3.1 Dry and wet season organic matter content (OMC) in sediments from Msimbazi River
Dry and wet season % OMC varied from 0.05 to 4.55, mean 0.93 and from 0.01 to 9.39, mean 1.63 respectively. The slightly higher % OMC during the wet season could have been due to heavy rains which lead to erosion and deposition of organic matter from human habitats to the river (Fig. 2A and B). A similar study by (Carreira et al. 2015) reported the influence of the Paraíba do Sul River in Brazil, as a source and distributor of organic matter in sediments. They found high organic matter content in river sediments during the wet season. Although the OMC in this study was slightly higher during the wet season, ANOVA showed that there was no significant difference for the average OMC content in the dry and wet season (p=0.58, α = 0.05). Since the analysis was performed at two different depths i.e. the upper and lower, it became imperative to correlate the two sampling levels. In the dry season, the % OMC ranged from 0.05 to 2.15 with a mean of 0.84 for the upper level sediments while it ranged from 0.05 to 4.55 with a mean of 1.033 for lower level sediments. In the wet season, the % OMC ranged from 0.03 to 8.16, with a mean 1.50 for upper level sediments and ranged from 0.01 to 9.39, a mean of 1.75 for lower level sediments. Although there was a slight difference between the (A) % OMC in the upper and lower level sediments, ANOVA showed that there was no significant difference in both the dry and the wet seasons (p=0.77 and 0.88, α = 0.05). This comfirmed that both the upper and lower organic matter originated from the same source and justified performing PCDD/Fs analysis for a homogenous mixture of the upper and lower sediments.
Downstream closer to the Indian Ocean in samples M6, M7, wM6 and wM7, the % OMCs were higher compared to up-stream samples in both the dry and the wet season. This could be due to the deposition and accumulation of organic matter from all tributaries. The same results were reported in the Paraíba do Sul River in Brazil study (Carreira et al. 2015). Presence of organic matter in sediments of the Msimbazi river, was indicative of the presence of other organic pollutants, which justified the analysis of PCDD/Fs in the same samples.
3.2 Dry and wet season PCDD/Fs concentration in sediments
The concentrations of 17 congeners of PCDD/Fs in sediments during the dry and the wet seasons ranged from 2.0 to 393.0 and 0.7 to 654.8 pg/g dry weight respectively as seen in Fig. 4 and 5. Since this was the first PCDD/Fs concentration study in Tanzanian rivers, comparison was made with results from studies elsewhere. The results were compared with the United States of America (USA) action level set at 1000 pg/g for PCDD/Fs and all were lower.
From Figures 3 and 4, congener concentrations varied between the sampling sites which indicated that the PCDD/Fs in the samples, originated from different sources. In the sample M0 collected upstream before discharged from suspected point sources, several congeners of PCDD/Fs were detected with PCDFs being predominant over PCDDs. This meant that the ratio of PCDDF to PCDD was greater than one and implied that their sources were mainly from open burning. This observation has been observed and reported by Everaerte and Baeyens (2002).
In both the dry and the wet season, regardless of the sampling point, the dominant congeners were 1,2,3,4,6,7,8-HepCDD, 1,2,3,4,6,7,8-HepCDF, 1,2,4,7,8-PeCDD, OctaCDD and TCDF. High concentrations of highly chlorinated-substituted PCDD/Fs such as Octa and Hepta signify atmospheric deposition as the major source of PCDD/F. Several studies have also reported the release of PCDD/Fs especially highly chlorinated substituted congeners from engines propelled by diesel (Chang et al. 2014; Laroo et al. 2011). It is also possible that, the detected PCDD/Fs were deposited in the Msimbazi river from diesel engine vehicles from garages, car wash stations and also high traffic roads such as Morogoro, Kawawa, SamNujoma and Mandela roads near or along the river or its tributaries at different points.
During the wet season, in addition to wM0, two more sites (wM6 and wM7) had their PCDDFs concentrations higher than PCDDs. This suggests atmospheric deposition as a mojor source of PCDD/Fs. The concentration of PCDD/Fs at the sampling site close to the Indian Ocean (wM7) in the wet season was higher than during dry season (M7). Since this was the lowest point of the river, this could have been due to sediments-bound PCDD/Fs which had eroded from the upper to the lower steam of the river. Movement of sediment bound -pollutants from upper to lower steams have been reported in the literature (Dinç et al. 2021;Verta et al. 2009; Weber et al. 2008). Concentration of PCDD/Fs in samples collected at Jangwani (M5) were relatively higher than those from other sampling points. This could be in addition to accumulation of sediments-bound PCDD/Fs from other streams such as Ubungo and Luhanga, heavy traffic along Morogoro road, car wash stations, parking slots and garages.
