**The Impact of Human Interventions and Changes in Climate on the Hydro-Chemical Composition of Techirghiol Lake (Romania)**

#### **Carmen Maftei <sup>1</sup> , Constantin Buta <sup>2</sup> and Ionela Carazeanu Popovici 3,\***


Received: 18 May 2020; Accepted: 21 July 2020; Published: 12 August 2020

**Abstract:** The aim of this study is to establish the potential effect of changes in climate and anthropic interventions made over time on the hydro-chemical properties of the Techirghiol Lake. Located in the littoral region of the Black Sea, Techirghiol Lake is the most hypersaline lake of Romania—well-known for the therapeutic properties of the saline water and sapropelic mud. Long-term time series of salinity and water level were investigated in relation to the lake water inputs (precipitation, overland flow and groundwater), to chemical parameters (pH, DO and BOD5) and also to the hydraulic works designed and built in the region. The obtained results reveal a degradation of this ecosystem in the period of 1970–1998, when the extensive irrigation practice in the proximity of the lake had a negative effect on the water budget of Techirghiol Lake (an increased freshwater input through runoff and seepage), followed by a major decrease of the lake's salinity.

**Keywords:** human intervention; changes in climate; salty lake

### **1. Introduction**

Climate change has a considerable impact on ecosystems, affecting air temperature, the amount of precipitation, the frequency and intensity of extreme events, the sea level, etc. In the past decades, many studies have been conducted on saline lakes which show not only the importance of saline lakes in the economies, but also the impact of climate change on the water level and chemical content [1–3]. The study conducted by Valero-Garces et al. [4] on the saline lakes from Spain highlights the influences of agricultural practices, particularly of irrigation, on the lake's hydrological behavior. Webster et al. [5] have examined the influence of the increasing trend of drought on semiconservative cations, Ca<sup>+</sup> and Mg+, in seven lakes from Northern Wisconsin, and have concluded that the high evaporation rates, combined with the decreased amount of precipitation, caused an increase of cation concentrations in all lakes. Recent studies conducted in Poland concerning the influence of many factors (climatic, hydrologic, morphometric) on lake temperatures have shown that the lake response to factor modifications depends on the local conditions and lake characteristics [6,7]. All over the world, the studies conducted had the same conclusion: the saline lakes are threatened by climate change and by the various anthropogenic activities, which lead to dramatic changes in lakes chemistry and dynamics [1,8–10]. The analysis of water chemistry in relation to environmental factors allows a better understanding of the process variability and is very useful for researchers and deciders in the field of water management and monitoring [7,11].

This study presents an analysis of the potential effect of changes in climate and anthropic interventions on the hydro-chemical properties of Techirghiol Lake, located in the littoral region of the Black Sea. We have combined historical knowledge of human activities and management of the lake and the surrounding areas with a compilation of data detailing precipitation, river discharge and more. The first part of this study is focused of the study area and its main characteristic elements (climatic, geologic, hydrogeologic and hydrologic). In the second part, the chemical composition and water quality of Techirghiol Lake were investigated in the context of climatic and anthropogenic impact using Romanian methodologies and regulation.

### **2. Materials and Methods**

This section is divided in two parts: the first part is dedicated to the presentation of the study area and its main characteristic elements (climatic, geologic, hydrogeologic, hydrologic and lake water chemistry) and the second one features the methods used.

### *2.1. Study Area and Its Characteristics*

Techirghiol Lake is a result of the latest paleogeographic evolution of the Black Sea, dictated by the evolution of the sea level over time [12–14], which contributed to the development of the present shoreline. Accordingly, the coastal development and the sand-belt formation have completely isolated Techirghiol Lake. Located on the Black Sea coast, 16 km south from Constanta City (Figure 1), Techirghiol Lake is mainly known for the curative properties of its sapropelic mud and hypersaline water. Here, a veritable tourism economy has developed since 1899, around balneological treatment and medical rehabilitation.

The catchment area of Techirghiol Lake is situated in the South Dobrogea Plateau, having a surface of approximately 160 km<sup>2</sup> . The lake is 8 km long, with the maximum width of 4.4 km and a water depth varying between 1.5 m and 9.5 m. The maximum water depth was recorded at 9.75 m and the average water depth is 3.6 m [15].

The studied area has a relief consisting in a not very tall plateau (+70–80 m), with a slope which descends to the sea that ends abruptly with a 30-m-high cliff (Figure 1).

The Techirghiol Lake area is situated in a temperate–continental climatic zone, which is influenced by the Black Sea. The region is characterized by an average annual temperature of approximately 11 ◦C and an annual rainfall amount of about 400 mm [16]. The data recorded at Constanta meteorological station were chosen in order to analyze the influence of climatic parameters on the lake0 s behavior.

From a geological point of view, Techirghiol Lake area is situated in the South Dobrogea Plateau. The South Dobrogea Plateau basement layer consists of granitic gneiss and crystalline shale. Above this basement layer, this sector integrates three main sedimentary geological systems: Sarmatian limestone, red clay mixed with gypsum and loess deposits. The presence of faults and the sedimentary structure of Techirghiol Lake area have determined the development of several deep complex aquifers, among which a free surface aquifer situated in Sarmatian limestone and a pressure aquifer located in limestone and dolomitic deposits [17].

From the hydrological point of view, Techirghiol Lake is situated at the confluence of several important valleys (Figure 2), most of them with an intermittent flow. In 1910, Pascu [18] identified four important valleys that drained the Techirghiol catchment: Carlichioi (Biruinta) Valley, Techirghiol Valley, Muzurat (today Urlichioi) valley and Tuzla Valley. In 1976, Breier identified three important valleys: Techirghiol, Tuzla and Carlichioi (Biruinta) valleys (Figure 2). Today, the hydrological regime of the main hydrographic networks is very different, and the important valleys are barred by different hydraulic works (dams and penstock) in order to prevent the entrance of freshwater in the lake (Figure 2). The hydrological features of the lake are related to the evolution of the lake's level, which is strongly influenced by the aquifer input and by the discharge of the valleys.

*Water* **2020**, *12*, x FOR PEER REVIEW 3 of 14

**Figure 1.** Location of Techirghiol Lake. **Figure 1.** Location of Techirghiol Lake.

The catchment area of Techirghiol Lake is situated in the South Dobrogea Plateau, having a surface of approximately 160 km2. The lake is 8 km long, with the maximum width of 4.4 km and a water depth varying between 1.5 m and 9.5 m. The maximum water depth was recorded at 9.75 m and the average water depth is 3.6 m [15]. The studied area has a relief consisting in a not very tall plateau (+70**–**80 m), with a slope which descends to the sea that ends abruptly with a 30-m-high cliff (Figure 1). The Techirghiol Lake area is situated in a temperate**–**continental climatic zone, which is influenced by the Black Sea. The region is characterized by an average annual temperature of approximately 11 °C and an annual rainfall amount of about 400 mm [16]. The data recorded at Constanta meteorological station were chosen in order to analyze the influence of climatic parameters on the lake′s behavior. From a geological point of view, Techirghiol Lake area is situated in the South Dobrogea Plateau. The South Dobrogea Plateau basement layer consists of granitic gneiss and crystalline shale. Above this basement layer, this sector integrates three main sedimentary geological systems: Sarmatian limestone, red clay mixed with gypsum and loess deposits. The presence of faults and the sedimentary structure of Techirghiol Lake area have determined the development of several deep complex aquifers, among which a free surface aquifer situated in Sarmatian limestone and a pressure aquifer located in limestone and dolomitic deposits [17]. From the hydrological point of view, Techirghiol Lake is situated at the confluence of several important valleys (Figure 2), most of them with an intermittent flow. In 1910, Pascu [18] identified Techirghiol Town is well known as a balneotherapy center due to the importance of sapropelicmud and saline water used in therapeutic treatment. Two important centers have been developed here: the Techirghiol Balneotherapy Center (in 1899) and the Eforie Balneotherapy Center (in 1923). To sustain the economic development of this area, water supplies for the localities were established around the lake in the period of 1953–1956. During this period, wastewater was discharged intothe lake. Since 1956, wastewater has been treated by a wastewater treatment plant built in South Eforie Town. Treated water is discharged first into the Tuzla pond, and then into Techirghiol Lake. In 1969, important hydraulic works were made in the Lake Techirghiol catchment: (i) 12 km west of the lake is situated the principal irrigation channel "Basarabi—Negru Voda", which loses 60% of thewater through infiltration; (ii) 8 km north of the lake is located the "Danube—Black Sea"-navigable channel. In 1971, an irrigation system built in the area was put into operation. Since 1976, water from the treatment plant has been introduced into the irrigation system. In order to eliminate the <sup>e</sup>ffects of irrigation on the lake's parameters—and due to the fact that the stoppage of irrigation wasincompatible with the state policies of that period (Dobrogea Region being an arid area where crops cannot grow in optimal conditions without irrigation)—the National Water Administration took atthat time a series of measures to limit the effects of irrigation. First, in 1972–1973 and then in 1983, water from the lake was pumped directly into the sea. The protection works were carried out inthree stages, which were completed in 2005. In the first stage during 1977–1979, all groundwater observation drillings were equipped with pumps and the wastewater discharge into the lake wasforbidden. During the second stage (1980–1983), another 11 groundwater drillings were equipped

four important valleys that drained the Techirghiol catchment: Carlichioi (Biruinta) Valley, Techirghiol Valley, Muzurat (today Urlichioi) valley and Tuzla Valley. In 1976, Breier identified three

with pumps (intercepted water was used for water supply) and on the rivers Biruinta, Izvoarele and Gospodarie were built dams (behind the dams were placed two pumping stations). The third stage began in 1988 with the construction of the Techirghiol dam—which ended in 1991—and the drainage of the freshwater from behind the dam through a pipe (diameter of 1400 mm and a length of 9.1 km) into Belona Lake (near Eforie Town). A number of small dams were also built in all of the small valleys to intercept the freshwater and evacuate it. As a result of all these hydraulic works, the water surface of Techirghiol Lake decreased. Now, the studied area is divided into three zones: the freshwater area—Biruinta, Izvoarele, Gospodariei lakes, the brackish water area—Zarguzon Lake and the saline water area—Techirghiol Lake (Figure 2). *Water* **2020**, *12*, x FOR PEER REVIEW 4 of 14 important valleys: Techirghiol, Tuzla and Carlichioi (Biruinta) valleys (Figure 2). Today, the hydrological regime of the main hydrographic networks is very different, and the important valleys are barred by different hydraulic works (dams and penstock) in order to prevent the entrance of freshwater in the lake (Figure 2). The hydrological features of the lake are related to the evolution of the lake's level, which is strongly influenced by the aquifer input and by the discharge of the valleys.

**Figure 2.** Digital Elevation Model of Techirghiol Lake basin. **Figure 2.** Digital Elevation Model of Techirghiol Lake basin.

Techirghiol Town is well known as a balneotherapy center due to the importance of sapropelic mud and saline water used in therapeutic treatment. Two important centers have been developed here: the Techirghiol Balneotherapy Center (in 1899) and the Eforie Balneotherapy Center (in 1923). The main characteristics investigated in this study which influence the Techirghiol water budget are annual precipitation, overland flow and groundwater. These data are obtained from government reports spanning the period of 1953–2015.

To sustain the economic development of this area, water supplies for the localities were established around the lake in the period of 1953**–**1956. During this period, wastewater was discharged into the lake. Since 1956, wastewater has been treated by a wastewater treatment plant built in South Eforie Town. Treated water is discharged first into the Tuzla pond, and then into Techirghiol Lake. In 1969, important hydraulic works were made in the Lake Techirghiol catchment: (i) 12 km west of the lake is situated the principal irrigation channel "Basarabi**—**Negru Voda", which loses 60% of the water The lake water chemistry parameters investigation that covers salinity, pH, dissolved oxygen (DO) and biochemical oxygen demand (BOD) is based on the data provided by the Romanian Water Administration—Dobrogea Littoral Branch. The data were obtained from various sources, such as government reports, old published papers or unpublished reports [19–25]. The systematic measurement started in 1993, but several government works [21,22] provide some values for these chemical parameters before this period. The investigated period is 1993–2015.

