This study was carried out in the agroecosystem of N’djili CECOMAF (CECOMAF stands for “Centre de Commercialisation des Produits Maraîchers et Fruitiers”, in French) which is located between N’djili and Kimbanseke communes at the eastern part of Kinshasa, the capital city of the Democratic Republic of the Congo. The agroecosystem of CECOMAF is located at coordinates S 04˚25436' and E 015˚21660', on the right bank of the N’djili River, upstream the point of water collection by the N’djili Water Distribution Authority for the supply of water for human consumption.

CECOMAF is a marketing center for vegetable and fruit products, created in 1972 as a cooperative market for agriculture products from N’djili commune. This CECOMAF site was created to supply the growing population with agriculture products. Due to the particularity of its geography, hydrography and soil, the N’djili commune/CECOMAF occupies the first place in vegetable production in Kinshasa and supplies also several markets of the capital with various agriculture products (Mokili, 1998).

The CECOMAF site is limited to the north by districts VI and VII of the commune of N’djili, to the south by the commune of Mont-Ngafula, to the east by the commune of Kimbanseke and to the west by the N’djili River and the commune of Kisenso (Figure 1(A)). The area is mostly flat and located at an average altitude of 290 m above sea level.

The aquatic system of the N’djili River in the CECOMAF section includes aquatic plants, invertebrates, and several fish species. The common North African catfish (Clarias gariepisnis Burchell, 1822) from the Siluridae family, is very abundant and may attain 1 m total length and near 40 kg wet weight. The population, fishermen, and farmers have free access to the N’djili River in CECOMAF area and the fish captures attain an estimated amount of 5 tons per year. Agriculture production from the N’djili/CECOMAF area encompasses several kinds of vegetables, such as the green amaranth (Amaranthus viridis). N’djili and Kimbanseke communes (Figure 1(B)) have a population of approximately 1.5 million inhabitants, some of whom are engaged in agriculture activities and freshwater fishing. Both agriculture and freshwater fishery products are consumed by

the local population and account for an important fraction of the diet, although no accurate statistics are available.

The sampling campaign was performed on February 2020. In the section of N’djili River at CECOMAF, bottom sediments were collected manually, using a plastic shovel and transferred into 1.5 L plastic bottle. From each sampling point, about 250 - 300 g of surface sediment was collected. The sediments were sampled at a distance of about 2 m from the shoreline and with about 1 m water depth. The sediment samples (n = 5) were labelled as Sed1, Sed2, Sed3, Sed4, and Sed5. Top soil (0 - 20 cm depth) samples (n = 2) were collected in agriculture areas by the CECOMAF site and labelled Sol1 and Sol2 (Figure 1, Figure 2).

Biota samples, including the common African catfish Clarias gariepisnis, (Figure 2(D)) from the agroecosystem of CECOMAF, and the earthworm Pontoscolex corethrurus from agriculture soils nearby (Figure 2(C)), were collected using a gill net and a shovel, respectively.

All samples were stored at 4˚C and transported refrigerated to the analytical platform of the University of Geneva for analysis.

Earthworms were washed, weighed, dissected and the tissues of several specimens

bulked to make a sufficiently large sample. Fish were washed, weighed, dissected and the muscle tissue (filet) from several adult specimens were combined in one sample. Biological samples were then frozen and stored at −20˚C in clean polypropylene bottles until acid digestion. The digestion of samples was performed, with minor modifications, as described by Dai et al. (2004), Rashed (2001), and Sivaperumal et al. (2007). In short, a portion of tissue from each sample was freeze-dried (Adolf Kühner, Birsfelden, Switzerland) and grinded to obtain a fine powder. Then, approximately 1 g of fish sample and 100 mg of earthworm sample was digested in a suprapur HNO₃ (Nitric acid 65% Suprapur^®, Merck KGaA, Darmstadt Germany)-HClO₄ (Perchloric acid 70%, Merck KGaA, Darmstadt Germany) mixture (3:1), in Teflon pressure vessels and heated overnight at 110˚C. The digested samples were cooled at room temperature and centrifuged to obtain a clear solution.

