Quantifying the contribution of sediment sources upstream of Anzali wetland in north Iran using the fingerprinting technique

Widespread sediment-related issues have been observed in many parts of the world, even in those parts with low soil erosion and sediment loads. Fine sediments are the main factor in controlling the transportation, deposition and release of nutrients and pollutants such as insecticides, heavy metals and sediment-interconnected organic pollutants (Ikeda et al. 2009; Walling & Collins 2016). All these pollutants drastically decrease the quality of aquatic habitats (Han et al. 2020; Ebrahimi et al. 2022a, b). These issues highlight a wide range of environmental and ecological importance of fine sediments and the need to consider effective sediment control strategies in watershed management programs (Koiter et al. 2013a). In watershed management programs, it is necessary to identify the major eroding areas accurately and, thus, the primary and dominant sources of sediment production. In many cases, expensive conservation strategies have been implemented to prevent soil erosion in areas that have not been prone to sediment production (Collins et al. 2017b).

Various methods are available for determining sediment sources, including visual evaluation, monitoring with profile meters and erosion pins and erosion plot measurements that simulate different land uses (James et al. 2010; Clapcott et al. 2011; Higa 2024; Mostafazadeh et al. 2024). However, these traditional methods often encounter spatial and temporal sampling limitations and high costs. As an alternative, sediment fingerprinting analyzes physical, chemical and isotopic properties to identify sediment sources such as agricultural fields or riverbanks. While sediment fingerprinting has made progress, it still struggles to provide a universal response due to the heterogeneity of field data. Nevertheless, by examining the geochemical and physical characteristics of sediment samples reaching designated watershed points, it overcomes challenges faced by conventional techniques (Koiter et al. 2013b; Collins et al. 2017a; Xu et al. 2022). Although it may not directly quantify sediment yield or pinpoint sources, when combined with water and sediment data, it provides quantitative insights on sediment loads from various sources across watersheds. Initially developed for agricultural catchments, sediment fingerprinting has been successfully applied in diverse landscapes, informing watershed restoration efforts and resource allocation strategies for maximum impact (Smith & Blake 2014; Cashman et al. 2018).

The sediment fingerprinting method, which utilizes various soil attributes as tracers, accurately identifies sediment sources within watersheds (Manjoro et al. 2016; Pulley et al. 2016; Huangfu et al. 2020; Vale et al. 2020). Recent studies have also focused on sediment production and numerical modeling in basins and reservoirs (Sarkar et al. 2022; Wei et al. 2022). Progress in sediment sourcing using the fingerprinting technique is ongoing, with recommendations for future critical research topics, including comprehensive tracer examination across different environments and scales (Collins et al. 2020). Gaspar et al. (2019) reviewed the sensitivity of multivariate mixing models, emphasizing the importance of artificial mixtures in evaluating model precision when considering tracer choices, source combinations and particle size fractions.

The Anzali international wetland, located in northern Iran along the Caspian Sea, is a critical natural habitat. It is home to diverse flora and fauna, including endangered species, and supports migratory birds, fish and aquatic plants. The wetland plays a crucial role in water purification, flood control and local climate regulation. It is vital to local livelihoods, providing fish and other resources and supporting traditional fishing practices. The Anzali wetland also attracts tourists and offers ecotourism opportunities. In addition, it has cultural and historical significance and is an important site for scientific research. Despite its importance, it faces threats such as pollution, land reclamation and invasive species. Conservation efforts are essential to maintain its ecological integrity. Construction, mining and improper land use practices increase soil erosion, resulting in higher sediment loads entering the wetland (Asadi 2016; Ebrahimi et al. 2022a, b, 2024). Effective watershed management practices such as reforestation, sustainable agriculture and erosion control measures are essential to reduce sediment production and its impact on the wetland. This study presents an innovative approach to the challenges of sediment source fingerprinting in the complex, mountainous Anzali wetland watershed, highlighting its novelty in addressing the complexities of diverse land use systems. By identifying primary sediment sources, the research provides a critical foundation for developing effective soil conservation strategies across the watershed. A key innovation lies in the identification of optimal fingerprint attributes that allow for more accurate, efficient and cost-effective identification of sediment sources in similarly sensitive wetland environments. This study not only advances the methodology of sediment fingerprinting by analyzing a wide range of tracers but also underscores the central role of wetlands in sustainable environmental management. The results provide groundbreaking insights for future conservation efforts in fragile ecosystems such as the Anzali wetland, paving the way for more targeted and effective soil and water resource management practices. Overall, the study identifies erosion-prone areas and sediment hotspots upstream of the Anzali wetland while recognizing the significant physical, chemical and biological characteristics in the artificial mixture for sediment source identification.