#### through infiltration; (ii) 8 km north of the lake is located the "Danube**—**Black Sea"-navigable channel. In 1971, an irrigation system built in the area was put into operation. Since 1976, water from the *2.2. Methodologies*

treatment plant has been introduced into the irrigation system. In order to eliminate the effects of irrigation on the lake's parameters**—**and due to the fact that the stoppage of irrigation was incompatible with the state policies of that period (Dobrogea Region being an arid area where crops cannot grow in optimal conditions without irrigation)**—**the National Water Administration took at that time a series of measures to limit the effects of irrigation. First, in 1972**–**1973 and then in 1983, The methodology used for the analysis of hydrological data is described by Kundzewicz and Robsson [26] and is based on the following steps: (1) obtainment and preparation of a suitable dataset; (2) exploratory analysis of the data and (3) application of statistical tests. Concerning the first step the datasets were performed by INHGA (Romanian National Institute of Hydrology and Water Management), so they are expected to be reliable and free of gross errors, given that the gauging process

water from the lake was pumped directly into the sea. The protection works were carried out in three stages, which were completed in 2005. In the first stage during 1977**–**1979, all groundwater observation drillings were equipped with pumps and the wastewater discharge into the lake was

with pumps (intercepted water was used for water supply) and on the rivers Biruinta, Izvoarele and

was supervised by professional personnel. A set of statistical tests which detect step-change in the mean or median of a series was used in previous studies [16,27–31], namely Pettitt, Buishand test, Lee and Heghinian test. In order to detect multiple changes in time series data, the segmentation procedure of Hubert and changing point analysis (based on CUSUM procedure) were used and presented in previous studies we have already mentioned. Some results will be provided in the following paragraph in correlation with other investigated parameters.

The chemical composition and water quality of Techirghiol Lake were investigated in the context of climatic and anthropogenic impact using Romanian methodologies and regulation [19,20]. According to the methodology, Techirghiol Lake is a heavily modified water body. In this respect, for each chemical element mentioned above, the methodology establishes the limits and the ecological status/potential. Three ecological potential classes are identified for heavily modified water bodies: (i) maximum ecological potential (PEM), (ii) good ecological potential (PEB), (iii) moderate ecological potential (PEMo). The range of variation of each class was developed by a series of research institutes and experts.

### **3. Results and Discussion**

valleys: Biruinta and Urlichioi.

To determine the effects of changes in climate in the Dobrogea region and thus on the behavior of Lake Techirghiol, the results obtained in the studies previously mentioned are capitalized [16,27–30]. To conclude: (T—temperature) a break point is identified in 1997–1998 and the mean annual temperature increased by 0.8 ◦C in the 1997–2015 period—compared to the period of 1953–1997, which is in concordance with the estimation made for Europe by different reports [32–35]. (P−precipitation) Figure 3 shows the variability of rainfall amount from 1953 to 2015. *Water* **2020**, *12*, x FOR PEER REVIEW 6 of 14

**Figure 3.** Precipitation variation (1953**–**2015 period). **Figure 3.** Precipitation variation (1953–2015 period).

The annual precipitation value varies between a minimum value of 255 and a maximum value of 825 mm. The multiannual rainfall value for the entire study period is 425 mm. According to [27,35], precipitations have a break point in 1994. For the period of 1953**–**1994 we observe a decreasing trend and after it an increasing trend (Figure 3), and the mean annual precipitation increased from 373 to 531 mm. Starting with 1995, the annual precipitation has been above the multiannual precipitation except the following years: 2000 and 2001, 2011 and 2013. The overland flow (OvF) is presented in Figure 4. This flow is provided by two major river The annual precipitation value varies between a minimum value of 255 and a maximum value of 825 mm. The multiannual rainfall value for the entire study period is 425 mm. According to [27,35], precipitations have a break point in 1994. For the period of 1953–1994 we observe a decreasing trend and after it an increasing trend (Figure 3), and the mean annual precipitation increased from 373 to 531 mm. Starting with 1995, the annual precipitation has been above the multiannual precipitation except the following years: 2000 and 2001, 2011 and 2013.

**Figure 4.** Overland flow variation [35].

According to [35], the overland flow time data series presents three break points in 1971, 1978 and 2000. The first was observed in 1970 when the irrigation system became operational (in the period of 1953**–**1970 the average overland flow rate was 48 mm). The third one is observed in 2000. After 2000 the overland flow value returned to the average value of the 1953**–**1970 period (41 mm). We *Water* **2020**, *12*, 2261 and after it an increasing trend (Figure 3), and the mean annual precipitation increased from 373 to

**precipitation (mm)**

The overland flow (OvF) is presented in Figure 4. This flow is provided by two major river valleys: Biruinta and Urlichioi. except the following years: 2000 and 2001, 2011 and 2013. The overland flow (OvF) is presented in Figure 4. This flow is provided by two major river

531 mm. Starting with 1995, the annual precipitation has been above the multiannual precipitation

**Figure 3.** Precipitation variation (1953**–**2015 period).

1953-1994 time series average 1953-1994 1995-2015 time series average 1995-2015

The annual precipitation value varies between a minimum value of 255 and a maximum value of 825 mm. The multiannual rainfall value for the entire study period is 425 mm. According to [27,35], precipitations have a break point in 1994. For the period of 1953**–**1994 we observe a decreasing trend

Linear (1953-1994 time series) Linear (1995-2015 time series)

min: 255 mm 1987;1993

avg: 531 mm

max: 825 mm

*Water* **2020**, *12*, x FOR PEER REVIEW 6 of 14

avg: 373 mm

1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

**year**

**Figure 4.** Overland flow variation [35]. **Figure 4.** Overland flow variation [35].

According to [35], the overland flow time data series presents three break points in 1971, 1978 and 2000. The first was observed in 1970 when the irrigation system became operational (in the period of 1953**–**1970 the average overland flow rate was 48 mm). The third one is observed in 2000. After 2000 the overland flow value returned to the average value of the 1953**–**1970 period (41 mm). We According to [35], the overland flow time data series presents three break points in 1971, 1978 and 2000. The first was observed in 1970 when the irrigation system became operational (in the period of 1953–1970 the average overland flow rate was 48 mm). The third one is observed in 2000. After 2000 the overland flow value returned to the average value of the 1953–1970 period (41 mm). We consider that this breakpoint is in relation to anthropic intervention: 1997–1998 was the last time freshwater was introduced into the main irrigation channel that crosses the lake's catchment: in 1991, the Techirghiol dam entered into operation. In the period of 1970–2000, the average overland flow increased to 632 mm (the increase was about 14 times relative to the previous period—580 mm). However, the maximum value of overland flow was recorded in 1995 (1058 mm).

The groundwater supply (GW) is presented in Figure 5. The values varied between a minimum value of 98.2 mm (2015) and a maximum value of 1206.9 mm recorded in 1985. It is noted that the groundwater input time series is divided in three subseries [35]. During the period of 1953–1969, the groundwater input value did not exceed 534 mm. Since 1970, this value has increased on average about 1.5 times. The average value for the period of 1970–2000 was 754.7 mm. Since 2000, as a result of finalizing the works proposed in the third stage, the groundwater input value decreased, reaching the minimum value (98 mm) in 2015.

In Figure 6 is represented the variation of the main inputs (P + OvF + GW) in Techirghiol Lake as average per period. The periods marked by the human interventions and the breakpoint in precipitation data series is highlighted. Analyzing the results obtained we could conclude that the hydraulic works built until 1970, especially the irrigation system, changed the water budget of Techirghiol Lake after 1971. In the 1971–1978 periods, the overland flow increased from 46 to 411 mm on average. In this period, the overland flow represented 31% of the total budget. The first protective works performed in 1977–1979 did not influence the overland flow and groundwater regime in the sense that it diminished. On the contrary, in the period of 1979–1983 (only five years), the average values of these parameters increased and represented 80% of the total water budget. We could conclude that the hydraulic works did not have the expected effect, given the average values of groundwater increase from 692 to 1040 mm approximately, in the period of 1984–1986.

*Water* 

measurement data are from 1909.

the minimum value (98 mm) in 2015.

consider that this breakpoint is in relation to anthropic intervention: 1997**–**1998 was the last time freshwater was introduced into the main irrigation channel that crosses the lake's catchment: in 1991, the Techirghiol dam entered into operation. In the period of 1970**–**2000, the average overland flow increased to 632 mm (the increase was about 14 times relative to the previous period**—**580 mm).

The groundwater supply (GW) is presented in Figure 5. The values varied between a minimum value of 98.2 mm (2015) and a maximum value of 1206.9 mm recorded in 1985. It is noted that the groundwater input time series is divided in three subseries [35]. During the period of 1953**–**1969, the groundwater input value did not exceed 534 mm. Since 1970, this value has increased on average about 1.5 times. The average value for the period of 1970**–**2000 was 754.7 mm. Since 2000, as a result

However, the maximum value of overland flow was recorded in 1995 (1058 mm).

**Figure 5.** Groundwater input variation. **Figure 5.** Groundwater input variation. **2020**, *12*, x FOR PEER REVIEW 8 of 14

**Figure 6.** Variation of the min input values (average/period) in the Techirghiol water budget. **Figure 6.** Variation of the min input values (average/period) in the Techirghiol water budget.

The hydraulic works performed in the period of 1983**–**1986 failed to bring new improvements to the Techirghiol water budget. The average values of overland flow are maintained at the level of the period of 1987**–**1994, while the average values of groundwater flow have decreased by 200 mm. As previously mentioned, starting with 1995 the value of precipitation increased. In the following period, the average values are maintained at 43% of the total water budget. The protective works started in 1988 and were finalized in 1991 and 2005, the stoppage of the irrigation activity (1998) causing an improvement of the water budget: the average of overland flow decreased to 39 mm and the average groundwater value to 264 mm. The hydraulic works performed in the period of 1983–1986 failed to bring new improvements to the Techirghiol water budget. The average values of overland flow are maintained at the level of the period of 1987–1994, while the average values of groundwater flow have decreased by 200 mm. As previously mentioned, starting with 1995 the value of precipitation increased. In the following period, the average values are maintained at 43% of the total water budget. The protective works started in 1988 and were finalized in 1991 and 2005, the stoppage of the irrigation activity (1998) causing an improvement of the water budget: the average of overland flow decreased to 39 mm and the average groundwater value to 264 mm.

Figure 7 shows the evolution of the water level in the lake between 1953 and 2015; some isolated

*Water* **2020**, *12*, 2261

Figure 7 shows the evolution of the water level in the lake between 1953 and 2015; some isolated measurement data are from 1909.

*Water* **2020**, *12*, x FOR PEER REVIEW 9 of 14

anthropic impact. (**a**) Salinity; (**b**) water level. **Figure 7.** Water level and salinity evolution of Techiorghiol Lake under changes in climate and anthropic impact. (**a**) Salinity; (**b**) water level.

It can be seen that in the investigated period, water levels in the lake rose from −150 cm to +153 cm (the measurements are relative to the Black Sea level). Between 1909 and 1952 the water level in the lake increased by an average of 0.8 cm/year, between 1954 and 1966 the water level increased by 6 cm/year. Since 1970 the water level in the lake has become positive (+9 cm), relative to the Black Sea reference level (±0.00). The increase in water level in the lake was accentuated after 1970, the average It can be seen that in the investigated period, water levels in the lake rose from −150 cm to +153 cm (the measurements are relative to the Black Sea level). Between 1909 and 1952 the water level in the lake increased by an average of 0.8 cm/year, between 1954 and 1966 the water level increased by 6 cm/year. Since 1970 the water level in the lake has become positive (+9 cm), relative to the Black Sea reference level (±0.00). The increase in water level in the lake was accentuated after 1970, the average value

value being 9 cm/year, as a result of land irrigation of the lake's catchment. From 1965 to 1989, the water level steadily increased to 133 cm, and in the period of 1996**–**1999 the level reached the highest values (+153 cm). As a result of the hydraulic works, there has been a trend of increasing the water being 9 cm/year, as a result of land irrigation of the lake's catchment. From 1965 to 1989, the water level steadily increased to 133 cm, and in the period of 1996–1999 the level reached the highest values (+153 cm). As a result of the hydraulic works, there has been a trend of increasing the water level in the lake, combined with the decrease of the salinity of the water. Since the land irrigation was stopped (in 1998) the lake's water level has been slowly decreasing in the following years, in 2015 the level being 22 cm above Black Sea level (Figure 7). *Water* **2020**, *12*, x FOR PEER REVIEW 10 of 14 stopped (in 1998) the lake's water level has been slowly decreasing in the following years, in 2015 the level being 22 cm above Black Sea level (Figure 7).

Increases of water inputs led to the severe decrease of salinity around 47 g‰ in 1992 and 1997. In the irrigation period of 1970–1997, even if a number of protective measures were introduced, a substantial increase in salinity was not possible. After the irrigation was stopped (1998), the salinity began to increase, reaching the value of 70 g‰ (in 2015). Increases of water inputs led to the severe decrease of salinity around 47 g‰ in 1992 and 1997. In the irrigation period of 1970**–**1997, even if a number of protective measures were introduced, a substantial increase in salinity was not possible. After the irrigation was stopped (1998), the salinity began to increase, reaching the value of 70 g‰ (in 2015).

