Geochemical Speciation, Ecological Risk and Assessment of Main Sources of Potentially Toxic Elements (PTEs) in Stream Sediments from Nile River in Egypt

Abstract

Studying and understanding the complexity and interactions of different factors influencing stream sediment quality is necessary for the development of successful water quality management strategies. This study aims to evaluate the level of contamination by potentially toxic elements (PTEs) (As, Co, Cr, Cu, Mn, Ni, Pb, V, Zn) of the stream sediments of the Nile River. During the spring of 2019, river sediments were sampled at 23 sites along the Nile River. For each sample, one aliquot was digested in aqua regia and analyzed by ICP-MS for pseudo-total concentration, while for another aliquot, sequential extraction procedures were applied to determine chemical speciation. Compositional data analysis (CoDa) and k-means were applied to recognize the contribution of natural and anthropogenic sources, while pollution indices (EF, RAC) and sediment quality guidelines (SQGs) were applied to assess the ecological risk to biotic species. The results reveal that elements such as Cr, Mn, V and Fe, found in high concentrations in almost all samples (Cr up to 739 mg/kg, Mn up to 1942 mg/kg, V up to 507 mg/kg, Fe up to 98,519 mg/kg), have a natural origin, while the concentrations of Cu (up to 69 mg/kg), Ni (up to 88 mg/kg), Co (up to 42 mg/kg) and As (up to 9.8 mg/kg) are linked to both natural and anthropogenic processes. Sequential extraction shows that Mn, Co, Ni and, in some sites, Cu and Zn, are the most bioavailable elements. These elements present a high risk of toxicity, while the remaining elements imply a low-to-moderate risk.

1. Introduction

Potentially toxic elements (PTEs) are chemical elements (Al, As, Ba, Be, Cd, Co, Cr, Cu, Fe, Hg, Mo, Mn, Ni, Pb, Se, Sb, V, Zn) that can be toxic to human health and ecosystems when they are present at high concentrations in the environment and are bioavailable in some molecular forms [1,2,3].

In rivers, PTEs can accumulate in microorganisms, flora and fauna and, hence, enter the human food chain, causing health problems such as damage to the nervous and immune systems and to enzyme function [4,5,6]. Because of their toxicity, bioaccumulative nature and non-biodegradability, PTEs have become a source of increasing public concern worldwide. Due to adsorption, co-precipitation and hydrolysis, only a small portion of PTE ions are dissolved in water, while more significant quantities are deposited in sediments [7]. For this reason, stream sediments represent a main reservoir of contaminants and, therefore, are relevant sources of information to assess man-made contamination in rivers. In accordance with this, a number of authors suggest the study of stream sediments to evaluate PTE pollution [8,9,10,11].

Moreover, through the Environmental Quality Standards Directive 2008/105/EC [12], the European commission has also recognized the importance of contaminated sediments as a water quality problem across European countries.

In recent decades, the anthropic pressure caused by industrial and agricultural activities has been intense and the PTE emissions have accelerated considerably. In some cases, the geological setting fosters favorable conditions for the natural enrichment of PTEs. As a result, in some geographic areas, the combination of natural and anthropogenic effects contributes to explaining the high concentration of PTEs in stream water and sediments and puts the ecosystem at serious risk [13,14,15,16,17,18]. Rocks represent a natural source of PTEs in stream sediments that are scattered into the environment due to erosion, rock–water interactions, topsoil leaching and riparian and floodplain interaction [19,20]. The main anthropogenic sources of PTEs are wastewater, forest fires, fossil fuel combustion and atmospheric pollutant deposition [21,22,23]. When the concentration of PTEs in stream sediments is linked to anthropogenic sources, they are present in a bioavailable form and generate a greater environmental risk to aquatic biota [24]. In these cases, it is essential to utilize chemical analysis of samples through sequential extraction procedures in order to evaluate PTE bioavailability. Nevertheless, in environmental studies, the chemical speciation of these elements is often neglected because it is costly and time-consuming to obtain [25]. Therefore, if the PTE chemical speciation is not known, the assessment of the environmental risk related to their concentration can lead to misleading results in complex environmental matrices such as soils and stream sediments.

The aim of the present study is to evaluate the degree of PTE contamination of the Nile River stream sediments through the study of their pseudo-total chemical composition, the study of bioavailability and the use of univariate and multivariate statistical data analysis within a compositional data analysis framework (CoDA).

Although the Nile River is one of the main sources of water supply in Egypt, there are not many works on the PTE concentration in its sediments. In Egypt, most population clusters are located along the Nile River, which is the main source of drinking, irrigation and industrial water. Many anthropic activities developed along the Nile banks are potential sources of PTEs, for instance, motor vehicle exhaust fumes, iron and steel factories, metal-containing fertilizers, synthetic anthropogenic chelating agents, municipal waste discharges, fuel combustion, drainage and a variety of urban operations [26]. This alone justifies the interest of environmental research aimed at determining the degree of pollution, ecological risk and main sources of PTEs along the Nile River. Both the increasing level of pollution and the decreasing volume of the Nile have become major issues in Egypt, especially after the construction of the Aswan High Dam. However, studies so far have been concerned with limited areas and have often neglected the molecular shape of the PTEs and the geochemistry of the outcropping rocks in the hydrographic basin [27,28]. In this context, the present work investigates the concentration, distribution, geochemical speciation and potential sources of PTEs in sediments of an approximately 800 km long segment of the Nile River, while taking into account the geological context. A variety of environmental assessment indices are applied to provide reliable information about the presence and characteristics of PTEs.

