The Application of a Combined Fingerprinting Method to Lake Sediments in the Central Highlands of Vietnam

1. Introduction

Currently, climate change, industry expansion, and agricultural cultivation are globally affecting water quality in inland aquatic ecosystems. In this regard, the assessment of tropical lake environments has become a major goal for researchers, since these water reservoirs are crucial for climate change management, clean water, good health, and social–economic activity of the population [1,2,3,4,5,6]. Geochemical studies of lake sediment contribute to the understanding of the geological background, while trace (TEs) and rare earth (REEs) elements are effectively utilized as geochemical tracers of natural elemental input and possible anthropogenic impact [7,8,9]. However, recent research has shown that environmental assessment and source appointment of sediments in inland water reservoirs is often complicated by a complex geological and geochemical composition. Therefore, they should be implemented with a particular focus on the diverse methods of multi-element data normalization and multivariate statistics analyses [2,8,10,11,12]. In recent years, there has been ever-growing interest in the environmental geochemistry of TEs and REEs in Vietnam due to the rapidly growing tourism, industry, and urban and agricultural activities [2,13,14,15,16,17,18]. Dak Lak province, in the Central Highlands of Vietnam, has the biggest share (36%) of Vietnamese Robusta coffee production. Because of climate change, more severe and longer droughts are occurring in this area, altering the balance of water resources [2,15,19]. Additionally, aggressive land use has caused significant soil erosion, resulting in landscape changes in the catchment [20]. In this context, water reservoirs such as natural and human-made lakes, are vital for clean water access and sustainable agricultural production in the province.

This study progresses from a recent survey of ten previously unstudied small water reservoirs in Dak Lak province, which primarily focused on establishing a baseline geochemical assessment of lake bottom sediments. The results revealed severe spatial variability in the chemical composition of lake sediments within the relatively small area of Dak Lak province [21,22]. Since the studied lakes were separated from industrial anthropogenic sources of pollution until the sampling period, the detected trace and rare earth element anomalies were related to the local lithology rather than to anthropogenic influence [10,23,24]. However, agricultural non-point source pollution and land-use effects should still be considered.

This work aims to develop and implicate a comprehensive approach for processing and interpreting contrast elemental abundances in sediments using ten small lakes in Dak Lak province as a case study. For this, a combined fingerprinting method was applied: (1) complex EF calculations (EF_Al^s, EF_Al^f, EF_Fe^s, and EF_Fe^f) were used for advanced risk element assessment; (2) Zr/Hf and Nb/Ta ratios were utilized as geochemical indicators of host rock origin; (3) cluster analysis (CA) was employed for sediment provenance discrimination; (4) principal component analysis (PCA) was employed for elemental fingerprint identification; (5) normalized rare earth element (REEs) distributions and fractionation patterns were implemented to determine the sediment sources.

This study provides a baseline for sustainable land use and clean water access in the Dak Lak region under climate change stresses and suggests a new approach for comprehensive environmental assessment of sediments of the inland water reservoirs, even with limited and contrasting geochemical data.

2. Materials and Methods

2.1. Study Area

Dak Lak, Vietnam’s vastest region for coffee-growing, spans an area of 13,030 km² and is home to 1.919 million inhabitants. The rainfall data ranges between 1300 to 2000 mm year^-1 and highlights an altitude dependency towards the mountains in the east and southeast. The province regularly goes through a dry season lasting from December until the end of April, often resulting in drought throughout the area [25].

The density of urbanization is highest within the central basaltic plateau, which is also the area of highest agricultural productivity. Most common land usages include rice, urban zones, and rubber forest, while the crop with the most significant role in the economy of the province is coffee. A strong correlation can be traced between land use and geology. The coffee plantations almost exactly map out the basaltic plateau, where the soil types are mostly Rhodic Ferralsols.

The complex geology of Dak Lak province is highlighted in Figure 1 [26].

