Discrimination of Heavy Metal Sources in the Sedrud Delta Coastal Lagoons, Caspian Sea, N Iran: A Statistical Approach

28 Amirkola (more than 500 years in age), Kiashahr and Zibakenar (a few decades in age) lagoons 29 are located on the Sefidrud Delta, the southern coast of the Caspian Sea. Evaluating the 30 pollution degree caused by heavy metals (including Cu, Cr, Co, Ni, Pb, Zn and V), 106 31 sediment samples and three sediment cores were taken from studied lagoons. Three indices, 32 including geo-accumulation index (I geo ), contamination factor (CF), and pollution load index 33 (PLI), were employed to determine the contamination degree in the lagoons. Based on 34 contamination indices, the Kiashahr and Zibakenar lagoons show significant to moderate 35 contamination with Co, Cu, Pb, Zn and V. Nonparametric statistical analysis (Two-step cluster 36 analysis, analysis of variance, and T-test technique) was used to discriminate the pollution 37 sources. Statistical methods indicated the unique interpretation of contaminants sources. There 38 is a significant difference in metal concentrations between Amirkola and two younger lagoons. 39 Despite the same geogenic origin of sediments in the Sefidrud Delta and lagoons deposits, Co, 40 Zn, and Pb show anthropogenic sources in the newly-formed Kiashahr and Zibakenar lagoons.


45
River deltas and their related coastal lagoons are influenced mainly by human activities (e.g.

97
The Kiashahr Lagoon is located in the eastern part of the delta and northeast of Kiashahr. The 98 Zibakenar Lagoon is located in the western part of the delta and the north of Zibakenar city 99 (Fig. 1). The water depth in these lagoons is generally less than 2 m and is controlled by 100 freshwater entering through precipitation and irrigation. 101

102
A total of 106 surface sediment samples (Amirkola: 56, Zibakenar: 22 and Kiashahr: 28 103 samples) and three core samples (one core sample for each lagoon) were collected from all 104 three lagoons (Fig. 1). Surface samples were taken by Van Veen Grab in a regular network and 105 packed in polyethylene bags. The core samples were taken by PVC pipe and hand auger. Two 106 ends of the tubes were sealed after taking core samples so that their contents remain intact.

107
Samples were sent to Bu-Ali Sina University and Geological Survey of Iran (GSI) laboratories 108 for various physicochemical and chemical analyses.

119
To determine the pH of samples, 2 g of dry samples were mixed with 10 ml of deionized water 120 (1:5 sediment to water ratio) and stirred for 2 hours (Segura et al., 2006). The mixture was then 121 let to settle, and the pH was measured from the clear liquid above the sediments. The titration 122 method was carried out to determine the samples' calcium carbonate content (Carver, 1971).

123
This process was repeated three times to ensure the results. The organic matter amount in 124 sediments was determined by the burning method, for which 1 g of the air-dried sample was      Table 2. Based on granulometry results, the texture of most sediment samples in Amirkola

189
Lagoon is fine to medium grain. The sediment samples were frequently classified as slightly 190 sandy mud to slightly muddy sand (Fig. 2). Some gravelly samples in this lagoon were mostly 191 close to the lagoon's barrier and sea-side. Most samples from Zibakenar and Kiashahr lagoons 192 have fine grain texture, so more than 55% of samples were classified as mud, sandy mud, 193 slightly gravely mud, and slightly gravely sandy mud based on Folk (1980). Currently, none of 194 these lagoons have major water inlets.

278
According to the mean PLI values, the samples taken from all three lagoons are unpolluted 279 with studied metals. As is evident in Table 5, the highest PLI value in the Amirkola Lagoon is    Table 6. indicating the exact behaviour of these elements.

328
In the Zibakenar Lagoon, the relationship between elements changes (Table 6)

385
If Bonferroni p is greater than one, the value of Bonferroni p is considered one. 386

387
The results of the independent samples t-test are presented in Table 9. There are significant             Results of two-step cluster analysis, a) Silhouette measure shows values close to the maximum amount, and cluster sizes including two distinct clusters; b) features from the most to the least predictor (variable) importance.