The changes caused by the increase of the freshwater inflow and the decrease of salinity caused quantitative changes in the lake0 s biotic community, especially in some organisms involved in the process of peloidogenesis. Some studies [21,36] reveal a decrease of green algae *Cladophora vagabunda* from 81.49 tons in 1978 to 42 tons in 1981. It is known [37] that the optimal salinity values in which this alga can develop are 73–83 g/L. This situation began to improve in 1987 when the second stage of protective works became operational. The completion of protective work (in 2005) and closure of the irrigation system (in 1998) has led to ecosystem regeneration. The changes caused by the increase of the freshwater inflow and the decrease of salinity caused quantitative changes in the lake′s biotic community, especially in some organisms involved in the process of peloidogenesis. Some studies [21,36] reveal a decrease of green algae *Cladophora vagabunda* from 81.49 tons in 1978 to 42 tons in 1981. It is known [37] that the optimal salinity values in which this alga can develop are 73**–**83 g/L. This situation began to improve in 1987 when the second stage of protective works became operational. The completion of protective work (in 2005) and closure of the irrigation system (in 1998) has led to ecosystem regeneration.

The water of Techirghiol Lake is alkaline; the average value for pH is 8.3 (Figure 8). Normally, there is a direct relationship between water pH and salinity; a higher value of pH is given by the high content of mineralization. Even if the salinity of Techirghiol Lake waters decreased to a value of 47 g‰, the pH value of the waters was never under the value of 7.9. Figure 8 shows that the lake water pH is situated in the range 6.5–9 pH, more accurate under 8.5 units, except the values from 1999 and 2000. The water of Techirghiol Lake is alkaline; the average value for pH is 8.3 (Figure 8). Normally, there is a direct relationship between water pH and salinity; a higher value of pH is given by the high content of mineralization. Even if the salinity of Techirghiol Lake waters decreased to a value of 47 g‰, the pH value of the waters was never under the value of 7.9. Figure 8 shows that the lake water pH is situated in the range 6.5**–**9 pH, more accurate under 8.5 units, except the values from 1999 and 2000.

**Figure 8.** Variation of pH over time. **Figure 8.** Variation of pH over time.

Figure 9 shows the variation of the dissolved oxygen (DO) time data series. It can be seen that the dissolved oxygen (DO) varies between 11 and 4.04 mg/L. In the period of 1975**–**1990 the DO values were over 8 mg/L and lake water could be included in the PEB category. The DO values have decreased after 1990 from an average of 8.91 mg/L to 6.24 mg/L. Correspondingly, the ecological potential has decreased, lake water could be included in the PEB/PEMo category. Three exceptions could be considered: 2009, 2013 and 2014, when water could be included in the PEM/PEB category. Figure 9 shows the variation of the dissolved oxygen (DO) time data series. It can be seen that the dissolved oxygen (DO) varies between 11 and 4.04 mg/L. In the period of 1975–1990 the DOvalues were over 8 mg/L and lake water could be included in the PEB category. The DO values have decreased after 1990 from an average of 8.91 mg/L to 6.24 mg/L. Correspondingly, the ecologicalpotential has decreased, lake water could be included in the PEB/PEMo category. Three exceptions could be considered: 2009, 2013 and 2014, when water could be included in the PEM/PEB category.

*Water* **2020**, *12*, x FOR PEER REVIEW 11 of 14

**Figure 9.** Variation of dissolved oxygen (DO). **Figure 9.** Variation of dissolved oxygen (DO). **Figure 9.** Variation of dissolved oxygen (DO).

The biologic oxygen demand (BOD) represents the mass concentration of dissolved oxygen consumed by microorganism or measures the chemical oxidation of inorganic matter in a given time (e.g., BOD5 stands for five days test). BOD5 affects the DO values. A greater BOD5 value means less oxygen for the microorganism′s activity. The variation of this indicator is presented in Figure 10. Generally during the 1975**–**1993 period, the BOD5 values were between 3 mg/L and 6 mg/L and the lake water could be included in the PEM/PEB ecological potential category. After 1994 the BOD5 values have generally increased above 6 mg/L. In fact, during the period of 1994**–**2012, the average was three times higher than the average of the previous period. The biologic oxygen demand (BOD) represents the mass concentration of dissolved oxygen consumed by microorganism or measures the chemical oxidation of inorganic matter in a given time (e.g., BOD5 stands for five days test). BOD5 affects the DO values. A greater BOD5 value means less oxygen for the microorganism's activity. The variation of this indicator is presented in Figure 10. Generally during the 1975–1993 period, the BOD5 values were between 3 mg/L and 6 mg/L and the lake water could be included in the PEM/PEB ecological potential category. After 1994 the BOD5 values have generally increased above 6 mg/L. In fact, during the period of 1994–2012, the average was three times higher than the average of the previous period. The biologic oxygen demand (BOD) represents the mass concentration of dissolved oxygen consumed by microorganism or measures the chemical oxidation of inorganic matter in a given time (e.g., BOD5 stands for five days test). BOD5 affects the DO values. A greater BOD5 value means less oxygen for the microorganism′s activity. The variation of this indicator is presented in Figure 10. Generally during the 1975**–**1993 period, the BOD5 values were between 3 mg/L and 6 mg/L and the lake water could be included in the PEM/PEB ecological potential category. After 1994 the BOD5 values have generally increased above 6 mg/L. In fact, during the period of 1994**–**2012, the average was three times higher than the average of the previous period.

**Figure 10.** Variation of Biochemical oxygen demand (BOD). **Figure 10.** Variation of Biochemical oxygen demand (BOD).

The ecological potential of Techirghiol Lake from the point of view of the investigated elements varies throughout the period investigated (Table 1). It can be seen that Techirghiol Lake water could be included in the PEB/PEM category for the period of 1975**–**1990 (in this period, a larger quantity of freshwater was introduced via the irrigation system). After 1990 (1993 for BOD5) lake water was The ecological potential of Techirghiol Lake from the point of view of the investigated elements varies throughout the period investigated (Table 1). It can be seen that Techirghiol Lake water could be included in the PEB/PEM category for the period of 1975**–**1990 (in this period, a larger quantity of The ecological potential of Techirghiol Lake from the point of view of the investigated elements varies throughout the period investigated (Table 1). It can be seen that Techirghiol Lake water could be included in the PEB/PEM category for the period of 1975–1990 (in this period, a larger quantity

freshwater was introduced via the irrigation system). After 1990 (1993 for BOD5) lake water was

included in the PEB/PEMo category.

included in the PEB/PEMo category.

of freshwater was introduced via the irrigation system). After 1990 (1993 for BOD5) lake water was included in the PEB/PEMo category.


**Table 1.** Ecological Potential Variation.

It is very complex to explain the multiple factors that play a role in changing water chemistry variation. Barbulescu and Barbes [32] consider that one of the direct consequences of the decrease in water salinity was the modification of the lake biodiversity. Some studies appreciated that the phytoplankton structure was modified during the irrigation period: the number of species diminishing to 14–18 in relation to the previous period, when 38 species were found [21,22,37]. Another indicator is *Artemia* Salina. A drastic decreasing of *Artemia* Salina during the irrigation period, compared to the reference period (1952–1960), when densities above 100 g/L were recorded [22]. We can conclude that the cause of the increase of BOD5 in the last period (after 1995) could be an increase of aquatic life forms (phytoplankton or/and zooplankton).

### **4. Conclusions**

An important aspect of Techirghiol Lake is its potential in the tourism industry, due to its unique properties: saline water and sapropelic mud. The malfunction of environmental protection measures and faulty or insufficient design (period of 1960–1987), in conjunction with the changes in climate has disturbed the normal functioning of the Techirghiol Lake ecosystem, finally resulting in a decrease in its capacity to yield economic values. It is therefore concluded that 1953 is considered as the last year in which the Techirghiol Lake system was under the influence of natural factors and the 1970 year is marked by the passage of the water level to positive values. Starting with 1971, the irrigation system became operational and the ecosystem degradation became aggressive. The rate of salinity decreased as a result of freshwater supply being 1.24‰ during the period of 1970–1987. The period of 1980–1987 is a critical one: overland flow increased (14 times the level of the period of 1953–1970 and the hydrological regime of Biruinta and Urlichioi tributary rivers became permanent); groundwater input increased by 7.2 mil.mc/year over the same period.

The most important challenge in the management of Techirghiol Lake basin is to integrate and balance the interest of the ecosystem and the economy. It is well known that the Dobrogea region is an arid area where crops cannot grow in optimum conditions without irrigation. New investigation in needed to provide the sound, scientific basis in order to find a balance between protecting the ecosystem, increasing the economy and designing hydrotechnical systems in the context of climate change.

**Author Contributions:** Conceptualization, C.M.; methodology, C.M., C.B. and I.C.P.; validation, C.M., C.B. and I.C.P.; investigation, C.M.; resources, I.C.P. and C.B.; data curation, C.B.; writing—original draft preparation, C.M.; writing—review and editing, C.M., C.B., I.C.P.; visualization, I.C.P.; supervision, C.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors would like to thank the Dobrogea Littoral Water Basin Administration for technical support and National Meteorological Agency—Dobrogea meteorological Center which provided the climatic data.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Multivariate Statistical Analysis of Water Quality and Trophic State in an Artificial Dam Reservoir**

**Md Mamun, Ji Yoon Kim and Kwang-Guk An \***

Department of Bioscience and Biotechnology, Chungnam National University, Daejeon 34134, Korea; mamun1006001@gmail.com (M.M.); jiyoonn20@naver.com (J.Y.K.)

**\*** Correspondence: kgan@cnu.ac.kr; Tel.: +82-010-6404-9844; Fax: +82-42-882-9690

**Abstract:** Paldang Reservoir, located in the Han River basin in South Korea, is used for drinking water, fishing, irrigation, recreation, and hydroelectric power. Therefore, the water quality of the reservoir is of great importance. The main objectives of this study were to evaluate spatial and seasonal variations of surface water quality in the reservoir using multivariate statistical techniques (MSTs) along with the Trophic State Index (TSI) and Trophic State Index deviation (TSID). The empirical relationships among nutrients (total phosphorus, TP; total nitrogen, TN), chlorophyll-a (CHL-a), and annual variations of water quality parameters were also determined. To this end, 12 water quality parameters were monitored monthly at five sites along the reservoir from 1996 to 2019. Most of the parameters (all except pH, dissolved oxygen (DO), and total coliform bacteria (TCB)) showed significant spatial variations, indicating an influence of anthropogenic activities. Principal component analysis combined with factor analysis (PCA/FA) suggested that the parameters responsible for water quality variations were primarily correlated with nutrients and organic matter (anthropogenic), suspended solids (both natural and anthropogenic), and ionic concentrations (both natural and anthropogenic). Stepwise spatial discriminant analysis (DA) identified water temperature (WT), DO, electrical conductivity (EC), chemical oxygen demand (COD), the ratio of biological oxygen demand (BOD) to COD (BOD/COD), TN, TN:TP, and TCB as the parameters responsible for variations among sites, and seasonal stepwise DA identified WT, BOD, and total suspended solids (TSS) as the parameters responsible for variations among seasons. COD has increased (R<sup>2</sup> = 0.63, *p* < 0.01) in the reservoir since 1996, suggesting that nonbiodegradable organic loading to the water body is rising. The empirical regression models of CHL-a-TP (R<sup>2</sup> = 0.45) and CHL-a-TN (R<sup>2</sup> = 0.27) indicated that TP better explained algal growth than TN. The mean TSI values for TP, CHL-a, and Secchi depth (SD) indicated a eutrophic state of the reservoir for all seasons and sites. Analysis of TSID suggested that blue-green algae dominated the algal community in the reservoir. The present results show that a significant increase in algal chlorophyll occurs during spring in the reservoir. Our findings may facilitate the management of Paldang Reservoir.

**Keywords:** multivariate statistical methods; Trophic State Index; water quality; empirical model; Paldang Reservoir

### **1. Introduction**

Although water is indispensable to life, it is one of the most threatened resources worldwide [1]. Clean and safe freshwater is a basic need for human health and economic development, but anthropogenic activities like industrialization, urbanization, and intensive agricultural farming have negatively impacted freshwater sources, hindering their use for drinking, irrigation, fishing, recreational, domestic, and industrial purposes [2–5]. Therefore, serious attention should be paid to protect freshwater resources. Among these, reservoirs are the most vulnerable due to high loads of pollutants, nutrients, organic matter, and suspended solids from the watershed [6,7]. For effective water management, gathering reliable information on reservoir water quality, evaluating spatial and seasonal water qual-

**Citation:** Mamun, M.; Kim, J.Y.; An, K.-G. Multivariate Statistical Analysis of Water Quality and Trophic State in an Artificial Dam Reservoir. *Water* **2021**, *13*, 186. https://doi.org/ 10.3390/w13020186

Received: 9 November 2020 Accepted: 10 January 2021 Published: 14 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

ity changes, detecting pollution sources, determining water quality status, and controlling water pollution in reservoirs are essential [1,3,8–11].