3.3 Toxicity potential of PCDD/Fs at different sampling points along Msimbazi river
TEQs indicate the magnitude of toxicity of PCDD/Fs congeners relative to 2,3,7,8-TCDD which is the most toxic congener assigned a Toxic Equivalent Factor (TEF) of 1. TEQs were calculated by summation of the products of the concentration of individual congeners at every sampling point to toxic equivalent factor (TEF) established by the World Health Organization (WHO)(WHO 2005). The total PCDD/Fs concentrations varied widely from one point to another along the river and its tributaries. In the dry season, the TEQ varied from 19.7 to 36.5, mean 27 pg TEQ WHO2005/g and in the wet season it varied from 2.0 to 38.72 mean 20.70 pgTEQ WHO2005 g−1 as seen in Figure 5A and B.
During the dry season, the lowest TEQ value was found up-stream of the river (M0). This was the point before the Msimbazi River passed through the suspected point sources of contamination. Furthermore, 86% of the TEQ in this site was contributed by 1,2,3,7,8-PeCDD. It was speculated that the PeCDD could have precipitated due to fair deposition since this was the point that was assumed not to be affected by point sources of PCDD/Fs. This observation corroborated with other studies which reported the existence of PeCDD in sediments as a result of air depositions (Sundqvist et al. 2009; Verta et al. 2007). The highest TEQ was found in Luhanga river, (M2), followed by M1 which is the main Msimbazi river at Vingunguti. All these points had their TEQs mainly contributed to, by low chlorinated PCDD/Fs such as PeCDD and TCDF. The elevated TEQ values in Luhanga river could be attributed to Nida textile factory which is located near the river. This also corroborates well with other studies that have reported the release of PCDD/Fs to the environment from textile industries (Huang et al. 2021; Klasmeier and McLachlan, 1998; Križanec and Majcen Le Marechal, 2006).
During the wet season, the lowest TEQ was estimated at Sinza river sampling site wM6 followed by sampling site wM7i.e. the last point before the river poured into the Indian Ocean. At this point, there was mixing of water from the ocean and the river especially during the wet season. The low TEQ of PCDD/Fs was attributed to wash out of sediment bound PCDD/Fs to the Ocean. The highest TEQ was estimated at Jangwani, sampling site wM5, a point where Ubungo and Luhanga merge with the main Msimbazi river downstream. The elevated TEQ at this point could be due to movement of sediment bound PCDD/Fs from up-stream and from other tributaries. At this point, there was high traffic and parking slots for the rapid bus transport.
3.4 Seasonal variation of TEQ of PCDD/Fs in sediments from Msimbazi river
Concentrations of PCDD/Fs were lower during the dry season compared to where they were detected during the wet season as seen in Figure 6.
During the dry season the water in the river had receded, and it was speculated that the sediments-bound PCDD/Fs were deposited at the riverbed. During the wet season, there was a big variation of PCDD/Fs from one sampling point to another. This could be attributed to the movement of sediment bound PCDD/Fs due to the rain water flow responsible for washing out sediments bound PCDDFs from one point to another.
However, ANOVA showed that there was no significant difference between PCDD/Fs TEQ values during the dry and the wet season, (P=0.08), α = 0.05). Notable was the detection of PCDD/Fs upstream of the Msimbazi River (M0), before the suspected PCDD/Fs point sources. This suggested that PCDD/Fs in Msimbazi river were not only from point sources such as industrial effluent but also atmospheric deposition as a result of open burning of wastes and biomass that are common practices in many places in Dar es Salaam.
Additionally, data was subjected to partial least squares-discriminant analysis (PLS-DA) using PLS_Toolbox – version 9.0 for MATLAB Release 23592, 2021Copyright (C) 1995 - 2021 Eigenvector Research, Inc. (PLS Toolbox 9.0, 2021). MATLAB R2021b for academic use was used in this instance. This was done in order to see if the two classes i.e. dry and wet seasons could be distinguished. The PLS-DA model was built using training data and exhibited a receiver operating characteristic (ROC) curve of sensitivity in the wet season concentrations of PCDD/Fs above a threshold of 0.8 with the area under the curve (AUC) 0f 0.9688 while in the dry season, the values were above 0.8 and 0.9531 respectively. These model parameters were used to predict test data from which the confusion matrix showed an overall classification error of 12.5% which was the same at 12.5% for the prediction results. The variance was significant only up to approximately 3 latent variables as deduced from the X variance captured (%) versus the latent variable. The scores plot of latent variable (LV) 1 with variance 22.9% versus latent variable (LV) 3 with variance 11.96% is shown in Figure 7.