The Anzali wetland, which encompasses an area of 200 km² on the southern coast of the Caspian Sea, is renowned for its diverse ecosystem and has been designated under the Ramsar Convention since 1975. Despite its ecological importance, the wetland is facing challenges, including the influx of sediment and the discharge of wastewater. Nine principal rivers discharge into the wetland, with the Pasikhan River exerting a considerable influence. The Pasikhan River watershed, which encompasses an area of 671 km², comprises mountainous uplands and flat lowlands. These areas experience elevated levels of sediment transport, particularly from the Mobarakabad sub-watershed (Ebrahimi et al. 2022a, b, 2024).

The sampling method was designed in accordance with the specific study objectives, the characteristics of the study area and the available resources. These procedures were implemented in order to ensure consistency and comparability with other research findings. Samples were collected from a variety of potential source locations, including areas that were suspected to contribute to pond contamination and the Anzali wetland. The locations were identified based on prior knowledge of potential pollution sources, including industrial sites, agricultural areas and urban developments. The samples were collected at regular intervals to capture any temporal variations in the composition and characteristics of the soil and sediment. To maintain consistency with previous studies, the same sampling techniques and equipment were used as Walling (2013) and Collins et al. (2017a) described. This methodology ensured that the collected samples could be directly compared with those from other studies, thereby facilitating a more comprehensive understanding of the pollution sources and their impacts on the pond and reservoir. In general, adherence to established procedures for sample collection is of paramount importance in scientific research, as it facilitates accurate comparisons between different studies and enhances the reliability of the findings.

A total of 93 samples were collected for this study, comprising 14 sediment samples (watershed outlet) and 79 soil samples from sediment-producing sources (Figure 1). The number of soil samples collected from each potential sediment source was determined by the overall spatial uniformity of that source, which was assessed through field surveys, evidence of soil erosion and aerial imagery. The number of soil samples collected from each sediment source was based on the overall spatial uniformity of that source. The sediment sources and their corresponding number of collected samples were as follows: forests (eight samples), paddy fields (seven samples), tea gardens (five samples), river walls (seven samples), degraded rangelands (23 samples), undisturbed rangelands (11 samples) and gullies (18 samples). Due to the presence of evidence of soil erosion in the forest and the uniformity of the cover, only a limited number of samples were collected from the forest area. Samples were collected from the surface soil (paddy fields, tea gardens, undisturbed rangelands, degraded rangelands and forests) at a depth of 0–2 cm, as well as from the subsurface soil of eroding riverbanks and gullies, which comprised multiple layers. Soil sampling from eroded areas commenced with a thorough cleaning of the soil surface to remove any contaminants that could compromise the integrity of the samples. Following cleaning, plant residues were carefully cleared to access the undisturbed surface layer of the soil. Soil samples were then systematically collected to a depth of 2 cm using an aluminum shovel. The selection of aluminum shovels was justified by their inert nature, durability, ease of cleaning and capacity to maintain sample purity and integrity throughout the sampling process. Sampling from river walls and gullies involved identifying and delineating distinct soil layers based on variations in color, root depth and morphological features observed within the watershed. Samples were meticulously collected from these stratified layers, ensuring that sections disturbed by previous activities or natural processes were excluded.

A sampling depth of 0–2 cm was chosen for surface soils because this depth is commonly used in sediment fingerprinting to capture the physical, chemical and biological properties most relevant to surface soil erosion and sediment transport, particularly in agricultural and rangeland areas where land use practices such as tillage, grazing and cultivation influence sediment movement and deposition. According to Collins et al. (1997a) and Owens et al. (1999), surface soils are most susceptible to erosion, making them a critical zone for understanding sediment dynamics. These soils also contain the highest concentrations of fine particles, organic matter and other key tracers essential for distinguishing sediment sources. In areas of more pronounced erosion, such as riverbanks and gullies, stratified sampling was used to account for different sediment characteristics at different soil depths, allowing for a better understanding of sediment dynamics and the long-term effects of erosion processes (Haddadchi et al. 2013; Li et al. 2020). To ensure sample integrity, aluminum shovels were used, as their inertness, durability and ease of cleaning prevent contamination and ensure the accuracy of sediment fingerprinting (Pulley & Collins 2018). Special attention was also given to flood-deposited sediments, as these sediments play a critical role in sediment transport dynamics. Flood events redistribute large amounts of sediment, and capturing these materials is key to understanding how sediment is mobilized during high-flow events and contributes to sediment budgets (Gellis et al. 2017). The combination of surface and stratified sampling, the use of specialized tools and a focus on flood-deposited sediments provides a reliable methodology for sediment fingerprinting, enabling detailed analysis of sediment sources and transport processes that are critical for developing effective soil and water conservation strategies in erosion-prone areas.

