Characterization of Synanthropic Habitats on Shallow Seabeds Using Map Clustering Techniques: A Case Study in Taranto, Apulia, Italy

Abstract

The Mar Piccolo is a transitional water system located in Taranto city (Southern Italy); it is a semi-enclosed basin affected by severe pollution issues due to the presence of various industrial, agricultural and other anthropic activities that require careful monitoring and management. The pollution levels reached over time have harmed marine biodiversity and human health, repeatedly requiring timely actions for its mitigation. Characterization methodologies and techniques today play a fundamental role in supporting the decision-making phase, processing large quantities of data and identifying complex patterns and correlations. An approach focused on gaining detailed knowledge of complex environmental contexts through clustering map techniques enables highly precise results, capturing even the smallest variations in the features of the study object and strongly correlating them with possible sources of pollution. The use of these techniques improves the precision of the analyses and can significantly contribute to improving the understanding of the environmental state in the Mar Piccolo area. This study addresses the issue of pollution in Mar Piccolo due to marine litter, which has led to the formation of synanthropic habitats on the seabed. It also highlights the value of clustering maps and other characterization techniques for achieving detailed insights at various levels of analysis. Data processing through the proposed methodology can generate very detailed mapping useful for planning precision reclamation interventions that also include species conservation actions, as well as a better understanding of how synanthropic habitats are distributed and evolve. In summary, this study demonstrates how it is possible to improve the precision of data processing, providing crucial details for the management and conservation of highly threatened marine ecosystems.

1. Introduction

Synanthropic habitats are environments heavily modified and influenced by human activity [1]. These habitats include urban, periurban and rural areas, like city parks, gardens, streets, buildings and industrial areas. Synanthropic habitats are often characterized by a complex interaction between natural and artificial elements; often these habitats host a variety of organisms, including animals, plants and microorganisms adapted to live in anthropogenic environments [2]. The study of synanthropic habitats is important for several reasons: first, these habitats have become a significant part of the landscape in many regions of the world and can host rare and urban-adapted species [3]. Furthermore, understanding how organisms interact with the anthropic environment can provide information on ecological resilience and species’ adaptation strategies to human presence. Synanthropic habitats can influence human health [4], since they can be sources of pollution and host disease vector species [5] and contribute to the quality of air and water in urban areas. Studying these habitats can therefore help mitigate negative impacts on public health and develop more sustainable urban planning practices and remediation strategies.

The use of data produced from environmental monitoring activities represents a precious source of information that allows us to understand the ecosystem’s evolution and changes over time, identifying trends [6,7], anomalies and emerging criticalities; the elaborated results offer a solid basis for making informed environmental management and conservation decisions.

The analysis of historical data can reveal the impacts of human activities on the environment over time, helping to evaluate the effectiveness of policies and corrective actions. Furthermore, it provides essential knowledge for predictive models and simulations, which are key to addressing complex environmental challenges.

In this context, the clustering map technique (CM) plays a vital role in processing large amounts of data. Clustering enables the extraction of meaningful insights from massive datasets in ways that would otherwise be impossible for humans; it is used to identify hidden patterns [8,9], trends and complex relationships among data, providing valuable insights to make informed decisions [10]. CM is fundamental in environmental management, as it permits real-time monitoring and prediction of changes in ecosystems [11,12], contributing to the analysis of climate data, the modeling of ecosystem services and the understanding of complex phenomena [13,14,15,16].

On the one hand, CM is widely used in several disciplines including statistics, biology, marketing and social sciences, demonstrating its versatility in multidisciplinary applications [17]. On the other hand, some limitations of the method are related to the choice of the number of clusters and the level of detail, which can influence the results, making the analysis sensitive to subjective decisions. The sensitivity of the scale can also affect the results, which is why it is necessary to standardize the data. The interpretation of clusters can also be subjective and depend on the analyst’s experience, leading to variable interpretative results; therefore, its application in areas sufficiently known by data analysts is recommended. Finally, the presence of noisy data or outliers can negatively affect the results; therefore, it is necessary to be able to recognize them and exclude them with appropriate reasons.

In this study, CM has been used as an automated classification tool (hierarchical clustering) to minimize the risk of subjective interpretation and human bias in data analysis.

