Geochemical and petrographic characteristics of sediments along the transboundary (Kenya–Tanzania) Umba River as indicators of provenance and weathering

1.1 Climate

The tropical climate of this region is influenced by the monsoon winds attributed to the proximity of the warm Indian Ocean and the north-south inter-annual migration of the Inter-Tropical Convergence Zone (ITCZ) [14,15]. This ITCZ migration and the resulting changes in wind directions coin two major seasons, namely the southeast monsoon (SEM) and the northeast monsoon (NEM). The NEM brings with it two rainy seasons, the “long rains” between March and June and the “short rains” between October and mid-December. Conversely, the SEM causes two dry seasons sandwiched between the two rainy (wet) seasons [14,16]. The mean annual precipitation in this region reaches 4,000 mm, and the average annual temperature is at 25°C. The “long rains” account for the majority of the annual rainfall, and consequently water discharge and annual sediment transport of the Umba River [17].

2.1 Study site

The Umba River originates in the tectonically uplifted Usambara Mountains in Mkinga, Tanzania, with an elevation of circa 2,000 m above mean sea level, and drains to the Indian Ocean in Vanga, Kenya (Figure 2). The mangroves of Vanga at the south coast of Kenya (Figure 2) receive the terrigenous sediments from the relatively short circa 200 km long Umba River. It drains a catchment of circa 8,000 km² [17]. Part of the study area lies within a proposed Transboundary Conservation Area (TBCA) between Kenya and Tanzania (Figure 2), which extends south from Diani within the Kwale County in Kenya (39°00′ E, 4°25′ S) to Tanga within the Mkinga District in Tanzania (39°40′ E, 5°10′ S) covering a north-south distance of circa 120 km and a narrow strip along the coastline in Kenya and Tanzania, with an estimated total area of circa 2,500 km². The TBCA intends to integrate and align the coastal conservation and management in an ecologically connected region transcending the Kenya–Tanzania border. The proposed TBCA will have significant ecological significance mainly because it is characterized by contiguous and connected marine and terrestrial habitats as well as similar and shared socio-economic status [15]. Sampling for the present study was done at seven stations along the Umba River from source to mouth. Stations S1, S2, and S4 are located on the Tanzanian side, while Stations S5, S6, S7, and S8 are on the Kenyan side (Figure 2).

Figure 2

The study site within the Umba River catchment and sampling locations along the Umba River. Stations S1, S2, S3, and S4 in Tanzania and S5, S6, S7, and S8 in Kenya.

3.1 Grain size analysis

Grain size analysis of the bulk river sediment samples was conducted in a laboratory at the Kenya Marine and Fisheries Research Institute, Kenya, using a Malvern© Mastersizer 2000 laser diffraction particle size analyzer. Sediment pre-analysis treatment including removal of organic matter using 30% of hydrogen peroxide and clastic grains >2 mm [26] was undertaken before measurement of grain sizes according to Hossain et al. [26]. Mean grain size (μm) was estimated according to Folk and Ward [27].

3.2 Petrographic analysis

Thin sections were prepared at the geology laboratory of Leibniz Centre for Tropical Marine Research (ZMT), Germany, according to protocols in refs [28,29]. Small blocks (4 cm × 4 cm) of dried sediment samples were bound together with Araldite 20/20 resin. The blocks were hardened and after 24 h, one side of the block was ground to obtain a plain surface without scratches. The plain surface was cleaned in an ultrasonic bath and dried. The dried block was fixed with resin on a glass slide and bound together using a press system. Once it was bound together and hard-dried again for another 24 h, the block was cut into 1 mm slices and polished parallel to the glass slide to obtain one single thin section. The 1 mm sample was then ground on the slide down to 100 µm and further ground to 28 µm suitable for mineralogical analyses. The thin sections were examined under an optical polarized light in LEICA DM500-EP microscope.