To assess the water quality of surface water resources, MSTs, TSI, and TSID have been widely used, and therefore have played a significant role in water resource management [2,9,11,12]. Multivariate statistical methods, such as discriminant analysis (DA), principal component analysis (PCA), factor analysis (FA), correlation analysis, and analysis of variance (ANOVA) facilitate the interpretation of complex water quality datasets [1,13,14]. These methods are also used to identify factors that influence surface water quality, serving as a valuable tool for effective surface water quality management [2,11]. These approaches can be used to evaluate temporal and spatial changes in surface water quality caused by natural and anthropogenic factors [2].

However, MSTs have some limitations when used alone [2]. Therefore, applying MSTs, TSI, and TSID in combination can be advantageous for assessing the water quality of reservoirs. To date, a few studies have used MSTs, TSI, and TSID together for surface water quality assessment of reservoirs [2,15]. The TSI and TSID were used to quantify the degree of eutrophication of a water body. Carlson [12] proposed a quantitative index to calculate the degree of eutrophication in lakes and reservoirs based on total phosphorus (TP), chlorophyll-a (CHL-a), and Secchi depth (SD). According to Carlson and others, TP is the best forecaster of algal growth, while CHL-a is the most reliable algal biomass indicator, and SD is the best proxy for water clarity in water bodies [4,16–18]. Moreover, TSI and TSID are used to evaluate spatial and seasonal changes in the water quality of reservoirs, and thereby provide useful information for reservoir management [19,20].

Seasonal rainfall patterns, hydrology, and watershed morphology are the major factors known to regulate water quality within a watershed [21]. These factors are closely related to the ecosystem's nutrient regime, water clarity, and algal growth. Rainfall is directly linked to inflow, outflow, depth, and water residence time (WRT), which control nutrient and suspended solids loads to the water body [22,23]. Empirical evidence suggests that phosphorus (P) is the key factor limiting CHL-a growth in freshwater systems [4,17,19]. Excessive concentrations of nutrients, especially P, may accelerate algal growth and cause eutrophication in reservoirs [24]. Total suspended solids are a potential source of P and play an essential role in the P cycle in reservoirs [2].

Paldang Reservoir is one of the largest reservoirs in South Korea, with a water volume of 255 <sup>×</sup> <sup>10</sup><sup>6</sup> <sup>m</sup><sup>3</sup> and a surface area of 28.9 km<sup>2</sup> [25,26]. The maximum depth of water at full supply level is 21 m, and the mean depth is 8.3 m [25]. It is a manmade lake formed after the construction of a hydroelectric dam in 1973 and is located in the central Korean Peninsula [25]. Paldang Reservoir has been used for fishing, irrigation, recreation, hydroelectric power, and drinking water purposes. Additionally, it serves as an essential water resource for people living in the Seoul metropolitan area and surrounding cities [26]. More than 24 million people (48% of the Korean population) rely on the Paldang Reservoir for drinking water [27]. Therefore, the water quality of the reservoir is of great importance to Korea. However, human activities in the watershed have increased, resulting in significant pollution problems in the reservoir. Urbanization, domestic and industrial wastewater discharge, intensive agricultural activities, waste from animal farms, and inflowing rivers are all major sources of water pollution in the reservoir [27–29].

For these reasons, comprehensive water quality assessments of the reservoir are needed. The purposes of the present study are to (1) determine the spatial and seasonal variations of water quality parameters and identify the key factors affecting water quality in the reservoir using MSTs, (2) assess the trophic status of the reservoir using TSI and TSID, (3) determine how water quality parameters are correlated with hydrology, and (4) develop empirical models of the CHL-a-TP, CHL-a-TN, TSS-TP, and TSS-TN in the reservoir. Thus, this study will assess the current status of water quality and aid the development of effective management and conservation strategies to protect water quality in Paldang Reservoir.

### **2. Materials and Methods**

### *2.1. Study Sites and Water Quality Parameters*

Paldang Reservoir is the most downstream reservoir in the Han River system, and is situated at the confluence of the North Han River, South Han River, and Kyoungan Stream (Figure 1; [30]). In this study, five reservoir sampling sites (S1–S5) were selected. Sites 1 and 2 were located in the South Han River part of the reservoir. In contrast, sites 3–5 were located at the North Han River, Kyoungan Stream, and dam, respectively. The water intake tower for Paldang Reservoir is located at S5 (Figure 1). Based on their hydrological characteristics, reservoirs can be divided into two types, namely, lake- and river-type reservoirs. Lake-type reservoirs are generally characterized by high depth and long water WRT, while river-type reservoirs have shallower depths and shorter WRTs [31]. Paldang Reservoir is considered a river-type reservoir due to its shallow depth (mean depth: 8.3 m) and short WRT (3–10 days) [25,32]. Paldang Reservoir does not fully stratify throughout the year [30]. The overall inflow and outflow rates of the reservoir are almost equal, resulting in very small annual water level fluctuations. The annual amount of rainfall and water inflow from the upstream watershed directly influence WRT in Paldang Reservoir [30]. *Water* **2021**, *13*, x FOR PEER REVIEW 4 of 19

**Figure 1.** The map showing the sampling sites of Paldang Reservoir.

**Figure 1.** The map showing the sampling sites of Paldang Reservoir. *2.2. Trophic State Index and Trophic State Index Deviation*  The trophic status of the Paldang Reservoir was determined using Carlson's TSI. The range of average TSI values designated Oligotrophic is 30–40, Mesotrophic is 40–50, Eu-Monthly surface water quality data for the Paldang Reservoir from 1996 to 2019 were obtained from the Ministry of Environment's national water quality measurement network (http://water.nier.go.kr). Monthly rainfall and inflow and outflow data were collected from the Korean Meteorological Administration and the Korean Water Resource Corporation, respectively. WRT was defined as the reservoir water volume divided by the inflow rate [33]. The loading data for TP, TN, TSS, BOD, and COD were calculated using a conversion factor derived from the corresponding concentrations.

TSI (CHL-a, µg L−1) = 10 × [6 − (2.04 − 0.68ln(CHL-a))/ln2] (1)

TSI (TP, µg L−1) = 10 × [6 − ln(48/TP)/ln2] (2)

TSI (SD, m) = 10 × [6 − ln(SD)/ln2] (3)

trophic is 50–70, and Hypereutrophic is >70 [18]. The following equations were used to

Using two-dimensional approaches, the TSID was defined using the relationships TSI (CHL-a)-TSI (SD) and TSI (CHL-a)-TSI (TP). This method has also been used frequently to quantify the degree of eutrophication and identify the limiting nutrient in res-

The Kolmogorov–Smirnov single-sample test was used to examine the distribution of water quality data prior to statistical analyses [1]. One-way ANOVA was performed to determine whether there were significant spatial and seasonal variations in the reservoir's

calculate TSI values for the Paldang Reservoir [12]:

ervoirs [18].

*2.3. Statistical Analysis* 

### *2.2. Trophic State Index and Trophic State Index Deviation*

The trophic status of the Paldang Reservoir was determined using Carlson's TSI. The range of average TSI values designated Oligotrophic is 30–40, Mesotrophic is 40–50, Eutrophic is 50–70, and Hypereutrophic is >70 [18]. The following equations were used to calculate TSI values for the Paldang Reservoir [12]:

$$\text{TSI} \,\text{(CHL-a, } \mu \text{g L}^{-1}) = 10 \times [6 - (2.04 - 0.68 \ln(\text{CHL-a}))/\ln 2] \tag{1}$$

$$\text{TSI} \,\text{(TP, } \text{\#g L}^{-1} \text{)} = 10 \times \left[ 6 - \ln(48/\text{TP})/\ln 2 \right] \tag{2}$$

$$\text{TSI} \text{ (SD, m)} = 10 \times [6 - \ln(\text{SD})/\ln 2] \tag{3}$$

Using two-dimensional approaches, the TSID was defined using the relationships TSI (CHL-a)-TSI (SD) and TSI (CHL-a)-TSI (TP). This method has also been used frequently to quantify the degree of eutrophication and identify the limiting nutrient in reservoirs [18].

### *2.3. Statistical Analysis*

The Kolmogorov–Smirnov single-sample test was used to examine the distribution of water quality data prior to statistical analyses [1]. One-way ANOVA was performed to determine whether there were significant spatial and seasonal variations in the reservoir's water quality parameter values. Pearson correlation analysis was used to analyze the relationships between various water quality variables. PCA/FA was conducted to determine the factors and pollution sources affecting the surface water quality [34]. Bartlett's sphericity test and the Kaiser–Meyer–Olkin (KMO) test were conducted first to determine the suitability of the data for PCA/FA [2]. DA was performed to assess both spatial and temporal variations in water quality and to identify water quality variables that could best distinguish among sites and seasons [11,34]. Standard and stepwise DA was applied to raw data. PCA/FA was applied to experimental data, standardized through Z-scale transformation, to avoid misclassification [2]. SPSS software (version 22.0; SPSS Inc., Chicago, IL, USA) was used for all statistical analyses. Bar, box, and scatter plots were prepared using SigmaPlot 14.0 software (Systat Software, Inc., San Jose, CA, USA). Interpolation of TSI values was conducted using QGIS 3.14 (QGIS Development Team, Gossau, Switzerland). Conditional plotting analysis was carried out with R 3.5.2 (R Development Core Team, Vienna, Austria).

### **3. Results and Discussion**

### *3.1. Spatial and Seasonal Variations*

The mean values of 12 water quality parameters recorded at five sampling sites in Paldang Reservoir are presented in Table 1. In this study, all variables except pH, dissolved oxygen (DO), and total coliform bacteria (TCB) showed significant spatial differences among sites (*p* < 0.05, Table 1). The spatial variations of these parameters indicate impacts of anthropogenic activities in the reservoir [25,32]. For example, BOD, COD, TSS, TN, TP, and CHL-a concentrations were significantly higher at site S4 than any other site; this site receives inputs from industrial and domestic wastewater [30]. Site S4 in Paldang Reservoir is affected by Kyoungan Stream. The water quality of this tributary stream is worse than that of the South Han River (Sites S1 and S2) and North Han River (Site S3), and thus, it may significantly impact the reservoir's water quality [25]. Water clarity (SD) was higher at Site S3 than other sites, indicating that the North Han River input is cleaner than the South Han River and Kyoungan Stream inflows [25]. The highest mean electrical conductivity (EC) was recorded at Site S1 due to agricultural activities and untreated household wastewater effluent.

*Water* **2021**, *13*, 186



111

The water quality parameters of the Paldang Reservoir showed significant heterogeneity (*p* < 0.05) among the four seasons (Table 1). Water temperature (WT), TSS, TP, and TCB exhibited significantly higher values in the summer, whereas pH, EC, BOD, COD, TN, and CHL-a were highest in the spring. TSS and TP concentrations were elevated during the summer due to intense precipitation (Supplementary Figure S1). The summer monsoon significantly influences the hydrology, nutrients, and suspended solids concentration in Korean reservoirs [35]. More intense rainfall contributes to increased TSS in Paldang Reservoir water. The daily loading data also showed that TP, TN, TSS, BOD, and COD were higher during the summer monsoon season than any other season (Supplementary Figure S2). This result supports the view that the summer monsoon is the main driver of high levels of nutrients, organic matter, and suspended solids in mid-latitude East Asian reservoirs, such as those in South Korea, Eastern China, and Japan [36]. The regression equation between TSS and TP indicates that TSS is associated with 45% of TP in Paldang Reservoir (Supplementary Figure S3). This result suggests that TSS may act as a nutrient carrier in the reservoir [37].

Organic matter (BOD and COD) in reservoirs can have either allochthonous or autochthonous origins. Allochthonous organic matter enters aquatic systems mainly via runoff derived from overland water flow during rainfall events, while autochthonous organic matter is produced through photosynthesis by phytoplankton and hydrophytes [32]. As Paldang Reservoir is a river-type reservoir, it experiences high flow rates during the summer season, and large amounts of allochthonous organic matter is introduced into the reservoir. Park et al. [32] showed that 69% of the total organic matter was allochthonous in Paldang Reservoir during the summer season. In contrast, a high autochthonous organic matter load was observed in the winter and spring due to low flow rates and increased WRT [32]. Park et al. [30] revealed that 73% of autochthonous organic matter loading occurs during the spring. The peak organic matter concentration coincides with the maximum production of algae (spring bloom). This finding suggests that autochthonous production by phytoplankton (CHL-a) during the spring period is critical to organic matter buildup in Paldang Reservoir; thus, the threat to the water quality of Paldang Reservoir is greatest in spring [30,32].