Soil and sediment samples were freeze-dried, grinded to a fine powder, homogenized, and sieved using a 63 μm sieve. They were then digested, as described by Thevenon & Poté (2012). In brief, approximately 10 - 15 mg of sample powder was completely digested using pure acids in Teflon vessels and a glass-ceramic hot plate. The procedure involves three heating steps with: 1) 1 mL HNO₃ (suprapure, 65%); 2) a mixture of 0.5 mL of HClO₄ (suprapure, 70%) with 0.5 mL HF (suprapure, 40%); and 3) additional treatment with 0.5 mL of HNO₃ (suprapure, 65%). The solvents were evaporated between each step and finally the samples were diluted in 10 mL of a 1% HNO₃ solution before chemical analysis. Analysis was performed within 24 hours after dilution. Acid dilutions were performed using ultrapure water (Millipore, Milli-Q, 18MW, Merck, Darmstadt, Germany).

Twelve heavy metals and metalloids, namely scandium (Sc), chromium (Cr), cobalt (Co), nickel (Ni), copper (Cu), zinc (Zn), arsenic (As), selenium (Se), cadmium (Cd), tin (Sn), antimony (Sb), and lead (Pb) were analyzed using Inductively Coupled Plasma-Mass Spectrometry (Agilent 7700x series ICP-MS, developed for complex matrix analysis). Methods used were described in detail elsewhere (Atibu et al., 2018). In short, a collision/reaction cell (Helium mode) and several interference equations were used to eliminate different spectral interferences. For the ICP-MS calibration were used standard Merk IV solutions with different concentrations (0, 0.2, 1, 5, 20, 100 and 200 μg·L⁻¹), the ICP multi-element standard solution, and other mono-element solutions (Se and Sb, Merck KGaA, Darmstadt Germany) (Thevenon et al., 2013). Standard deviations of three replicate measurements were below 5%. The limit of detection (LOD) was calculated as 3 times the standard deviation of the analytical blanks, and was less than 0.001 μg·L⁻¹ for all analyzed elements. The chemical blanks for the procedure were less than 2% of the sample signal. The sensitivity of the device and the reliability of the results were verified through the repeated use of the certified reference material LKSD4 (CANMET, Canada). The analysis of samples by ICP-MS was performed after dilution of sample digests with suprapur 1% HNO₃ (Nitric acid 65% Suprapur^®, Merck KGaA, Darmstadt Germany).

Results of metal concentrations in soils and sediments are expressed in ppm (mg·kg⁻¹ of sample dry weight). The results for metal concentrations in biological samples are expressed in ppm (mg·kg⁻¹ wet weight) calculated with average values of water content in fish and earthworm muscle tissues as described in Garcia-Bravo et al. (2011).

The Advanced Mercury Analyser (AMA 254, Altecs.r.l., Czech Rep.) which is a mercury specific Atomic Absorption Spectrometer (AAS), was used for the total Hg analysis in soil and sediment samples according to the method described by Roos-Barraclough et al. (2002). The method consists on the sample combustion, and mercury amalgamation on a gold trap, followed by the gaseous mercury measurement by AAS. The limit of detection (3 SD blank) value was determined at 0.005 mg·kg⁻¹, and the reproducibility was better than 5%. The Hg concentrations are expressed is mg·kg⁻¹ of dry weight (ppm).

The distribution of metals between water and sediment phases was determined as the partitioning coefficient, Kd, being:

Kd = [metal concentration in the sediment (mg·kg⁻¹ of

dry weight)/metal concentration in the water (mg·L⁻¹)]. (1)

The accumulation of metals in the fish muscle was determined as the metal bioconcentration factor (BCF), being:

BCF = [metal concentration in fish (mg·kg⁻¹ of wet weight)/

metal concentration in the water (mg·L⁻¹)]. (2)

Several indices have been proposed to assess the contamination of soils and sediments by metals and potential environmental risks (Mavakala et al., 2022; Aja et al., 2021; Akanchise et al., 2020, Atibu et al., 2016).

According to Maanan et al. (2004), the natural concentrations and anthropogenic pollution of soil and sediment can be assessed through the determination of the Geo-accumulation Index (Igeo) and the Enrichment Factor (EF), respectively.