2. Materials and Methods

2.1. Study Area

The Nile River is Egypt’s most important water resource, and it has two main sources: the equatorial region around Lake Victoria and the Ethiopian highlands. More than 85% of the Nile’s water originates from the Ethiopian highlands, while the remaining 15% originates from the East African lake plateau. With a catchment area of approximately 3.1 million km² and a length of approximately 6850 km, it is considered the longest river in the world, contending the primacy of the Amazon River. The Nile River’s sediments mostly derive from the Ethiopian basaltic highlands, Precambrian basaltic rocks (PBRs) of the Arabian–Nubian shield, Saharan metacraton, Phanerozoic carbonate sedimentary succession and the local contribution of aeolian sources [29,30]. Other relevant contributors are basalts related to Oligo-Miocene volcanism (Figure 1), outcropping in central-northern Egypt [31]. The Red Sea Hills area in Egypt also falls within the drainage basin of the Nile River [32,33].

From a geochemical point of view, both PBRs and Oligo-Miocene basalts (OMBs) represent natural sources of PTEs. In particular, PBRs present regional enrichment of Cu, Co and Ni associated with gabbro and ophiolite, and local enrichment of Cr and V associated with chromites and black shales, respectively [34]. The geochemistry of OMBs is very similar to that of PBRs, showing an enrichment of V (up to 500 ppm) when compared with PBR basalts [35,36]. However, V is fairly abundant in Egypt, and it is often associated with Fe in black sands and iron oxide deposits. Moreover, the presence of V-rich black sands throughout the Nile Delta suggests that vanadium concentration is also important in Nile River stream sediments [34,37]. Egypt is also home to several other mineral deposits, including stratiform massive sulfide (Pb, Zn, Cu) and banded iron formation (BIF) deposits [37]. Among these, the iron oxide deposits are those exhibiting the widest distribution in the whole study area (Figure 1).

Figure 1. Geological map (modified after Abdelsalam [38]) and sampling points [34,37].

Figure 1. Geological map (modified after Abdelsalam [38]) and sampling points [34,37].

In addition to the natural enrichment of PTEs, there is anthropogenic input derived from the numerous industrial sites located along the Nile’s riverside. The Nile River supplies approximately 65% of the industrial water needed and receives most of the industrial waste [39]. It also collects enormous amounts of agricultural wastewater, hence carrying various chemical pollutants related to the widespread use of fertilizers and pesticides. The situation has worsened after the construction of the Aswan High Dam [40,41].

2.2. Sampling and Analysis

Stream sediment samples were collected in September 2019 at 23 sites from Aswan to Cairo (Figure 1) using a grab sampler (Ekman type). They consisted of composite samples from five points over a stream stretch of 100 m. They were preserved in clean polyethylene bags, stored at 4 °C and transported to ALS CHEMEX (Guangzhou, China) for chemical analysis. An aliquot of all samples was dried at 70 °C and sieved to the <150 μm fraction. Analyses of 9 potentially toxic elements (As, Co, Cr, Cu, Mn, Ni, Pb, V, Zn) were carried out on 0.25 g of dry sediment after digestion in 2 mL HCl, 6 mL HNO₃ and 2 mL HF. The digestion solutions were then analyzed by inductively coupled plasma atomic emission spectroscopy (ICP-AES) and inductively coupled plasma mass spectrometry (ICP-MS). The solution was taken to a final volume of 300 mL with 5% HCl. Aliquots of sample solution were aspirated into a Jarrel Ash Atomcomp 975 ICP-Emission Spectrometer and a Perkin Elmer Elan 6000 ICP-Mass Spectrometer.

Reagent blanks, standard reference soil samples and four duplicate samples were measured to meet quality assurance and control requirements. The tolerated relative difference across replicates was set to ±5% relative to the mean.