3.3 Major and trace elements

In the laboratory, wet sediment samples were dried at 60°C for 48–72 h, pulverized and homogenized using a Fritsch^© Planetary Micro ball-mill Pulverisette 7. Seven hundred milligram of each sample was mixed with 4,200 mg of di-Lithium tetraborate (Spectro-melt A10, B₄Li₂O₇) and pre-oxidized with 1,000 mg of ammonium nitrate at 500°C. The sample mixtures were then digested (melted) at 1,340–1,350°C in a CIMREX^TM 10 microwave accelerated reaction system, using a combination of HF + HNO₃ + H₃BO₃ for major and trace element analyses [30,31,32] and then fused to homogeneous glass pellets (beads). The pellets were then analyzed in PANalytical^© AXIO Max (3 kW) by wavelength-dispersive X-ray fluorescence. Major elements (as oxides) were reported in wt%, while the trace elements were given in ppm [33]. The accuracy of the analytical procedures was verified by analyzing three certified standard reference materials namely PACS, MESS, and PS-S. The major elements measured included Si, Ti, Al, Fe, Mn, Mg, Ca, K, Na, and P. The oxides of major elements were also used to constrain and assign the sediment provenance using first and second discriminant functions (DF1 and DF2) analysis [9,34,35]. Large-ion lithophile elements including Rb, Ba, Sr, Th, U, and Ce were measured. The high field strength elements including Zr, Y, and Nb were also measured, as were the transition trace elements, i.e., Cr, Co, Cu, Ni, Zn, Ga, Mo, and Pb.

3.4 Degree of weathering and chemical alteration

Several weathering indices are available to evaluate mineralogical maturity, chemical alteration, and element mobility [34,36,37,38]. The underlying principle of weathering indices is the comparison of mobile elements with immobile elements such as Al as the ref. [39]. Such immobile elements are less susceptible to alteration and weathering. They, therefore, retain their chemical characteristics over longer periods and longer distances of transportation by rivers. In this study, the chemical index of alteration (CIA) was used to evaluate the degree of feldspar alteration, calculated using the equation [40]:

The CIA is represented in molecular proportions, where CaO* represents the Ca in silicate fraction only, adjusted for other Ca-bearing minerals such as carbonates [41]. The abundances of the major elements were used to evaluate the chemical composition and provenance of Umba River sediments following the rationale described in refs [6,34,39]. To further evaluate the effect of transport and weathering, ternary plots were generated from the major element compositions of Umba River sediments [40]. Specifically, ternary plots were created showing the concentration on Al₂O₃ – (CaO + Na₂O) – K₂O, hereafter referred to as A-CN-K, and Al₂O₃ – (CaO + Na₂O + K₂O) – (Fe₂O₃ + MgO), hereafter referred to as A-CNK-FM. The position of sediments within the ternary diagrams indicates the dominant minerals, degree of weathering, mineral sorting, and thereby, the source geology. Specifically, A-CN-K and A-CNK-FM plots assist in the understanding of feldspar weathering and the effects of sorting on weathering of Fe-Mg bearing minerals, respectively [6,34].

3.5 Provenance

The compositional data of the Umba River sediments were also used to constrain their sedimentary provenance. Discriminant function plots express oxides of various elements as ratios with the immobile Al₂O₃ [34]. DF analysis based on oxide ratios distinguishes the provenance of sediments between four fields, namely mafic igneous, felsic igneous, intermediate igneous, and quartzose (recycled) sedimentary provenance [26,34]. DF analyses based on raw oxides are represented by the following equations;

3.6 Statistical analyses

Statistical analyses were performed using inbuilt R-program (v. 1.1.463) libraries and packages (R Core Team, 2020). Significant difference tests of major and trace element concentrations between sediment type (riverbank or bottom sediments) and environment (riverine, estuarine, and marine) were assessed by performing an analysis of variance (ANOVA). Data conformity to normal distribution was examined using the Shapiro-Wilk test of normality.

4.1 Grain size distribution

The Umba River sediments are composed of sand, silt, and clay in different proportions (Table 1; Figure 3). The mean grain size (μm) of the riverbank and bottom sediments is 172 and 266 μm, respectively (Table 1). The mean dominant fractions are sand and silt, ranging between 34 and 78% and 0.72 and 62% of total volume percentages, respectively. Clay constitutes less than 3% in both the riverbank and bottom sediments with an overall range between 0.07 and 2.90%. The grain size gradient is typical for rivers and transitions from predominantly coarse sandy sediments in the upper catchment (S1, S2, S4, S5, and S6) to sandy-silt-sized with clay in the estuary (S7) and silty-sand-sized sediment at the marine site (S8) (Figure 3 and Table 1). Clay-sized particles were most abundant in the estuarine stations (S7) compared to all the other stations.