1) Geoaccumulation Index (Igeo)

The following equation was used to calculate the geoaccumulation index (Igeo):

Igeo = Log₂[(C_n)/1.5(B_n)] (3)

In the above equation, “C_n” represents heavy metal (n) concentration in the sample; “B_n” represents heavy metal (n) concentration in the geochemical background; “1.5” represents the background matrix correction due to lithogenic effects.

2) Enrichment Factor (EF)

The following equation was used to calculate the Enrichment Factor, EF:

EF = (Metal/Sc) Sample/(Metal/Sc) Background (4)

where “Metal” represents the concentration of any heavy metal in the analyzed sample and in the geochemical background; “Sc” represents the scandium concentration in the sample and in the geochemical background. Therefore, a geochemical normalization is performed using scandium (Sc) as a normalization factor. The UCC (Upper Continental Crust) concentration values were used as background values of heavy metals (McLennan, 2001).

3) Contamination Factor (CF)

The contamination factor (CF) was determined to quantify the contamination level of heavy metals in soil and sediment samples. The following equation was used for CF calculation (Rubio et al., 2000; Förstner et al., 1989; Hånkanson, 1980):

CF = (C_n/B_n) (5)

where “C_n” represents the concentration of metal n in sample and “B_n” represents the concentration of the same metal in the geochemical background.

4) Polymetallic Contamination Degree (CD)

An a priori evaluation of the polymetallic contamination level for each sample was made by the calculation of the contamination degree (CD) using the following formula (Hånkanson, 1980):

CD = SCF_i (6)

where “i” represents a specific heavy metal, and “CF_i” represents the contamination factor of the heavy metal i. Eight heavy metals were considered for the CD calculation.

5) Ecological risk index (Eri)

To evaluate the harmful impact of heavy metals on the environment and humans, the Ecological risk index (Eri) was calculated. This parameter reflects and combines the ecological sensitivity and the toxicity of the pollutants (Hånkanson, 1980; Suresh et al., 2012). The Eri is determined through the following equation:

Eri = Tr_i × CF_i (7)

where “CF_i” represents the contamination factor of a specific heavy metal i, and “Tr_i” represents the toxic-response factor for a given heavy metal (or the biological toxic factor) of a given heavy metal. The Tr_i values used were 40; 30; 10; 5; 5; 5; 2 and 1 for Hg; Cd; As; Co; Cu; Pb; Cr and Zn, respectively (Hånkanson, 1980; Islam et al., 2014).

6) The potential ecological risk index (RI)

The potential ecological risk index (RI) for polymetallic contamination was calculated by summing the single ecological risk index Eri. This RI index considers the synergy (combined effect) of several parameters, namely the toxic level, the concentration of heavy metals, and the ecological sensitivity of biological communities to heavy metals (Singh et al., 2002). The RI was computed using the following equation:

RI = ∑Eri (8)

where “Eri” represents the ecological risk index for a specific heavy metal i.

Spearman’s rank order correlation was performed using XLSTAT (New York, USA. https://www.xlstat.com) in order to explore the relationship among compounds and their potential sources.

The metal and metalloid (Sc, Cr, Co, Ni, Cu, Zn, As, Se, Cd, Sn, Sb, Pb, and Hg) concentrations in soil and sediment samples are shown in Table 1.

^aTotal variation coefficients for triplicate measurements are smaller than 5% for ICP-MS analysis. The recovery values from measurements for reference material (LKSD 4) were above 97.5% for all elements the ICP-MS triplicate. In bold, values of the heavy metal’s concentration high than the recommended probable effect concentration according to International Sediment Quality Guidelines for the Protection of Aquatic Life recommendation (CCME, 1999). ISQG—Interim freshwater sediment quality guidelines; PEL—probable effect levels (sediment dry weight). ^bFAO/WHO permissible level for toxic metals in fish (Joint FAO/WHO Expert Committee on Food Additives, 2000). ^cEU permissible level for toxic metals in fish (EU, 2005).

Heavy metal concentrations in the soil samples did not differ from concentrations in sediments from the N’djili River section by CECOMAF. These concentrations are significantly higher than concentrations determined in soil samples from a reference site, not receiving urban and farm wastewater discharges (Ngweme et al., 2020), thus giving a first indication that there is environmental contamination by heavy metals in the CECOMAF area (Table 1).