The proposed CM application has been used in the Mar Piccolo area (Taranto city, Italy) as a data elaboration strategy to manage a big dataset originating from a multidisciplinary monitoring activity of marine litter and coastal wastes started in 2015 [18].

The objective is to segment large datasets into homogeneous groups of synanthropic habitats on the Mar Piccolo seabed, testing the applicability of CM techniques to handle extensive data and improve habitat management. This approach aims to ensure sustainable fishing practices, restore heavily degraded areas, and protect rare species by addressing the unique characteristics of each habitat.

2. Materials and Methods

2.1. Study Area

The Mar Piccolo of Taranto is a semi-enclosed basin characterized by a semi-lunar shape and its calm waters; it is separated from the Mar Grande by a thin isthmus and both areas are located on the Ionian coast of the Puglia region in Southern Italy. The Mar Piccolo extension is approximately 20 km² with varying seabed depths, from very shallow areas to deep depressions of over 10 up to 14 m. It is characterized by two different areas called I and II Seno which are smaller basins belonging to the transitional water bodies category: in the I Seno there are two submarine springs (Galeso and Citrello) and two intermittent rivers flowing into it. The study area for the application of the proposed methodology is the I Seno (Figure 1). The figure shows the main characteristics of the study area, the location of river mouths and springs and the names of the areas into which it is conventionally divided. The isobaths every 2.5 m of depth are also shown.

The Mar Piccolo represents a relatively small marine environment rich in biodiversity and therefore is classified as a protected area identified by Directive 92/43/CEE as code Site of Community Importance IT9130004; moreover, due to its strategic position, the area has an important role in the naval and commercial history of the city of Taranto [19].

Over the years, the Mar Piccolo of Taranto has attracted the interest of numerous studies, reflecting its scientific, historical and environmental relevance. Environmental research has investigated the biodiversity of transitional waters [20], highlighting on the one hand the presence of unique species and on the other the high impact of human activities on these fragile ecosystems [18,21,22,23].

Studies on the health of the Mar Piccolo have revealed environmental issues, including the presence of oil pollution [24] and heavy metals [25,26,27], which also require economic efforts for reclamation and ecological restoration.

Not only the Mar Piccolo but also the entire province of Taranto city is facing various environmental emergencies. One of the main ones is pollution due to the presence of an important industrial area. Emissions of pollutants, including hydrocarbons and heavy metals, have caused serious damage to the air [28,29], soil [30,31] and water resources [32,33]. The epidemiological study “Sentieri” [34] also highlighted how specific impacts caused by pollutant emissions, including hydrocarbons and heavy metals, on environmental resources and public health have led to an increase in respiratory diseases linked to air quality and an association between cancer incidence and residence near the industrial area of Taranto [35]. The Mar Piccolo is particularly impacted by uncontrolled discharges, which compromise the health of marine ecosystems and threaten food security linked to fishing activities [36].

This results in the loss of marine biodiversity and ecosystem services, as the balance of marine ecosystems is irreversibly disrupted by human activities [37], and it underscores the need for significant investment in environmental restoration, land [38,39] and water reclamation if polluting emissions are not reduced and sustainable practices are not promoted [40,41].

Financial resources have long been allocated to the recovery of the Mar Piccolo area and its protection from industrial pollution that has caused serious environmental impacts. Its ecological uniqueness and historical importance make it a critical area of interest for the conservation and sustainable management of marine resources requiring swift intervention. On one hand, sediment contamination with heavy metals [25] has led to bioaccumulation in marine species, threatening both biodiversity and human health; on the other, habitat loss for fish species [42] has resulted in ecosystem imbalances.

2.2. Data Acquisition

2.2.1. Marine Litter and Species Mapping

The submarine inspections, in the areas with the highest concentration of marine litter, were carried out in the period July–September 2015 by two expert diving biologists with underwater drones (ROV) [43], that recorded a continuous video following 10 m × 100 m shore-to-offshore transects and recording the data on depth with a depth gauge, while the type of marine litter and biological species were recorded by biologists on the support boat [41]. During each transect inspected, the seabed was video-recorded for all the paths allowing a detailed visual observation of marine litter identified and assessing the presence of biological species associated with such litter within a radius of 1–2 meters. The points of interest identified in each inspected transect were recorded in the geodatabase created in the WGS84/UTM33N reference system.