This meticulous approach was designed to preserve sample integrity and provide an accurate representation of sediment sources in the study area. Sediment sampling focused on collecting both suspended sediments and sedimented particles. This included retrieving sediment deposits found behind rubble and debris deposited during flood events. Special attention was paid to sediment accumulation within depressions formed by floodwaters. This methodological detail is crucial for understanding sediment dynamics and the impact of flood events on sediment transport and deposition processes.

The hydrometric method was employed to ascertain the particle size distribution of particles smaller than 2 mm. Hydrometer readings were conducted at 30-s intervals and at 1, 3, 10, 30, 60, 90, 120, 300, 600, 1,440 and 2,880 min. Following the hydrometer readings, the cylinder contents were poured into a 53 μm sieve in order to separate the sand particles from the clay and silt. Then, particles were categorized with respect to their diagonal size: very fine sand (VFS) (0.05–0.1 mm), fine sand (FS) (0.1–0.25 mm), medium sand (MS) (0.25–0.5 mm), coarse sand (CS) (0.5–1 mm) and very coarse sand (VCS) (1–2 mm) (Gee & Dani 2002). The sand sample is then subjected to a drying process and subsequently sieved using a set of sieves with a range of mesh openings. The sieves are arranged in descending order of mesh size, with the finest sieve positioned at the base. The sand sample is then placed on the top sieve and shaken for a specific duration, allowing the particles to settle into their respective fractions. Subsequently, the material retained on each sieve is collected and weighed, and then the weight of each fraction is determined as a percentage of the total weight of the sample. This methodology allows for the determination of the proportion of sand particles within specific size ranges. The organic carbon content (OC) was measured using the Walkley and Black method (Walkley & Black 1934). The electrical conductivity (in 5:1 water–soil suspension) was measured through the Rhodes approach (Rhodes 1996), and soil acidity was determined by using the Thomas method (1996). Tabatabaei's (Tabatabaei et al. 1994) method was used to measure phosphatase enzyme in samples. Available phosphorus was measured by the Olsen method at 880 nm with a spectrophotometer (Olsen 1954). The total amount of Cu, Zn, Ni, Cr, As, Pb, Cd, Hg, Al, Ba, Ca, Co, Fe, Na, Mn, Mg, Li and K in soil and sediment samples were measured using inductively coupled plasma mass spectrometry (ICP-MS) (US EPA 2007) at the soil and sediment extracts by HCl:HNO₃, 3:1, v/v.

In the first step, the outlier data were identified and removed using the outlier code in Excel software. To remove outliers from the calculation of the first and third quartiles, the outliers are identified with the help of conditional formatting. The main reason for removing outlier data in sediment fingerprinting methods is to ensure the accuracy and reliability of the results. Outliers are data points exhibiting significant deviation from the expected pattern or trend of the dataset. These outliers can be caused by various factors such as measurement errors, sampling biases or extreme events.

Additionally, removing outliers helps to improve statistical analyses and modeling techniques used in sediment fingerprinting. Outliers can disproportionately impact statistical calculations, leading to biased estimates and incorrect interpretations. By removing them, researchers can obtain more robust and reliable results. The outlier's data was removed using Excel 2016 software. The interquartile range (IQR) method was employed to eliminate outlier data. The IQR was computed by determining the difference between the third quartile (Q3) and the first quartile (Q1). Subsequently, data points below Q1 – 1.5 × IQR or above Q3 + 1.5 × IQR were categorized as outliers. In the second step, the Kolmogorov–Smirnov approach was used to test the normality of variables. Parameters that were not normally distributed (such as clay, fractal dimensions and phosphatase) were normalized using normal functions.

In the third step, multicollinearity among the variables was examined, and the aligned parameters were opted out of the process (Hair et al. 1998). After that, the significance of the parameters was examined using the Wilks' lambda statistical approach, and the parameters that were not significant were removed. Both multicollinearity and Wilks' lambda are important statistical approaches because they help ensure the validity and reliability of statistical analyses. Multicollinearity allows us to identify potential issues with predictor variables, while Wilks' lambda helps determine if there are significant differences between groups on multiple dependent variables. In the fourth step, a stepwise method was employed for discriminant analysis, resulting in the creation of various models (Collins et al. 1997a). In stepwise regression, all measured variables are entered into the model, and the one with a negligibly small effect on the output is removed from the model (Collins et al. 2017b).

