The Effect of Size Distribution on the Geochemistry and Mineralogy of Tropical River Sediments and Its Implications regarding Chemical Weathering and Fractionation of Alkali Elements

The weathered products of silicate rocks deposited off river mouths have been extensively used to reconstruct the past chemical and physical weathering history of continents at different time scales [1–7]. The large peri-Himalayan river basins of southeast (SE) Asia have the highest physical and chemical denudation rates in the world [3, 8, 9], and the resulting sediments account for up to 70% of the sediments deposited into oceans [2, 10, 11]. Therefore, these river basins represent a key region for better understanding the factors controlling chemical and physical weathering on land. Moreover, river sediments deposited into marginal seas (e.g., South China Sea, Bay of Bengal, and Arabian Sea) can be used to reconstruct paleoenvironmental and paleoclimatic changes associated with the evolution of the Asian monsoon or sea level changes (e.g., [1, 2, 12–16]). Studies have been conducted on the chemical weathering of river sediments in the basins of several SE Asian rivers, such as the Mekong River, Red River, Pearl River, Taiwanese mountain rivers that flow into north-western South China Sea (SCS) [5, 17, 18], and major rivers in Luzon in the eastern SCS [18], as well as the basins of small rivers flowing into the southern SCS in Sumatra, Borneo, and the Malay Peninsula [2, 3]. The results of these studies suggest that chemical weathering is intensive in tropical SE Asia (humid and warm climate) but that it also depends on tectonic activity in river basins [2, 3]. However, most studies have focused on the intensity and controlling factors of chemical weathering, whereas only a handful have investigated the fractionation of alkali elements due to the transportation and sorting of river sediments in tropical regions with intense hydrolysis conditions [19]. The effects of grain size on geochemical and mineralogical compositions are yet to be determined.

The division of mineralogical components into diverse grain-sized fractions could help us to understand the geochemical variations that arise during the process of transportation and sorting of river sediments [20–22]. Coarse-grained sediment fractions are rich in elements such as Zr and Th, with instances of higher relative contributions of zircon and monazite [23, 24]. Fine-grained sediments undergo intense chemical weathering [21, 25, 26], which may be attributed to the small particle size, as suggested by several studies [22, 27, 28]. Only a small number of studies have explored the relationship between grain-size fractions and chemical weathering in tropical river basins. Several studies aimed at understanding chemical weathering have focused on the <63 μm fraction, considering a negligible effect of grain size and hydrodynamic sorting on the mineralogical and geochemical tracers [2, 29]. Lupker et al. [27] suggested that the ratios of mobile to immobile elements should be normalized into a common Al/Si ratio in Himalayan River systems to allow a better comparison of the chemical composition of sediments subjected to sorting. It is important, therefore, to better constrain the relationship between mineralogical and geochemical proxies of chemical weathering intensity and particle size for the <63 μm fraction of river sediments from tropical regions.

Alkali elements such as Na, K, Rb, and Cs are highly soluble, but their geochemical behavior is more complex than expected [28]. The fractionation of alkali elements typically occurs during weathering of silicate rocks [30, 31]. This fractionation results from the ionization of alkali elements and their later sequestration on the weathering products [28, 30]. This process can be better understood by analyzing the alkali elemental distribution in the different grain-size fractions of river sediments [22, 27].

In this study, major and trace elements and mineralogical compositions were investigated for bulk sediments and seven different grain-size fractions of riverbed samples collected from two major tropical rivers of the Malay Peninsula (the Pahang River and Kelantan River). The aims are to (1) better constrain the relationship between particle size and elemental and mineralogical compositions and (2) reevaluate the reliability of proxies commonly used to reconstruct weathering intensity and/or sources of sediments deposited in the South China Sea and in the ocean more generally.

The Malay Peninsula has an area of ~131,554 km² [32] (Figure 1(a)). It is a part of the Sundaland region and has been tectonically stable since the Mesozoic [2, 33]. Its bedrock is mainly composed of Paleozoic-Mesozoic granite, granodiorite, and Paleozoic mudstone, sandstone, and andesitic-rhyolitic volcanic rocks [34] (Figure 1(b)). The climate of this region is chiefly controlled by East Asian-Australian monsoon. Although temperature variations (from 25.7 to 28.4°C) are not significant throughout the year, precipitation rates differ significantly between summer and winter [2, 3]. From September to February, the region is under the influence of the south-westerly East Asian monsoon and thus experiences a wet season with high precipitation (precipitation ranges from 140 to 598 mm), whereas from March to August, the region is under the influence of the north-easterly Australian monsoon and thus experiences a dry season with reduced precipitation (precipitation ranges from 150 to 178 mm) [2].