In the present study, the water quality status of each sampling site and season was assessed by comparing the mean values of water quality parameters with those listed in the Korean Lakes and Reservoirs Surface Water Quality Regulation, 2015 (Supplementary Table S1). As shown in Table S1, Site S4 had class III water quality (contaminated water), while all other sites had class II water quality (lightly contaminated water) in terms of COD. Based on TSS, all sites fell into the class III water quality category except site S3. Sites S1 and S4 were in the class IV water quality (somewhat poor; contaminated water) in terms of TP. Site S4 had class IV water quality in terms of CHL-a. During spring, algal growth increased in the reservoir, and water quality was somewhat poor (class IV; contaminated water). TP concentrations were higher during summer due to surface runoff, and the water body was in class IV. All sites and seasons had class Ia water quality in terms of pH and DO. Notably, Site S5 (near the water intake tower) was in class III (average; contaminated water) in terms of CHL-a, TSS, and TP.

### *3.2. Correlation Analysis*

Pearson correlation analysis was used to evaluate the relationships among 12 water quality parameters (Supplementary Table S2). As anticipated, DO was negatively associated with WT, as oxygen is more soluble in cold water [1]. High BOD and COD levels indicate organic pollution in the reservoir [34,35]. Increasing nutrient concentrations (TP and TN) lead to elevated organic matter concentrations (BOD and COD) in the reservoir [1,15]. EC showed significant positive relationships with BOD (r = 0.40, *p* < 0.01), COD (r = 0.59, *p* < 0.01), and TN (r = 0.55, *p* < 0.01). TN, TP, and BOD showed strong positive relationships with each other, demonstrating that their sources were analogous. TSS showed a significant positive relationship with TP (r = 0.47, *p* < 0.01), indicating that suspended particles have a

tendency to adsorb P [38]. During rainfall events and stream bank erosion in high-flow periods, agricultural and industrial runoff can contribute to high levels of TSS and TP in the reservoir [38]. CHL-a was positively correlated with TP (r = 0.70, *p* < 0.01) and TN (r = 0.53, *p* < 0.01), which are key factors affecting phytoplankton growth in this water body [39]. The reservoir's water clarity decreased with increase in the TP, TN, CHL-a, TSS, and BOD concentrations.

### *3.3. Annual Variations of Water Quality*

Annual data can provide information about long-term water quality dynamics in Paldang Reservoir. The results showed that TP (R<sup>2</sup> = 0.27, *p* < 0.01), BOD (R<sup>2</sup> = 0.26, *p* < 0.01), and CHL-a (R<sup>2</sup> = 0.33, *p* < 0.01) have decreased significantly since 1996 (Figure 2). The loading data for TP (R<sup>2</sup> = 0.21, *p* < 0.02), TN (R<sup>2</sup> = 0.19, *p* < 0.03), and BOD (R<sup>2</sup> = 0.35, *p* < 0.00) also showed a decreasing pattern in Paldang Reservoir (Supplementary Figure S4). COD (R<sup>2</sup> = 0.63, *p* < 0.01) and SD (R<sup>2</sup> = 0.37, *p* < 0.01) have increased in Paldang Reservoir since 1996. Moreover, the loading pattern for COD has changed. BOD concentrations in most Korean lakes and reservoirs are continuously decreasing, while COD concentrations have been increasing in most cases, indicating that high concentrations of nonbiodegradable organic matter in the influent may be inefficiently degraded by the biological effluent treatment process [30,40]. COD represents both biodegradable and nonbiodegradable organic pollution in water systems. However, increases in the COD level suggest increased nonbiodegradable organic loading from wastewater treatment plants (WWTPs) and urban sewage systems to the water body [41]. A previous study conducted in the United States found an increase in the occurrence and persistence of inorganic solid loading from a WWTP to a water body [42]. Industries may not strictly comply with environmental regulations, and thus may contribute large amounts of nonbiodegradable compounds to aquatic systems [43]. Water quality management strategies in Korean reservoirs likely need to be re-evaluated with a focus on water pollutant management, especially for organic matter.

### *3.4. Hydrology, Nutrients, and Chlorophyll-a*

Inflow, outflow, and WRT are major drivers of the distributions of nutrients, suspended solids, and CHL-a in aquatic environments. Compared to TN and CHL-a, inflow, outflow, and WRT were more significant determinants of the concentrations of TP (R<sup>2</sup> = 0.30, 0.29, 0.30, *p* < 0.01) and TSS (R<sup>2</sup> = 0.39, 0.36, 0.39, *p* < 0.01) (Figure 3). The present findings were similar to previous studies in Korean reservoirs. Previous research in various parts of the world has shown that external loadings of TP and TSS are highly correlated with inflow, outflow, and WRT in the watershed, and this conclusion was supported by the present study [23,44,45]. Many studies have reported effects of WRT on algal growth in aquatic systems [46,47]. However, the results of the present study did not concur with some previous studies. Thus, WRT may not always be linked to algal growth in the reservoir. This may indicate that release of autochthonous nutrients regulates algal growth in Paldang Reservoir. Lee et al. [36] suggested that algal chlorophyll growth was influenced by nutrients in Paldang Reservoir.

*Water* 

**2021**, *13*, x FOR PEER REVIEW 9 of 18

The empirical models based on log-transformed CHL-a-TP and CHL-a-TN relationships are shown in Figure 4. Nutrients more strongly influenced chlorophyll growth in the reservoir's ambient water, and the concentration of TP (R<sup>2</sup> = 0.45, *p* < 0.01) better explained algal growth than that of TN (R<sup>2</sup> = 0.27, *p* < 0.01), indicating a P-limited system. When two predictors are strongly correlated (R<sup>2</sup> > 0.70), collinearity problems may arise that impede determination of the nutrient limiting algal growth. The present results showed that TP and TN (R<sup>2</sup> = 0.55) are moderately correlated in Paldang Reservoir. To avoid these problems, conditional plots have been used to identify limiting nutrients in aquatic systems [35,48]. Conditional plots showed that the association between CHL-a and TP was relatively steady in Paldang Reservoir, as indicated by the smooth lines on the four panels with similar slopes, which suggested that the effect of TP on CHL-a is consistent irrespective of the level of TN, in turn indicating a P-limited reservoir (Supplementary Figures S5 and S6). In addition, the conditional plot shows no interaction between TP and TN, further verifying that Paldang Reservoir is a P-limited system. ships are shown in Figure 4. Nutrients more strongly influenced chlorophyll growth in the reservoir's ambient water, and the concentration of TP (R2 = 0.45, *p* < 0.01) better explained algal growth than that of TN (R2 = 0.27, *p* < 0.01), indicating a P-limited system. When two predictors are strongly correlated (R2 > 0.70), collinearity problems may arise that impede determination of the nutrient limiting algal growth. The present results showed that TP and TN (R2 = 0.55) are moderately correlated in Paldang Reservoir. To avoid these problems, conditional plots have been used to identify limiting nutrients in aquatic systems [35,48]. Conditional plots showed that the association between CHL-a and TP was relatively steady in Paldang Reservoir, as indicated by the smooth lines on the four panels with similar slopes, which suggested that the effect of TP on CHL-a is consistent irrespective of the level of TN, in turn indicating a P-limited reservoir (Supplementary Figures S5 and S6). In addition, the conditional plot shows no interaction between TP and TN, further verifying that Paldang Reservoir is a P-limited system.

The empirical models based on log-transformed CHL-a-TP and CHL-a-TN relation-

*Water* **2021**, *13*, x FOR PEER REVIEW 10 of 18

**Figure 4.** Empirical relationship of CHL-a, TP, and TN. TP—total phosphorus, TN—total nitrogen, CHL—chlorophyll-a. **Figure 4.** Empirical relationship of CHL-a, TP, and TN. TP—total phosphorus, TN—total nitrogen, CHL—chlorophyll-a.

#### *3.5. Trophic State Index and Trophic State Index Deviation 3.5. Trophic State Index and Trophic State Index Deviation*

The trophic state of Paldang Reservoir, based on TP, TN, CHL-a, and SD, showed heterogeneity among sites and seasons, all of which were categorized as mesotrophic to eutrophic (Supplementary Table S3) [49,50]. These results are similar to the findings of previous trophic state classification studies in Korean reservoirs [4,51]. The primary sources of nutrients for Paldang Reservoir are agricultural fertilizer, animal manure, municipal sewage, and industrial effluents [25]. Based on TP concentrations, all sites and seasons were under eutrophic conditions, except for Site S3 and the winter season. Notably, we found that Paldang Reservoir was in a eutrophic state in all sites and seasons, on the basis of TN, CHL-a, and SD. Considering the present results, measures should be taken to control eutrophication in Paldang Reservoir. The trophic state of Paldang Reservoir, based on TP, TN, CHL-a, and SD, showed heterogeneity among sites and seasons, all of which were categorized as mesotrophic to eutrophic (Supplementary Table S3) [49,50]. These results are similar to the findings of previous trophic state classification studies in Korean reservoirs [4,51]. The primary sources of nutrients for Paldang Reservoir are agricultural fertilizer, animal manure, municipal sewage, and industrial effluents [25]. Based on TP concentrations, all sites and seasons were under eutrophic conditions, except for Site S3 and the winter season. Notably, we found that Paldang Reservoir was in a eutrophic state in all sites and seasons, on the basis of TN, CHL-a, and SD. Considering the present results, measures should be taken to control eutrophication in Paldang Reservoir.

Assessing the potential of a water source to support cyanobacterial blooms or bluegreen algae is essential for water resource management [52]. WT, TP, CHL-a, and SD are essential factors for determining potential cyanobacterial growth in a reservoir [53]. The concentrations of TP and CHL-a, along with SD, in Paldang Reservoir indicate a moderate level of risk for cyanobacterial exposure (Supplementary Table S4). CHL-a is a good indicator of overall phytoplankton biomass, and monitoring CHL-a is a direct method for semiquantitative estimation of cyanobacterial biomass in aquatic systems [20]. For South Korean reservoirs supplying drinking water, a cyanobacteria watch is issued when the concentration of CHL-a exceeds 15 µg L−1. Furthermore, an alert is issued when the CHL-Assessing the potential of a water source to support cyanobacterial blooms or bluegreen algae is essential for water resource management [52]. WT, TP, CHL-a, and SD are essential factors for determining potential cyanobacterial growth in a reservoir [53]. The concentrations of TP and CHL-a, along with SD, in Paldang Reservoir indicate a moderate level of risk for cyanobacterial exposure (Supplementary Table S4). CHL-a is a good indicator of overall phytoplankton biomass, and monitoring CHL-a is a direct method for semiquantitative estimation of cyanobacterial biomass in aquatic systems [20]. For South Korean reservoirs supplying drinking water, a cyanobacteria watch is issued when the concentration of CHL-a exceeds 15 µg L−<sup>1</sup> . Furthermore, an alert is issued when the CHL-a concentration is greater than 25 µg L−<sup>1</sup> . Once a watch or alert has been issued, additional

treatment processes are required at water treatment plants until the watch or alert is cleared. Additionally, when an alert is issued, water intake below that at which algae can inhabit and analysis of cyanotoxin in the treated water, are required [54]. The results of the present study indicate that all sites and seasons (except site S3 and winter) were under watch conditions. Previous studies of Paldang Reservoir have suggested that cyanobacterial blooms occur during the spring season, which is in line with our findings [30,32]. cleared. Additionally, when an alert is issued, water intake below that at which algae can inhabit and analysis of cyanotoxin in the treated water, are required [54]. The results of the present study indicate that all sites and seasons (except site S3 and winter) were under watch conditions. Previous studies of Paldang Reservoir have suggested that cyanobacterial blooms occur during the spring season, which is in line with our findings [30,32]. Analysis of TSI and TSID provides valuable information on algal chlorophyll devel-

a concentration is greater than 25 µg L−1. Once a watch or alert has been issued, additional treatment processes are required at water treatment plants until the watch or alert is

*Water* **2021**, *13*, x FOR PEER REVIEW 11 of 18

Analysis of TSI and TSID provides valuable information on algal chlorophyll development, nutrient variability, and other parameters in lakes and reservoirs [4,12]. TSI and TSID were estimated based on TP, CHL-a, and SD in Paldang Reservoir, and their values showed spatial and seasonal variations (Figures 4 and 5). The mean TSI (TP), TSI (CHL-a), and TSI (SD) values indicate a eutrophic state during all seasons and at all sites (Figure 5, Supplementary Figure S7). These consistent eutrophic conditions may reduce DO and hamper ecosystem functions. The mean TSI (CHL-a) indicated more eutrophic conditions during spring and summer than the fall and winter. Water quality was worse in terms of TSI at Site S4 than other sites, and this site influenced water quality at the drinking water tower (Supplementary Figure S7; Site S5). Park et al. [32] revealed that Kyoungan Stream (Site S4) has a significant impact on the quality of drinking water in Paldang Reservoir. opment, nutrient variability, and other parameters in lakes and reservoirs [4,12]. TSI and TSID were estimated based on TP, CHL-a, and SD in Paldang Reservoir, and their values showed spatial and seasonal variations (Figures 4 and 5). The mean TSI (TP), TSI (CHLa), and TSI (SD) values indicate a eutrophic state during all seasons and at all sites (Figure 5, Supplementary Figure S7). These consistent eutrophic conditions may reduce DO and hamper ecosystem functions. The mean TSI (CHL-a) indicated more eutrophic conditions during spring and summer than the fall and winter. Water quality was worse in terms of TSI at Site S4 than other sites, and this site influenced water quality at the drinking water tower (Supplementary Figure S7; Site S5). Park et al. [32] revealed that Kyoungan Stream (Site S4) has a significant impact on the quality of drinking water in Paldang Reservoir.