In this study, discriminant analysis served as the cornerstone of stepwise diagnostic analysis, providing a systematic approach to tracer selection and sediment source differentiation. The process began with an initial screening of all potential tracers to ensure that they met the assumptions of statistical analysis, including normality (via Shapiro–Wilk tests) and homogeneity of variances (via Levene's test). This step was critical to avoid bias in subsequent analyses. Tracers that did not meet these criteria were excluded to maintain the robustness of the dataset. The remaining tracers underwent a stepwise inclusion or exclusion procedure based on their Wilks' lambda values, a measure of how well each tracer discriminated between sediment sources. Lower Wilks' lambda values indicated higher discriminatory power, and tracers that contributed most to source separation were retained. This iterative process ensured that the final set of tracers was both statistically optimal and scientifically relevant. Once the tracer set was finalized, canonical discriminant analysis (CDA) was used to develop six canonical functions. These functions reduced the multidimensional tracer dataset to a smaller number of variables that maximized variance between sediment sources while minimizing variability within sources. The canonical functions provided a clear separation of sediment sources in multidimensional space, making it easier to assign sediment samples to their respective origins. To validate the robustness of the model, classification accuracy was tested by reassigning sediment samples to their sources based on the canonical functions. High classification accuracy confirmed the reliability of the selected tracers and demonstrated the effectiveness of the CDA. Finally, the canonical functions were integrated into a multivariate mixing model that quantified the proportional contributions of each sediment source. The model used linear optimization techniques to minimize residuals between observed and predicted tracer values, ensuring consistency with field observations.

The diagnostic analysis is an approach to test the hypothesis that the group average for a set of independent variables is equal for two or more groups. To achieve this, the diagnostic analysis multiplies each independent variable by its corresponding weight and then sums them up. The result is a composite diagnostic Z-score for each observation in the analysis. The group average is calculated by averaging the diagnostic scores of all observations in each group (sediment production sources). This group average is referred to as the center of gravity. The significance of diagnostic functions can be quantified through the application of various statistical methodologies. In this study, the Wilks’ lambda statistical approach was employed.

In this equation, is the estimated value for ith properties (i = 1, 2, …, m), a_ij is the mean value of ith properties in the jth sediment source, b is the contribution of the sediment source, n is the number of sediment sources and m is the number of sediment source properties.

The above equation should be solved for each source property. Therefore, the multivariate tracing model will be generated with a set of equations with an array dimension of the number of properties. The contribution of each different source in sediment production can be calculated by solving this system of equations. Two conditions in the following should be met to solve the equations (Collins et al. 2001):

These equations are also called objective functions. The optimal solution for the contribution of sediment sources is obtained by opting and minimizing one of the above equations through iterations of trial and error. To do so, different values are selected for the contribution of sediment sources, b_j, and the value of the objective function (R or E) is calculated. This process continues until the minimum value for the objective functions is found. Then, the corresponding values for the source's contribution, b_j, are selected as the optimal solution. This process is repeated for all sediment samples, and then, the average of sediment source contributions is calculated. Finally, the RI of the source was determined by dividing the contribution of each sediment source by the percentage of its area. In this study, both R and E criteria were used, and the Solver tool in Excel software was used for optimization. Multiple alignment and selection of detectors (diagnostic analysis) were also investigated using SPSS 22 software through the stepwise method.

Table 3 presents the results of the group average equality tests for all independent variables. Wilks' lambda is the first statistical test employed to ascertain the ratio between the sum of the intragroup squares and the sum of the total squares. The statistical analysis reveals the proportion of variance in the combination of dependent variables (diagnostic scores) that is not explained by the differences among the groups. Consequently, the lower the value of Wilks' lambda, the greater the model's explanatory power. The value of Wilks' lambda is constrained between zero and one. Values approaching zero indicate a difference between the group means, whereas values approaching one indicate equality of these means. The results indicated that the Wilks' lambda range for the parameters under study was between 0.11 and 0.86. The Wilks' lambda values are inversely proportional to the F statistics. It was demonstrated that all parameters at this stage were meaningful, with the exception of Na, which will be addressed in subsequent steps.

Supplementary Table S1 represents the result of the diagnostic analysis. As can be seen from the consecutive steps, phosphatase was initially employed in the model. Then, OC, VFS, Mg, VCS, Hg and Cd were fed to the model step by step.