Among these species were considered those of Community Interest (e.g., Pinna nobilis Linnaeus, 1758, Geodia cydonium Linnaeus, 1767, Hippocampus hippocampus Linnaeus, 1758, H. guttulatus Cuvier, 1829, Cymodocea nodosa (Ucria) Ascherson, 1870). Moreover, the presence of turf-forming macroalgae and macro-zoobenthic community (filter- and suspension-feeders such as sponges, hydrozoans, polychaetes, bryozoans, molluscs, crinoids and tunicates) were also assessed. These data are critical for creating detailed maps of the seabed that can be used for a variety of purposes, including marine resource management, scientific research and remediation planning [41].

The typologies of marine litter and waste on river banks were noted using the code from the European Waste Catalogue (EWC) [44]. EWC codes are a series of numerical codes used in Europe to classify and identify specific types of waste. These codes serve to standardize the categorization of waste based on their origin, composition and danger. The main objective is to facilitate the management, monitoring and treatment of waste in an effective and ecologically sustainable way. The EWC codes are made up of six digits; the first two digits range from 00 to 20 and identify the production process and the industrial sector from which the waste or substance originates, and the second pair of digits ranging from 01 to 09 identifies the production activity with the specific process while the third pair of digits ranging from 01 to 99 indicates the substance contained within the waste. For our purposes, we stopped at the characterization of the first two pairs of digits. In Table S1, the EWC codes and descriptions of the waste are found.

2.2.2. Acquisition and Mapping of Other Parameters

In addition to the data acquired in the survey campaigns, further data referable to the following descriptive variables correlated to the distribution of synanthropic habitats were used:

(1) Seabed depth (DE) was obtained through the reconstruction of the bathymetry with Inverse Distance Weighted interpolation with 10 m resolution using depth isolines with a distance of 1 m acquired with single-beam sonar [40,45] and the recorded depth values recorded by divers at the survey points because the depth of the seabed plays a crucial role in influencing the distribution of marine biocenoses.

The following distance measurements were acquired through the use of QGIS software (v. 3.28, Firenze, Italy), having available various thematic maps coming from the regional information system or created in previous field activities.

(2) Distance from discharge channels of water treatment plants (D_P) because an excess of nutrients in treated wastewater, such as nitrogen and phosphorus, can cause marine eutrophication.

(3) Distance from the point of entry of water into the sea (D_U) because the surface runoff of water contaminated by waste that flows into the sea can cause serious environmental problems. All input points recorded in previous activities [45] were considered input data and were updated with the new inspections carried out in 2016 along the shores of the Mar Piccolo. These discharges can vary widely; they may be artificial or natural points of water confluence resulting from the terrain’s shape, asphalt slopes, or water drainage pipes.

(4) Distance from the coast (D_C) because human activities along the coast have a significant impact on marine biocenoses near the coast due to various factors. Dumping waste and pollutants into coastal waters can alter water quality and damage marine ecosystems.

(5) Distance from abandoned waste (small illegal landfills) along the coast (D_W) because this type of waste along the coast can generate devastating impacts on marine biocenoses. This waste, such as plastic, glass, metals, paints, and electrical components, can be transported into the sea by currents and directly harm marine life.

The results of 2016 field inspections mapping abandoned waste along the Mar Piccolo shores were used as input data. These inspections identified and quantified waste-affected areas in a GIS, with waste classified by EWC codes per Italian Legislative Decree 152/2006, noting origin (urban or special waste) and hazardous characteristics.

(6) Distance from freshwater coming from river mouths and submarine springs (D_FW) because these sources can influence the chemistry of the surrounding seawater, creating salinity and nutrient gradients that can encourage the growth of specific organisms. The underground freshwater sources were mapped through remote sensing activities carried out via aerial flights equipped with the single-band TABI 320 thermal sensor (8–12 um). The calculations were supported by temperature acquisitions on the surface of the water of the I Seno of the Mar Piccolo.

(7) Distance from channels for the exchange of salt water (D_SW) in the Mar Grande because they have significant effects on the biocoenoses of the seabed of the Mar Piccolo.