As seen in Figure 7, two classes were clearly distinguished for the sediment PCDD/Fs concentrations in the dry and wet seasons. Further, a biplot i.e. scores plot with latent variable 1 (LV1) carrying the same variance at 22.9% and LV3 carrying 11.96% superimposed on the loadings plot is shown in Figure 8. From Figure 8, the wet season showed higher concentrations of PCDD/Fs which were closely correlated. The observation underscores the fact that there was movement of sediment from the sampling points where PCDD/Fs were detected in high concentrations during the dry season, to the sampling points where they were detected in higher concentrations during the wet season.
The Variable Importance in Projection (VIP) scores versus latent variables showed the following as the most important variables: (i) 1,2,3,7,8-PeCDF (ii) 1,2,3,4,7,8-HexCDF (iii) 2,3,4,7,8-PeCDF (iv) 1,2,4,7,8-PeCDD and (v) 1,2,3,6,7,8-HexCDD, 1,2,3,4,6,7,8-HepCDD, OctaCDD and 1,2,3,7,8,9-HexCDF. These PCDD/Fs contributed to the classification observed in the scores, loadings plots and qualified the observation made in Figures 3 and 4 where the dominant analytes were a subset of these latent variables.
In this study, all the sediments collected along Msimbazi river exceeded the Interim Sediment Quality Guideline (ISQG) in Canada and USA for dioxins quoted 2,3,7,7-TCDD toxicity equivalents which are 0.85 and 2.5 ng/kg (dw) respectively. There is a possibility that the sediments could cause detrimental effects to sensitive organisms at all trophic levels (Canadian Council of Ministers of the Environment, 2002; Cook 1993). Fifty percent (50%) of dry and wet season sediments presented in this study exceeded the American guideline specified as a high risk to sensitive species 25 ng TEQ/kg dw for mammalian and wildlife. Except for M0, wM1, Wm4, Wm6,and wM7 all other stations for both wet and dry seasons had their TEQs exceed the upper limit of 20 pg WHO- dw set by the European commission for Sediments.
PCDD/Fs reported in this study were lower than the levels reported in River Nile in Egypt. PCDD/Fs in the river Nile were suggested to be high due to commercial formulation of DDT in which PCDD/Fs are found as impurities (El-Kady et al. 2007). This could be the reason why their TEQs were very high. PCDD/Fs reported in this study were relatively higher than the TEQs reported in South Africa, but their sources were closely related (open burning of wood and domestic wastes small contribution from industrial activities (Nieuwoudt et al. 2009). The River Elbe, in Central Europe was reported to have higher PCDD/Fs concentrations compared to the levels in Msimbazi river reported in this study (Götz et al., 2017).
3.5 PCDD/Fs in catfish collected from Msimbazi river
Concentrations of PCDD/Fs congeners detected in fish samples are presented as log10concetrations for clarity in Figure 9. Out of 17 congeners, four namely 1,2,4,7,8-PeCDD, 1,2,3,4,6,7,8,-HepCDD and 1,2,3,4,,7,8,9-HepCDF were most dominant. Concentrations of these congeners were also high in sediments analyzed from Msimbazi river, indicating a common source of PCDD/Fs in sediments and in fish.
PCDD/Fs in fish collected from Msimbazi river presented in WH02005 fresh weight ranged from 9.3 to 145.2, mean 61.2 pg WHO2005-TEQg−1as seen in Figure 10.
The TEQs estimated in fish were very high compared to values reported in sediments. This could be due to long time accumulation of PCDD/Fs in fish fat tissue compared to accumulation in sediments. The mean concentration of PCDD/Fs was higher than the European Commission regulation value (3.5 pg WHO-TEQ pg/g) fw for muscle meat of fish and fishery products, (European Commission 2011) suggesting a health risk for consumers of fish from Msimbazi river. 2,3,7,8-TCDD was detected in all fish samples ranging from 3.5 to 12.7 with a mean of 8.1 pg/g. The minimum risk level for hazardous substances established by the Agency for Toxic Substance and Diseases Registry prescribes a minimum residue level (MRL) of 0.0002 ugkg−1day−1 of 2,3,7,8-TCDD to be eaten on average by a person of 70 kg (Chou et al.1998). With a mean value of 8.1 pg/g TCDD obtained in this study, even less than 1g of fish per person per day would still exceed the MRL. This surfaces a probable high risk of consuming catfish from Msimbazi river. About 50% of fish samples had their PCDFs concentrations higher than PCDDs, so the ratio of PCDF to PCDD was higher than 1. This suggested that the PCDD/Fs, which accumulated in fish, could have been from open burning of waste and biomass.
Results obtained from this study were higher than values obtained in lake whitefish from Canadian Great lakes (22-54 TEQ g−1) (Bhavsar et al. 2008) and those reported in fish from Lake Victoria in East Africa i.e. 0.001 to 0.09 WHO-TEQ pg/g (Ssebugere et al., 2013)