Figure 3

Ternary diagram showing textural distribution of sediments during the transition period (July 2019). With S1–S6 being upstream riverine stations and S7 and S8 being downstream estuarine and marine stations, respectively.

4.2 Mineralogy of Umba River sediments

The mineralogical composition identified by petrographic analyses is similar for the upstream riverine sediments and the downstream estuarine and marine sediments (Figures 4 and 5). All sediments along the entire course of the Umba River (Figure 2) contain mainly K-feldspar, lamellar and striated plagioclase, quartz, muscovite, biotite, and hornblende. Accessory minerals include opaque and clay minerals that were not further identified here. However, from comparison of site-relevant literature [18,20], we infer that the opaque minerals include magnetite, ilmenite, and maghemite. Additionally, clay minerals such as kaolinite and illite are also reported for the Umba River catchment.

Figure 4

Photomicrographs of individual lithoclasts in upstream sediments (left to right: stations S1 and S4 (top row) and stations S5 and S6 (bottom row)). K = K-feldspar, BPlg = Banded plagioclase, G = Garnet, Q = Quartz, Hbl = Hornblende, B = Biotite, and Ol = Olivine.

Figure 5

Photomicrographs of individual mineral lithoclasts in downstream sediments embedded in epoxy resin. Top row S7 (estuary) and bottom row S8 (marine setting). K = K-Feldspar, Plg = Banded plagioclase, G = Garnet, Q = Quartz, Hbl = Hornblende, and B = Biotite.

The quartz crystals occur as elongated and deformed lenticular grains. They are also present in lithoclasts, where they exhibit anhedral shapes adjacent and surrounding feldspar, biotite, and other minerals. In the lithoclasts, the feldspars (K-feldspars and plagioclase) exhibit tabular and irregular crystals [42]. The sediments collected in the downstream stations (S7 and S8) show evidence of river transportation or high energy environments, characteristic of intertidal areas due to their relatively rounded shapes (Figure 5). From source to mouth, a decrease in lithoclast sizes is observed that goes along with an increase in the monocrystalline grains (as opposed to lithoclasts made up of multiple crystals) (Figures 4 and 5). Downstream sediments (Figure 5) exhibit poor grain sorting and varying degrees of roundness of the individual clasts.

4.3 Major elements

The sediments exhibit a relatively narrow SiO₂ range in both the riverbank sediments (53.30–70.43 wt%, with an average of 59.89 wt%) and bottom sediments (55.57–76.22 wt%, with an average of 65.76 wt%) (Table 1). Highest SiO₂ contents coincide with coarse-grained sediments, for example, in station S4 for riverbank and S8 for bottom sediments. Low SiO₂ content (55.67 wt%) is observed in the estuarine silty sands collected in station S7 (Table 1). The Al₂O₃ content varies between 10.45 and 19.06 wt% (average of 15.97 wt%) in riverbank sediments, and between 9.90 and 17.12 wt% (average of 13.64 wt%) for bottom sediments (Table 1 and Supplementary Information 1). Fe₂O₃ and TiO₂ contents were >5 wt% and <2 wt%, respectively, in both the riverbank and bottom sediments. The immobile Al₂O₃ shows a clear negative relation with SiO₂ in both the riverbank and bottom sediments (Figure 6a). There is no particular trend on Al₂O₃ versus TiO₂ in riverbank sediments but a positive relation in the bottom sediments (Figure 6b). Al₂O₃ also shows a positive relation with Fe₂O₃ and P₂O₅ in both the riverbank and bottom sediments (Figure 6c and d).

Figure 6

Major oxides (wt%) plots (a) SiO₂, (b) TiO₂, (c) Fe₂O₃, (d) P₂O₅, (e) CaO, (f) K₂O, (g) MgO and (h) Na₂O for the Umba River riverbank and bottom sediments referenced to the immobile Al₂O₃. Strong relations are observed between the abundances of SiO₂ (negative), Fe₂O₃ (positive), and P₂O₅ (positive) with Al₂O₃ (a, c, and d respectively).