(8) Comparison with the results of modeling of the marine circulation and the marine climate of the Mar Piccolo sea through circulation simulation [46,47]. To simulate realistic conditions, we chose to reproduce the annual circulation generated by measured inputs, such as wind, tide, wave motion, thermohaline gradients and interface flows. The annual circulation was consulted to identify possible correlations with habitat distribution.

2.3. Data Elaboration

2.3.1. Pre-Processing

A geodatabase was created to store each record from the mapping activities, including coordinates, acquisition date and processed information on the variables detailed in Section 2.2.1 and Section 2.2.2.

Normalization and Encoding for EWC Codes and Biological Data

Clustering algorithms use the evaluation of distance between points in an n-dimensional space. To group the most similar records in a geodatabase, we are forced to evaluate how “close” these records are in terms of numerical differences between the values present in the various columns. The Euclidean distance between the various points is not calculated directly from the distance values. This is because the columns represent different quantities with different units of measurement, which could lead to an imbalanced calculation and incorrect evaluations. To avoid this possible imbalance, we first normalized the data, i.e., reduced the range of values within a certain interval, or predefined range, with the minimum-maximum method. All the values of the parameters linked to anthropic or environmental factors of the dataset columns now vary in the range 0–1, i.e., in the normalized range of distances where 0 is the minimum and 1 is the maximum.

Similarly, for categorical variables, such as the presence of a living organism or the type of marine litter at the mapping point, we had to proceed with a transformation of these categorical features into vectors of numbers to improve the comprehensibility of the data. For these variables, we recursively applied the one hot encoder technique [48] assigning the value 0 to presence (intended as maximum proximity and minimum distance) and the value 1 to absence.

2.3.2. Data Processing

This section contains the main data processing procedures. For our purposes, we utilized several data processing techniques. Hierarchical clustering allowed us to group similar synanthropic habitats on shallow and highly anthropized seabeds. The scatter matrix aided in visualizing relationships between variables, facilitating pattern recognition. Additionally, the Pareto chart helped prioritize factors influencing habitat distribution and composition, enhancing our understanding of ecological dynamics in these environments.

Hierarchical Clustering

Hierarchical clustering can be used to divide geographical areas into environmentally homogeneous regions, simplifying land-use planning and natural resource management. Producing a result that can be classified at various levels (each cluster identifies a habitat with characteristics different from those nearby) is well suited to supporting the decision-making process by providing visual insights through hierarchical trees.

We chose hierarchical clustering because unlike k-means clustering, in which the output is an assignment of a point to a specific cluster, hierarchical clustering returns a broader result, i.e., a hierarchical structure of concentric clusters that leads to a discussion that can consider all the different variables to be considered and, in any case, the environmental complexity of the study area. This method enables the exploration of nested relationships within the environmental data, making it particularly suited to our multi-level analysis without the need for rigid cluster definitions required by other methods. While in k-means we must indicate a priori the number of clusters to create (specifying the hyperparameter k) [13], through hierarchical clustering we have the possibility of evaluating the belonging of each element in various scenarios, by varying the number of clusters considered (levels of in-depth analysis), and interpreting the results according to the variables that most influence the creation of different numbers of clusters. Hierarchical clustering was performed with Scikit Learn [49] through Python scripts [50].

Scatter Matrix

The scatter matrix is used in the analysis of environmental data to examine the relationships between variables, thus explaining the result of a cluster analysis.

This analysis can reveal significant associations between variables, helping to identify the main drivers of environmental change and guide decisions on environmental management and conservation. The scatter matrix was created with the plotly graphing libraries [51].

Pareto Chart

The Pareto chart is a graph that represents the importance of the differences caused by a certain phenomenon. It is composed of a bar graph and a line graph, where each factor is represented by bars placed in decreasing order and the line instead represents a cumulative distribution (called the Lorenz curve). On the Cartesian plane, the values of the standard deviation of each normalized parameter of the geodatabase, for each cluster identified at various levels, are represented on the abscissa and the relative cumulative quantities on the ordinate. This type of graph is used to evaluate which are the greatest dispersion values of the variables that can influence the clustering results.

Representation of Data on the Geographic Information System

A GIS system allows you to integrate data from different sources, providing a holistic view of the environment and the relationships between variables. This data integration is essential for land-use planning, resource management and environmental impact assessment. QGIS software was used to represent the data [52].