Fractured Neogene to Quaternary basaltic rocks, primarily of tholeitic and sub-alkaline olivine composition, form a morphologically slightly undulating plateau in the central part of the area. In the southern part of the region, intrusive Jurassic to Paleogene rocks outlines the major topographic highs of the province. The Precambrian to Paleogene magmatic and metamorphic rocks crop out in the northeastern, southern, and western parts of the area. The Jurassic sedimentary rocks crop out along a northwestern–southeastern axis across the province. Quaternary sediments crop out along the major streams and rivers.

2.2. Sampling

Bottom surface sediment samples were collected from 10 previously unexplored water reservoirs in Dak Lak province of Central Vietnam in August 2010 (Figure 1). Surface sediment samples (0-5 cm) were collected using a plastic piston corer and placed in pre-cleaned polyethylene containers using stainless steel spatulas. One portion of each sample was kept frozen until grain size analysis. The other portion was dried to a constant weight at 60⁰C for chemical element analysis. All sampling, sampling transportation, and preparation procedures were performed using standard clean techniques according to the manual of Loring and Rantala [27].

2.3. Laboratory Analyses

The sediment was classified by particle size distribution using wet-sieving with precision sieves. The obtained fractions were: gravel (≥2 mm), sand (62.5 μm – 2 mm), silt (4 – 62.5 μm), and clay (<4 μm). The total organic carbon (TOC) content in the sediment samples was determined using a TOC 5000-V-CPH analyzer (Shimudzu Co., Japan), provided by the Ocean Chemistry Laboratory of the Shirshov Institute of Oceanology (Russian Academy of Science, RAS) [28,29].

Elemental analyses of the sediment samples were conducted at the Analytical and Certification Center of the Institute of Microelectronic Technology and High-Purity Materials (RAS). Inductively coupled plasma–atomic emission spectrometry (ICP-AES) (ICAP-61, Thermo Jarrell Ash, USA) and inductively coupled plasma–mass spectrometry (ICP-MS) (X-7, Thermo Elemental, USA) were utilized for elemental analyses of solutions obtained by total dissolution of sediment samples in HNO₃ + HClO₄ (3:1 by volume, Merck) using an autoclave system (Аnkon-АТ-2, Russia) [30]. ICP-AES was used to measure the major (Na, Mg, P, S, K, Ca, Al, Ti, and Fe) and trace elements (Li, V, Cr, Mn, Co, Ni, Cu, Zn, and Sr). ICP-MS was only utilized for the analysis of trace elements (Li, Be, B, Sc, Cr, Ni, Cu, Zn, Ga, As, Se, Rb, Sr, Y, Zr, Nb, Mo, Sn, Sb, Cs, Ba, Hf, W, Hg, Tl, Pb, Bi, Th, and U) and rare earth elements (La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu) [31]. The measurement error was less than 10–15 rel. % for ICP-AES and less than 10–30 rel.% for ICP-MS, regarding the element content. The precision and validity of the analytical determinations were assessed using certified reference materials (CRMs): Andesite, AGV-2 (United States Geological Survey), and Essexite Gabbro SRM-2A (Russian Geochemical Standard). In each case, the difference between the certified and measured element contents was within the limits of the confidence intervals.

A more in-depth description of the analytical determinations is given in Koukina et al. [24,32].

2.4. Calculation of the Enrichment Factor

The enrichment factors (EF_Al and EF_Fe) were calculated as follows:

$$\mathrm{E}\mathrm{F}={\displaystyle \frac{\left(\mathrm{M}\right)/\left({\mathrm{M}}_{\mathrm{R}\mathrm{E}\mathrm{F}}\right)\mathrm{s}}{\left(\mathrm{M}\right)/\left({\mathrm{M}}_{\mathrm{R}\mathrm{E}\mathrm{F}}\right)\mathrm{b}}}$$

(1)

where (M)/(M_REF)_s is the ratio of the metal concentration to the reference metal (Al or Fe in this study) in the sample, and (M)/(M_REF)_b is the corresponding in the reference materials, which included average World Shale [33,34] and average World Riverbed Sediments (WRSed) [35].