**Figure 5.** Seasonal Trophic State Index and its deviation. **Figure 5.** Seasonal Trophic State Index and its deviation.

Analysis of TSID showed that blue-green algae were predominant in the reservoir during all four seasons based on the relationships of TSI (CHL-a) with TSI (SD) and TSI (TP) (Figure 5). Blooms of blue-green algae are associated with eutrophic conditions [18]. Previous research identified the following genera of cyanobacteria in Paldang Reservoir: *Anabaena, Aphanocapsa, Chroococcus, Coelosphaerium, Dactylococcopsis, Microcystis, Merismopedia, Phormidium, Oscillatoria,* and *Pseudoanabaena* [26]*.* The occurrence of cyanobacteria is affected by light, temperature, pH, and nutrients. The concentration of TP is a major Analysis of TSID showed that blue-green algae were predominant in the reservoir during all four seasons based on the relationships of TSI (CHL-a) with TSI (SD) and TSI (TP) (Figure 5). Blooms of blue-green algae are associated with eutrophic conditions [18]. Previous research identified the following genera of cyanobacteria in Paldang Reservoir: *Anabaena, Aphanocapsa, Chroococcus, Coelosphaerium, Dactylococcopsis, Microcystis, Merismopedia, Phormidium, Oscillatoria,* and *Pseudoanabaena* [26]. The occurrence of cyanobacteria is affected by light, temperature, pH, and nutrients. The concentration of TP is a major factor influencing the cyanobacterial contribution to total algal biomass [55]. Moreover,

the biomass of cyanobacterial genera, such as *Aphanizomenon, Anabaena,* and *Microcystis*, is strongly influenced by the levels of TP and TN [55].

Nonalgal turbidity was observed during the summer and fall due to surface runoff from the watershed. Such turbidity is a common occurrence in Asian lakes and reservoirs during the monsoon period [35,56]. Small amounts of zooplankton grazing and P-limited small particles were observed in the reservoir. In addition, the TSID data indicated that TSI (CHL-a) was significantly higher than TSI (TP) during spring and winter, demonstrating that algal productivity was higher than expected and highlighting the controlling effect of P [4,18]. The water's trophic state must remain oligotrophic to mesotrophic for drinking water purposes according to the United States Environmental Protection Agency and Korean Ministry of Environment guidelines. The reservoir water intake towers face substantial bloom problems, impeding access to the water supply for local residents.

### *3.6. Discriminant Analysis*

To study spatial and seasonal variations of water quality, DA was performed on the raw dataset. The spatial discriminant functions (DFs) and classification matrixes (CMs) used in this study are provided in Tables 2 and S5, respectively. Spatial standard and stepwise DFs with 14 and 8 discriminant variables, respectively, were used to generate CMs, which assigned 100% of cases correctly (Tables 2 and S5). The stepwise spatial DA results suggest that WT, DO, EC, COD, BOD/COD, TN, TN:TP, and TCB are the most important variables for explaining spatial variations in water quality in Paldang Reservoir among the five sites. The DFs indicated that WT, COD, BOD/COD, and TN were higher at Site S4 than other sites due to industrial and domestic wastewater effluents. These results are in accordance with previous findings in Paldang Reservoir [30]. Therefore, the spatial variations of these water quality parameters were mainly related to anthropogenic activities in the watershed.

**Table 2.** Classification functions for discriminant analysis of spatial variations in water quality of the reservoir. pH hydrogen ion concentration, WT—water temperature, DO—dissolved oxygen, EC—electrical conductivity, BOD—biological oxygen demand, COD—chemical oxygen demand, TSS—total suspended solids, TN—total nitrogen, TP—total phosphorus, CHL—chlorophyll-a, SD—Secchi depth, TCB—total coliform bacteria.


Fisher's linear discriminant functions.

Seasonal DFs and CMs are shown in Tables 3 and S6, respectively, and were used to evaluate seasonal changes in water quality in Paldang Reservoir. Seasonal standard and stepwise mode DFs using 14 and 3 discriminant variables, respectively, generated CMs that assigned 100% of cases correctly (Tables 3 and S6). Temporal stepwise DA findings showed that WT, BOD, and TSS were the most important factors in temporal variations in the water quality of Paldang Reservoir among the four seasons. The DFs indicated that WT and TSS were higher during summer than other seasons. WT was highest in summer and lowest in winter due to the impact of climate seasonality [1]. TSS concentrations were higher during summer due to summer monsoon effects [35]. In contrast, BOD was highest in spring. Previous research revealed elevated organic matter concentrations during spring in Paldang Reservoir [32].

**Table 3.** Classification functions for discriminant analysis of seasonal variations in water quality of the reservoir. pH hydrogen ion concentration, WT—water temperature, DO—dissolved oxygen, EC—electrical conductivity, BOD-biological oxygen demand, COD—chemical oxygen demand, TSS—total suspended solids, TN—total nitrogen, TP—total phosphorus, CHL—chlorophyll-a, SD—Secchi depth, TCB—total coliform bacteria.


Fisher's linear discriminant functions.

Varol et al. [2] studied surface water quality variations in Keban Reservoir, Turkey, using the DA method, and found that eight and three variables successfully explained the temporal and spatial variations, respectively, among 19 water quality parameters. Chen et al. [14] studied surface water quality variations in Danjiangkou Reservoir, China, using the DA method, and their results indicated that six and four variables effectively explained spatial and temporal variations, respectively, among 11 water quality parameters. Mustapha et al. [57] studied surface water quality variations in the upper reach of the Kano River, Nigeria, using the DA method and successfully identified 7 variables, among 23 tested, having a statistically significant effect on the spatial variations. Singh et al. [9] showed that DA allows for data reduction, where only six and two variables were sufficient to discriminate spatial and temporal variations, respectively, in the Gomti River, India. Similarly, Zhang et al. [58] applied this method to evaluate spatial-temporal variations of water quality in the southwest New Territories and Kowloon, Hong Kong, and revealed

that four and eight parameters could support 84.2% and 96.1% correct assignment in temporal and spatial analysis, respectively [58]. Furthermore, they suggested that the number of monitoring variables (and the associated cost) could be reduced, as their method allowed for considerable reduction of the dimensionality of the large dataset. Overall, DA led to a considerable reduction in the present research dataset and helped determine the parameters responsible for spatial and temporal variations.

### *3.7. Principal Component Analysis Combined with Factor Analysis (PCA/FA)*

Urbanization, domestic sewage, industrial wastewater effluents, intensive agricultural activities, and waste from animal farms and inflowing rivers are the primary sources of water pollution in Paldang Reservoir. Bartlett's test and KMO were performed to examine the suitability of the data for PCA/FA. In the present study, the KMO value was 0.59, and Bartlett's test was significant (*p* < 0.000), indicating that the Paldang Reservoir data were suitable for PCA/FA and that meaningful relationships were present among the water quality variables. PCA/FA with varimax rotation identified five varifactors (VFs), which explained 82.32% of the total variance (Table 4). Varifactor 1 (VF1) represented 25.82% of the total variance and showed a strong positive loading (>0.70) for TP, strong negative loadings for TN:TP and SD, and moderate positive loadings (between 0.5 and 0.7) for TSS, TN, and CHL-a (Table 4). This VF represents inputs of nutrients and suspended matter from untreated domestic sewage, industrial effluents, and agricultural runoff. Nutrient inputs influence algal growth in Paldang Reservoir. The negative contribution of SD to this VF is related to high levels of nutrients, suspended solids, and algal growth [4,11]. VF2 showed strong positive loadings for pH and BOD/COD, and a moderate positive loading for BOD. This VF represents organic matter concentrations in the reservoir. VF3 (17.85% of the total variance) showed strong positive loadings for WT, EC, and COD and a moderate positive loading for BOD. This VF indicates the contributions of ions and organic matter input to the reservoir from untreated domestic sewage, industrial effluents, and agricultural runoff. VF4 (9.65 of the total variance) had a strong positive loading for DO, while VF5 (9.61% of the total variance) displayed a strong positive loading for TCB. The PCA/FA findings suggest that most of the variation in reservoir water quality can be attributed to nutrients and organic matter (anthropogenic), suspended solids (both natural and anthropogenic), and ionic concentrations (both natural and anthropogenic), which are regulated by both natural and anthropogenic activities.

**Table 4.** Varimax rotated component matrix for water quality parameters (Kaiser–Meyer–Olkin (KMO) = 0.59, Bartlett's test was significant (*p* = 0.000), extraction method: principal component analysis, and rotation method: varimax with Kaiser normalization, and bold and italic values represent strong and moderate loadings, respectively). pH—hydrogen ion concentration, WT—water temperature, DO—dissolved oxygen, EC—electrical conductivity, BOD-biological oxygen demand, COD—chemical oxygen demand, TSS—total suspended solids, TN—total nitrogen, TP-total phosphorus, CHL chlorophyll-a, SD—Secchi depth, TCB—total coliform bacteria.



**Table 4.** *Cont.*

PCA/FA is a dimension-reduction technique that provides information about the most significant factors through simplification of the data. Therefore, this method has been utilized in various studies exploring the pollution sources affecting a water system. For example, PCA/FA was employed by Lim et al. [59] to identify sources of pollution in the Langat River, Malaysia. Four components were extracted in group 1, explaining 85% of the total variance, while six components were extracted in group 2, explaining 88% of the total variance. Based on these data, they determined that seawater intrusion, agricultural and industrial pollution, and geological weathering were mainly responsible for the river pollution. In addition, Tanriverdi et al. [60] used PCA/FA to analyze and assess the surface water quality of Ceyhan River and suggested that stations near cities were strongly affected by household wastewater, while other stations were influenced by agricultural facilities. Moreover, Jha et al. [61] identified major pollution sources influencing physicochemical variables in Aerial Bay, Andaman Islands, using the FA technique, which included rivulet flux into the bay, land run-off, prevailing biological processes, and tidal flow. Haji Gholizadeh et al. [11] identified five and four potential pollution sources to the Miami Canal in South Florida during the wet and dry seasons, respectively, which affected water quality variables. PCA/FA was used to distinguish four potential pollution types, namely, organic pollution, nutrient pollution, chemical pollution, and natural pollution, in Danjiangkou Reservoir, China, revealing that the study area was primarily influenced by industrial effluent and domestic sewage [14]