Supplementary Table S2 shows the result for canonical discriminant functions. The results show that implementing the discriminant analysis led to the identification of six canonical discriminant functions. The variance percentage in different functions shows the highest variation was in the first function, 66 and 16.1% of the total variance in the second function. Considering the 95% of total variance covered by the first three functions, it can be interpreted that they can distinguish the sources of sediments.

The results indicate that a favorable combination of chemical, physical and biological parameters of soil has been selected as tracers in the study. As previously stated, the phosphatase enzyme has been incorporated into the model as the initial variable, which underscores the significance of this parameter in differentiating the sediment sources. One particularly useful phosphatase enzyme for the purpose of sediment fingerprinting is alkaline phosphatase. Alkaline phosphatase is an enzyme that catalyzes the hydrolysis of phosphate esters in an alkaline environment. Alkaline phosphatase is a common enzyme found in a variety of organisms, including bacteria, fungi, plants and animals. In sediment fingerprinting studies, alkaline phosphatase can be employed to evaluate the presence and activity of microorganisms. The enzyme is capable of down organic phosphorus compounds present in the sediment into inorganic phosphate ions, which can then be measured using spectrophotometric methods. By analyzing alkaline phosphatase activity in sediment samples from different locations or time points, researchers can gain insights into microbial activity and nutrient cycling processes. This information can assist in the identification of sources of sediment pollution, track changes in environmental conditions and the evaluation of the impact of human activities on aquatic ecosystems. Overall, alkaline phosphatase is an essential enzyme for fingerprinting sediment as it provides valuable information about microbial communities and their role in sediment biogeochemistry. Enzymes are produced by microbial activity, which is sensitive to temporal changes. However, several studies have shown that enzyme activity is often stable in the soil environment (Tabatabai & Dick 2002; Tate 2002). Furthermore, the adsorption of enzymes onto clay particles or organic and inorganic materials affects the mobility and stability of the enzyme (Tate 2002). It affects their separation and transport by water in the erosion process so that organic matter in soil protects enzymes from microbial degradation. Soil enzymes play a pivotal role in sediment fingerprinting, as they reflect microbial activity and root interactions, which are crucial indicators of sediment sources (Nosrati et al. 2011, 2012). In sediment fingerprinting studies, phosphatase enzymes have been identified as significant tracers due to their sensitivity to environmental changes and their role in soil nutrient cycling (Nannipieri et al. 2011; Eivazi et al. 2018; Janes-Bassett et al. 2021; Yao et al. 2022). Numerous studies have shown that enzymatic activity is significantly correlated with organic matter in topsoil and soil profiles (Chaer et al. 2009). Therefore, enzymes are located in a network of organic–mineral complexes (Tabatabai & Dick 2002) to maintain their stability for a long time. The transport of enzymes is associated with compounds of fine soil particles, such as clay, which change less in concentration during the erosion process. This process has been proven to have almost the same effect on the source of sediment and sedimentation area. Fox & Papanicolaou (2008) concluded that the biochemical detectors could also be a tool in sediment source tracking.

The use of soil particle size distribution as a tracer enhances the precision of sediment fingerprinting models. Particle size data provides valuable information about the characteristics of sediment sources and their contributions to the sediment mixture. The particle size distributions of different sediment sources are distinct due to their geological origins, weathering processes and land use histories. The incorporation of particle size data into fingerprinting models enhances the ability to discriminate between sources. The integration of particle size distribution with other source characteristics enhances the complexity and accuracy of the model, thereby facilitating a more comprehensive source characterization. Previous studies by Chapman et al. (2001) and Gaspar et al. (2019) have shown that incorporating particle size data is effective in identifying sediment sources and improving model efficiency. This study further emphasizes the importance of including particle size distribution in sediment fingerprinting for reliable source identification. The size of sediment particles plays a significant role in their transportation and deposition. For instance, the study on the drained shear behavior of dense fluvial sand indicates that particle size can influence the shear strength of the sediment, which in turn affects its stability and mobility (Deng et al. 2021). Similarly, the characteristics of sediment particle size distribution in the Pisha sandstone area show that there is a selectivity phenomenon in the transportation process, with coarse sediment particles being transported preferentially from slopes to gullies (Zhang et al. 2023). This selectivity is influenced by the effects of vegetation and complex erosion, which can significantly impact sediment production and transportation. The shape of sediment particles also affects their transportation. The effects of particle size–shape correlations on the steady shear strength of granular materials demonstrate that particle shape, particularly elongation, can exert a significant influence on shear strength (Carrasco et al. 2022). This study emphasizes that the correlation between particle size and shape due to geological formation and weathering processes can impact the connectivity and orientation of particles, influencing load transmission and, consequently, the transportation of sediments. The size of particles in the soil is a significant factor in the distribution of geochemical elements throughout the soil matrix. The elemental compositions and distribution patterns of larger particles tend to differ from those of smaller particles. The variation in particle size affects the movement and retention of elements, including metals, nutrients and organic compounds, within the soil environment (Gaspar et al. 2022).