2.4. Post-Processing

Final Post-Processing Mapping for Detail Refinement

This phase includes the procedure used to obtain a detailed map of the distribution of the synanthropic habitats of the seabed of the I Seno of the Mar Piccolo after having been classified in the various hierarchical clusters. The mapping is carried out through the creation of Voronoi polygons for the identification of homogeneous areas and more similar elements, and then with a smoothing procedure, through the application in GIS environment of the Chaikin’s corner-cutting algorithm [53], of the geometries to make them “softer” and similar to environmental contexts.

3. Results

3.1. Marine Litter Mapping

On the seabed of the Mar Piccolo, 1344 points with marine litter have been mapped in the 10 m × 100 m shore-to-offshore transepts. As expected in some areas, there is a greater concentration of marine litter, i.e., in the areas Vasto descent, Cariati dock, ex-Marigenimil pier (Figure 2, Figures S1 and S2, Table S2) but the presence of numerous biological species was also found (Figure 3). In total, it was possible to identify n. 52 species. In the area called “170 ha”, although we have many points mapped, we have little information regarding the species present both because the area is affected by maneuvers by Navy ships which can make the development of biocoenosis more difficult and because the submarine acquisitions were made without biologists. In this area, only the Pinna nobilis has been mapped.

3.2. Mapping of Waste on the River Banks

In the coastal areas close to the first Seno of the Mar Piccolo, no. 49 areas with an average extension of 256 m², a height of 0.7 m and an average volume of 187 m³ of waste per site have been found. These areas along the coast are represented by their centroid which is the element used to calculate distances. The types of waste predominantly present in the area are plastic waste resulting from fishing activities (EWC 0201), plastic packaging (EWC 1501), waste resulting from demolition and construction activities, including wood and glass (EWC 1701 and 1702) and unsorted municipal waste (EWC 2003). These product classes overall represent almost 90% of the estimated volume of abandoned waste present in the areas of the I Seno. The remaining approximately 10% is represented by end-of-life tires (EWC 1601), plastic of urban origin (EWC 2001), electrical and electronic waste and so-called Waste from Electrical and Electronic Equipment (WEEE, EWC 1602). Asbestos cement materials were also found with a certain frequency, some drums containing unidentifiable and presumably dangerous materials, glass wool and other drums containing oily material spilled on the ground (Figure 4).

The number of entries of water into the sea, following inspections carried out in the period from July 2015 to June 2016, is 180 of which include 87 ducts, 70 artificial channels, 5 natural channels, and 18 descents. In addition, there are three input points which are discharges from purifiers with regular authorization (Figure 5).

The saltwater inlets are known and are represented by the communication channels with the Mar Grande. They are no. 2 (Figure 1).

3.3. Cluster Analysis

The multi-level hierarchical clustering results are reported in Figure 6. The images represent a tree diagram showing how the mapped points are grouped during the hierarchical clustering process. The branches in the dendrogram represent the clusters formed at different levels of similarity. The horizontal lines in the dendrogram represent the points where clusters can be extracted based on the desired levels of similarity. These clusters represent groups of similar data within the dataset. Each cluster can be labeled or identified based on the data it contains or its common characteristics.

The hierarchical clustering obtained at level one of depth identifies four clusters, each made up of 271, 283, 215 and 565 nodes, respectively, whose main discriminating component is the salinity parameter. In this case, the size of the fourth cluster is slightly larger than the other three (about double). The first two clusters are more similar to each other than those of the third and fourth clusters; this can be deduced from the fact that the vertical distance lines are shorter (Figure 6a).

Considering the second level of depth, the clusters double and, with the total number of nodes remaining unchanged, dichotomous subdivisions are observed whose main discriminating component appears to be the parameter distance from coastal waste and distance from saltwater exchange channels (Figure 6b).

These aspects will be verifiable with the scatter matrix and the Pareto chart.

In the matrix shown in Figure 7, we find a scatter plot, obtained from the first level of analysis, for each combination of variables relating to the factors of anthropic and natural origin that can influence the presence and distribution of biological species and synanthropic habitats. Scatter plots along the diagonal of the matrix are not meaningful because they simply show the distribution of the variables with respect to themselves. We will use the scatter matrix in Figure 7 to graphically interpret the clustering results.

