1 INTRODUCTION

Silicon (Si) is the second most abundant element in the Earth’s crust (Ding et al., 1994; Wedepohl, 1995). It is a key constituent of diatoms (DeMaster, 1981), which is closely related to the global carbon cycle (Tréguer et al., 1995; Ragueneau et al., 2000), thereby, plays a key role in marine ecosystem material cycling. The amount of silicon fixed by terrestrial plants is 55– 113 Tmol Si/a (Carey and Fulweiler, 2012), which is in the same order as the total BSi production by the Earth’s marine systems, which is estimated at 240 Tmol Si/a (Tréguer et al., 1995). Many studies show that BSi is easy to dissolve in the water column and sediment in comparison with silicate weathering (Alexandre et al., 1997; Loucaide et al., 2008). It, therefore, has become the main and most direct source of dissolved silicate (DSi) in rivers (Derry et al., 2005), thereby playing an essential role in linking terrestrial and aquatic ecosystems worldwide (Conley, 2002; Tréguer and De La Rocha, 2013). In recent years, many studies have focused on silicon transport and outflows as well as the factors that control these processes (Ragueneau et al., 2000; Conley, 2002), but BSi composition and sources are still poorly understood.

The rate of silicate weathering in tropical regions is thought to be higher than in other regions (Jennerjahn et al., 2006), with a faster Si turnover in these environments (Lucas et al., 1993; Rose et al., 1993). Alexandre et al. (1997) showed that the about 92% of annual BSi production was rapidly recycled in rain forest environments, which means that most BSi is reactive silica and controls the silica biogeochemical cycle. Although phytoliths have been observed in river water (Olivié-Lauquet et al., 2000; Cary et al., 2005; Ran et al., 2015), they have not been well quantified in terms of their contribution to the BSi reserve in rivers worldwide. Such quantification would bring new insights regarding the processes involved in the biogeochemical silicon cycle. Analyzing the form and composition of BSi present in tropical ecosystems will enable a deeper understanding regarding the sources of BSi and its contribution to marine silicate volumes. This study examines the form and composition of BSi found in river sediment samples from Peninsular Malaysia, focusing on quantitative BSi characterization. The main objective of the research is to systematically investigate the transfer and composition of silicon in tropical rivers, with the aim of elucidating its biological role in these systems.

2 MATERIAL AND METHOD 2.1 Study area

The Pahang, Pontian and Endau Rivers are located in the state of Pahang, on the east of the Malaysian Peninsular. The Pahang River is about 459 km long with a catchment area of 25 600 km². It flows into the South China Sea and has an average discharge of 596 m³/s. The sampling area is covered with rubber and oil palm plantations, rice paddies and other tropical plants, but the coastal plain is mostly swampy. The eastern coastal states of Malaysia are strongly influenced by the Northeast Monsoon Season, which often results in severe floods that wash the catchment’s topsoil into the South China Sea.

2.2 Sampling

Eight sediment samples were collected along the Pahang, Pontian and Endau Rivers in May 2012 (Fig. 1). Samples were preserved at -20℃ for subsequent laboratory analysis with respect to the content and composition of BSi, particulate organic carbon (POC), sediment grain-size and organic carbon in phytoliths (PhytOC).

2.3 Analytical methods

Each sample was divided into three parts after drying. One part was used for BSi determination, which was achieved using the alkaline extraction method (DeMaster, 1981). To this end, 40.0±1.0 mg of dry sediment was extracted using 40 mL of 1% Na₂CO₃ solution at 85℃ for 8 h. Meanwhile, a second sample was extracted by a modified two-step method, which includes a mild acid pretreatment (0.1 N HCl at~22℃ for 18 h) and a subsequent alkaline extraction (using 1% Na₂CO₃ at 85℃ for 8 h) to estimate the total amount of reactive silica and the extent of reverse weathering in the sample (Michalopoulos and Aller, 2004; Presti and Michalopoulos, 2008). The Si concentration in the extract was analyzed by a QuAAtro Autoanalyzer, using the silicomolybdic blue method with a precision of 5%–10% at < 1– 10 μmol/L, and 1%–5% at >10 μmol/L (Ran et al., 2013). The operational Si pools are defined as follows: Si-HCl (mild acid leachable), Si-Alk (mild alkalineleachable after acid pre-treatment), ∑Si (Si-HCl+SiAlk, representing total reactive silica), and BSi (mild alkaline-leachable without acid pre-treatment) (Michalopoulos and Aller, 2004).