On the remaining sample portion, sequential extraction procedures were applied to determine the chemical speciation, mobility and origin of PTEs as previously proposed [42,43,44,45]. Chemical fractionation of PTEs in sediments was performed according to the European “Community Bureau of Reference” (BCR) sequential extraction procedure [46,47]. The BCR methodology was embedded into a four-step scheme following Delgado et al. [43] and García-Ordiales et al. [44]. This allowed the identification of four chemical speciation fractions: The first fraction (F1) was extracted by a solution of 0.11 M of acetic acid and allowed the extraction of the exchangeable fraction, which included weakly adsorbed elements retained on the solid surface by relatively weak electrostatic interactions, elements released by ion-exchange processes and elements co-precipitated with carbonates. The second fraction (F2) was extracted by a solution of 0.1 M of hydroxylamine hydrochloride and consisted of reducible or bound to Fe and Mn oxides. The third fraction (F3) was extracted by a solution of 17.8 M of hydrogen peroxide (adding 8.8 M twice 1 h apart, after shaking at room temperature and heating at 85 °C) and 1 M of ammonium acetate, and it included easily oxidizable species (bonded to sulfides and organic matter). Finally, the fourth fraction (F4) was extracted by a mixture of acid (HNO₃ + HF + HClO₄) and included low-solubility species related to residual fraction (mainly silicates). Before applying the BCR procedure, the sediments were completely dried in an oven at 40 °C for around 48 h. The sediments were agitated at room temperature for 16 h using a shaker. The fractions extracted from residue at each step were centrifuged at 3000 rpm for 20 min in a polyethylene centrifuge tube. The residue was rinsed with 20 mL deionized water for 15 min in a mechanical shaker and then centrifuged at 3000 rpm for 20 min. The whole procedure was conducted at the advanced geochemistry laboratory of the China University of Geoscience and validated using the BCR-701 sediment certified reference material, which included certified and indicative extractable concentrations of Cd, Ni, Cu, Pb, Cr and Zn [48]. Finally, note that ICP-MS was used to determine the metal content in each fraction.

2.3. Assessment of Stream Sediment Contamination

To determine the pollution status of the Nile River and evaluate potential environmental health risks related to the aquatic sediment systems, two different methods were applied: enrichment factor (EF) [49] and risk assessment code (RAC) [50]. These methods are based on total and sequential extraction analyses, respectively. The EF (Equation (1)) allows us to determine the degree of contamination of stream sediments by comparing the concentration of each element to that of a reference element. This indicator is computed as follows:

$$EF=\frac{\frac{C_x}{C_{ref}}sediment}{\frac{C_x}{C_{ref}}background},$$

(1)

where the numerator is the ratio of the concentration of the element of interest to that of an element adopted as a reference in the sample, and the denominator is the ratio of the concentration of the element of interest to that of an element adopted as a reference in the background. A reference element must be a notably stable element in the sediments, i.e., characterized by no degradation and low mobility. The concentration in the environment of the chosen element must not be altered by anthropic activities. Elements typically used for this purpose are Al, Fe, Mn and Rb, as well as Sc, Zr and Ti. In this study, Fe was chosen as the reference element and Fe in the Upper Continental Crust (UCC) was chosen as the background reference element [51]. The values of EF are indicative of different levels of enrichment. Thus, EF ≤ 1 suggests that the concentration of elements may derive entirely from crustal materials or natural processes [49]. Conversely, EF > 1 indicates that a part of the elements derives from non-crustal materials, i.e., they originate from other sources such as anthropic activities. The RAC defines the risk of toxicity for aquatic species associated with each PTE considering only the bioavailable fraction (F1). This concentration can be quickly released by small changes in pH and become easily bioavailable. The RAC is computed as the percentage of each individual PTE in the F1 fraction. RAC ≤ 1 indicates no risk, whereas RAC ≥ 50 indicates very high risk [50]. Finally, PTE concentrations were compared with sediment quality guidelines (SQGs) to assess their toxicity and the potential hazards posed to aquatic life. Sediments are generally recognized as a sink for many substances in aquatic systems, including contaminants. Consequently, SQGs have emerged as an important and critical consideration for the protection of benthic ecosystem health, fisheries conservation and protection of surface water quality (for more on this topic, see also Aradpour et al. [3]). The distribution of the total concentrations and of the exchangeable fraction of each element were plotted on cumulative frequency curves, including threshold concentrations. These graphs show the reference concentration values of the different indices considered: lowest effect level (LEL), threshold effect level (TEL), probable effect level (PEL) and severe effect level (SEL) [13,52]. Note that guidelines for Mn, Co, Fe, V in stream sediments do not exist. Thus, two threshold values proposed by Li et al. [53] for manganese were used: a threshold effect concentration (TEC) and a probable effect concentration (PEC).