The Pahang River (459 km long), whose upper reaches are located in the Titiwangsa Main Range, is the longest river in the Malay Peninsula (Figure 1). The river has two main tributaries, namely, the Jelai and Tembeling Rivers, which converge at Kuala Tembeling, located 300 km away from the estuary. The drainage area of the Pahang River covers 29,300 km², with annual runoff and precipitation of 1500 mm and 2170 mm, respectively [2]. The annual sediment discharge of the Pahang River is 20.4 Mt [35]. The upland areas are separated, with elevations ranging from 1000 m to 1500 m and with some peaks exceeding 2000 m. The topography is less rugged towards the main drainage area at the center of the river basin, with low hills and a maximum elevation of 75 m. Two types of soil are found in the upper reaches of the Pahang River: ferralitic soil, formed by the weathering of granitic rocks, and secondary lithosols. In the middle reaches, ferralitic soil formed due to the weathering of sedimentary rocks is the main soil type. The lower reaches are mainly composed of ferralitic and organic soils [36].

The Kelantan River, with a length of 335 km, is the second largest river in the Malay Peninsula (Figure 1). The drainage area of the Kelantan River is 12,691 km², with average annual runoff, precipitation, and sediment discharge of 1500 mm, 2500 mm, and 13.9 Mt, respectively [2, 35]. The river originates near Gunong Korbu in the Malay Peninsula “Main Range” at an altitude of 2100 m, with a gradient of <100 m over the last 100 km of the range [37]. Approximately 95% of the Kelantan River basin area comprises steep mountains, whereas the remaining 5% comprises lowland plains [38, 39]. The main type of soil in the upper reaches of the Kelantan River is ferralitic soil, which is mostly formed from granitic parent rocks and lithosols. In the middle reaches, ferralitic soil formed from sedimentary rocks is the main soil. The soil in the lower reaches is mainly ferralitic soil with riverine alluvium [36].

In October 2016, we harvested 13 samples of river sediment from the middle/lower reaches of the Pahang River (P22, P29, P31, P33, P34, and P36) at depths of <1 m and from the lower reaches of the Kelantan River (K02, K03, K04, K05, K06, K07, and K09) (Figure 1). The sampling sites were free from pollutants and riverbank sediments. The samples were collected in polyethylene bags and immediately stored at 4°C; subsequently, they were pretreated and analyzed at the Key Laboratory of Marine Geology and Metallogeny of the Ministry of Natural Resources (MNR) of China.

The river sediment samples were grouped into seven fractions based on grain size (<63, 32-63, 16-32, 8-16, 4-8, 2-4, and <2 μm) differing at intervals of 1Φ. Firstly, particles larger than the silt size fraction (>63 μm) were separated out by wet sieving. Thereafter, particles smaller than 63 μm were further separated out by sedimentation, based on Stokes’ law [40, 41]. Settling times of 245, 123, 31, 8, and 2 min were observed at liquid levels of 5, 10, 10, 10, and 10 cm, respectively, to separate fractions corresponding to <2, 2-4, 4-8, 8-16, 16-32, and 32-63 μm, respectively. We repeated each step several times to fully extract each grain-size fraction from the riverbed samples.

Each sediment fraction was ground into powder using an agate mortar and then dried in an oven at 60°C. Thereafter, 50 mg of each fraction was dissolved in HNO₃-HF (1 : 1) in Teflon vessels and then dried at 190°C for 48 h. Subsequently, the fractions were individually allowed to react with 3 mL of 50% HNO₃ in an oven at 150°C for at least 8 h before analysis [42]. The concentrations of major and trace elements were determined through inductively coupled plasma optical emission spectrometry (ICP-OES; Al, Ca, Fe, K, Mg, Mn, Na, P, Ti, Ba, Sr, V, Zn, and Zr) and ICP mass spectrometry (ICP-MS; Sc, Cr, Co, Ni, Cu, Cr, Pb, Rb, Cs, and Th). The elemental concentrations of the GSD-9 reference standard were measured to confirm the accuracy of the analyses. The relative errors in the obtained concentrations of major elements and most trace elements were less than 2% and 5%, respectively.