The third extraction, of amorphous silica particles, was performed after a wet extraction procedure (Wang and Lu, 1993) to determine BSi composition. The procedure involved the following steps: (1) dissolution of carbonates using 1 N HCl; (2) oxidation of organic matter using 30% H₂O₂ at 90℃; (3) removal of the clay fraction ( < 2 μm) by decantation based on gravitational sedimentation; (4) densimetric separation of opal particles in heavy ZnBr2 liquid (with a density of 2.35 g/cm³) (Wang and Lu, 1993). The extracted particles were observed under an optical microscope (Nikon Eclipse E100) and PhytOC was determined. Approximately 300–500 BSi particles were counted on each station’s microscopic slide, and their biovolume was determined. The phytoliths were classified according to the International Code for Phytolith Nomenclature (ICPN). General morphological assessments of the particles were performed using a scanning electron microscope (SEM, Quanta 200, FEI Company, Hillsboro Oregon, USA). The SEM was equipped with an energy dispersive spectrometer (EDS) system which was used for detailed morphological and chemical studies. Both the POC in the sediments and the PhytOC in the BSi were analyzed by the Euro EA3000, with a precision of≤0.3%. Before measurement, sediment samples were decalcified with 6 mol/L HCl and rinsed to neutral for analysis (Fischer et al., 2002). Sediment grain sizes were analyzed using the Mastersizer 2000 (Malvern Ltd., Malvern, UK). The measurement range of this instrument is 0.02–2 000 μm, with a precision of < 3% and a resolution of 0.01 φ. The samples were pretreated with 30% H₂O₂ (15 mL for 24 h) and 3 mol/L HCl (5 mL for 24 h), respectively, in order to remove organic matter and calcareous cement. The samples were then rinsed to neutral. All samples were fully desalted and dispersed before the grain size analysis took place (Liu et al., 2014).

3 RESULT 3.1 Biogenic silica content

The reactive Si contents of the Pahang River samples are shown in Table 1. The BSi content in these sediment samples ranged from 5.17 to 10.2 mg/g, with a mean value of 8.55±1.67 mg/g. The content of Si-Alk, similar to BSi, was 8.55±2.64 mg/g, ranging from 4.06 to 12.2 mg/g. The mean value of Si-HCl and the total reactive silica value were 3.29±1.16 mg/g and 11.8±3.69 mg/g, respectively. The minimum value for every form of Si was recorded at St5 in the Pahang River estuary. The average proportions of the different forms of Si present in these sediment samples were Si-HCl/∑Si~0.28, SiAlk/∑Si~0.72, BSi/∑Si~0.75 and BSi/Si-Alk~1.05, respectively.

3.2 Composition of biogenic silica

As shown in Table 2, the BSi in the sample sediments comprised phytoliths, diatoms and sponges, whose mean proportion values were 76.5%, 21.4% and 3.34%, respectively. The composition of BSi in the Pahang River was fairly different to those in the Pontian and Endau Rivers.

Phytoliths were the most abundant BSi components (92.8%–98.3%) at sample stations St1–St6. Elongate phytoliths were the most abundant form (35.9%), followed by globular (33.1%) and globular echinate (10.0%) phytoliths. The proportions of other measured phytolith forms (i.e., cuneiform, ovate and saddle) were low and contributed little to total BSi values. At stations St1–St6, the pennatae diatoms were generally the dominant diatom form, representing 3.71% of the total BSi and 96.4% of diatom BSi. In contrast, diatoms dominated the BSi forms at St7 and St8, with an average value of 74.1% (these stations were on the Pontian and Endau Rivers, respectively). Centricae diatoms were the most abundant diatom BSi forms at both of these stations, accounting for 59.9% of the total diatom BSi.

Most of the phytolith sizes ranged 10–200 μm. The largest particles (>30 μm) were those of the elongate, cuneiform and cylindric sulcate tracheid forms. The surfaces of these large particles tended to be pitted. By contrast, diatom and sponge BSi particles were generally 20–150 μm and were mostly in the form of debris. Examples of the most typical geometric BSi forms are presented in Figs. 2 and 3.

3.3 Particulate organic carbon in sediments and BSi

As shown in Table 1, the average POC content in the samples was 1.08%. The minimum POC content (0.17%) was recorded at St5 and the maximum POC content (3.16%) was recorded at St8. The BSi carbon content (PhytOC) ranged from 1.85% to 10.8%, with an average of 4.79%. The amount of PhytOC extracted from the BSi particles was low at stations St4–St6, but relatively high at station St1.