Standardized graphs of REE distribution were developed with respect to North American Shale Composite (NASC) [36].

2.5. Statistical Analysis

Descriptive statistics and Pearson correlation analysis were conducted using Microsoft Excel 2019 software. Multivariate statistics (Cluster (CA) and principal component (PCA) analyses) were carried out using the MATLAB R2018a computing environment (The Math Works, Inc., USA). Clustering was performed using an agglomerative hierarchical approach for correlation-based pairwise distances (where distance is defined as 1 minus the sample correlation coefficient of points) and average linkage function. PCA was performed using the single-value decomposition algorithm. All analyses were performed on datasets normalized using Z-scores (where chemical elements were treated as variables).

3. Results

3.1. Lake Bottom Sediment Characteristics

The age of artificial lakes at the time of sampling was 7 to 37 years. Organic carbon content in lake sediment samples varied in a wide range from 0.71% of dry weight in the Ea Knop to 8.41% of dry weight in the natural lake Lak, which indicates differences in organic matter influx, sedimentation duration, and rate. The dominant granulometric fractions were sand and silt, which varied from 0.12 to 83.26 and from 11.84 to 97.31% of dry weight, respectively. The finest silty sediments were discovered in the Lak, Buon Tria, Buon Triet, Ea Kao, Ea Nhai, and Buong Dong lakes, while the coarsest sandy sediments were found in Eadar and Ea Knop (Table 1).

3.2. Abundance and Risk Assessment

The mean bulk concentrations of major elements, TEs, and REEs in the surface sediments of Dak Lak lakes are presented in Table 2 (also see Supplementary Materials for bulk element content distribution in Table S1). The mean contents of most elements correspond to the reference material values in the World Shale [33,34] and World Riverbed Sediments (WRSed) [35].

According to Pearson correlation analysis (see correlation matrix in Table S2 of Supplementary Materials), total organic carbon (TOC) was not a controlling geochemical factor of distribution of most elements, which is consistent with the findings of Awankwaa et al. [5] in sediments from a crater lake in Bosomtwe. However, TOC showed positive correlations with bulk Li, S, Sn, Cs, W, and Bi, suggesting their affinity to sedimentary organic matter. The granulometric fractions showed positive (in the case of silt) and negative (in the case of sand) correlations with bulk contents of some major (Al and Fe), trace (Be, Zn, Ga, Se, and Y), and rare earth (Eu, Dy, Ho, Er, Tm, Yb, and Lu) elements.

The enrichment factor (EF) is widely utilized in environmental sedimentary studies as an effective indicator of anomalous elemental contents [6,8,11,20,32,37]. However, in EF calculation, the implication of a single reference element in combination with a single reference material may complicate risk assessment and element source determination [10]. To reduce uncertainty in interpreting contrasting elemental abundances, in this study, the four EF series were calculated using Al and Fe as reference elements, and World Shale and World Riverbed Sediments as reference materials, as follows: EF_Al^s, EF_Al^f, EF_Fe^s, and EF_Fe^f (Table 2).

The results generally confirm that the lake bottom sediments were not contaminated in relation to most TEs and REEs, with a few exceptions. Average selenium EF values exceeded 2 in all calculations, indicating minor enrichment: EF_Al^f (2.98) > EF_Al^s (2.77) > EF_Fe^f (2.37) > EF_Fe^s (2.24), while moderate Se enrichment corresponded to Eadar (EF_Al^s of 5.03 and EF_Al^f of 5.41) and Ea Knop (EF_Fe^s of 5 and EF_Fe^f of 5.27) sediments. Furthermore, minor enrichment was detected for Ti (average EF_Al^s of 2.14), Ta (average EF_Al^s of 2.44), and Nb (average EF_Al^s of 3.17, EF_Fe^s of 2.20). The heightened selenium content is likely related to its abundance in the magmatic source rocks of Dak Lak province (Figure 1). This is consistent with the study of [38], in which Se abundance in the topsoil of Hainan Island in China was inherited from the underlying basalt and granite, while Se content in both basalt and granite increased with chemical weathering intensity. Accordingly, minor to moderate enrichment in lithophile Ta, Nb, and Ti in sandy sediments in Eadar and Ea Knop is likely associated with underlying magmatic and metamorphic rocks [39] (Figure 1).