### **4. Conclusions**

MSTs, TSI, and TSID were combined to assess the water quality of Paldang Reservoir. All variables except pH, DO, and TCB showed significant spatial variations due to the effects of anthropogenic activities. The mean values of TSI (TP), TSI (CHL-a), and TSI (SD) indicated a eutrophic state, and TSID showed that blue-green algae dominated the reservoir. PCA/FA results revealed that the concentrations of TP, TN, BOD, COD, TSS, and EC were generally linked to both anthropogenic activities and natural processes. Stepwise DA provided better results for both spatial and temporal analyses. Thus, this study demonstrated that MSTs, TSI, and TSID are effective approaches for evaluating reservoir water quality, and that these methods can be used in combination as useful water quality management tools. Relative to US EPA and MOE guidelines, the reservoir is in a eutrophic state in terms of CHL-a, which is unfavorable for drinking purposes. To improve the water quality of this reservoir, nutrient and organic matter loads from the watershed should be limited.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2073-4 441/13/2/186/s1, Figure S1: Seasonal and Total rainfall pattern of Paldang watershed (Spring: March–May, Summer: June–August, Fall: September–November, Winter: December–February, and TRF: total rainfall), Figure S2: Loading Data of TP, TN, TSS, BOD, and COD in the Paldang Reservoir (TP—total phosphorus, TN—total nitrogen, TSS—total suspended solids, BOD—biological

oxygen demand, COD—chemical oxygen demand, Spring: March–May, Summer: June–August, Fall: September–November, and Winter: December–February), Figure S3: Empirical relations among TSS, TP, and TN (TP—total phosphorus, TN—total nitrogen, TSS—total suspended solids), Figure S4: Yearly loading data of TP, TN, TSS, BOD, COD (TP—total phosphorus, TN—total nitrogen, TSS—total suspended solids, BOD—biological oxygen demand, COD—chemical oxygen demand), Figure S5: The relationship between CHL-a and TP is plotted conditional on the range of TN (TP total phosphorus, TN—total nitrogen, CHL—chlorophyll-a), Figure S6: The relationship between CHL-a and TN is plotted conditional on the range of TP (TP—total phosphorus, TN—total nitrogen, CHL—chlorophyll-a), Figure S7: Trophic State Index of Paldang Reservoir at five different sites, Table S1: Water quality classes of Paldang Reservoir based on sites and seasons according to the Korean Ministry of Environment water quality standards for reservoirs and lakes (pH—hydrogen ion concentration, DO—dissolved oxygen, COD—chemical oxygen demand, TSS-total suspended solids, TN—total nitrogen, TP—total phosphorus, CHL—chlorophyll-a, TCB—total coliform bacteria, Ia: very good (high-quality water), Ib: good (high-quality water), II: somewhat good (lightly contaminated water), III: average (contaminated water), IV: somewhat poor (contaminated water), V: poor (highly contaminated water), VI: very poor (highly contaminated water)), Table S2: Pearson correlation analysis of water quality parameters (units mg L−<sup>1</sup> , except pH, WT (◦C), EC (µS cm−<sup>1</sup> ), TP (µg L−<sup>1</sup> ), CHL-a (µg L−<sup>1</sup> ), SD (m), and TCB (MPNML−100)). pH—hydrogen ion concentration, WT—water temperature, DO—dissolved oxygen, EC—electrical conductivity, BOD—biological oxygen demand, COD—chemical oxygen demand, TSS—total suspended solids, TN—total nitrogen, TP—total phosphorus, CHL—chlorophyll-a, SD—Secchi depth, TCB—total coliform bacteria, Table S3: Trophic state criteria based on TP, TN, CHL-a, and SD from Nurnberg (1996) for Paldang Reservoir (TN—total nitrogen, TP—total phosphorus, CHL—chlorophyll-a, SD—Secchi depth, M: mesotrophic, E: eutrophic, and H: Hypereutrophic), Table S4: Thresholds of risk associated with potential exposure to cyanobacteria in Paldang Reservoir (adopted from WHO, 2015, LRE: lower risk of exposure, MRE: moderate risk of exposure and HRE: higher risk of exposure, TP—total phosphorus, CHL—chlorophyll-a, SD—Secchi depth), Table S5: Classification matrix for discriminant analysis of spatial variations in water quality of the reservoirs, Table S6: Classification matrix for discriminant analysis of seasonal variations in water quality of the reservoirs.

**Author Contributions:** Conceptualization, M.M.; methodology, M.M.; software, M.M. and J.Y.K.; formal analysis, M.M.; data curation, M.M. and J.Y.K.; writing—original draft preparation, M.M.; writing—review and editing, M.M. and K.-G.A.; visualization, M.M. and K.-G.A.; supervision, K.- G.A.; funding acquisition, K.-G.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the "Korea Environment Industry & Technology Institute (KEITI)" through the "Aquatic Ecosystem Conservation Research Program" funded by the Korean Ministry of Environment (Grant number: 2020003050004).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The datasets presented in this study are available on reasonable request from the corresponding author.

**Acknowledgments:** The authors would like to acknowledge the Korean Ministry of Environment for their assistance.

**Conflicts of Interest:** The authors declare that they have no conflicts of interest.

### **References**


### *Article* **Evaluation of the Gulf of Aqaba Coastal Water, Jordan**

#### **Ahmed A. Al-Taani 1,2,\* , Maen Rashdan <sup>2</sup> , Yousef Nazzal <sup>1</sup> , Fares Howari <sup>1</sup> , Jibran Iqbal <sup>1</sup> , Abdulla Al-Rawabdeh <sup>2</sup> , Abeer Al Bsoul <sup>3</sup> and Safaa Khashashneh <sup>2</sup>**


Received: 7 July 2020; Accepted: 21 July 2020; Published: 27 July 2020

**Abstract:** (1) Background: The Gulf of Aqaba (GoA) supports unique and diverse marine ecosystems. It is one of the highest anthropogenically impacted coasts in the Middle East region, where rapid human activities are likely to degrade these naturally diverse but stressed ecosystems. (2) Methods: Various water quality parameters were measured to assess the current status and conditions of GoA seawater including pH, total dissolved solids (TDS), total alkalinity (TA), Cl−, NO<sup>3</sup> <sup>−</sup>, SO<sup>4</sup> <sup>2</sup>−, PO<sup>4</sup> <sup>3</sup>−, NH<sup>4</sup> <sup>+</sup>, Ca2+, Mg2+, Na+, K+, Sr, Cd, Co, Cr, Cu, Fe, Mn, Pb, and Zn. (3) Results: The pH values indicated basic coastal waters. The elevated levels of TDS with an average of about 42 g/L indicated highly saline conditions. Relatively low levels of inorganic nutrients were observed consistent with the prevalence of oligotrophic conditions in GoA seawater. The concentrations of Ca2+, Mg2+, Na+, K+, Sr, Cl−, and SO<sup>4</sup> <sup>2</sup><sup>−</sup> in surface layer varied spatially from about 423–487, 2246–2356, 9542–12,647, 513–713, 9.2–10.4, 22,173–25,992, and 317–407 mg/L, respectively. The average levels of Cd, Co, Cr, Cu, Fe, Mn, Pb and Zn ranged from 0.51, 0.38, 1.44, 1.29, 0.88, 0.38, and 6.05 µg/L, respectively. (4) Conclusions: The prevailing saline conditions of high temperatures, high evaporation rates, the water stratification and intense dust storms are major contributing factors to the observed seawater chemistry. The surface distribution of water quality variables showed spatial variations with no specific patterns, except for metal contents which exhibited southward increasing trends, closed to the industrial complex. The vast majority of these quality parameters showed relatively higher values compared to those of other regions.

**Keywords:** water quality; coastal area; metals; pollution source; Gulf of Aqaba; Jordan; Red Sea

### **1. Introduction**

The Gulf of Aqaba (GoA) is the upper northeastern segment of the Red Sea [1]. It is a partially-isolated, narrow and deep coastal water body. The Strait of Tiran connects GoA with the Red Sea (Figure 1). Despite the extreme environmental conditions, the GoA supports unique aquatic ecosystems and biodiversity, and is a habitat for one of the world's richest coral communities [2,3].

*Water* **2020**, *12*, x FOR PEER REVIEW 2 of 17

**Figure 1.** Location map of the Gulf of Aqaba (GoA) and sampling sites. **Figure 1.** Location map of the Gulf of Aqaba (GoA) and sampling sites.

The GoA is one of the high anthropogenically impacted coasts in the Middle East region [4]. The expansion in economic and industrial activities in the Gulf's bordering countries have contributed to the degradation of naturally stressed coastal and marine ecosystems. They are subjected to various impacts and sources of pollution including dredging and reclamation activities, coastal construction development, industrial waste, ports, oil spills, and domestic The GoA is one of the high anthropogenically impacted coasts in the Middle East region [4]. The expansion in economic and industrial activities in the Gulf's bordering countries have contributed to the degradation of naturally stressed coastal and marine ecosystems. They are subjected to various impacts and sources of pollution including dredging and reclamation activities, coastal construction development, industrial waste, ports, oil spills, and domestic sewage, among others [5].

sewage, among others [5]. GoA is the only marine port for Jordan, and is highly vulnerable to pollution, where all marine-related activities are concentrated within a few kilometers of the coast (27 km). In addition, many economic, industrial, and recreational activities are taking place along the Jordanian coastline, many of which are of potential environmental impacts [6–9]. Additionally, the region plans to have a number of large coastal projects (such as the Red-Dead Sea conduit, new resorts, and ports relocation), which will certainly accelerate the degradation cycle of GoA is the only marine port for Jordan, and is highly vulnerable to pollution, where all marine-related activities are concentrated within a few kilometers of the coast (27 km). In addition, many economic, industrial, and recreational activities are taking place along the Jordanian coastline, many of which are of potential environmental impacts [6–9]. Additionally, the region plans to have a number of large coastal projects (such as the Red-Dead Sea conduit, new resorts, and ports relocation), which will certainly accelerate the degradation cycle of existing environmental conditions and threaten these unique marine communities [7]. Signs of human impacts were reported [10–12].

existing environmental conditions and threaten these unique marine communities [7]. Signs of human impacts were reported [10–12]. In addition to human impacts, the GoA is subject to regular dust storm events that contribute metals and other chemicals to the GoA coastal water [8,9]. Aeolian dust flux to GoA is likely to influence seawater chemistry [13], where atmospheric dry deposition in the GoA is considered an important external source of trace metals [8,9,14–16]. The mineral dust rate on In addition to human impacts, the GoA is subject to regular dust storm events that contribute metals and other chemicals to the GoA coastal water [8,9]. Aeolian dust flux to GoA is likely to influence seawater chemistry [13], where atmospheric dry deposition in the GoA is considered an important external source of trace metals [8,9,14–16]. The mineral dust rate on GoA region is one of the highest on Earth [8,9,13,17]. It is believed that the frequency of dust storm events will become more common in the GoA, due to increase in regional aridity and dust fluxes [18,19].

GoA region is one of the highest on Earth [8,9,13,17]. It is believed that the frequency of dust storm events will become more common in the GoA, due to increase in regional aridity and dust fluxes [18,19]. The relatively small volume and absence of significant wave action along with the low rate of water circulation and renewal (between GoA and the Red Sea), render the Gulf particularly vulnerable to pollution. The residence time of water in the Gulf averaged 1–3 years [20,21].

The growing concern over the sustainability of these unique aquatic ecosystems of GoA has recently gained momentum and became a priority issue in Jordan. The impact of intense and widespread human activities from the Gulf's bordering countries poses imminent threats to GoA coast, which requires a proper monitoring plan. The objective of the present study is to assess the current status of surface water quality along the coastal region of the Jordanian GoA coast. It also intends to evaluate the spatial distribution of a variety heavy metals and to identify potential sources of contamination. This assessment will help develop a sustainable management plan for coastal water resources.

### **2. Materials and Methods**

### *2.1. Description of Study Area*

The GoA is the Red Sea's northeastern extension. It is a partially-enclosed, narrow and deep coastal water body. The Strait of Tiran connects the Red Sea to the GoA (Figure 1). The GoA extends approximately 180 km southward with a width ranging from 5–25 km (the average of 16 km maximum), and a maximum depth of about 1800 m (the average is about 800 m). Only 27 km of the eastern coast belong in Jordan, and the remaining coastline, unpopulated, and largely underdeveloped, lie in the Saudi territory.

The GoA is influenced by prevailing subtropical conditions with extremely high temperatures, high evaporation rate (about 400 cm/year) and negligible rainfall (of less than 2.2 cm/year) [22]. Surface water flow in the Gulf is nonexistent or limited solely during rare intense rainstorms occurring as flash floods in winter. The average water temperature in the upper 200 m varies seasonally from 20 ◦C in winter to 28 ◦C in summer, whereas the average air temperature ranges between 32.20 ± 3.16 ◦C in summer and 17.60 ± 3.46 ◦C in winter [23]. The maximum sea level of 154.30 cm was recorded in 2013 [23].

These conditions result in a high salinity in surface water layer, ranging from 40.3 to 40.8‰ in winter and from 40.5 to 46.6‰ in summer [9,24–26]. The surface coastal water of the GoA is extremely oligotrophic, because of its nutrient-poor water originating from the Red Sea surface waters through the Straits of Tiran. The surface water is shallow with stable thermocline throughout the year, except in wintertime, when a wind-driven convective mixing occurs between the deep (nutrient-rich) and surface waters. Water stratification occurs in spring. However, the oceanographic characteristics of extensive solar irradiance, high transparency, deep sunlight penetration, and warm water created unique aquatic ecosystems and biodiversity, with one of the world's richest coral communities [2,3].

The northerly wind, with a high speed and activity during summertime, is the prevailing wind direction and is responsible for the majority of aeolian deposition events in the region. However, Khamaseen winds blowing in springtime account for most sand and dust storms in southern Jordan and the adjacent areas [27]. They deliver dust from the interior of the Sahara Desert in north Africa.

### *2.2. Sampling and Analysis*

Surface water samples were collected in September 2017 from 30 different locations along the coastal areas of GoA, Jordan (from north to south), sampling sites are presented in Figure 1. Coastal water samples were collected in 1-L precleaned polyethylene containers pre-rinsed with 10% HCl and 2 mL of HNO3, Samples were labeled and measured in the field for pH, electrical conductivity (EC, mS/cm at 25 ◦C), and total dissolved solids (TDS) using pH-meter (Sensions 5, HACH portable case), and EC/TDS-meter (ECOSCAN-hand held series, EUTECH instruments). Water samples were kept refrigerated at 4 ◦C and transported to a water laboratory (Yarmouk University, Jordan) for subsequent chemical analyses. Sample preparation and analysis followed APHA [28] procedures.