The tolerance number is 1 − R², in which R is the multiple correlation coefficient between the current variable and those previously added. The small values for the tolerance number indicate the negligible effect of the parameter on the analysis. As can be inferred from the data in Table 3, the tolerance numbers for studied parameters were high, indicating the significant role of each in distinguishing the sediment resources.

OC plays a crucial role in sediment fingerprinting studies due to its distinct properties and function as a tracer for sediment sources. The content of OC varies among sources such as soils, vegetation and forests, allowing for differentiation between potential contributors to sediment mixtures. Its stability remains after transportation, aiding in reliable source identification. Unique chemical signatures in OC are influenced by factors such as vegetation type, land use and soil properties. Changes in land use can impact OC content, affecting sediment source vulnerability and export potential. Monitoring fluctuations in OC can provide valuable information for land management and erosion control. Incorporating OC into fingerprinting models can improve the accuracy of source identification and determination of mixing proportions. Previous studies by Chen et al. (2016), Pulley & Collins (2018) and Du et al. (2023) support the suitability of OC as a tracer for sediment origin, confirming the findings of this study. It is observed that the most important sources of sediment production in the studied watershed are gullies and river walls. These two sources account for 97.17% of the RI of producing sediment. The incorporation of organic matter and OC is a fundamental aspect of sediment fingerprinting models. These elements serve as effective tracers, influence particle size distribution, provide insight into biogeochemical processes, and contribute to the identification of sediment sources and the dynamics of sediment transport. It is of paramount importance to consider the significance of organic matter and OC in order to ensure the reliability of sediment source attribution and to facilitate an understanding of the fate and transport of sediment-associated contaminants in aquatic ecosystems (Nhan et al. 2021; Wiltshire et al. 2021, 2022).

Quaternary deposits are prevalent along the principal riverbanks within the basin, as illustrated on the geological map. These deposits are susceptible to lateral mass movements and shallow landslides caused by river scour. The erosion of river channels results in the transport of a considerable quantity of sediment into the mainstream. Observation of the river margins has revealed numerous instances of slope movement. Gullies are identified as erosion ‘hotspots’, or areas experiencing accelerated erosion due to the confluence of several factors, including steep slopes and concentrated flow. Liu et al. (2016) demonstrated the effectiveness of the discriminant analysis method in identifying sediment sources within the Bull Creek basin, highlighting the significant contribution of silt to sediment generation. Riverbanks, vulnerable to erosion from flowing water and hydraulic forces, play a critical role in sediment production, as supported by previous studies (Owens et al. 2000; De Rose et al. 2005; Wilkinson et al. 2005; Kronvang et al. 2013; Lu et al. 2015).

Gullies and riverbanks are important sources of sediment in the watershed under study, followed by tea gardens, primarily because of their location on steep slopes, which facilitate erosion and the transport of particles during rainfall. Krishnarajah (1985) and Wang et al. (2023) have highlighted the increased erosion in tea gardens compared with coconut gardens and forests, which is attributed to the steeper slopes. In contrast, forests experience minimal erosion due to canopy protection and high soil permeability. Degraded rangelands produce more sediment than undisturbed ones, as surface erosion is worsened by excessive livestock grazing (Bayat et al. 2017). While forests and undisturbed rangelands have lower sediment yields due to protective canopy and soil permeability, active gullies and areas without vegetation show significant erosion, supporting the findings of sediment fingerprinting studies.

The area under study is subject to significant flooding during the winter and autumn months, due to the combination of steep terrain and heavy rainfall. This results in damage to riverbanks and the deposition of sediment. Floodwaters can saturate the soil, leading to slope instability and riverbank collapses. Li et al. (2020) found that forests minimally contribute to sediment, which is consistent with the findings of this study. Collins et al. (1997a) evaluated sediment sources in UK watersheds, highlighting the dominance of surface erosion over rangeland erosion. Collins et al. (2001) used the quantitative multiparameter fingerprinting technique in the Kalia Watershed in southern Zambia to estimate the RI of potential sediment sources. During the study period (1997–1999), the weighted average of relative contribution at the watershed outlet was as follows: industrial agriculture (2%), grassland (17.1%), canal walls and ditches (17.2%) and traditional agriculture (63.7%).