For the riverbed samples, the mineral compositions of the fractions with different grain sizes were determined through X-ray diffraction (XRD) using a Rigaku D/max-2500 diffractometer. The samples were firstly dried at 60°C and then ground into powder. Thereafter, the sediment samples were placed in a holder for pretreatment. XRD was conducted under CuKα radiation at a current and voltage of 100 mA and 40 kV, respectively. In the diffraction patterns, 2θ ranged from 3° to 75°, varying at a rate of 2.0° min⁻¹ and with a step of 0.02°. The Jade 6.5 software was used for semiquantitative analysis of the mineral compositions.

The clay mineralogical composition of the clay fraction (with particle $size<2$ μm) was determined through XRD by varying the scanning angle between 3° and 30°. The analysis of the samples was performed under three conditions, namely, air-dried, ethylene glycol saturated, and after heating for 2 hours at 490°C, and the fractions were scanned at a rate of 2.0° min⁻¹ and in steps of 0.02°. Using the Jade 6.5 software, an ethylene glycol curve was obtained in order to semiquantitatively analyze the clay mineralogical composition; the curve comprised basal reflection peaks corresponding to the major clay minerals: smectite (15–17 Å), illite (10 Å), and kaolinite/chlorite (7 Å). Thereafter, the relative kaolinite and chlorite percentages were calculated based on the 3.57/3.54 Å peak height ratio. The analytical error was estimated at 2% by the replicate analysis of samples. Additionally, the illite chemistry index of the fraction was determined as the ratio of peak areas corresponding to illite at 5 Å and 10 Å illite in the XRD pattern. Furthermore, illite crystallinity was determined by measuring the midheight width of the peak at 10 Å.

In addition, the mineral compositions of the 32-63 μm fractions of samples P33 and K09 were determined through smear slide observations using a polarizing microscope [24].

The XRD patterns of the different grain-size fractions from the Pahang and the Kelantan Rivers sediments are shown in Figure 2. For the Kelantan River samples, all of the size fractions display two distinct peaks occurring at 26.7° (2θ) and 21° (2θ), which are assigned to quartz and mica, respectively. A sharp peak at 27.5° (2θ), corresponding to K-feldspar, is observed in the coarser silt fraction (32-63 μm) (Figure 2). The relative intensity of the K-feldspar peak becomes higher with increasing grain size. A small peak at 28° (2θ) indicates the presence of plagioclases. The diffraction peaks at around 8.7° (2θ) and 12.3° (2θ) are attributed to illite-micas and kaolinite, respectively. A significant increase in the relative proportion of kaolinite can be observed with decreasing grain size. In all the size fractions of both river system sediments, our results show the presence of quartz, illite-micas, K-feldspar, plagioclases, and kaolinite in different proportions (Figure 2). The mineralogical resemblance indicates the similarity in the weathered soils in both river basins.

The relative concentrations of minerals in the different fractions are shown in Figure 3. In the sediment samples from both the Kelantan and Pahang Rivers, the kaolinite concentration decreases from approximately 60% to 5% with increasing size fractions (Figure 3(a)). The most significant decrease in concentrations (approximately 30%) is observed between the fractions with a grain size of <2 μm and those with grain sizes of 2-4 μm, indicating that kaolinite concentrations are highest in the fraction with the smallest grain size. For both rivers, quartz concentration is directly proportional to the grain size (Figure 3(b)). Furthermore, illite-micas concentrations increase from approximately 20% to 30% in the grain-size fractions between <2 μm and 2-4 μm, then decrease from approximately 30% to 10% in the grain-size fractions between 4-8 μm and 32-63 μm (Figure 3(c)). K-feldspar concentrations are relatively similar in the fractions with smaller grain sizes (e.g., <2 μm, 2-4 μm, and 4-8 μm), but vary proportionally with grain size in the fractions with grain sizes larger than 8 μm (Figure 3(d)). Plagioclase concentrations in Pahang River samples are similar for different grain sizes, whereas in the Kelantan River, the concentrations are similar in fractions with smaller grain sizes (e.g., <2 μm, 2-4 μm, and 4-8 μm) and vary in proportion to the grain size in fractions with large grain sizes (e.g., >8-16 μm) (Figure 3(e)).