The alkaline oxides, i.e., K₂O, Na₂O, and MgO, are relatively high (>1 wt%) in all stations and in both riverbank and bottom sediments (Table 1; Supplementary Information 1). However, on average K₂O and MgO are slightly higher in riverbank sediments compared to bottom sediments (1.46 and 1.68 wt%, respectively). Na₂O content is highest (2.14 wt%) in bottom sediments (Table 1). Additionally, the total alkali content (K₂O + Na₂O) and the alkali ratios (K₂O/Na₂O) in all the stations are >2.5 and >0.5%, respectively, in both the riverbank and bottom sediments (Table 1). The sediments exhibit a wide range of CaO values in riverbank (1.88–4.04 wt%, average of 3.01 wt%) and in bottom sediments (1.52–4.45 wt%, average of 3.03 wt%) (Table 1). A negative relation between Al₂O₃ and CaO (Figure 6e) and K₂O (Figure 6f) is observed in the riverbank as well as in the bottom sediments. Additionally, CaO is relatively depleted in estuarine sediments (S7) compared to riverine and marine sediments. Both MgO and N₂O show a positive but scattered relation with the immobile Al₂O₃ (Figure 6g and h, respectively). Major element ratio Al₂O₃/SiO₂ vs Fe₂O₃/SiO₂ exhibits a wide variation in all the grain size fractions and covers the entire quartz–clay spectrum (Figure 7). The larger grain size fractions (e.g., S4, S6, and S8) show similar enrichment of Fe and Al, as exhibited by the fine fraction and relatively clay-rich estuarine sediments (S7).

Figure 7

Variation in Al₂O₃/SiO₂ vs Fe₂O₃/SiO₂ ratios in Umba River sediments showing the distribution of Fe and Al elements in estuarine mud/silt sediment collected in station S7 across the entire quartz-clay spectrum.

4.4 Trace elements

Average concentrations of the trace elements Cr, Ni, and Zn are all enriched in riverbank sediments compared to bottom sediments (Table 1, Supplementary Information 1). Cr concentration ranges between 25 and 346 ppm with an average of 107.60 and 34.59 ppm in riverbank and bottom sediments, respectively. Ni concentration ranges between 5 and 73, again showing a higher average of 34.59 ppm in riverbank sediments compared to bottom sediments (24.86 ppm) (Table 1; Supplementary Information 1). Cr and Ni concentrations are depleted in marine bottom sediments collected in station S8 (Table 1). Zr concentration shows values ranging between 169 and 541 ppm (average value of 314 ppm) for river bank sediments compared to bottom sediments that show slightly lower values ranging between 238 and 349 ppm (average of 280 ppm) (Table 1). On the contrary, Sr concentration is enriched in marine (472 ppm) and riverine sediments (majority >300 ppm) (Table 1), while they are depleted in bottom estuarine sediments (232.80 ppm). The Ba content is slightly enriched in both riverbank and bottom sediments (majority >800 ppm), while the lowest Ba content of 635.33 ppm is recorded in estuarine bottom sediments (Table 1). The Rb/Sr ratios are similarly low (<0.5) in all the stations and in both the riverbank and bottom sediments, ranging between 0.06 and 0.26 with average values of 0.13 and 0.11 in the riverbank and bottom sediments, respectively (Table 1). Such low Rb/Sr ratios indicate low to moderate chemical weathering.

A less distinct and scattered relation is observed between Al₂O₃ and Sr (Figure 8a), while Al2O3 vs Zr shows a distinct negative relation (Figure 8b) in both the riverbank and bottom sediments. On the contrary, strong positive relations are observed between Al₂O₃ and the immobile elements Ni, Zn, Cr, and Rb (Figure 8c, d, e, and f, respectively). Relatively high positive relations are also observed between Sr and CaO as well as between the immobile Ba and K₂O (Figure 8g and h, respectively) in both the riverbank and bottom sediments. Generally, a low concentration of Sr and Ba in the bottom sediments collected in the estuarine station S7 is evident (Figure 8a, g, and h).

Figure 8

Shows the selected trace element variations with respect to the immobile Al₂O₃ wt% in Umba River sediments. Sr shows positive but scattered (a) relation compared with Zr which shows negative relation with Al₂O₃ (b). Ni, Zn, Cr, Rb, and Rb (in ppm) show good positive relation with Al₂O₃ in both the riverbank and bottom sediments (c–f, respectively). Sr (g) and Ba (h) maintain a positive relation with alkali CaO and MgO, respectively, with relatively low Sr and Ba (in ppm) in the bottom estuarine sediments.