Regarding the relative concentrations in soil and sediments (mg·kg⁻¹ dw), in general heavy metals followed the order: Zn (910.6) > Pb (309.9) > Cu (303.7) > Cr (17.3) > Se (5.2) > As (3.7) > Co (1.8) > Ni (1.6) > Cd (1.5) > Hg (1.5) > Sb (0.5) > Sn (0.2) > Sc (0.1).

The concentrations of Cu, Zn, As, Cd, Pb and Hg in sediments were higher than the reference values set by the sediment quality guidelines (SQGs) for protection of aquatic life and their probable effect levels (CCME, 1999). These high concentrations of Cu, Zn, As, Cd, Pb and Hg in samples are attributed to anthropic sources around the site, such as untreated urban water runoff, runoff from farmlands, fertilizer applications, etc. (Tshibanda et al., 2014; Ngweme et al., 2020). Such level of contamination is likely to put a threat to the health of the aquatic system.

Table 1 indicates also the concentrations of Sc, Cr, Co, Ni, Cu, Zn, As, Se, Cd, Sn, Sb and Pb in biological matrices, the fish (Clarias gariepisnis) and the earthworm (Pontoscolex corethrurus). Comparing metal concentrations in the fish filet against the reference limit values set by both the Food and Agriculture Organization (FAO) and the European Union (EU), it may be noted that metals in fish did not exceed the concentration limits recommended by FAO, but exceed the most stringent values adopted by the EU for protection of the human health. Therefore, humans consuming regularly this fish are exposed to a relatively high intake of Cd and Pb.

The earthworm P. corethrurus displayed heavy metal concentrations generally higher than those of fish C. gariepisnis. High metal concentrations in earthworms indicate availability of these metals from agriculture soils (Dai et al., 2004). There is no similar data for earthworms from a reference site in the DRC to make a comparison.

The computed values for metals partitioning between the water and sediment phases (Kd) and metals bioconcentration in fish muscle (BCF) are shown in Table 1. Although Kd values have been reported for many metals in freshwater ecosystems, many values have not been validated in recent literature reviews and, therefore, comparisons of our results are limited to a few chemical elements (IAEA, 2010). The same is valid for BCF values in freshwater fish, and large variations in BCF values reported have been noticed (IAEA, 2010). Many determinations were made under the conditions of temperate environments and may not apply to tropical ecosystems, in particular due to the presence of organic matter in higher amounts in tropical aquatic systems and higher temperature, which may affect the distribution and bioaccumulation of metals. Nevertheless, for the metals with available data for comparison, the determinations made in this study are in the same order of magnitude or in the range of values reported in literature (Table 1). It must be noted that high Kd values (e.g., 10⁴ - 10⁵) correspond to strong sediment-water partitioning, i.e., low water solubility of the metal species with most of the metal bound to the sediment. Low Kd values (10⁰ - 10²) indicate high water solubility of the metal and a minor immobilization of the metal in sediments. The BCF values calculated were noticeably high for some metals and indicate that a significant metal transfer to humans may occur through fish consumption.

The calculated Igeo and EF parameter values are shown in Table 2. The results for both of them were interpreted according to classification scales developed in previous studies which are shown under Table 2 (Atibu et al., 2018, Thevenon et al., 2013). These scales allow for an easier ranking and interpretation of results.

The Igeo values ranged from −1.0 to 3.0 (Cu), 0.0 to 3.0 (Zn), 0.0 to 1.0 (As), 2.0 to 3.0 (Cd), 1.0 to 3.0 (Pb) and 0.0 to 4.0 for Hg. The Cr and Co Igeo values were less than −1.0 for all sampling sites, i.e., much lower than the average concentration of these two elements in the upper continental crust. This can be due

to high content of the samples in organic matter, deviating their composition from the average geochemical background. In general, samples were graded in class 0 and 1 (“practically unpolluted” and “unpolluted to moderately polluted”, respectively) for Cr, Co and As. Some samples were graded in classes 2 and 3 (“moderately polluted” and “moderately to heavily polluted”, respectively) namely for Cu, Zn, Cd, Pb, and Hg (Table 2), indicating that these elements were present in the agroecosystem in concentrations much higher than the average geochemical background.