All of the minerals detected using XRD are mostly found in granitic rocks, as well as in their weathered forms [43]. Granitic rocks are extensively distributed throughout the Malay Peninsula [33], and the mineralogical compositions of the samples in this study are similar to those of the weathered profiles of granites in the region [44]. In the Malay Peninsula, granite and granodiorite, with abundant alkaline elements, are susceptible to intensive weathering, which results in the formation of kaolinite [2]. Plagioclase is more easily weathered compared with other primary minerals found in granitic rocks [25, 26]. This explains why the plagioclase concentration in the collected river sediments is relatively low (3%-11%). In addition, the difference in the variations of plagioclase concentration between the Pahang and Kelantan Rivers may be due to differences in the topography of the river basins. The Pahang River drains a larger area of lowland plain than the Kelantan River, which allows plagioclase in the Pahang River sediment to undergo prolonged hydrolysis and sorting during transportation over a larger distance, resulting in its lower concentration. Conversely, quartz is more resistant to chemical weathering and sediment transportation, and so it appears in relatively high concentrations (14%-57%) in all fractions.

The clay mineral assemblage of the Kelantan River is dominated by kaolinite (57%-65%; average 62%), followed by chlorite (20%-30%; average 24%) and illite (12%-15%; average 14%). Smectite is present in trace amounts and thus is negligible. Previous studies have also reported very low smectite contents in Malay Peninsular rivers [2, 3]. The Pahang River has clay mineral assemblages similar to those of the Kelantan River (Table 1). The illite chemistry index of the samples of both rivers, which exceeds 0.40, indicates intensive chemical weathering; Kelantan River samples have values of >1.0, which are higher than those of Pahang River samples (0.52-0.83). The illite crystallinity of Kelantan River samples ranges from 0.34 to 0.42 (mean value of 0.38), whereas that of Pahang River samples displays a slightly lower range of 0.24-0.31 (mean value of 0.27) (Table 1).

Distributions of major and trace element concentrations in different grain-size fractions from the Kelantan and Pahang River sediments are shown in Table 2. The concentrations of Al₂O₃, Fe₂O₃, and K₂O are high, whereas those of other major elements are less than 1%. Typically, major element concentrations are similar in fractions with the same grain size between the Kelantan and Pahang Rivers. For fractions with a large grain size, the Al₂O₃ concentrations are low, whereas Na₂O concentrations vary inversely with grain size (Figure 4). Our XRD and element concentration data suggest Al₂O₃ concentrations are likely related to secondary minerals with higher Al₂O₃ in the finer grain-size fractions, which are characterized by high clay mineral percentages (Al-rich minerals). The concentrations of mobile alkaline elements such as Na (the most leachable element) are found to be low. According to previous studies on Malay Peninsular Rivers, clay particles have higher Al₂O₃ and Fe₂O₃ concentrations than coarse sediments, whereas Na₂O content is lower [2, 3]. Fe accumulates easily on weathered substances and results in the formation of red soils, which are prevalent in tropical regions due to intensive chemical weathering and constitute the main source of Fe in fluvial sediments [2].

The Kelantan and Pahang River samples exhibit similar trace element compositions. The concentrations of Ba, V, Zr, and Rb are relatively high, whereas the concentrations of Sc, Co, and Cd are low (Table 2). The concentrations of V, Sc, and Cs vary inversely with grain size, whereas that of Zr is directly proportional to the grain size (Figure 4).

The elemental concentrations normalized to upper continental crust (UCC) are reported in Figure 5 [46]. With the exception of TiO₂, most major elements (Al₂O₃, CaO, K₂O, MgO, MnO, and Na₂O) in the different grain-size fractions are depleted for both rivers. CaO, MgO, and Na₂O display the most intensive depletion, and their concentrations have no correlation with grain size. Al₂O₃ and Fe₂O₃ have a trend that approaches UCC (elements/$UCC=1$⁠) with decreasing grain size.

The concentrations of trace elements such as Pb, Cs, and Th are relatively high compared with UCC, whereas Sr and Cd are significantly depleted (Figure 5). The concentrations of V and Cs tend to be high for small grain sizes; the Pb concentration of the Pahang River samples varies similarly with V and Cs. The concentration of Zr does not vary with grain $size<8$ μm, but it varies significantly in proportion to grain size in the case of the 8-16 μm, 16-32 μm, and 32-63 μm grain sizes (Figure 5). This indicates that Zr enrichment mainly occurs in the coarse-grained fractions. Furthermore, the degree of Th enrichment is low for grain sizes of <8 μm but high for grain sizes in the 16-32 μm and 32-63 μm ranges. This result indicates that Th enrichment also mainly occurs in the coarse-grained fractions. Variations in Sr, Sc, Cd, Ba, and Rb with respect to grain size are relatively moderate. In the case of <2 μm fractions, we observe that, relative to UCC, they are enriched in Cr, Co, Ni, and Cu. The Zn content of Pahang River sediments shows an obvious grain-size effect, with decreasing enrichment and with increasing grain size, while the Zn content of Kelantan River sediments fluctuates around UCC.