The figure should be read by observing the variation in the variables compared two by two with the cluster number, identified with a different color. For example, it is possible to observe how the parameters D_SW and D_FW vary together along a diagonal, while D_SW and D_U do not vary together and are very dispersed.

Comparing the scatter matrix with the cluster graph, we can observe that the nodes in the first two clusters are situated very close to the coast, where habitats are densely populated and salinity levels are elevated, regardless of depth, primarily in the southwestern area of the Mar Piccolo’s first Seno. The third and fourth clusters appear under even more diverse conditions, both in salinity levels and depth, frequently near freshwater sources. The main distinction between the third and fourth clusters is that the fourth leaf includes points mapped at greater depths and closer to freshwater springs or waters with lower salinity and a different type of marine litter (EWC 16) than those in the third cluster (EWC 17).

At the second level of detail, the scatter matrix identifies situations that can be defined as “ultra-detailed” and on which in-depth analysis may be necessary for the design of reclamation, environmental restoration and species protection plans (this could be the subject of future activities) (Figure S3).

The results reported in the form of Pareto charts can help to establish which are the major factors that influence a given phenomenon and are useful for interpreting the results obtained with cluster analysis. By exploiting these observations, it is possible to analyze a set of data to determine the few variables (among the many under examination) that significantly influence the final results of a given phenomenon (called Pareto analysis).

The diagram shows us that at the first level of in-depth analysis, the greatest variability is attributable to the degree of salinity determined mainly by the distance from the communication channels with the Mar Grande (D_SW), followed by the distance from coastal waste (D_W) and distance from the various entry points along the coast (D_U) (Figure 8a). These parameters alone explain about 65% of the synanthropic habitat diversity. Observing the Pareto chart relating to the variables’ EWC codes and biological species, we observe that the Lorenz curve is very close to the bisector indicating a certain equity in diversity even if we can observe a slightly greater influence of the variables’ presence of waste along the coast mainly attributable to EWC codes 0201, 1601 (Figure S4a). Following the same reasoning, from the point of view of biology, the main variability is recorded in the presence of Paracentrotus, Codium bursa and Antedon mediterranea. Considering the Pareto chart at the second level of detail (Figure 8b), it is possible to infer that the dichotomous subdivision of the clusters can be identified mainly in the variables, in order of importance, distance from coastal waste (D_W), distance from the communication channels with the Mar Grande (D_SW) and distance from the discharge channels of water treatment plants (D_P) for approximately 55% of the total variability. From a biological point of view, Codium bursa and Antedon mediterranea contribute more than the first level of analysis. A particular aspect to note at this classification level is the variability of waste with EWC 1704 which is much more decisive than the first level of classification in determining the formation of clusters. On the other hand, wastes with the EWC 2003 code identified mainly in the wrecks are not very variable in their distribution and consequently have little influence on the diversity of the clusters even if it is known that they favor the growth and development of biocoenoses (Figure S4b).

Below are the detailed maps at the level of depth “one” of some areas of the I Seno of Mar Piccolo reporting the results obtained with the applied methodology (Figure 9). The map shows the presence of different synanthropic habitats with a distribution that can be defined as patches. The map shows the very widespread presence of a habitat belonging to clusters n. 4 of Figure 6a; this is in line with the acquisitions made: the yellow areas are areas in which the mapping density is not as dense as other areas because, looking at Area 170 ha, only the presence of Pinna nobilis has been mapped (as mentioned before). Cluster n. 1 certainly identifies very specific situations, such as proximity to the coast and the presence of higher degrees of salinity, a condition that occurs in waters with limited marine currents (inside piers and more sheltered areas). Cluster no. 2 contains very diversified and densely spaced habitats at greater depths than cluster no. 1 and is made up of different types of waste. Cluster no. 3 identifies areas where synanthropic habitats are present at lower density due to the intense passage of boats and various human activities and whose salinity conditions are very varied.

The map at the second level of detail with which it is possible to verify the achievability of applying this methodology level of detail can be viewed in Figure S5.