2.4. Identification of Main PTE Sources

As PTEs emitted by human activities are essentially hosted in the exchangeable fraction (F1) and in the organic fraction (F3) [24], we grouped F1 and F3 data together in reference to possible anthropic sources. Likewise, we grouped F2 and F3 data together in reference to natural sources. All statistical analyses were conducted on both databases using the R system for statistical computing [54]. To identify different PTE sources in sediment samples, the data were analyzed using multivariate statistical methods within a compositional data analysis (CoDA) framework. CoDA provides a consistent methodology to extract relevant information from data defined on a relative scale (mg/kg in our case). It has, thus, found a fruitful area of application in the geosciences, where relative data are commonplace. In brief, CoDA focuses the analysis on log-ratios between the elements constituting the geochemical compositions that, unlike the raw data, are independent of the measurement units used (scale invariance) and of the size of the composition (subcompositional coherence). Moreover, using log-ratios facilitates the application of ordinary statistical methods and models. Namely, in this work, we expressed the original geochemical compositions in so-called isometric log-ratio (ilr) coordinates before proceeding with statistical analyses [55,56,57,58,59,60]. Accordingly, the well-known k-means algorithm was applied on ilr coordinates in order to reveal clusters of samples in the data set, which may be associated with different PTE sources. The number of clusters was formally determined using the NbClust package in R [61]. The groupings obtained defined a factor variable that was used as input in multivariate analysis of variance (MANOVA), which allowed us to verify whether the differences in mean composition between groups were statistically significant at the usual 5% significance level, following the CoDA formulation detailed in Martín-Fernández et al. [62]. Post hoc pairwise MANOVA comparisons between clusters were subsequently applied, using the Benjamini–Hochberg procedure to adjust p-values for multiple comparisons [63]. Subsequently, compositional robust principal component analysis (rPCA) was performed following Filzmoser et al. [64] to explore the multivariate structure of the data and the geochemical associations for each cluster in low dimensions. A planar biplot graphical display was produced based on the first two principal components (i.e., those accounting for the largest portion of the total data variability as usual). Using a robust version of PCA helped to downplay the influence of outlying samples on the results. Finally, the ArcGIS software package allowed us to produce a cluster distribution map including lithology and the main anthropogenic sources.

3. Results and Discussion

3.1. PTE Distribution in Stream Sediments of the Nile River

Basic summary statistics of the PTEs are shown in

Table 1. Summary statistics for the investigated elements in stream sediments of the study Nile River area (in mg/kg). Instrumental detection limit (IDL), quantile (Q), median absolute deviation (MAD), robust coefficient of variation in % (%rCV).

	As	Co	Cr	Cu	Fe	Mn	Ni	Pb	V	Zn
IDL	0.2	0.1	1	0.2	100	5	0.2	0.5	1	2
Min	6.6	32	240	42	53,751	940	66	8	280	113
Q25	7.6	33	323	49	71,948	1191	74	10	324	125
Mean	8	37	425	54	61,084	1366	77	19	367	136
Median	8.1	37	378	54	74,776	1350	77	12	345	135
Geometric mean	8	37	405	53	77,519	1347	77	15	362	135
Q75	8.6	38	538	58	85,200	1518	81	20	413	146
Q95	9.4	41	667	60	96,804	1735	86	52	474	158
Max	9.8	42	739	69	98,519	1942	88	63	507	161
MAD	0.64	2.09	115	4.45	8654	187	4.2	11.2	52.8	12.3
Skewness	0.3	0.11	0.64	0.29	0.12	0.11	0.04	1.65	0.56	0.12
rCV (%)	7.9	5.7	30.4	8.2	11.6	13.9	5.5	93.3	15.3	9.1

. It can be observed that the highest concentrations corresponded to Cr, Fe, Mn and V. Note that these values also stand out when compared with those found in previous studies, both regarding the Nile River and other rivers worldwide (

Table 2. Comparison of arithmetic mean PTE concentrations (in mg/kg) in the studied area with other local and worldwide rivers in previous studies. Maximum and minimum values are shown in parentheses, ND stands for not detected.

	As	Co	Cr	Cu	Fe	Mn	Ni	Pb	V	Zn
Cairo–Aswan (Nile) (This study)	8 (6.6–9.8)	37 (32.2–42.4)	425 (240–739)	54 (42–69)	61,084 (53,751–98,519)	1366 (940–1834)	77 (66.3–88)	19 (8.6–63.4)	367 (280–507)	136 (113–161)
Aswan–Giza (Nile) [67]	ND	29 (25–36)	ND	36 (30–41)	20,432 (18,420–23,570)	2468 (550–5800)	ND	45 (34–60)	ND	170 (91–270)
Middle Delta (Nile) [68]	5 (1–14)	31 (10–40)	183 (84–336)	61 (24–139)	62,583 (16,786–76,724)	1099 (434–1549)	70 (36–94)	30 (12–182)	192 (96–240)	143 (88–689)
Nile (Egypt) [69]	ND	58 (31–79)	3 (1–8)	36 (19–53)	11,347 (8398–14,118)	257 (106–548)	22 (5–40)	33 (13–79)	ND	51 (14–114)
East Attica (Greece) [70]	41 (8–273)	18 (4–39)	285 (53–1389)	31 (8–80)	27,169 (7600–41,800)	716 (271–2673)	172 (53–512)	217 (17–2611)	66 (26–121)	170 (23–1331)
Tigris (Turkish sector) [17]	5 (2–18)	37 (5–389)	84 (28–163)	344 (28–5075)	ND	682 (282–1657)	132 (74–288)	264 (62–566)	ND	202 (60–2396)
Luan (China) [71]	5 (2–13)	ND	71 (29–152)	45 (6–178)	ND	ND	ND	21 (8–38)	ND	75 (21–161)
Euphrates (Iraq) [72]	ND	28 (22–39)	58 (36–120)	18 (10–30)	2249 (928–3441)	228 (136–312)	67 (40–103)	22 (8–32)	ND	48 (15–130)
World average [73]	ND	20	100	100	48,000	1050	90	150	ND	350

). The concentration ranges of As, Co, Cu, Ni, Pb and Zn appeared to be completely normal even when compared with previous studies. The elements were ranked according to their robust coefficient of variation (rCV% = MAD/Median × 100) as follows: Pb > Cr > V > Mn > Fe > Zn > Cu > As > Co > Ni. Note that this was relatively high for Pb (93.3%) and Cr (30.4%), whereas it remained between 5 and 15% for the remaining ones. According to previous research [22,65,66], anthropogenic factors may be responsible for higher rCV% values. This suggests that the Pb distribution in Nile River sediments, even though it did not show very high concentrations (8 to 63 mg/kg), may be mostly associated with anthropogenic activity in some areas such as Aswan (63.4 mg/kg) and Edfu (53.3 mg/kg).