River sediments are mixtures of drainage basin weathering products [21, 24]. Due to differences in weathering intensity and duration, surface samples exhibit different mineralogical and chemical components along with different grain size distributions [23]. Such changes in element contents depending on grain size may be attributed to the changes that occur in mineralogical compositions during the processes of sorting and weathering [21, 23]. Therefore, we have focused on the effects of mineralogical changes on major and trace element contents. Smear slides of typical minerals are shown in Figure 6.

Illite-micas show positive correlations with the Al₂O₃ (⁠$r=0.63$⁠) and Cs (⁠$r=0.72$⁠) contents and negative correlations with Na₂O (⁠$r=−0.81$⁠) and CaO (⁠$r=−0.65$⁠) (Table 3). Al₂O₃ and Cs are typically concentrated in clay fractions (Figure 4), and illite-micas are likewise mainly concentrated in fine size fractions (Figure 3). Na₂O and CaO are mobile elements during weathering. Thus, their concentrations are very low in the Malay Peninsula riverbed sediments because of intense chemical weathering, particularly in clay or fine fractions [2]. Micas may contain Na₂O, which is typically concentrated in plagioclases and rarely exists in fine size fractions. Plagioclases may occur in silt fractions due to the ablation of sand-size grains Tanaka and Watanabe [28]; as a result, Na₂O and CaO contents may also be enriched in coarser fractions. Concentrations of Al₂O₃, Fe₂O₃, P₂O₅, V, Zn, Cr, Co, Cu, Pb, and Cs are well correlated with the relative abundance of kaolinite (Table 3). Kaolinite is mainly concentrated in clay-sized fractions and can easily adsorb many elements due to its large specific surface area Tanaka and Watanabe [28]. Al₂O₃ and Fe₂O₃ are both residual elements from weathering and are widely distributed in clay minerals of the Malay Peninsula rivers owing to intense chemical weathering [2, 3]. To a large extent, quartz shows a negative correlation with most elements (Table 3), which reflects its dilution effect on element contents. K-feldspar displays an obvious grain-size effect, with concentrations increasing with increased grain size (Figure 4).

During weathering, plagioclases are more easily hydrolyzed than K-feldspars. Consequently, K-feldspars display a higher occurrence than plagioclases in weathering products. K-feldspars are mainly found in coarse-grained components and show obvious positive correlations with Zr, Cd, and Th, suggesting concentration in coarse-grained fractions. Zr and Th enrichment in coarser fractions has been related to greater amounts of heavy minerals such as zircon and monazite [47]. Zr mainly exists in the form of zircons which are remarkably resistant to chemical weathering [48]. Zircons are commonly found throughout the Malay Peninsula [49], and smear slide observations have revealed the existence of coarser grain-sized zircons (Figure 6). Monazite is an important high-grade Th resource in the soils of the Malay Peninsula [32]. Plagioclases show a positive correlation with CaO, Na₂O, Ba, and Sr (Table 3). Na in sediments is mainly concentrated in feldspars, especially plagioclases [50]. High CaO and Sr contents are typical of carbonate and silicate minerals such as calcite, dolomite, and plagioclase. However, XRD analyses of sediments from both rivers show an absence of carbonates. Carbonates such as calcite and dolomite are not common in the Malay Peninsula due to strong dissolution conditions. There is a large volume of granitic rocks (mostly hornblende-bearing biotite granitoids), rich in plagioclases (usually 10%-30%), in the Kelantan and Pahang drainage basins [33, 45] [51]. Consequently, we can hypothesize that the Ca and Sr present in the studied riverbed samples are mainly associated with plagioclases.

CIA and α^AlNa values plotted versus grain-sized fractions indicate that, for both rivers, chemical weathering intensity shows a decreasing trend with the increase in the riverbed sediment particle size (Figure 7). This is because coarser fractions have higher contents of primary minerals (e.g., plagioclase, K-feldspar, quartz, and heavy minerals) owing to the physical erosion of granites and sedimentary rocks of river catchments and less pronounced chemical weathering. The CIA value for <63 μm fractions is between that of the <2 μm and 32-63 μm fractions, while the value for bulk sediment is less than that for the <63 μm fractions (Figure 7).