4.5 Elemental ratios and weathering indices

The major and trace elements were normalized to the upper continental crust (UCC) using values from Rudnick and Gao [43]. The relative averages of the major elements show similarity with UCC values; however, the relative Na₂O, K₂O, and MgO (<0.70) concentrations are generally low compared to UCC and relative to the immobile Al₂O₃ (Figure 9a). Additionally, SiO₂ shows a marked similarity with UCC with no discernible enrichment. The average trace element compositions of the sediments show a variability that is quite comparable to other documented values from the same region [18,20]. The general pattern shows a strong depletion in Rb relative to Ba, and depletion of Th and U relative to Sr (Figure 9b).

Figure 9

Major element oxides (a) and trace element (b) spider plot of Umba sediments normalized against the UCC values from Rudnick and Gao [43]. Major elements are normalized as oxides and trace elements as ppm. Mobile major elements (MgO, CaO, Na₂O, and K₂O) and trace elements (Rb, Th, and U) are depleted relative to UCC.

The average UCC has a CIA value of 50, and weathered residual clay has a CIA value close to 100 [43,44]. The CIA of the Umba River sediments show a wide range, with CIA values ranging between 39 and 81 (Figure 10 and Supplementary Information 1). Within this range, the mean CIA values are higher in riverbank sediments ranging between 53.98 and 73.13 compared to the bottom sediments with values ranging between 52.04 and 68.01 (Table 1). The marine bottom sediments (collected in station S8) have the lowest CIA (Figure 10 and Table 1). The riverbank as well as the bottom river sediments show a positive relation between CIA and Al₂O₃ and Fe₂O₃ ≥50 (Figure 10a and b). They also show elevated Al₂O₃, Fe₂O₃, and TiO₂ at higher CIA values of >70, while in contrast the bottom marine sediments have lower Al₂O₃, Fe₂O₃, and TiO₂ at CIA values of <50 (Figure 10a, b, and c). The bottom marine sediments are K₂O-enriched compared to the bottom estuarine sediments; however, the riverine riverbank sediments exhibited a scattered non-distinct variation at varying CIAs (Figure 10d). There is also a strong relation between Cr concentration and CIA values in the Umba River sediments especially for downstream (estuarine and marine) bottom sediments (Figure 10e).

Figure 10

Plots of selected major elements namely (a) Al₂O₃ , (b) Fe₂O₃, (c) TiO₂, (d), K₂O and trace element, (e) Cr versus the chemical weathering index (CIA) of Umba River sediments. CIA are defined as (Al₂O₃/[Al₂O₃ + CaO + Na₂O + K₂O]) x 100 in molecular proportions [41,45]. Bottom estuarine samples (S7) show high major oxides variation at different values of CIA. Immobile and less soluble elements (Al₂O₃, TiO₂, and Cr) increases with the increase in CIA (a, c, and e, respectively).

On the A-CN-K plot, the sediments are plotted across the plagioclase – K-feldspar field (Figure 11a), while also maintaining a linear trend parallel to the A-CN edge towards the smectite field. The trendline also starts at the pristine CIA value of ∼40 to a highly weathered value of ∼80. The sediments are plotted below the smectite and K-feldspar field and around the K-feldspar – chlorite and K-feldspar – biotite boundaries in the A-CNK-FM (Figure 11b) with a discernible trend away from the CNK-FM boundary.

Figure 11

Ternary diagrams showing the common primary and secondary minerals and weathering trends of Umba River sediments. The CIA of Umba River sediments range from 39–81 with majority >50. (a) On A-CN-K, Umba River sediments plot along a linear trend on the A-CN edge. (b) On A-CNK-FM, the sediments plot below the smectite field. A = Al₂O₃, CN = CaO* + Na₂O, K = K₂O, CNK = CaO* + Na₂O + K₂O, and FM = Fe₂O₃ + MgO. Average values of basalt, UCC, and clinopyroxene taken from ref. [3].

Based on the calculated DF1 and DF2, the Umba River sediments are plotted across three provenance fields, i.e., mafic igneous, intermediate igneous, and quartzose sedimentary provenance fields (Figure 12). The majority of the sediments are plotted within the mafic igneous provenance field, which is consistent with the geology of the catchment and the associated dominant minerals.