Generally, based on the results of complex EF analysis, we support the employment of several methods of normalization in risk element assessment. Specifically, in tropical lake sediments derived from Rhodic Ferralsols, the application of Fe as a reference element and average fluvial sediment (WRSed) as a reference material appears to be the most appropriate approach.

3.3. Geochemical Indicators

In multiple studies, element ratios such as Zr/Hf and Nb/Ta are widely utilized as geochemical fingerprints of rock origin and evolution [40,41]. Since sediments inherit their geochemical composition from parent source rocks, Zr/Hf and Nb/Ta ratios may add to sediment provenance discrimination.

Zr/Hf and Nb/Ta ratios are generally constant in basalts and equal to their chondritic values (Zr/Hf=36.6±2.9, Nb/Ta=16.8±3.1). In granitic rocks, Zr/Hf and Nb/Ta ratios are used as the geochemical indicators of fractional crystallization, since their values generally decrease from earlier to later intrusive phases [41,42,43]. In the Dak Lak lakes sediments, Zr-Hf and Nb-Ta were strongly positively linked pairwise (r=0.99), while their Zr/Hf and Nb/Ta ratios were strongly positively correlated (r=0.90). Both Zr/Hf and Nb/Ta varied widely (from 28.2 to 47.4 and from 11.9 to 22.2, respectively) within the relatively small study area of the province, reflecting a pronounced geochemical heterogeneity of sedimentary material. The relatively low Zr/Hf (28.2) and Nb/Ta (11.9) values in sediments of Lak lake indicate a moderate degree of fractional crystallization in the source intrusive rocks, corresponding to high-silica granites [42]. The near-chondritic Zr/Hf and Nb/Ta values of 32.6 and 14.2 in Dak Minh, 35.3 and 15 in Ea Knop, 37.4 and 14 in Buon Tria, 38.6 and 13.7 in Buon Triet, and 39.4 and 15.9 in Ea Sup Thuong, respectively, correspond to the continental material values [44]. The highest Zr/Hf and Nb/Ta ratios were detected in Ea Nhai (47.0 and 22.2), Ea Kao (42.7 and 19.1), Buon Dong (42.7 and 18.4), and Eadar (47.4 and 18.8). According to [44], extremely high “super-chondritic” Zr/Hf and Nb/Ta values suggest a very low degree of fractionation in the primary magmatic source rocks associated with the central basaltic plateau of Dak Lak province (Figure 1).

Overall, the distribution of Zr/Hf and Nb/Ta highlights the geochemical differentiation of Dak Lak lake sediments, conditioned by the provenance of source rocks.

3.4. Multivariate Analyses

In recent sedimentary research, various multivariate techniques, such as cluster analysis (CA) and principal component analysis (PCA), have been widely used to identify elemental fingerprints and discriminate sediment provenance [3,4,8,20,32]. In this study, cluster analysis and principal component analysis were applied to three datasets: bulk element content, EF_Al^s, and EF_Fe^s.

CA enabled the classification of lakes into hierarchical clusters. Figure 2 shows cluster dendrograms for bulk element content (a), EF_Al^s (b), and EF_Fe^s (c).

Summarizing the three dendrograms, we can identify two main groups that remain stable through all three datasets: Ea Kao, Ea Knai, and Buong Duong (1); and Lak, Buon Tria, Dak Minh, and Ea Sup Thuong (2). Lake Eadar was associated with the first group for both EF datasets, while for the bulk content, it was closer to the second group. Lake Ea Knop exhibited distinctive behavior for both EF datasets, forming a separate single-element cluster in the first three agglomerative stages, and was associated with the second group of lakes in the fourth stage. For the bulk dataset, the same pattern was observed for Ea Knop paired with Buon Triet.