3.4. Modeling of the Marine Circulation

The simulation of marine circulation on representative seasonal situations was conducted by the Polytechnic of Bari Department DICATECH [54]. The main observations useful for characterizing the area are the withdrawal of the steel industries’ water pump and booster flow, the cyclical character of the circulation due to the effect of cyclic tide and homogeneous wind, the flow inversion in the channels connecting Mar Piccolo and Mar Grande between high and low tide, the presence of stationary vortex structures on the I and II Seno, modest variations over time but decreasing intensity towards the bottom. The simulation indicates that the average annual surface circulation is mainly outgoing from the Mar Piccolo towards the Mar Grande and the Gulf of Taranto with very low average speeds (up to 20 cm/s). The circulation at the bottom indicates an exchange dynamic between the two water bodies of the Mar Piccolo with an average bottom flow entering from the Mar Grande to the I Seno and from the I to the II Seno. The average renewal time for the I Seno is estimated at 25 days.

4. Discussion

Producing detailed maps of synanthropic habitats on the Mar Piccolo seabed is crucial, especially considering the reduced costs associated with this methodology. Typically, seabed debris receives far less investigation due to the high costs and logistical challenges of sampling, making these detailed, cost-effective maps highly valuable [55].

Compared to other developed methods aiming to identify correlations between sources of marine litter [56,57], this methodology deepens the knowledge of the distribution of synanthropic habitats to individuate the best reclamation techniques applicable while respecting the biological species. This study responds to the expressed need for detailed knowledge on the presence of a relationship between the state of ecosystems and the environmental damage created [56]. It also seeks to meet remediation needs through sustainable practices, transforming this challenge into a true circular economy opportunity [58]. This approach minimizes harm to fragile synanthropic habitats and may even incorporate citizen science initiatives [59].

The spatial association between marine litter and living organisms enables precise risk analysis and the identification of appropriate management practices for specific areas or groups. In a GIS environment, selecting a species can instantly generate maps showing its presence, while choosing a specific type of marine litter highlights all areas requiring targeted reclamation. This approach also allows for the selection of synanthropic habitat clusters to implement focused remediation projects, providing a crucial foundation for developing intervention priority indices essential to risk analysis [60,61].

In a polluted environment such as the seabed of the Mar Piccolo of Taranto, knowing in detail the distribution of these synanthropic habitats is important to implement correct reclamation procedures by treating the areas with different approaches depending on the density, distribution [56] and natural factors [62] which influence the distribution of biological species.

Looking at data processing, based on the obtained results we can deduce that the synanthropic habitats are distributed mainly based on the salinity gradient conditioned by the exchange of water with the Mar Grande (especially in the southern canal where there is a swing bridge) and by the presence of underwater springs and river mouths in the northern area of the I Seno. This aspect indicates that despite the strong anthropic pressures present in the area, environmental conditions (degree of salinity) are still the main factor responsible for the distribution of habitats. Other aspects indicating the main differentiations in the distribution of synanthropic habitats include the diversity of marine litter present, water depth, circulation patterns affecting salinity levels, and the type and distribution of waste, particularly if generated by aquaculture and fishing (EWC 0201), end-of-life vehicles (EWC 1601), and metal waste (EWC 1704).

The results of the representative simulation of seasonal average conditions influencing the hydrodynamic structure of the Mar Piccolo through the Canal Grande indicate continuous exchanges of water flows that occur predominantly through the Canal Grande towards the Punta Pizzone channel, the two main connecting channels between Mar Grande and I Seno and I with II Seno. The other areas are somewhat less affected by these exchanges, with salinity levels primarily influenced by freshwater inflows from rivers and springs, water extraction from pumping stations, and the presence of anthropogenic structures such as small harbors and boat docks. For this reason, vortex motions are also created with salt gradients that vary slightly locally as they are not affected by exchanges with the Mar Grande.