3.4 Grain size

The sediments in this area were mainly classified as sandy silt and silty sand (according to Fokker nomenclature), with a mean range of 3.51–6.57 φ. The proportions of gravel (>63 μm) and powder (4– 63 μm) particles were higher than that of clay ( < 4 μm). The average proportions of each of these fractions were 42.0%, 45.0% and 13.1%, respectively.

3.5 SEM observations

The results of SEM observations and EDS chemical analyses are presented in Fig. 4. Detailed SEM observations indicated that significant portions of the phytoliths’ surfaces had been corroded. Additionally, parts of the phytoliths and sponge spicules were covered by a remnant of surface coatings. EDS measurements showed that the phytoliths consisted of cation-rich aluminosilicate phases, such as Si, Al, K, Mg and Fe (Fig. 4a). In general, phytolith elements consisted primarily of Si and O, with C and Al contributing little to their weights.

4 DISCUSSION 4.1 Content and composition of biogenic silica

The mean BSi content of the sampled sediments was 8.55 mg/g (0.85% by weight), which is equivalent to the BSi content in soil samples from other areas, such as the tidal freshwater marsh at Scheldt estuary in Belgium (0.9%–1%) (Struyf et al., 2005), the Baiyangdian wetland in China (0.80%) (Li et al., 2013a) and a riparian wetland in Poland (0.2%–0.8%) (Struyf et al., 2009). This indicates a high primary production in the study basin. In addition, a significant positive correlation between BSi content and sediment particles with a grain size < 40 μm (R²=0.60, n=8, P=0.01) proved that BSi contents are considerably affected by sediment texture and are fairly stable in sediments comprising large proportions of fine particles.

Results show that the BSi content was lowest at station St5 in the Pahang River estuary. This may be due to removal processes such as resuspension, dissolution, burial and transformation (DeMaster, 1981; Conley and Malone, 1992; DeMaster, 2002) of BSi in the estuary environment.

In this study, the mean ratio of BSi/∑Si was 0.75, which suggests that the BSi pool represents a substantial proportion of the sediments’ total Si content (Presti and Michalopoulos, 2008). The BSi/ Si-Alk ratio for the samples approached a value of 1, which is very close to the result (0.95) obtained for suspended matter in Mississippi River (Presti and Michalopoulos, 2008). This indicates that the degree of authigenic alteration to BSi in the study area is reasonably low. The mean (∑Si-BSi)/∑Si ratio was 0.25, which suggests that approximately 25% of the reactive silica in the sediments was subject to reverse weathering processes as described by Michalopoulos and Aller (2004). In addition, the reverse weathering of reactive silica in the study area was lower than that for the Amazon River delta (65.7%, Michalopoulos and Aller, 2004) and that for the Mississippi River delta (59.5%, Presti and Michalopoulos, 2008), suggesting a high silica turnover rate in the area. In general, the vegetation in humid tropical environments is fairly efficient at recycling silicon and other elements (Lucas et al., 1993), and a high biomass turnover may weaken the reverse weathering process.

4.2 Chemical composition of phytoliths and organic carbon in sediments and BSi

The EDS results (Fig. 4) revealed that quite a significant amount of carbon was fixed in the sampled phytoliths. Other elements, namely, Al, Fe, K, Mg and Cl, were also found in the phytoliths. Michalopoulos et al. (2000), and Michalopoulos and Aller (2004) found that BSi particles have undergone conversion to aluminosilicate phases by combination with cations such as Al³⁺, Fe³⁺, K⁺in water during the transportation process. This may also be a potential sink for Si, K, Fe and Al.

The results of PhytOC indicate that about 5% of organic carbon is sealed in tropical BSi, which is relatively higher than the estimates presented by Zuo and Lv (2011) and Li et al. (2013b). The average percentage of PhytOC present in millet, which is a typical example of a dry-farming crop cultivated in China, is estimated to be 2.21%, (Zuo and Lv, 2011), and it is estimated that 1.55% of PhytOC is sequestrated in subtropical wetland plants (Li et al., 2013b). In this study, the range of average BSi carbon contents measured at stations St1–St6 indicate that the degree of carbon trapping varies amongst phytoliths and is largely a function of the type of plants that produced the phytoliths, and the temperature zones in which these plants grew. There is no doubt, however, that the phytoliths have sequestrated significant quantities of carbon and should, in the future, be incorporated into global carbon cycle models.