In the Malay Peninsula, the CIA values of nonweathered granites, unaltered plagioclases and K-feldspars, smectite and illite, and kaolinite are 48-51 [44], 50 [52], 70-85 [43, 52], and approximately 100 [52], respectively. CIA values increase with the formation of clay minerals due to the leaching of Na, K, and Ca. Thus, the increasing CIA values are consistent with the increase in clay mineral proportions with decreasing grain sizes (Figure 3).

Weathering trends for sediments with diverse grain-size fractions in the Kelantan and Pahang Rivers are illustrated in the ternary diagram of Al₂O₃–(CaO + Na₂O)–K₂O (A–CN–K) [54, 55] (Figure 8). Compared with coarse fractions, fine fractions are located near the Al₂O₃ apex, which indicates that CaO, Na₂O, and K₂O are preferentially leached, whereas Al₂O₃ enrichment occurs when clay minerals are formed. The A-CN-K diagrams of both rivers show an increasing trend of chemical weathering as grain size decreases. However, the results also reveal different trends in the A-CN-K diagrams obtained for the Kelantan and Pahang River sediments. For sediments from the drainage basin of the Kelantan River, the main trend of silicate weathering exhibits the preferential leaching of CaO and Na₂O and enrichment of Al₂O₃ from the 32-63 μm, 16-32 μm, to 8-16 μm fractions, whereas the K₂O content is constant. Subsequently, the main silicate weathering trend is parallel to the A–K line from the 4-8 μm, 2-4 μm, to <2 μm fractions, suggesting the complete leaching of CaO and Na₂O, followed by the active leaching of K₂O (Figure 9(a)). For the Pahang River, the main trend of silicate weathering in the drainage basin is parallel to the A–K line in all grain-sized fractions, indicating that CaO and Na₂O are completely leached and that there was active leaching of K₂O (Figure 8(b)). As revealed by Nesbitt et al. [25, 30], there was weak K leaching in parent rocks in early moderate weathering periods. Yang et al. [50] proposed that in the Yangtze River Basin, plagioclases were selectively weathered first, while K-feldspars remained almost intact. The plagioclase content of Kelantan River sediments is relatively high in coarse-grained components, such as the 16-32 μm and 32-63 μm fractions, owing to the greater physical erosion of granite in the highlands of the river drainage basin (Figure 3); as a result, CaO and Na₂O are firstly leached along the A-CN line. Then, the plagioclase content in the fine-grained components is almost exhausted, and K₂O is leached along the A-K line. By contrast, the extensive plain of the Pahang River basin is characterized by very low plagioclase concentrations in the different grain-sized fractions (Figure 3), suggesting that plagioclase is almost exhausted and that K₂O leaching from K-feldspars has been initiated.

Samples collected from the Mekong River, Red River, and Pearl River mostly leach Na and Ca parallel to the A-CN line [2, 3], indicating that the sediments from the Malay Peninsula have greater leaching capacity and undergo relatively more pronounced chemical weathering. Wang et al. [3] reported that Na and Ca in river samples from the Malay Peninsula are almost completely leached because of a high degree of chemical weathering. Moreover, the chemical weathering trend is in the process of gradually leaching K and other elements. Such differences may be attributed to the fact that previous studies have focused on the clay fractions and sediments within the <63 μm fraction. Consequently, changes in mobile elements within different grain-size fractions due to chemical weathering processes could not be fully determined.

The lower reaches of the Pahang and Kelantan Rivers are mainly comprised of ferralitic soils. The predominance of granitic rocks in these river basins and the humid and warm East Asian–Australian monsoon climate, along with the tectonic stability in the Malay Peninsula, combine to induce the formation of ferralitic soil that is characterized by high proportions of kaolinite [56]. Thus, a high kaolinite content in the clay-sized fraction indicates intensive chemical weathering in the Malay Peninsula (Table 1), which is consistent with results from previous studies on dominant kaolinite contents of river sediments from the Malay Peninsula [2, 3]. Kaolinite and CIA show a positive correlation (⁠$r2=0.42$⁠; $n=12$⁠) (Figure 9(a)), suggesting that both can be used to estimate the chemical weathering state of river sediments.