The isolated clusters indicate similar geochemical compositions of the sediments within the separated lake groups inherited from similar geological sources. The first group included three lakes (Ea Kao, Ea Knai, and Buong Duong) located on the central plateau, which is composed of Neogene–Quarternary basalts of tholeitic and sub-alkaline olivine composition (Figure 1). The second group included two lakes of the southwestern area (Lak, and Buon Tria) and two lakes of the northwestern area (Dak Minh, and Ea Sup Thuong) of the province. These areas are mostly associated with a patchwork of Quarternary and Jurassic sediments that are cut and metamorphosed by Jurassic and Cretaceous intrusive rocks. Ea Knop sediment is likely associated with the Precambrian to Paleogene magmatic and metamorphic crystalline underlying rocks, while Eadar lake is located on the border of this formation and the plateau.

According to the CA results of the three datasets, the first group of lakes was most clearly isolated from the other lakes due to a different geochemical composition of the sediments. Furthermore, the CA results are supported by the Pearson correlation analysis results shown in Figure 3 (for EF_Fe^s). Among the studied lakes, Ea Kao, Ea Knai, and Buong Duong were strongly positively correlated to each other and positively correlated to Eadar.

PCA was implemented to isolate multi-element associations based on similarities of variable behavior. Figure 4 shows the PCA plots for the bulk element content (a), EF_Al^s (b), and EF_Fe^s (c).

In all three plots, the first three PCs accounted for 82-83% of the total variation (see Supplementary Materials for correlation coefficients and respective p values (Tables S2 and S3), principal component loadings, and scores (Tables S4 and S5)). In each plot, the x and y coordinates correspond to the first two principal component loadings by variables (elements). Site scores were normalized to fit the plot. Scores and loadings for the third principal component are displayed in color.

In plot (a) for bulk element content, the first component (explaining 45% of variance) showed high positive loadings of Al, P, Fe, Ti, V, Cr, Mn, Co, Cu, Ni, Mo, Zn, Ga, Se, Y, Sc, Zr, Nb, Hf, Ta, and REEs (Sm-Lu), and high positive scores for Ea Nhai and Buon Dong (sites 9 and 10). The second component (explaining 24% of variance) showed high positive loadings of Li, Be, K, Rb, Sn, Sb, Cs, As, W, Pb, Tl, Pb, U, Th, and Bi, and high positive scores for Lak, Buon Tria, and Buon Triet (sites 5, 6, and 7). The third component (explaining 14% of variance) showed high positive loadings of Ca, high negative loadings of Sr and light REEs (La-Sm), and high negative scores for Ea Knop and Buon Triet (sites 2 and 7).

In plot (b) for element enrichment in relation to Al (EF_Al^s), the first component (explaining 39% of variance) showed high positive loadings of Ca, Ti, V, Cr, Mn, Co, Mo, Se, Y, Sc, Ag, Cd, Ba, Zr, Nb, Hf, Ta, Hg, and REEs (Eu, Tb, Dy, Ho, Er, Tm), and a high positive score for Eadar (site 1). The second component (explaining 29% of variance) showed high positive loadings of Be, K, Rb, Sr, Tl, Pb, Th, U and high negative loadings of Fe, Ni, and Cu. Ea Knop and Buon Triet (sites 2 and 7) had the highest positive scores, while Ea Kao, Ea Nhai, and Buon Dong (sites 8, 9, and 10) had the highest negative scores along PC2. The third component (explaining 14% of variance) showed high positive loadings of Li, Mg, As, Sn, Sb, Yb, and Lu.

In plot (c) for element enrichment in relation to Fe (EF_Fe^s), the first component (explaining 52% of variance) showed high positive loadings of Be, Al, K, Rb, Zn, Ga, Sr, Se, Y, Sc, Mo, Ag, Cd, Ba, Tl, Pb, Th, U, and most REEs (except Eu, Yb, and Lu), high positive scores for Ea Knop and Buon Triet (sites 2 and 7), and high negative scores for Ea Kao, Ea Nhai, and Buon Dong (sites 8, 9 and 10). The second component (explaining 19% of variance) showed high positive loadings of P, Ti, V, Cr, Mn, Co, Ni, Nb, and Ta, high negative loadings of Li, S, and Cs, a high positive score for Eadar (site 1), and high negative scores for Lak, Buon Tria, and Buon Triet (sites 5, 6 and 7). The third component (explaining 11% of variance) showed high positive loadings of Zr, Hf, and Mg.