In relation to the other few studies conducted regarding the Egyptian section of the Nile River, the PTE concentrations found in the current study are not particularly different to those in Moalla et al. [67] and Khatita [68]. However, they do differ from those in Goher et al. [69], especially in regard to Cr and Mn. Compared to rivers in Eastern Attica (Greece), where there is a natural enrichment of Cr, this same element shows significantly higher concentrations in Nile stream sediments. This is due to the higher abundance of mafic rocks in the study area and the geochemical affinity of source rock in the Ethiopian basaltic highlands. Low average concentrations of Cu, Zn and Pb were measured in Nile River stream sediments when compared to the Tigris River, which is seriously affected by pollution related to copper mines and mining waste. Moreover, our samples included higher average concentrations of Co, Cr and Cu when compared with the Luan and Euphrates rivers, where moderate pollution with, respectively, Cu and Cr on the one hand, and Co and Cr on the other hand, was found.

EF values were useful to investigate whether PTE concentrations were relatively high as a consequence of anthropogenic pollution [74], when compared to concentrations in unpolluted samples. Based on EFs (

Table 3. Enrichment factors (EFs) of stream sediments at the different studied sites. EF legend: ≤ 1 deficiency, > 1 ≤ 2 low enrichment, > 2 ≤ 5 moderate enrichment.

Sampling Site	As	Co	Cr	Cu	Mn	Ni	Pb	V	Zn
Giza	0.7	1.1	3.1	1.2	1.1	1.1	0.6	1.6	0.9
Cairo	0.9	1.3	2.3	1.2	0.8	1.1	1.2	1.6	1.1
Helwan	0.7	1.2	2.4	1	0.8	1.1	0.4	1.6	0.9
Naser	0.8	1.2	2.8	1	1	1.1	0.3	1.7	0.8
Beni Suef	0.8	1.2	2.1	1	1.1	1.2	0.3	1.6	0.8
Biba	0.8	1.3	2.7	0.8	1	1.2	0.3	1.7	0.8
Minya	0.7	1.2	4	0.8	1	1	0.3	2	0.9
Beni Mazar	0.8	1.3	2.3	1.2	0.9	1.2	0.7	1.6	0.9
Samalut	0.7	1.3	2.4	1.1	0.9	1.3	0.3	1.8	0.9
Asyut	0.5	1.1	3.2	0.8	0.9	0.9	0.3	1.8	0.8
Abu Tij	0.8	1.4	2.2	1.1	1	1.2	0.4	1.5	0.9
Sidaf	0.7	1.2	2.8	1.2	1	1.2	0.3	1.7	0.9
Girga	0.8	1.2	2.6	1.1	1.2	1.2	0.4	1.6	0.9
Sohag	0.7	1.1	3.3	1	0.9	1.1	1.2	1.7	0.9
Tahta	0.5	1.2	3.1	0.8	1	1	0.2	1.8	0.8
Nagaa Hammadi	0.6	1.2	3.4	0.9	1.1	1	0.2	1.8	0.9
Qena	0.5	1	4	0.7	0.9	0.9	0.2	1.7	0.7
Luxor	0.7	1.2	3.6	1	1	1.2	0.5	1.8	0.8
Armant	0.5	1.2	2.3	0.9	0.8	1.1	0.3	1.6	0.8
Esna	0.8	1.3	2.1	1.3	0.9	1.3	0.7	1.5	1.1
Edfu	0.5	1.2	3.4	0.7	0.9	1.1	1.4	1.8	0.8
Kom Ombo	0.7	1.3	2.5	1.1	0.7	1.1	0.3	1.8	0.9
Aswan	0.9	1.7	2.4	1.4	1.1	1.5	2.7	1.9	1.5

), all samples showed Cr enrichment ranging from low (Aswan, Cairo) to high (Qena, Minya). Vanadium showed moderate enrichment in many sediments similarly to Co, Ni and Cu. However, Pb only did so in the sample taken near Aswan. All other elements had EF < 1 (or just slightly higher), thus indicating deficiency or low enrichment. Figure 2 shows the mean distribution (as % of the total) of single elements included in the different fractions.

Results of the sequential extraction suggest that As (74.2%), Cr (67.3%), Zn (43.8%), Fe (43.6%) and V (36.1%) were mostly concentrated in F4. This suggests that outcropping rocks were the main source of such elements. Moreover, geochemical speciation data suggest that the high percentage of some elements (Co (50.7%), Mn (76.2%) and Ni (36.2%)) in F1 might have had a detrimental impact on aquatic biota. These elements have high affinity with carbonates and, therefore, can co-precipitate within carbonate minerals when released from their source [26].