Collins et al. (1997a, b) investigated sediment sources in UK watersheds using composite fingerprinting techniques. Their research highlighted the importance of particle size distribution, OC and phosphorus content as key tracers for identifying sediment sources. The study identified surface erosion from agricultural land as a significant contributor to sediment loads, while bank erosion varied in importance depending on land use. Similar to this study, Collins et al. (1997a, b) identified OC and particle size as critical input parameters. However, the current study extends this methodology by incorporating biological indicators such as phosphatase enzyme activity, providing a more comprehensive understanding of sediment sources, particularly the significant role of stream banks (69.17%) and gullies (12.14%). Pulley & Collins (2018) developed an open source tool, SIFT (Sediment Fingerprinting Tool), to facilitate the fingerprinting of sediment sources. Their study included geochemical tracers such as heavy metals (e.g., cadmium, mercury) and particle size distribution, and found that degraded riparian zones, particularly riverbanks, contributed over 50% of the sediment load. The results are consistent with the conclusion of this study that stream banks are the dominant sediment source. However, while Pulley & Collins (2018) focus primarily on geochemical tracers, this study uniquely integrates biological parameters and highlights the second-largest contributor, gullies (12.14%), underscoring the importance of this type of erosion in sediment dynamics. Gaspar et al. (2019) evaluated the sensitivity of multivariate sediment fingerprinting models using artificial sediment mixtures. Their research relied on particle size fractions, geochemical markers and elemental concentrations, including aluminum and iron, to distinguish sediment sources. While Gaspar et al. (2019) effectively highlighted particle size as a reliable indicator, this study advances this approach by introducing a stepwise diagnostic analysis that optimizes parameter selection. In addition, the inclusion of phosphatase enzyme activity in the current study fills a gap in the methodology of Gaspar et al. (2019) and improves the accuracy of sediment source differentiation. Haddadchi et al. (2013) reviewed sediment source discrimination methods in fluvial systems and emphasized the need to integrate physical, chemical and biological tracers. Their research highlighted the limitations of relying solely on chemical tracers and advocated the incorporation of biological indicators to improve source differentiation. This study closely follows these recommendations by combining particle size, elemental concentrations and biological tracers such as enzyme activity. In addition, the use of diagnostic analysis in this study enhances its ability to identify stream banks and gullies as dominant sediment sources, providing actionable insights for erosion control measures. Fox & Papanicolaou (2008) used biochemical tracers such as nitrogen stable isotopes and organic matter content to track sediment sources at the watershed scale. Their results demonstrated the value of biochemical markers in distinguishing sediment sources in complex watersheds. This study builds on this by incorporating phosphatase activity, a novel biochemical tracer, into the analysis. While Fox & Papanicolaou (2008) focused on biochemical tracers in general, this study's emphasis on this specific enzyme provides a more detailed understanding of sediment sources, identifying stream banks as the primary contributor and gullies as the second-largest source. Kumar et al. (2024) highlighted the significant impact of human activities on riverbank erosion, identifying factors such as land use change, urban expansion and resource extraction as critical contributors to the destabilization of river systems. They emphasized that sustainable management practices, including vegetation restoration, regulated resource use and strategic planning, are essential to mitigate erosion and enhance riparian stability. These findings are consistent with this study, which examines the identification of sediment sources and highlights the dominant role of anthropogenic factors, such as deforestation and agricultural practices, in sediment production. Both studies emphasize the need for integrated management approaches to address anthropogenic erosion and ensure the sustainability of watershed ecosystems. Basha et al. (2024) used the InVEST model to study the effects of land use change on annual water yield in the Upper Ganga Basin, India. They found that urbanization, deforestation and agricultural expansion have disrupted the hydrology of the region, increasing surface runoff and reducing water sustainability. Similar to the present study on sediment dynamics in the Anzali wetland watershed, this research highlights the critical role of land use management in mitigating human-induced pressures on ecosystems. Gupta et al. (2023) reported that fine sediment intrusion poses significant challenges to river ecosystems, affecting both physical and biological processes. The review highlighted that excessive fine sediment deposition reduces habitat quality by clogging river beds, limiting oxygen exchange and disrupting the life cycles of aquatic organisms, particularly fish and macroinvertebrates. Anthropogenic activities such as deforestation, agriculture and urbanization have been identified as major drivers of sediment accumulation. Gupta et al. (2023) emphasized the need for integrated management approaches, including riparian vegetation restoration, erosion control and sustainable land use practices, to mitigate sediment intrusion and ensure the long-term health and resilience of river ecosystems.