Summarizing the three PCA plots, we can identify two main element groups that remain stable through all three plots and can be interpreted as multi-element fingerprints of the dominant source rocks in lake sediments. The Ti-V-Cr-Fe-Mn-Co-Zr-Hf-Nb-Ta fingerprint is associated with detrital heavy minerals, such as zircon and titanite, derived from effusive basalts of the central basaltic plateau. A similar pattern has also been observed in basaltic effusives in other regions [37,41,43,45]. The Be-K-Rb-Tl-Pb-Th-U fingerprint is affiliated with felsic igneous source rocks, such as intrusive granites derived from a patchwork of magmatic and metamorphic rocks of the southwestern and northwestern parts of the province [40,46]. Additionally, the Al-Zn-Ga-Se-Y-Sc fingerprint is associated with the fine-grained silt- and clay-sized aluminosilicate host fraction, which is present in all lake sediments in different quantities. In fact, bulk Al-Zn-Ga-Se-Y-Sc exhibited a significant positive correlation with the silt fraction content, while Al and Ga showed very strong correlation to each other (r=0.99) and to the silt fraction (r=0.90) (Table S2). This is consistent with the study of [47], showing that Ga is concentrated with Al in clay minerals during the weathering process in stream and floodplain sediment and soils.

3.5. REE Patterns

Rare earth elements (REEs) are a group of metallic elements that are resistant to chemical mobilization and strongly partitioned into the particulate phase. Geochemical patterns of REEs are widely implemented to discriminate the provenance and depositional history of surface sediments in aquatic environments [8,11,37,48,49,50,51].

In Dak Lak sediments, the ΣREEs (summed from La to Lu) varied from 75 μg g⁻¹ in Dak Minh to 272 μg g⁻¹ in Buon Dong, which corresponds to the WRSed values (Table 2, Table S1). The light rare earth elements (LREEs, La - Nd) comprised 79 to 91% of the total REEs sum, medium rare earth elements (MREEs, Sm – Dy) comprised 6 to 10%, and heavy rare earth elements (HREEs, Ho-Lu) comprised 3 to 11% (Figure 5). LREE enrichment was evident in all sediment samples, which, according to Wang et al. [48], indicates terrestrial material input.

The coherent REE behavior during geochemical processes has been widely observed in lake sediments, suggesting their co-existence during weathering, erosion, and deposition [8,11,37,49,50]. Interestingly, in the Dak Lak lakes, Pearson correlation analysis results highlighted a high fractionation degree among the LREEs, MREEs, and HREEs (see Figure 6 for EF_Fe^s). LREEs and MREEs were strongly positively correlated with each other (with exception of Eu) and weakly or insignificantly correlated with HREEs. The distinct HREE distribution may be due to its affinity to the fine-grained fraction of the sediments. Earlier in this study, bulk HREEs were shown to be positively correlated with the silt fraction (Table S2). A similar trend was observed by [48,52] in surface sediments from Dongting Lake in China, while in the study of Prabakaran et al. [53], the HREE distribution in sediments in the tropical Baram River was related to mineralogical fractionation of the solid fluvial load. Furthermore, considerable fractionation occurred within the HREE group. Yb, and especially Lu, showed no significant correlation with other HREEs, suggesting the partitioning of HREEs during source rock evolution and weathering.

Additionally, the distinct HREE behavior, along with the positive correlation of HREEs with Fe, suggest a principal role of Fe-oxides as a HREE-carrier phase in tropical sediments. This effect was described by [54] in bottom sediments derived from tropical ferrolsols in the Western Atlantic.