Figure 12

Major element provenance DF plot. Major elements-based oxide ratios for the provenance signatures of Umba River sediments [35] plotted across three provenance fields with majority in the mafic and intermediate igneous provenance.

5.1 Influence of hydrodynamic and sedimentary sorting

Sediment sorting during fluvial transport, settling, and deposition can influence the mineral and chemical compositions of the sediments. Therefore, the influence of grain size sorting on sedimentary chemical composition has to be evaluated prior to any provenance determination [6,26,44,46]. For example, major and trace elements, including Al, Fe, Mg, Ni, and Cr, are known to be abundant in fine fractions of sediments rich in silt and clays, while Si is abundant in sand fractions [47,48]. In the Umba River sediments, grain sizes are distributed within a relatively narrow range between coarse sand (riverine), silty sand (marine), and sandy silt (estuarine) (Table 1; Figure 3). With minimal clay content of <3% in both the riverbank and bottom sediments, these sediments fall within the broad class of sandy sediments (Table 1). The index of compositional variability (ICV) – which is used to indicate maturity of sediments – is >1 in all riverbank and bottom sediments (Table 1). Clay minerals have been shown to exhibit an ICV <0.8 [26]. Therefore, the ICV values derived here for Umba River sediments are consistent with the observation of a very low clay fraction (Table 1). Therefore, the effects of granulometric fractions and hydrodynamic sorting on the chemical composition of the sediments would be minimal. This is also reflected by the distribution of Fe and Al in the mud/sandy-silt fraction of sediments collected in the estuarine station (S7) across the entire quartz-clay spectrum in the Al₂O₃/SiO₂ vs Fe₂O₃/SiO₂ ratio plot (Figure 7), reflecting the minimal effect of sorting [6,47]. This is because the relative enrichment of Fe and Al in coarse sand (e.g., S1, S4, and S8) as well as the sandy silt (S7) means that there is no control of grain size on their enrichment. This implies that the enrichment of Fe and Al in Umba River sediments is dominantly attributable to the weathering inputs from the catchment. The enrichment of Fe (>0.20) and Al (>0.40) in estuarine (S7) samples could be attributed to the increased input of these elements from the weathering of quartzo-feldspathic and hornblende-pyroxene rich downstream Karoo Supergroup and younger lithologies ( Figure 1b). These lithologies and their associated minerals are rich in Fe and Al elements in their chemical structure [6,18]. Additionally, the contribution of Karoo Supergroup lithologies to the increased input of Fe and Al would be further promoted by a hot and humid climate due to the proximity to the Indian Ocean relative to upper catchment stations [6]. The similarity of SiO₂ – the lack of either enrichment or depletion – relative to UCC further indicates the reduced influence of hydrodynamic sorting on mineralogical and chemical compositions of Umba River sediments [26,44,54].

5.2 Weathering indices and source area characteristics

The strong but negative relation between SiO₂ and Al₂O₃ (Figure 6a), the positive relation between Fe₂O₃ and Al₂O₃ (Figure 6c), and the positive relation between both Al₂O₃ and Fe₂O₃ with CIA (Figure 10a and b), indicate a shift that is expected from the replacement of easily weathered minerals towards a dominance of less mobile minerals, e.g., kaolinite composed of mainly the immobile Al³⁺. The positive relation between Al₂O₃ and high CIA (Figure 10a) values in both the riverbank and bottom sediments could be attributed to the initial weathering of sediments before the grains enter the fluvial system. Generally, fluvial sediments exhibit CIA values of ∼50 at the onset of weathering of a predominantly metamorphic and igneous source geology similar to the Umba River catchment. This value increases with reworking processes during transportation [40,41]. The UCC has a mean CIA of ∼47 [44]; however CIA values >70 have also been reported to characterize intense weathering of source rock [49]. According to Campodonico et al. [39], reworked fluvial sediments – through weathering, transport, deposition, and diagenesis – have average CIA values of ∼80. Clay minerals, which are an indicator of extensive weathering and chemical alteration, have CIA values close to 100 [37,39]. The large variation in CIA of the Umba River sediments ranging between 39 and 81, with a majority >50 (Figure 11), signifies a low to high degree of weathering. This can be attributed to varying physico-chemical parameters including temperature, salinity, and pH that vary along the longitudinal cross-section of the Umba River [6,51]. The large variation in CIA values can also be attributed to the supply of pristine material to the fluvial system, supported by the smaller catchment of the Umba River and the short transport from source to mouth. However, the few high alterations recorded can be attributed to the alteration and input of feldspathic and clay minerals dominant in the downstream geology comprised of the Karoo Supergroup. The varying climate from the cold mountainous upstream to the hot and semi-arid midstream, and the hot and humid climate near coastal downstream could explain the variation in CIA [6]. Low CIA values have been reported to be related to low weathering at source due to cooler conditions [34] characteristic of higher elevations, e.g., mountainous regions and semi-arid conditions [6]. These conditions are similar to the Umba River catchment near the source of the river in the Usambara and Pare mountains (Figures 1b and 2). The likely presence of clay minerals, e.g., illite and kaolinite in the Umba River sediments as discussed in Section 4.2 and as observed in the grain size distribution (Table 1), could further explain the CIA variation. Additionally, the observed decrease in lithoclasts and the increase in grains made up of individual mineral crystals from source to mouth (Figures 4 and 5) can indicate the gradual weathering with time exacerbated by river transport. The different roundness of grains within the downstream sediments (Figure 5) indicate that some clasts are incorporated in the bulk sediment later than others during fluvial transport. Part of these sediments, thus, have been minimally influenced by river transport; however, in the marine river mouth, wave action is expected resulting in increasing roundness of the grains.