Strong hydrolysis conditions induce an opening of the illite structure and leaching of Fe/Mg from the illite crystal lattice. Such conditions mainly produce Al-rich illite (muscovite) characterized by a high illite chemical index with values typically >0.40 [56]. By contrast, low chemical weathering conditions induce the formation of Fe-Mg-rich illite (biotite) in soil, characterized by a low illite chemistry index with values typically <0.15. The illite chemistry index has been adopted for determining illite chemical weathering intensity in rivers and deep-sea sediments in several previous studies [2, 6, 12, 57]. Results obtained for sediments from the Kelantan and Pahang Rivers indicate a strong correlation between illite chemistry index and CIA (⁠$r2=0.70$⁠; $n=12$⁠) (Figure 9(b)), which is consistent with results obtained in previous studies [3]. The illite chemistry index for both rivers is in the range 0.52 to 1.30, suggesting the presence of Al-rich illite due to intense hydrolysis conditions.

Typically, low values of illite crystallinity indicate a high degree of crystallinity characterized by weak hydrolysis in soils [2, 56]. A positive correlation is observed between CIA and illite crystallinity values (⁠$r2=0.71$⁠; $n=12$⁠) (Figure 9(c)). The fact that CIA increase is associated with an increase in the illite crystallinity index (i.e., poor crystallinity) indicates that the strengthening of hydrolysis induces degradation of the illite crystal lattice. The mean illite crystallinity is 0.32°Δ2θ (0.24°-0.42°Δ2θ), which is close to the 0.30°Δ2θ value obtained from the Pearl River [58], indicating relatively strong hydrolysis.

In brief, kaolinite proportions, illite chemistry index, and illite crystallinity all exhibit significant positive correlations with CIA, suggesting that these mineralogical parameters can be used to estimate the degree of chemical weathering of riverbed sediments in tropical environments.

Chemical weathering intensity (CIA) is found to be strongly negatively correlated with an increase in grain-size fraction (Figure 7). Thus, it is possible to adopt size-fractioned particle sequencing for constraining alkali elemental fractionation (K, Rb, and Cs) that occurs during sediment transfer and chemical weathering under tropical climate conditions. Al constitutes a major component of aluminosilicates of granitic rocks which is resistant to chemical weathering [52]. Consequently, it is possible to utilize the alkali element/Al ratio for evaluating the amounts of alkali elements released in primary minerals Tanaka and Watanabe [28]. A significant negative correlation is observed between CIA and K/Al (⁠$r2=0.94$ and 0.94, respectively) (Figure 10(a)), indicating that K in sediment samples is continuously leached during chemical weathering, with the greatest depletion occurring in the finest particles. Similar to the K/Al ratio, the Rb/Al ratio also shows a significant negative correlation with CIA (⁠$r2=0.91$ and 0.86, respectively) (Figure 10(b)), which indicates the release of K and Rb in primary minerals, mainly K-feldspars and micas from granitic rocks, during chemical weathering (Figure 3) [28, 50]. By contrast, CIA is significantly positively related to the Cs/Al ratio (⁠$r2=0.75$ and 0.90, respectively) (Figure 10(c)), which increases with an increase in CIA. This result indicates the gradual accumulation of Cs in fine particles during chemical weathering (Figure 10(c)). The increase in the Cs concentration in fine particles can be attributed to the reabsorption of Cs that is generated in primary minerals and subsequently adsorbed onto the interlayer sites and frayed edge sites of clay minerals during chemical weathering [28, 59]. Some studies have revealed that Cs exhibits high adherence on interlayer sites in clay minerals by forming complexes [28, 60]. The finest fractions of the Kelantan and Pahang River sediments, which are relatively rich in kaolinite due to strong chemical weathering and which also have high Cs concentrations, may be attributed to their binding with kaolinite. Granite, the most important parent rock in both river drainage basins, can be categorized as eastern granite and main range granite. Both are plotted near the fit line of K/Al with CIA for the Kelantan River (Figure 10(a)), and eastern granite is plotted between the fit line of Rb/Al with CIA for the Pahang and Kelantan Rivers (Figure 10(b)). However, they are not close to the fit line of Cs/Al with CIA for riverbed sediments (Figure 10(c)), a fact that may be attributed to the complicated geochemical behavior of Cs during weathering and transfer.

Rb/K and CIA show a significant positive correlation (⁠$r2=0.91$ and 0.89, respectively) (Figure 11(a)). The ratios of Rb/K increase with a decrease in particle size and an increase in chemical weathering (CIA), indicating a lower release of Rb than K. Moreover, Cs/K is significantly positively related to CIA (⁠$r2=0.84$ and 0.87, respectively) (Figure 11(b)), indicating that the release of Cs in sediments is lower than that of K with increasing chemical weathering. Therefore, the mobility of alkali elements declines during chemical weathering, and this occurs in the order K > Rb > Cs. In addition, their fractionation levels may be adopted for evaluating the chemical weathering intensity in felsic rocks Tanaka and Watanabe [28]. Granites from drainage basins are not close to the fit lines of Rb/K and Cs/K with CIA (Figures 11(a) and 11(b)), which indicates possible mineral sorting and geochemical fractionation during weathering and sediment transfer.