Summarizing the NASC-normalized REE patterns presented in Figure 7, the studied lakes were divided into four groups. Lak, Buon Tria, and Buon Triet exhibited a generally even REE distribution around the NASC values, with a declining trend towards HREEs (a). Both Ea Knop and Eadar exhibited pronounced HREE depletion, while Ea Knop revealed the highest Eu depletion. Ea Sup Thuong, and especially Dakh Minh (c), were characterized by relative LREE depletion and a relatively even HREE distribution. The most cohesive group comprised Ea Kao, Ea Nhai, and Buon Dong (d), which were characterized by a vivid Eu elevation. Generally, Eu anomalies indicate differentiation between Eu and other REEs in the processes of sediment transportation and deposition [52]. As feldspars are the primary host for Eu, positive Eu anomaly suggests an abundance of feldspar minerals in the sediments of Ea Kao, Ea Nhai, and Buon Dong [7,54]. In contrast, negative Eu anomaly indicates depletion in feldspar minerals during the evolution and weathering of the source rocks in Ea Knop sediments [9,55].

4. Provenance Discrimination

Summarizing the results of multivariate analysis of the bulk content, EF_Al, and EF_Fe of 57 elements, geochemical indicators, and REE distribution patterns, two main facies of lake sediments were identified and linked to three isolated multi-element fingerprints.

Buon Dong, Ea Nhai, and Ea Kao comprise the most isolated group of lakes. In these lakes, surface bottom sediments are characterized by «super-chondritic» Zr/Hf and Nb/Ta, a cohesive REE distribution pattern with a positive Eu anomaly, and a Ti-V-Cr-Fe-Mn-Co-Ni- Zr-Hf-Nb-Ta elemental fingerprint. These sediments are rich in detrital feldspar and originate from effusive tholeiitic basalts of the central basaltic plateau of Dak Lak province.

In Lak, Buon Tria, and Buon Triet, surface bottom sediments are affiliated with near-chondritic Zr/Hf and Nb/Ta values, near-NASC REE distribution pattern, and a K-Rb-Sr-Tl-Pb-Th-U elemental fingerprint. These sediments are derived from host intrusive granites incorporated in a patchwork of Quarternary and Jurassic sediments that are cut and metamorphosed by Jurassic and Cretaceous intrusive rocks.

The Al-Zn-Ga-Se-Y-Sc element fingerprint is related to the fine-grained silt- and clay-sized material inherited from the host sedimentary rocks that are largely composed of clay minerals, siltstones, shales, marls, and clayey limestones [26]. In Ea Knop, the detected negative Eu anomaly also indicates sedimentary rock origin.

Overall, the high degree of fractionation between the LREEs and HREEs suggests partitioning of the REEs during source rock evolution and weathering, while HREE fractionation is more pronounced than that of LREEs.

5. Conclusions

Multivariate analyses (CA and PCA) of three datasets (bulk content, EF_Al, and EF_Fe) combined with geochemical indicators (Zr/Hf and Nb/Ta) and REE distribution patterns effectively determined the provenance of bottom sediments of 10 small lakes in Dak Lak province in Vietnam. Notably, a group of three lakes (Buon Dong, Ea Nhai, and Ea Kao) was clearly separated from the other lakes due to super-chondritic Zr/Hf and Nb/Ta values, a distinctive REE pattern with positive Eu anomaly, and a Ti-V-Cr-Fe-Mn-Co-Ni- Zr-Hf-Nb-Ta elemental fingerprint in the surface bottom sediments derived from tholeiitic basalts of the central basaltic plateau of Dak Lak province. The implemented comprehensive approach proved effective in linking elemental fingerprints to geochemical provenance when applied to limited geochemical data and contrasting element abundancies.

This study provides a baseline for sustainable land use, optimized irrigation, and clean water availability in Dak Lak province, which is experiencing climate change stresses. Generally, detected TE and REE anomalies in the studied sediments most likely relate to the local lithology and do not infer anthropogenic influence. However, in view of clean water use, agricultural non-point source pollution must be considered in future studies.