quartz distribution within the region is mainly affected by aeolian transport and bottom current processes (Leinen et al., 1986). Closely associated with quartz, plagioclase can exist as terrigenous debris or the weathering debris of basement basalts (Peterson and Goldberg, 1962; Humphris and Thompson, 1978). Pyroxene is one of the major rock-forming minerals of both basalt (Eggleton et al., 1987) and peridotite (Snow and Dick, 1995), and its widespread distribution indicates that it is produced by the bedrock weathering process. Many scholars have noted that the formation of serpentine, magnetite, talc and hornblende near-vent is associated with the alteration (Kelley et al., 2001; Allen and Seyfried, 2003; D’Orazio et al., 2004) and metamorphism (Bonatti et al., 1975) of bedrocks in hydrothermal circulation. Epidote and sphalerite occur at the same time as two stations in this area, which is similar to the results of previous studies on hydrothermal sulfides (Alt et al., 1996; Bowman et al., 1987). Therefore, we suggest that the epidote is likely to be formed at the same time as sphalerite after surrounding rocks suffer from hydrothermal replacement and finally fallout into the sediments. The sulfides distributed around the hydrothermal vent undoubtedly belong to hydrothermal deposits of the 49.6°E hydrothermal field. Tao et al. (2011), Ye et al. (2011), Sun et al. (2014) and Jia et al. (2017) conducted detailed mineralogy studies on chimneys, massive sulfides, upper suspended water, and surrounding sediments, respectively, in this region, and all found pyrite and sphalerite. Barite may be the authigenic deposit most closely associated with primary productivity in the marine sedimentary environment (Sun, 2011).

The above analyses of mineral sources show that EM1 reflects a significant contribution of terrigenous materials of the aeolian and bottom current inputs, although it also contains a few suspended materials related to seafloor weathering and hydrothermal circulation. According to previous research on simulating dust deposition flux using global models, the desert dust aerosol from southern Africa that is dominated by particles of diameter 0.1 to 10 μm, with the mean size being approximately 2 μm could reach the study region in the Southwest Indian Ridge (Jickells et al., 2005).

EM2 has a bi-segment pattern (Fig. 6b) that consists of the bed and suspended load, with a cut-off of 4Φ (62.5 μm). The bed load section, which corresponds to the major mode of EM2, has a higher proportion of up to 70% and slope of approximately 60°, suggesting a slightly better sorting. This end member also has some suspended particles whose the size is the same as that of EM1.

As mentioned above, siliceous bio-clasts are enriched in the fraction of 125–250 μm. This size range is consistent with previous research on the size of radiolarian skeletons in the subtropical region by Schmidt et al. (2006). Through SEM examination, several common species, including Lamprocyclas maritalis, Rhopalastrum hexaceros, Stylacontarium octatignum, Spongocore polyacantha, Octopyle stenozona, Acrosphaera spinosa, Theopilium tricostatum, and Hexalonche philosophica, were identified in this area. Radiolarians are planktonic protozoa widely distributed throughout the water column in oceans, which are adapted to a drifting existence. Conversely, sponges are important contributors to benthic biomass in predominantly deep-sea environments and are frequently found on firm surfaces such as rocks or soft sediments. When these creatures die, their skeletal remains make up a part of the cover of the ocean floor as siliceous bio-clasts. Thus, this biogenic component in the area belongs to local biogenic deposits. In addition, considering the intact morphology of the shells described previously, we suggest that the study region is in a weak hydrodynamic environment.

Compared with EM2, EM3 has 10% gravel-grained particles that have not experienced any transport process. The bed load section has a wider size range and lower content (58%), whereas the content of the suspended load is unchanged. The secondary part can also be observed in the suspended load (Fig. 6c). This may reflect that the hydrodynamic environment or provenance of this end member has changed.