The relatively low weathering in the downstream sediments compared to the upstream sediments, could be attributed to in situ and proximal deposition devoid of fluvial transportation. Additionally, the occurrence of poorly rounded quartz mineral grains, being physically and chemically resistant to weathering (Figures 4 and 5), is consistent with moderate reworking by river transport (Table 1 and Figure 3). The persisting presence of detrital feldspars (plagioclase and K-feldspars), a precursor of clay minerals, is an indication of low weathering of the gneissic granite, igneous, and meta-igneous rocks sourced from the catchment.

The Umba River sediments are plotted across the plagioclase- K-feldspar line and reveal a narrow near-linear trend along the A-CN edge (Figure 11a). This trend has been attributed to possible higher removal rates of Na⁺ and Ca²⁺ from plagioclase compared to K⁺ removal rates during weathering, transport, and other diagenetic processes [34,49]. However, the linear trend parallel to the A-CN axis could be attributed to the supply and input of relatively pristine sediments to the fluvial system [50]. This is consistent with the relatively short course of the Umba River and the small catchment it drains, consequently allowing faster delivery of newly weathered and pristine material into the fluvial sediments. On the A-CNK-FM, the sediments are plotted below the smectite and K-feldspar field and away from the CNK-FM edge (Figure 11b) further indicating a strong variable degree of weathering of the Umba River sediments. This corroborates the other findings that indicate high variability in weathering and reworking, attributed to progressive sediment input into the river and to varying climatic conditions along its course.

5.3 Source characteristics from bulk geochemical composition

In order to make inter-oxide comparisons, the abundance of Al₂O₃ was used as a reference factor, mainly because it is immobile and less soluble [34]. High positive inter-oxide relations are observed in the element bivariate plots between the immobile oxide Al₂O₃ and SiO₂, Fe₂O₃, and P₂O₅ (Figure 6a, c, and d) in both the riverbank and bottom sediments. This can be attributed to less preferential fractionation during various sedimentary processes including weathering, transportation, and deposition [7,52,53]. It can also be the result of a relatively uniform source geology [26] that would have similar mineralogy as reflected by these elements and observed in the mineral distribution (Figures 4 and 5). This preposition is further supported by the aforementioned influence of the relatively short course of the Umba River that drains a small catchment, limiting the amount of rock lithologies it crosses and therefore, providing a uniform mineralogical and elemental composition. Short rivers tend to transport fresh sediments from source to mouth in a relatively short time, consequently reducing the degree of alteration of the chemical composition. The positive relation between Al₂O₃ and Fe₂O₃, on one hand, and P₂O₅ on the other, can be further attributed to abundances of these elements in platy and sheet phyllosilicates. The observed presence of muscovite, biotite, apatite, and clay minerals in the Umba River sediments (Figures 4 and 5) supports this observed trends [18,47].