In particular, the high content of Mn in F1 is most likely related to its similarity to Ca in terms of ionic rays. This similarity allows Mn to replace Ca in the carbonate phase. Among the other fractions, the reducible fraction (Fe and Mn oxyhydroxides) showed a notable percentage of Pb (35.74%), whereas the oxidizable fraction (organic matter and sulfides) presented a larger percentage of Cu (48.45%). Copper and Pb are the elements with the highest affinity for adsorption and their prominent presence in F2 and F3 is often associated with polluted sites [26,75]. In particular, the high affinity of Cu²⁺ ions for soluble organic ligands can notably increase their mobility in sediments, thus favoring its dispersion. As already highlighted above, the geochemical speciation of Zn is consistent with the fact that it showed the highest concentration in the residual fraction (43.8%). However, it also exhibited a non-negligible concentration (28.06%) in the exchangeable fraction. This suggests that a large amount of Zn was probably released locally from anthropogenic sources.

According to the RAC classification (

Table 4. Risk assessment codes (RAC) of Nile River stream sediments at the different studied sites. RAC legend: ≤ 1 no risk– > 1 ≤ 10 low risk– > 10 ≤ 30 moderate risk– > 30 ≤ 50 high risk– > 50 very high risk.

Sampling Site	As	Co	Cr	Cu	Fe	Mn	Ni	Pb	V	Zn
Cairo	10.3	56	12.4	27.4	18.9	91.2	46.4	20.8	30.7	32.6
Giza	10.4	44.2	11.9	18.5	4.7	68.1	37.3	10	7.9	27.2
Helwan	18.4	56.2	8.4	21.2	10.9	89.6	43.2	12.1	21.3	26.4
Naser	15.9	63.8	9.5	21.4	15.7	91.6	48.1	8.3	26.9	24.6
Beni Suef	9.4	60.4	6.8	21.7	13.1	89.2	42.4	10.8	20.6	27.2
Biba	3.2	30	2.7	11.9	4.3	75.4	15.9	10.3	6.1	16.9
Minya	15.3	41.6	3	30	4.5	71.2	26.5	14.8	4.9	32.5
Beni Mazar	10	66.5	8.8	18	18.3	86.8	48.7	10.3	31.2	26.8
Samalut	4.6	51.8	4.5	34.4	6.8	76.5	40.0	12.5	14.7	13.9
Asyut	5.9	38	3.5	35.3	7.7	72.6	26.1	17.9	10.2	26.6
Abu Tij	9.9	55	6.1	20.2	10	75.4	41.7	9.3	24.7	18.1
Sidaf	12.6	51.3	6.3	18.6	4.9	78	33.6	6.9	8.7	24.3
Girga	12	56.2	8.4	22.9	12.1	88.6	39.5	7.4	20.5	30
Sohag	10.4	45.3	4.2	23.4	4.1	66	32.9	11.3	10.2	34.1
Tahta	4.1	26.6	2.6	19.1	4.8	59.9	16.9	17.3	4.1	20.6
Nagaa Hammadi	10.3	59.7	6.1	23.8	9.8	89.2	39.9	12.1	11.7	25.5
Qena	8.3	33	1.7	43.1	3.9	45.1	22.1	18.9	13.6	33.2
Luxor	33	65.9	12.6	21	16.4	92.1	46.4	16.5	24.6	44.1
Armant	10.9	43.4	3.8	14.1	6.2	52.6	30.9	7.6	19.9	18.4
Esna	10.5	64.2	18.1	35.3	10	82.1	43.2	11.2	21.3	20.2
Edfu	5.4	44	1.8	31	5	62.1	31.9	13.3	5.9	17.3
Kom Ombo	7.1	46.4	3.7	22.7	5.6	68.8	34.5	27	6.6	65.3
Aswan	7.8	66	7.6	27.6	17	80.2	45.2	13.9	28.6	39.4

), Mn, Co and Zn revealed a very high risk in 22, 13 and 2 studied sites, respectively. The RAC for Mn reached values of 92.1 in Luxor, 91.6 in Naser and 91.2 in Cairo. Ni showed a high RAC in most of the studied sites (18 out of 23), and Co did so in 8 sites. For both Ni and Co, the highest RAC values were obtained near Beni Mazar. Note that a high risk was also observed for Cu and Zn in some sites. Moreover, chromium, which is poorly bioavailable, had RAC values indicating a low risk in almost all the studied sites (in spite of high EF values). Based on geochemical speciation, the cities with the highest environmental risk (average bioavailability > 30%) were Luxor, Giza, Aswan, Beni Mazar, Naser, Esna and Beni Suef. At the other end, Tahta and Biba were the sites with the lowest RAC values.

3.2. Sediment Quality Guidelines

The cumulative frequency curves in Figure 3 report, for each single element, the distribution curves of total concentration (black line) and exchangeable fraction (red line).