The 500–1 000 μm fraction sediments are mainly composed of plagioclase, volcanic ash and volcanic glass. EDS analyses show that the coarser plagioclase consists mostly of basic plagioclase (An: 80.20%–85.57%, Ab: 14.57%–19.80% and Or: 0%). Thus, the plagioclase mineral mainly originate from the submarine weathering of basalt rocks. Previously mentioned volcanic activities of ridge segments in this region are intense and of relatively quiet overflow eruption to form seafloor basalts. Consequently, volcanic glass and ash are also the result of volcanic activities or the weathering of basalt rocks.

The remarkable characteristic of EM4 is that the gravel-grained sediments without transport account for 70% of the total. In contrast, the contents of the bed load and suspended load are very low, 20% and 10% (Fig. 6d), respectively. This directly reflects that the water mass movement in this region has no ability to reform gravel-grained or much coarser sediments. Therefore, the distribution of these coarser particles should be closer to their source area.

The gravel-grained fraction in this end member mainly consists of volcanic matter, and with few hydrothermal alteration products, serpentine. The source material between EM3 and EM4 is very similar. It also shows indirectly that the input of volcanic materials makes a large contribution to the coarser-grained sediments in this area. Therefore, EM4 could be interpreted as representing the coarsest volcanic matter from the seafloor bedrocks of the SWIR.

The relative contributions of four end members show distinct spatial variations. All the samples are divided into three types according to their end member contents (Fig. 7). Type 1 includes 55% samples that are mainly composed of EM1. In Type 2, EM2 is the dominant end member. Type 3 is marked by a high proportion of EM3 and EM4.

From the wide distribution of samples in Type 1, it is observed and accepted that the aeolian and Antarctic bottom current play the key role in the carbonate-free fine sediments transport in SWIR. In addition, the content of EM1 has an increasing tendency with the distance from the hydrothermal activity field. This trend is thought to be mainly controlled by sediment provenance. On the one hand, the dilution of particles associated with hydrothermal circulation near-vent is relatively clear. These particles contain not only fine suspended settling materials but also coarse hydrothermal alteration products. Further, the coarse particles of volcanic and biogenic sources in EM2 and EM3 also have a dilution effect on EM1. For example, SW5 and SW11 have higher EM2 content. Moreover, the flat areas of the seafloor create beneficial conditions for the accumulation of suspended materials, including hydrothermal components. Due to the lack of actual observation data about topography, there is no further discussion about this aspect.

By combining Type 1 with Type 2, the results show that there was no significant regularity in the spatial distribution of the biogenic component. For how it was created, there are various factors. The biogenic fraction content in marine sediments is related to the supply of biological skeletons in upper or bottom seawater and has a closed relationship with ocean biogenic production controlled by temperature, salinity and hydrological properties of the upwelling current and ocean current. Additionally, siliceous skeletons usually dissolve because of the unsaturated dissolution of silicon dioxide in modern oceanic water. Furthermore, other pelagic sediments can also have a dilution effect on the relative content of biogenic sediment.

The samples in Type 3 have a high content of the coarser (>500 μm) volcanic materials. According to the grain size compositions of four end members, there is nearly 20% basement weathering debris of the coarsest grained materials without transport or reform in the carbonate-free fraction. We believe that such samples provide a better estimate of the sources and average content of the gravel particles in the noncarbonated fraction sediments from the SWIR. Therefore, the concentrated input of endogenous materials, which refers to the volcanic origin, is the main cause of the different spatial distributions of EM3 and EM4.

In summary, the distinct spatial variations of the end members are mainly related to different provenances. Sediments from terrigenous, submarine weathering process of bedrocks, biogenic production and hydrothermal input come together to form a mixed population, such as the carbonate-free fraction of surface sediments in the SWIR. Each origination supplies characteristic mineral suites that contain special grain size information. Therefore, multi-provenance is a major factor influencing the grain size pattern in this fraction. Second, since the entire region has weak hydrodynamic action owing to its deep-sea environment, only the Antarctic bottom current that flows to the northwest can provide a velocity component (Mantyla and Reid, 1995) to transport some of the surface sediments and control their deposition in the seabed. Combined with other indirect transport processes in this region, such as wind, the particle sizes ultimately present a slightly decreasing tendency from southeast to northwest. In addition, other factors such as topography may influence the grain size composition of SWIR surface sediments.