For a specific heavy metal, the EF value can be used to identify whether site contamination has occurred and, because of the normalization introduced with the use of Sc, this parameter provides information that is more robust than Igeo. For 0.5 ≤ EF ≤ 1.5, the concentration of metal in samples is considered to acceptably match heavy metal natural concentrations in crustal sources. However, when EF ≥ 1.5 a metal enrichment has occurred and may be attributed to input of the heavy metal from anthropogenic sources to the study area (Feng et al., 2004; Zhang & Liu, 2002). In the study area, the EF values of all samples ranged between “no enrichment” (EF < 1) to “moderately severe enrichment” (EF 5-10) for Cu and Co, while for Cu, Zn, As, Cd, Pb and Hg the EF values ranged from “severe enrichment” (EF 10 - 25) to “extremely severe enrichment” (EF > 50) (Table 2).

The heavy metal EF values in samples followed this order: Hg (2946) > Pb (1519) > Cu (1336) > Cd (1235) > Zn (1151) > As (205) > Cr (11.9) > Co (6.6). According to these results, the anthropogenic activities did originate a clear enhancement of heavy metal concentrations in the studied site. The first five elements of this roll based on EF are the same spotted by Igeo in classes 2 and 3, and thus both parameters provide convergent information.

The results for potential ecological risk parameters, such as the Contamination Factor (CF), the Ecological Risk Factor (Eri), the Polymetallic Contamination Degree (CD) and the Ecological Risk Index (RI) are shown in Table 3.

According to the CF parameter, most samples displayed a very high contamination factor (6 < CF) for Cu, Zn Cd, Pb, and Hg. Soils and sediments displayed moderate contamination for As (1 < CF < 3). Low sediment contamination (CF < 1) was displayed only for Cr, Co (and some samples for Cu, as well).

Based on the parameter values for Polymetallic Contamination Degree (CD), all samples displayed either high or very high contamination (Table 3).

Based on the parameter values for Ecological Risk (Eri), the results indicated low or moderate risk for Cr, Co, Cu, Zn, As, and Pb, while for Cd and Hg the results indicated high, or very high ecological risk.

Parameter values for the Synergistic Effects index (RI) indicated very high ecological risk from all soil and sediment samples.

Previous analysis of heavy metals in the region included also areas non-significantly impacted by human activities that can be used for comparison with current results for the CECOMAF site (Ngweme et al., 2020). This comparison showed that metal concentrations in soil and sediments from CECOMAF were

one to 44 times higher than the reference soil labelled “CESCO” by Ngweme et al. (2020). Therefore, the overall results from the ecological indices for the CECOMAF site are in line with the relative level of metal concentrations in soils and sediments.

In view of these results, the populations living near the CECOMAF agroecosystem site, namely those who use the section of the N’djili river at CECOMAF site for domestic needs, crop irrigation, and consume fish from the river are exposed to heavy metals contamination.

The Spearman’s rank-order correlation values, which were calculated to investigate heavy metal possible sources and pathways in soils and sediments, are presented in the Table 4.

Correlation coefficients have been calculated using the log value of the parameter contents to normalize their distribution (n = 7, statistically significant coefficients, p < 0.05, are in bold).

In one hand, positive and significant correlation (p < 0.05) was recorded between some metal concentrations, such as Cr/As (r = 0.955), Cr/Se (r = 0.775), Cr/Cd (r = 0.649), Co/As (r = 0.864), Co/Se (r = 0.745), Ni/As (r = 0.927), Ni/Cd (r = 0.673), Zn/Pb (r = 0.607), As/Se (r = 0.700), As/Cd (r = 0.718) and, Sb/Hg (r = 0.876). The positive correlations observed among these metal pairs suggest that these metals could have originated from common sources and may have similar transport pathways. In the other hand, some negative and significant correlations (p < 0.05) were observed, namely between metal pairs Hg/Sc (r = −0.775), Hg/Cr (r = −0.685), Hg/Fe (r = −0.811), Hg/Co (r = −0.545) suggesting that Hg did not originate from the same source than other metals and may have followed different environmental pathways (Poté et al., 2008; Haller et al., 2009; Atibu et al., 2021).