This study employed sediment fingerprinting in the Mobarakabad watershed, situated in northern Iran, through the analysis of a range of physical, chemical and biological soil properties. The results indicated that these properties were effective in distinguishing different sediment sources. Specifically, the phosphatase enzyme, reflecting microbial and root interactions, and organic matter, resistant to degradation, played a pivotal role in differentiating sediment sources. The incorporation of diverse sand fractions, particularly those comprising very fine and coarse grains, enhanced the precision of sediment source identification. The study identified river walls and gullies as significant contributors to sediment in the watershed, underscoring the need for effective sediment management strategies. The erosion of these areas has a substantial impact on sediment production, which in turn affects downstream processes such as alterations to channel morphology, reductions in water quality, degradation of habitats and increases in flood risk. In contrast, forests, agricultural lands and rangelands exhibited a relatively minor contribution to sediment production. To mitigate erosion and sediment production in river walls and gullies, the study recommends the implementation of protective measures, including concrete barriers, riprap stone covers, vegetation planting and wooden check dams. Additionally, gabions have been identified as an effective method for the protection of rivers and gullies. The implementation of these measures can facilitate the promotion of sustainable land use practices and enhance the overall health of the watershed. The strength of this study is its use of a simple, effective model for identifying changes and understanding sedimentation. Complex models can be impractical for management. Clear presentation of results aids comprehension and avoids watershed issues. Despite minimal contribution from agricultural lands, their large area makes them significant. Human activities such as deforestation and land conversion have worsened erosion, making agriculture a key factor in sediment production after riverside and gully erosion. This study in the Mobarakabad watershed in northern Iran used sediment fingerprinting to analyze a wide range of soil properties. It identified key factors such as the phosphatase enzyme and organic matter in distinguishing sediment sources, highlighting the role of river walls and gullies in sediment production.

Sedimentation in the Anzali wetland is a critical driver of its degradation, with far-reaching impacts on the region's hydrology, water quality and biodiversity. The large volume of sediment transported by upstream soil erosion has progressively reduced the wetland's water storage capacity, filling its water beds and altering its hydrological balance. This process disrupts natural water flow patterns, reduces the depth of the wetland and creates stagnant zones prone to further ecological degradation. Sediments entering wetlands often carry chemical pollutants, heavy metals and nutrients such as nitrogen and phosphorus. These substances degrade water quality and accelerate eutrophication, which leads to excessive growth of aquatic plants and invasive algae. This phenomenon not only destroys habitats for native aquatic species but also depletes dissolved oxygen levels, threatening the survival of sensitive species. As a result, invasive species replace native species, drastically altering the biodiversity composition of the wetland. These changes in biodiversity are not limited to aquatic species but extend to wetland-dependent wildlife, including migratory birds, which suffer from habitat loss and reduced food availability. In addition, the hydrological disruptions caused by sedimentation impair essential ecosystem services such as flood regulation, maintenance of the hydrological cycle and natural water purification.

A combination of structural, vegetative and management practices can be used to reduce soil erosion in stream banks and gullies. Structural measures include the construction of riprap, gabions or retaining walls to stabilize stream banks and reduce the impact of water flow. In channels, check dams, silt traps and terracing can slow water movement, reduce sediment transport and promote sediment deposition. Vegetative approaches involve planting deep-rooted vegetation such as grasses, shrubs and trees along stream banks and within gullies to bind soil and provide a natural barrier against erosion. Riparian buffers of native plants can further stabilize soils while filtering sediments. Management practices include controlled grazing, reduced logging and no-till farming near these areas to minimize soil disturbance. The combination of these strategies provides effective erosion control, improves soil stability and promotes sustainable land use. The study's simple, effective model enhances understanding of sedimentation and underscores agriculture's role in sediment production. Effective management of river walls and gullies could regulate sediment entering the Anzali wetland, benefiting similar watersheds in the area. The management of river walls and gullies can effectively regulate the entry of sediment into the Anzali wetland, thereby facilitating its regeneration. The findings of this study are applicable to other watersheds within the Anzali wetland catchment area.

This study was funded by the Iran National Science Foundation (ISNF, Grant No. 96003440).

The authors have considered the subject of plagiarism, and this article is without a problem.

E.E. and H.A.: Conceived of the presented idea. E.E., H.A. and M.R.: Developed the theoretical framework. E.E., H.A. and E.A.: Developed the theory and performed the computations. H.A., E.E. and E.A.: Verified the analytical methods. E.E. and M.R.: Carried out the experiments. All authors discussed the results and contributed to the final manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.