4.3 Biogenic silica forms and their implications

The elongate phytolith type showed a positive correlation (R²=0.49, n=8, P=0.03) with larger sediment particles (63–2 000 μm), and a negative correlation with smaller ones [i.e., 4–63 μm (R²=0.48, n=8, P=0.03) and < 4 μm (R²=0.50, n=8, P=0.03)]. This may indicate that these phytoliths had a relatively short retention time in the river water and tended to be deposited on larger sediment particles due to the fairly faster sedimentation rate of larger particles.

Phytolith type is widely used to identify the plant species that produced the phytoliths. In this study, elongate phytoliths showed were significantly positively correlated with elongate echinate (R²=0.72, n=6, P=0.02) and cuneiform (R²=0.64, n=6, P=0.03) phytoliths, while being negatively correlated with the globular echinate (R²=0.49, n=6, P=0.08) and globular (R²=0.67, n=6, P=0.03) phytoliths. This suggests that the former group originated from the same plant type whereas the latter originated from other plant types. It has been found that elongate and cuneiform phytoliths tend to develop in leaf cells and bulliform cells, respectively (Wang and Lv, 1993). Globular and globular echinate phytoliths, however, are uniquely associated with the Palmae family of tropical plants (Piperno, 1988; Wang and Lv, 1993). Furthermore, a correlation of the phytolith assemblages found at each of the sampling stations where phytoliths contributed the largest portion of BSi (i.e., stations St1–St6), showed that station St1 was significantly positively correlated with station St2 (R²=0.92, n=6, P < 0.01), station St3 (R²=0.84, n=6, P < 0.01), and station St4 (R²=0.79, n=6, P=0.01). Likewise, a significant positive correlation was found between the assemblages at stations St5 and St6. This indicates that phytoliths found at stations St1–St4 originate in similar vegetation types which are different to those that produced the phytoliths found at stations St5 and St6. Riparian vegetation in the vicinity of stations St1–St4 comprised mostly oil palm plantations, forest and rice paddies, whilst stations St5 and St6 were situated in a wetland forest near the Pahang River estuary. It therefore seems clear that the different phytolith assemblages found in the river sediments can be attributed to variations in the local vegetation. Unlike the phytolith assemblages found in the Yellow River basin (Ran et al., 2015) and the Changjiang River basin (unpublished data), those found in this study included the unique globular echinate phytolith form, but not the cross-shaped phytolith form. The cross-shaped form is typically formed in the short cells of Panicoideae species (e.g. corn), whilst the globular echinate form only originates in palm plants (Wang and Lv, 1993). This further indicates that there is a link between BSi in sediments and the type of vegetation in a river basin.

Alexandre et al. (1997) showed that the quantity of silica released into rivers via phytoliths is about double that released through silicate weathering. As shown in Figs. 2 and 3, indicators of phytolith corrosion as well as diatom fragmentation were observed in sediment samples. This suggests that a portion of the BSi dissolved and was released into the river water as a source of dissolved silicate. Thus, the combination of the vast biomass present in tropical ecosystems and a relatively high dissolution rate could mean that terrestrial plants supply a substantial quantity of silica to river systems.

This study suggests that the BSi found in sediments along the rivers of the Malaysian Peninsular can primarily be attributed to terrestrial plants, and that the sizable silica pool in tropical environments may represent a significant component of the coastal biogeochemical silicon cycle. Additional research is required with respect to the three-dimensional morphology and size of phytoliths in order to determine exactly the plants from which they originate, and to further elucidate their role in the silicon cycle.

5 CONCLUSION

The BSi extracted from sediment samples from the Malaysian Peninsula comprised fractions of phytoliths, diatoms and sponge spicules. BSi was present in sediments at a concentration of 8.55 mg/g. Phytoliths comprised 92.8%–98.3% of the total BSi found in the Pahang River sediments. The dominant phytolith form in the samples was the elongate type, followed by the globular and globular echinate types. Together, these represent a typical phytolith assemblage originating from Palmae plant species.

Diatoms present in the river sediments were predominantly of the Pennatae type. These played a relatively minor role in the BSi composition along the Pahang River. However diatoms accounted for 68.8% and 79.3% of total BSi found in sediments from the Pontian and Endau Rivers, respectively, where the dominant diatom type was Centricae.

The carbon content of the phytoliths ranged from 1.85% to 10.8%, with an average content of 4.79%, and BSi plays a major role in controlling permanent carbon burial.