The bulk chemical composition of the sediments shows a comparable trend with average UCC chemical compositions (Figure 8a) and is similar to earlier studies within the region [18]. The relative depletion of Na₂O and K₂O observed here (Figure 8a) can be attributed to the result of increased leaching of the highly mobile elements and enrichment of the immobile and less soluble elements. Additionally, alkali elements including Na, Mg, Ca, and K are largely lost as clay minerals are formed during weathering [47]. Alteration of Al- and Mg-bearing minerals including plagioclase and phyllosilicates due to water percolation into rock fractures and crevices could accelerate the removal of mobile elements including Ca, Na, K, and Mg [6]. This is also consistent with moderate to high CIA as proxy for chemical weathering, with higher elemental values of these elements. Additionally, the elevated MgO and CaO contents (Figure 6e and g, respectively) coupled with a scattered relation between K₂O and Al₂O₃ (Figure 6f), also reflect a mixed source composition and varying degrees of weathering [26]. However, the observed high K₂O + Na₂O (>2.5%) and the relatively low K₂O/Na₂O ratio (<1) in the Umba River sediments (Table 1) indicate an almost equal contribution of K-rich feldspar and Na-rich plagioclase [34,35,42].

Trace element concentrations vary along the longitudinal course of the Umba River in both the riverbank and bottom sediments (Table 1 and Supplementary Information 2). The less distinct and negative relation of Sr and Zr with Al₂O₃ (Figure 8a and b) can be attributed to the mixed geology from source to mouth of the Umba River. The weathering of the crystalline rocks of downstream geology (Figure 1) – rich in quartzo-feldspathic rock composition – would introduce increased amounts of Sr-rich K-feldspars and/or biotite to the sediments [55]. Positive covariation and enrichment of Ni, Zn, and Cr with Al₂O₃ in Umba River sediments (Figure 8c, d, and e) can be attributed to the dominance of ferromagnesian elements. These elements are rich in mafic minerals dominant in the mafic granulites [6,18,23] observed in the upper reaches of the Umba River in the Usambara Mountains (Figure 1). Cr/Ni has been used to indicate the mafic and ultramafic provenance [6]. Low Cr/Ni values between 1.3 and 1.5 represent an ultramafic provenance. Both the riverbank and the bottom Umba River sediments exhibit Cr/Ni > 2, indicating a dominantly mafic to felsic provenance. The moderate to high abundances of both Ni (Figure 8c) and Cr (see Figure 8e) in both the riverbank and bottom sediments, suggest a mixed source of Umba River sediments. This mixture is occasioned by the changing lithologies and the subsequent variation in mafic and felsic minerals from source to mouth. The relative depletion and significant variation in Sr (Figure 8g) at low CaO and the depletion of Ba (Figure 8h) at moderate CaO in the estuarine bottom sediments have been attributed to the physicochemical variations (salinity, dissolved oxygen, and temperature) and the grain size [46,56,57,58]. This is also supported by the significant variation (p < 0.05) in the two elements with environmental setting (Supplementary Information 1). However, the observed depletion of Rb, Th, and U in the Umba River sediments (Figure 9b) signifies a magmatic source geology with geochemical affinities of a meta-igneous granulite [18,19,24]. The sandstone lithology dominant in the lower reaches of Umba River (Figure 1b) is known to be low in Rb and Sr [34], therefore explaining the depletion of Rb relative to UCC.

5.4 Provenance

The sediments are plotted across three provenance fields, i.e., mafic igneous, intermediate igneous, and quartzose sedimentary provenance fields (Figure 12) with the majority located in the mafic to intermediate igneous field. This indicates a strong influence of the upper and middle catchment geology, and a subordinate contribution of the lower catchment geology that is primarily composed of quartz-rich lithologies (Figure 1b). The Umba River catchment has pockets of mafic to ultra-mafic granulites (in the mountainous Mng’aro region) and mafic hornblende-pyroxene facies within the Karoo [18,24] further downstream (Figure 1b). On the contrary, the lower catchment is dominated by quartzo-feldspathic-rich sandstones and siltstones that would explain the placement of some Umba River sediments in the quartzose provenance field (Figure 12). Further, the plot in the quartzose sedimentary provenance quadrant suggests the derivation of the Umba River sediments from a recycled orogen [34,35] – presumably, because the upstream Umba River is characterized by the geologically uplifted Usambara mountains [18,20,21].