Studies have shown that >90% of K is found in dissolved forms in river waters, whereas Cs is present mostly in particulates Tanaka and Watanabe [28]. The Rb content in fine-particle fractions lies between K and Cs contents. With an increase in the chemical weathering intensity, minerals interact with meteoric water, resulting in the release of K and Rb ions into the water. Fine suspended sediment particles are the main carriers of Cs; therefore, Cs is transported with fine particles in the river system [28, 61]. Alkali elemental fractionation occurs on the surfaces of minerals during chemical weathering; therefore, different grain-size fractions may contain different chemical components (Figures 4 and 5). In a river system, fine particles are transported as suspensions and have different transportation patterns compared with coarser particles, which indicates that the chemical compositions of bulk river sediments are different from those in source materials because of size fractionation of sediment particles during transportation [20, 21]. According to a study, the sediment chemical components from the Ganges and Brahmaputra River systems are mostly influenced by the sediment sorting effects that occur during transport and deposition processes [22, 27]. In order to rule out the influence of grain size caused by the sorting effect and to discuss only changes in state due to chemical weathering, Lupker et al. [27] suggested normalizing of mobile to immobile element ratios to a common Al/Si ratio so as to correct the sorting effect on elemental ratios during transportation [7]. As confirmed by the new results obtained for Pahang and Kelantan River sediments, the impact of particle size fractionation on sediment geochemistry and on the chemical weathering process should be carefully considered when the chemical compositions of river sediments are used to identify sediment source materials and to evaluate chemical weathering.

Major and trace elements and mineralogical compositions have been analyzed on different grain-size fractions of riverbed samples collected from the Kelantan and Pahang Rivers in the Malay Peninsula in order to assess the effect of sediment transportation on mineralogical and geochemical proxies of chemical weathering.

• (1)

  Fine particles, such as clay minerals, contain more elements than coarse fractions. However, heavy minerals considerably influence some element concentrations (e.g., Zr and Th), leading to extremely high concentrations in coarser fractions. Geochemical element concentrations and distributions in diverse grain-size fractions in riverbed sediments are associated with changes in different minerals due to mineral sorting and weathering

• (2)

  The degree of chemical weathering increases with a decrease in the grain size of sediments in both rivers, a pattern that is similar to the results obtained in rivers in the temperate zone. However, the chemical weathering trend in these two rivers varies with a decrease in particle size due to different plagioclase and K-feldspar contents, especially in the coarser size fractions (16-32 μm and 32-63 μm fractions). Kaolinite proportions, illite chemical index, and illite crystallinity all show significant positive correlations with CIA, suggesting that these clay mineral indices can be reliable proxies of chemical weathering intensity

• (3)

  During the process of chemical weathering, mobility decreases in the order of K, Rb, and Cs. K and Rb are preferentially released into the water, whereas most of Cs is retained in fine-grain weathering products as it is reabsorbed on clay minerals, resulting in large-scale fractionation of alkaline elements at different particle sizes

• (4)

  The sorting effect of different particle sizes during transport considerably influences element fractionation. Therefore, the influence of particle size fractionation on sediment geochemistry should be carefully considered when using the chemical compositions of river and deep-sea sediments to evaluate the extent of chemical weathering

The authors confirm that the data supporting the findings of this study are available within the article.

The authors declare that there is no conflict of interest regarding the publication of this paper.

We acknowledge the staff of the National University of Malaysia and First Institute of Oceanography (FIO), Ministry of Natural Resources (MNR) of China for sample collection. We are grateful to Ying Zhang, Xiaojing Wang, and Jingjing Gao from FIO, MNR, for their assistance with the elemental analyses. This work was supported by the National Program on Global Change and Air-Sea Interaction (GASI-GEOGE-03, GASI-02-SCS-CJB01), the Natural Science Foundation of China-Shandong Joint Fund for Marine Science Research Centers (U1606401), the Scientific and Technological Innovation Project financially supported by Qingdao National Laboratory for Marine Science and Technology (2015ASKJ03), the China–Malaysia Cooperation Project “Effect on variability of seasonal monsoon on sedimentary process in Peninsular Malaysia waters”, and the Taishan Scholar Program of Shandong (tspd20181216).