Considering only the exchangeable fraction, As, Pb and Zn had concentrations well below all the thresholds given by the guidelines. Looking at the total concentration instead, all samples presented concentrations higher than LEL and TEL for As, 70% of the samples did so for Zn and it was only four samples that trespassed these thresholds in the case of Pb (at the Aswan, Edfu, Sohag and Cairo sites). Note also that the concentrations of these three elements never exceeded the PEL and SEL thresholds. As to copper, the curve of total concentrations exceeded the LEL and TEL in all samples, but it fell notably lower than the PEL and SEL thresholds.

The exchangeable fraction distribution curve only exceeded the LEL at five locations: Qena, Esna, Asyut, Samalut and Cairo. Nickel was present in high concentrations and, looking at the distribution curve for total concentration, more than 50% of the samples exceeded the SEL threshold. Focusing on the exchangeable fraction, all the samples had values lower than the PEL threshold, with the exception of Beni Mazar.

The total concentrations of Cr were markedly above all SQGs. However, in relation to the exchangeable fraction, all samples had concentrations lower than the SQGs, except for Luxor, Giza and Esna, where LEL and TEL thresholds were exceeded. Similarly, Mn was generally present in high concentrations in Nile River stream sediments. However, this was larger in the exchangeable fraction and, therefore, most of the samples passed the PEC threshold. Consequently, considering only the exchangeable fraction and based on SQGs, there is a potential risk to aquatic species linked only to the presence of high concentrations of Mn. However, when the total concentration is taken into account, the risk is linked not only to Mn, but also to Ni, and particularly to Cr.

3.3. Identification of Main Sources

The identification of main PTE sources was statistically carried out by rPCA and k-means clustering, aiming to recognize both geogenic and anthropogenic contributions. Due to the natural abundance of Fe–Mn oxide in the soils of the Nile River basin, the reducible fraction (F2) was considered mainly of natural origin along with the residual fraction (F4). Therefore, the database obtained from F2 + F4 was used to identify natural sources (Figure 4A). Furthermore, considering that organic matter represents one of the main traps of PTEs emitted by industrial activities [26,75,76], the PTEs contained in the oxidizable fraction (F3) were considered mainly of anthropogenic origin, along with the exchangeable fraction (F1). Therefore, the database obtained from F1 + F3 was used to identify anthropogenic sources (Figure 4B). Three clusters were determined as the optimal partition, and they were obtained by the k-means algorithm on ilr coordinates for both databases. MANOVA supported a statistically significant overall difference between mean compositional profiles by cluster (p-value < 0.01). Post hoc pairwise MANOVA comparisons between clusters concluded statistically significant differences in mean between all pairs (adjusted p-values < 0.01). These results prompted the need to identify areas with different natural background contents. They also suggested that the concentration of PTEs in F1 and F3 might originate from different sources. In order to investigate the different nature of the sources, rPCA was applied to both databases. The associated biplot displays were obtained based on the first two PCs (Dim1 and Dim2 in the graphs) using different colors to distinguish the previously determined clusters (Figure 4A,B).

For natural sources (F2 + F4), the biplot represents 90.5% of the total data variability and depicts three geochemical associations corresponding to each of the three clusters. Cluster 1 is characterized by the enrichment of Cu, Co and Ni. This is attributable to Nubian shield rocks and Oligo-Miocene basalts. Cluster 2 is linked to a higher concentration of Mn, which is attributable to sediments dominated by the presence of Mn oxides. Finally, the geochemical association characterizing cluster 3 (Fe, V, Cr, Pb, Zn, As) appears to be related to sediments dominated by Fe oxides. Note that this latter cluster includes a wider range of elements because Fe oxides can derive from both alteration of sulfides (enrichment in Zn, Pb, As) and alteration of mafic volcanic rocks (enrichment in V and Cr). The corresponding distribution map (Figure 4A bottom) shows that cluster 1 is mainly located in the central-southern sector of the study area, where Oligo-Miocene basalts and Nubian shield rocks are located. However, when samples are close to Fe oxide deposits, they may be relatively depleted in Co, Cu and Ni, even where there are rocks belonging to the Nubian shield (see, e.g., Aswan).

For anthropogenic sources (F1 + F3), the biplot accounts for 79% of the total variability and depicts four geochemical associations consisting of Zn-Mn, Pb-Cu, Fe-V-Cr and As. Cluster 1 is characterized by the geochemical association Fe-V-Cr, which, according to the previous analysis, seems to have a natural origin.

Furthermore, these elements were mainly available in the residual fraction, which implies that their occurrence was mostly driven by the contribution of geogenic factors. In fact, stream sediments act as a sink, removing ions released by rock dissolution from the water column, rather than as a pool for supply [77].