Unfortunately, the natural conditions are overwhelmed by other aspects strongly deriving from the presence of man: distance from waste on the coasts whose leaching of contaminants due to runoff water creates further inputs of pollutants into the sea, especially if close to one of the various mapped entry points into the sea, proximity to the coast with all the resulting consequences such as pollution from fuel, detergents, food waste and others. Biological species that could be used as indicators of certain types of synanthropic habitats are Paracentrotus, Codium bursa and Antedon mediterranea; the species Pinna nobilis seems less indicative. Pinna nobilis is an endemic Mediterranean species protected by the Habitats Directive and the Barcelona Convention that in recent decades, in the northwestern Mediterranean, has suffered a decrease in its distribution area and the densities of individuals due to water pollution which affects the larval stages of the species and collection for ornamental purposes. Furthermore, forms of illegal fishing with grazing gear lead to the destruction of this species [63]. These aspects raise the possibility that this species could be considered a good indicator of the environmental state in the study area. In addition, this study was carried out when the Pinna nobilis population was in full vegetative growth; therefore, the distribution map of Pinna nobilis (Figure S6) is now a useful tool for researchers to evaluate the distribution and population dynamics of the species, also using predictive analysis based on machine learning models [64].

Some limitations can be deduced from the fact that an excessively detailed analysis would lead to increasingly complicating the cognitive framework; therefore, we must be careful in identifying the level of in-depth analysis useful according to the objective we want to achieve. In addition, the data used are not very recent; however, they remain highly relevant for understanding the environmental conditions of the Mar Piccolo, providing valuable insights for identifying pollution sources and serving as a fundamental baseline for future research efforts.

Compared to other studies that used mainly a multibeam echosounder and side-scan sonar devices [18], the primary distinction lies in the level of mapping detail achievable and the amount of meaningful information that can be conveyed within a single map.

Seabed maps obtained with this methodology are crucial for understanding anthropogenic impacts on the marine environment [65], like pollution [66] and coastal erosion dynamics, essential factors for the sustainable management of marine resources and mitigating the negative effects of human activities.

Similar phenomena are also seen in other areas. A recent assessment of seabed waste, carried out in 2022, reports the results of the state of the marine environment in the northeast Atlantic [67]. The results indicate a widespread presence of waste on all the seabeds examined, with a particular prevalence of fishing-related and plastic items. Some monitored areas such as the Bay of Biscay and the Iberian coast present a greater probability of waste accumulation than other areas [68]. The reasons for this distribution could be attributable to various factors, including greater and continuous contributions of litter of anthropic origin with flows coming from rivers, winds and marine currents. This study also shows that floating waste and marine litter follow different paths and accumulate in different areas. In our study areas, it is conceivable that part of the waste present on the coasts is linked to this phenomenon.

If synanthropic habitats in the Mar Piccolo remain undisturbed in the future, we can expect an increase in biodiversity adapted to urban life given that the main factors of growth and development of biocenosis are still conditioned by salinity gradients. Species that manage to thrive in anthropized environments could develop further adaptations that favor their survival and reproduction in urban contexts. This could lead to the expansion of new populations of species that have become generalist and opportunistic over time, as often happens for some species of birds and insects. Synanthropic habitats could become real refuges for species that struggle to survive in undisturbed natural environments. On the other hand, if the contributions of disturbing factors were to continue over time, there could be negative consequences for some native species and the biodiversity of the study area, as invasive and exotic species could establish themselves and be successful.

5. Conclusions

This study reports the methodology adopted, based on hierarchical clustering applied to registered associations of biological species and marine litter, to create a detailed map of synanthropic habitats present on the seabed of the Mar Piccolo of Taranto. Through the application of this methodology, it was possible to identify the main determinants of the distribution of synanthropic habitats, to evaluate the impacts of the presence of marine litter and to identify best ecosystem management practices. In highly disturbed environments such as the one studied, variations in salinity are still the major determinants of the distribution of habitats which remain strongly threatened by other purely anthropogenic factors, such as distance and type of waste.

This indicates that the conditions for a total recovery of the marine ecosystems of the Mar Piccolo still exist as long as we act quickly to remove all the disturbing factors present, especially those that cause contamination, unless Pinna nobilis is giving us an early warning alarm.

The relationship between types of marine litter and marine organisms is complex and requires a holistic approach to environmental management. It is essential to promote sustainable waste disposal practices, educate the population on the importance of reduction and recycling and implement environmental policies that protect the health of ecosystems.

This could also be one of the objectives set by the European Nature Restoration Law which requires the restoration of degraded habitats by 20% by 2030 and by 2050 by 100%.

The results and method of this study can be applied in other similar environmental contexts, such as Mediterranean coastal areas or other marine areas subject to strong anthropic pressure.