Cluster 2 is mostly characterized by As, while cluster 3 involves both the Zn-Mn and Pb–Cu geochemical associations. These associations seem to be disconnected from those that naturally characterize rocks and sediments. Therefore, they can be considered of anthropogenic origin. The distribution map shows that cluster 2 is located upstream or away from the main industrial areas, suggesting a different anthropogenic origin. Moreover, As is mostly present in the residual and reducible fraction; therefore, anthropogenic sources only had a minor influence on its concentration in Nile River sediments. Identifying the source responsible for As enrichment is difficult based only on the current data, even though pesticides used in agricultural practices might be one of the potential sources. As to cluster 3, it appears to be strongly associated with the presence of industrial activities according to the distribution map (Figure 4B bottom). Moreover, the elements involved (Mn, Zn, Pb, Cu, Ni and Co) were more available in the exchangeable and organic matter fractions, which stresses the great environmental impact derived from industrial activity. The cities included in clusters 2 and 3 are those where PTEs were mainly emitted by a single anthropogenic source; therefore, they may not correspond to the most polluted areas. In fact, the most polluted areas are those that receive a high contribution from all the anthropogenic sources present in the Nile River basin.

Figure 4. Compositional rPCA-based biplot and distribution maps of clusters of samples. (A) Results for residual and reducible fraction, where cluster distribution is compared to lithology and main mineral deposits; (B) results for exchangeable and oxidable fraction, where cluster distribution is compared with main industrial activity. Modified after El-Kammar et al. [78].

Figure 4. Compositional rPCA-based biplot and distribution maps of clusters of samples. (A) Results for residual and reducible fraction, where cluster distribution is compared to lithology and main mineral deposits; (B) results for exchangeable and oxidable fraction, where cluster distribution is compared with main industrial activity. Modified after El-Kammar et al. [78].

4. Conclusions

This work contributes to the scarce literature on the distribution of PTEs in stream sediments of the Nile River. It discusses the total concentration, bioavailability and main sources of PTEs, as well as the ecological risks related to their presence, through the application of compositional data analysis (CoDA) in a multivariate perspective. We found that the concentrations of Cr, Mn, V and Fe were, on average, higher than those found in other rivers worldwide. This can be explained by the abundance of these elements in rocks present within the Nile River basin.

Indices such as the enrichment factor and risk assessment code, along with comparison with sediment quality guidelines, were used to assess ecological risk. Based on the enrichment factor, we concluded that most elements showed deficiency or low enrichment, except for Cr, which showed moderate enrichment. The risk assessment code, based on results from sequential extraction, showed different results. The most bioavailable elements having a high risk of toxicity were Mn, Co, Ni and, in a few sites, Cu and Zn. The remaining elements were associated with low-to-moderate risk. Additionally, according to this index, the highest environmental risk (RAC > 30) was near Luxor, Giza, Aswan, Beni Mazar, Naser, Esna and Beni Suef. However, no risk was concluded near Tahta and Biba. Comparison with sediment quality guidelines was performed using both the total concentration and exchangeable fraction only. As would be expected, the results were markedly different. All samples except one exceeded the probable effect concentration for Mn, whereas many of them exceeded the severe effect level for Cr and Ni when total concentration was considered. For the exchangeable fraction, only Mn exceeded the probable effect concentration in 50% of the sampled sites. These results show that (i) the use of one or another index can rather change the assessment of the degree of pollution at a given site; and (ii) high toxicity may be associated with relatively low concentrations if the total concentration and not the bioavailable concentration is considered. This suggests that indices based on total chemical analyses only are an unreliable tool to assess the toxicity associated with PTEs in various environmental compartments, especially if the natural background contents of the various chemical elements are unknown.

Finally, we showed that industrial activity is a main anthropogenic source causing the enrichment of Mn, Zn, Pb and Cu. Another anthropogenic source, probably linked to agriculture, involved the enrichment of As, while Co and Ni seemed to be linked with both sources. However, the highest pollution level was found in downstream cities, where the combination of more anthropogenic sources takes place. Moreover, our results also suggest that even natural sources involve significant variation, hence justifying the need to define different background contents in the studied area.

The methodological approach shown in this paper helps to better understand the many factors that drive stream sediment quality and the complexities and interactions between these factors. This knowledge could be used to better define management strategies for the environmental quality of rivers and to develop better predictive models of river water quality.

In conclusion, this work seeks to highlight some fundamental steps to perform a correct analysis of the environmental health status of a river:

• The use of different contamination indices such as the enrichment factor (EF), geoaccumulation index (Igeo) and contamination factor (CF) is only indicative if there is no precise knowledge of the natural background contents of the chemical elements. In some areas, the background values can be very high in a natural way, hence the need for environmental geochemical mapping for the entire terrestrial globe;
• Pseudo-total chemical analysis of samples alone can be misleading. It is necessary to know the bioavailability of PTEs through sequential extraction procedures;
• It is important to correctly use multivariate statistical analysis according to the compositional approach (CoDA). Therefore, based on the CoDA framework, this is one of the first works that presents the results of multivariate statistical analysis of PTEs in stream sediments. This work shows that considering the compositional nature of data leads to some useful information about the origin of elements and reveals the capabilities of this approach for investigating the geochemistry of polluted stream sediments. The extent of any mineralization, weathering, diagenesis, contamination and a combination of these factors can be effectively detected following log-ratio transformation of the analytical data obtained from stream sediments.