In the laboratory, all samples were filtered by Whatman filter paper (No. 42) and analyzed for total alkalinity (TA), Cl−, NO<sup>3</sup> <sup>−</sup>, SO<sup>4</sup> <sup>2</sup>−, PO<sup>4</sup> <sup>3</sup>−, NH<sup>4</sup> <sup>+</sup>, Ca2+, Mg2+, Na+, K+, Cd, Co, Cr, Cu, Fe, Mn, Pb, Sr, and Zn, as follows: 50 mL of filtered samples were used to determine the concentrations of Na+, K+, Ca2+, Mg2+, and Cd, Co, Cr, Fe, Mn, Pb, Sr, Zn using flame atomic absorption spectrophotometer FAAS (NOVAA 300 Analytica JENA AJ with detection limits varying from 0.001–0.02 µg/L. Each sample was analyzed in duplicate. The accuracy and precision of the analytical method was evaluated by the analysis of a reference material (NASS-5), with recoveries ranging between 98.02–104.01%. A total of 5 mL of filtered samples was used to measure NH<sup>4</sup> <sup>+</sup>, Cl−, NO<sup>3</sup> <sup>−</sup>, SO<sup>4</sup> <sup>2</sup>−, PO<sup>4</sup> <sup>3</sup><sup>−</sup> by ion chromatography (IC) (Dionex ICS 1600, Thermoscientific). A total of 25 mL of filtered samples were titrated with 0.02 N H2SO4, using phenolphthalein and methyl orange as indicators to determine total alkalinity (TA) of the water samples. There are two pH endpoints corresponding to the above indicators at 8.3 and at 4.3. TA was calculated using the following equation: Mn, Pb, Sr, and Zn, as follows: 50 mL of filtered samples were used to determine the concentrations of Na+, K+, Ca+2, Mg+2, and Cd, Co, Cr, Fe, Mn, Pb, Sr, Zn using flame atomic absorption spectrophotometer FAAS (NOVAA 300 Analytica JENA AJ with detection limits varying from 0.001–0.02 µg/L. Each sample was analyzed in duplicate. The accuracy and precision of the analytical method was evaluated by the analysis of a reference material (NASS-5), with recoveries ranging between 98.02–104.01%. A total of 5 mL of filtered samples was used to measure NH4+, Cl−, NO3−, SO4−2, PO4−3 by ion chromatography (IC) (Dionex ICS 1600, Thermoscientific). A total of 25 mL of filtered samples were titrated with 0.02 N H2SO4, using phenolphthalein and methyl orange as indicators to determine total alkalinity (TA) of the water samples. There are two pH endpoints corresponding to the above indicators at 8.3 and at 4.3.

*Water* **2020**, *12*, x FOR PEER REVIEW 4 of 17

instruments). Water samples were kept refrigerated at 4 °C and transported to a water laboratory (Yarmouk University, Jordan) for subsequent chemical analyses. Sample

for total alkalinity (TA), Cl−, NO3−, SO4−2, PO4−3, NH4+, Ca+2, Mg+2, Na+, K+, Cd, Co, Cr, Cu, Fe,

preparation and analysis followed APHA [28] procedures.

TA = (volume of acid used \* normality of acid \* 50,000)/volume of sample. TA was calculated using the following equation: TA = (volume of acid used \* normality of acid \* 50,000)/volume of sample.

The average ionic mass balance for water quality data was −0.02% indicating a high level of accuracy. The average ionic mass balance for water quality data was −0.02% indicating a high level of accuracy.

### **3. Results and Discussion 3. Results and Discussion**

The results of dissolved metals and physicochemical properties of coastal water are tabulated in Table 1, and presented in Figures 2–5. The pH values of surface water layer ranged between 8 and 8.49, with a mean value of 8.26 (Table 1 and Figure 2). They indicate a slightly basic coastal water. They showed a slight spatial variability with no distinct trends. This is likely related to the calcium carbonate buffering capacity of water [23]. The results of dissolved metals and physicochemical properties of coastal water are tabulated in Table 1, and presented in Figures 2–5. The pH values of surface water layer ranged between 8 and 8.49, with a mean value of 8.26 (Table 1 and Figure 2). They indicate a slightly basic coastal water. They showed a slight spatial variability with no distinct trends. This is likely related to the calcium carbonate buffering capacity of water [23].

**Figure 2.** The electrical conductivity (EC), total dissolved solids (TDS), and pH of GoA coastal water. **Figure 2.** The electrical conductivity (EC), total dissolved solids (TDS), and pH of GoA coastal water.



**Table 1.** Characteristics of surface seawater layer, GoA, Jordan.


**Table 1.** *Cont.*


*Water* **2020**, *12*, x FOR PEER REVIEW 5 of 17

*Water* **2020**, *12*, x FOR PEER REVIEW 5 of 17

**Figure 3.** NO3<sup>−</sup>, NH4+ and PO4−3 in coastal water of the GoA, Jordan. **Figure 3.** NO<sup>3</sup> <sup>−</sup>, NH<sup>4</sup> <sup>+</sup> and PO<sup>4</sup> <sup>3</sup><sup>−</sup> in coastal water of the GoA, Jordan. **Figure 3.** NO3<sup>−</sup>, NH4+ and PO4−3 in coastal water of the GoA, Jordan.

**Figure 4.** The spatial distributions of Na+, K+, Mg+2, Sr, Ca+2, SO4−2 and Cl<sup>−</sup> concentrations in surface seawater of GoA, Jordan. **Figure 4.** The spatial distributions of Na+, K+, Mg+2, Sr, Ca+2, SO4−2 and Cl<sup>−</sup> concentrations in surface seawater of GoA, Jordan. **Figure 4.** The spatial distributions of Na+, K+, Mg2+, Sr, Ca2+, SO<sup>4</sup> <sup>2</sup><sup>−</sup> and Cl<sup>−</sup> concentrations in surface seawater of GoA, Jordan. **Figure 4.** The spatial distributions of Na+, K+, Mg+2, Sr, Ca+2, SO4−2 and Cl<sup>−</sup> concentrations in surface seawater of GoA, Jordan.

**Figure 5.** Metal contents in surface seawater in GoA, Jordan. **Figure 5.** Metal contents in surface seawater in GoA, Jordan. **Figure 5.** Metal contents in surface seawater in GoA, Jordan.

**Figure 5.** Metal contents in surface seawater in GoA, Jordan. In addition, these pH values are probably attributable to low growth levels, and production of algal biomass that would contribute organic acids to coastal water when decomposed. While the weather conditions of high temperatures and abundant sunlight allow phytoplankton to grow in abundance, our sampling campaign coincided with a period of nutrient-depleted and stratified water, where photosynthetic activity was at its lowest levels. Manasreh et al. [23] recorded the lowest chlorophyll-a levels in summertime and highest In addition, these pH values are probably attributable to low growth levels, and production of algal biomass that would contribute organic acids to coastal water when decomposed. While the weather conditions of high temperatures and abundant sunlight allow phytoplankton to grow in abundance, our sampling campaign coincided with a period of nutrient-depleted and stratified water, where photosynthetic activity was at its lowest levels. Manasreh et al. [23] recorded the lowest chlorophyll-a levels in summertime and highest In addition, these pH values are probably attributable to low growth levels, and production of algal biomass that would contribute organic acids to coastal water when decomposed. While the weather conditions of high temperatures and abundant sunlight allow phytoplankton to grow in abundance, our sampling campaign coincided with a period of nutrient-depleted and stratified water, where photosynthetic activity was at its lowest levels. Manasreh et al. [23] recorded the lowest chlorophyll-a levels in summertime and highest during winter. Microbial decomposition of dead phytoplankton, algae and other flora, In addition, these pH values are probably attributable to low growth levels, and production of algal biomass that would contribute organic acids to coastal water when decomposed. While the weather conditions of high temperatures and abundant sunlight allow phytoplankton to grow in abundance, our sampling campaign coincided with a period of nutrient-depleted and stratified water, where photosynthetic activity was at its lowest levels. Manasreh et al. [23] recorded the lowest chlorophyll-a levels in summertime and highest during winter. Microbial decomposition of dead phytoplankton, algae and other flora, produces humic substances, organic acids and amino acids that

during winter. Microbial decomposition of dead phytoplankton, algae and other flora,

during winter. Microbial decomposition of dead phytoplankton, algae and other flora,

raise the seawater acidity. Additionally, higher temperature during summertime will reduce dissolved CO<sup>2</sup> levels in coastal water and increases the pH value.

TA (total alkalinity) concentrations ranged between 128 and 162 mg/L, with an average value of 146 mg/L. These high values are consistent with the coastal water's buffering capacity (due to high contents of calcium carbonates) of the GoA's water. The TDS varied from about 41.22–42.74 g/L with an average of 41.95 g/L. EC varied from 51.42 to 52.87 mS/cm with average of 52.17 mS/cm. Similar values of TDS were reported along the Saudi GoA with an average of 41.4 g/L [9].

The spatial pattens of TDS in water showed insignificant variations (Figure 2). The elevated levels of TDS indicate highly saline conditions in the GoA and are primarily attributable to its geographic location in a subtropical desert region, with very high evaporation rates, very low precipitation, and negligible freshwater input. Manasreh et al. [23] reported an evaporation of 2 m/year in the GoA with an increasing salinity toward the north. In addition to these salinity raising factors, the high TDS values are linked to water stratification and poor water circulation during the sampling period. These factors created unique environmental conditions of higher temperature, evaporation, and salinity than average, compared to the average range for oceans. Lack of input of freshwater into the coastal water contributes to high salinity water. A negligible supply of terrigenous sediments into the water results in clear water conditions with high transparency.

Higher water density is often observed in summertime in response to high salinity (due to poor water mixing). From July–August, a stratified water column dominates with thermocline and pyncnocline, occurring at about 250 m in 2013 and near 350 m in 2014 and 2015 [23]. The northwards currents drag warm and saline waters to GoA from the Red Sea [25,29]. The flow of surface water from the Red Sea to the GoA is triggered by the high evaporation rate, where the flowing water offsets the evaporation loss [30]. The GoA surface seawater temperature is 2 ◦C lower than that of the Red Sea, where the flowing water brings heat that increases temperature, evaporation rates and salinity of surface seawater layer.

In addition, the TDS values become higher in August, corresponding to the summer season, a period of high dust storm events. The atmospheric dust input is an significant source of salts (and metals) to GoA water [8,9]. The GoA is located in a desert-belt region with frequent dust storms, where Negev and Sinai Deserts are in the west and the Arabian Desert is in east. It is believed that large quantities of dust aerosols delivered to GoA is originated from adjacent deserts [31]. Dust deposition will significantly influence the composition of GoA seawater, where the deposition rate of dust in GoA is one of the highest on Earth [13] ranging between 28 g/m<sup>2</sup> /year in the northwestern part [13] to about 34.68 g/m<sup>2</sup> /year in Aqaba city at the northeastern corner of GoA [8]. TDS was significantly correlated with Ca2+, Mg2+, Na+, Sr and Cl<sup>−</sup> with r = 0.86, 0.88, 0.81 and 0.70, respectively (Table 2). These ions are major contributors to seawater salinity.

Inorganic nutrients (nitrate, ammonium, phosphate) are minor constituents of seawater, but are essential for marine ecosystem productivity and growth. Relatively low levels of inorganic nutrients (NH<sup>4</sup> <sup>+</sup>, PO<sup>4</sup> <sup>3</sup>−, NO<sup>3</sup> −) (Table 1 and Figure 3) were observed in surface water layer, consistent with the findings of others [23,32–35]. The coastal water in GoA is in extremely oligotrophic conditions, with very limited nutrients supplied to Gulf's water through terrestrial runoff.

Nitrates are present in all water samples, where the concentrations increased slightly in some locations, although not all. NO<sup>3</sup> − concentrations ranged between 12.22 mg/L up to 15.50 mg/L, with overall mean and median levels of about 13.85 and 13.78 mg/L, respectively (Table 1 and Figure 3). Ammonium levels fluctuated from 13.08 and 16.91 mg/L, with mean and median values of about 15 mg/L (Table 1 and Figure 3). While the nutrient levels generally varied, their variations showed no spatial trends. Nitrate and ammonium showed relatively similar ups and downs and were significantly correlated with r = 0.65 (Table 2).



**Table 2.** Correlation matrices for surface water quality parameters.

1.00 0.11 −0.04 0.00 0.50 **Zn** 1.00 −0.22 −0.16 −0.09 **pH** 1.00 0.94 −0.14 **EC** 1.00 −0.16 **TDS** 1.00 **TA**

Nitrate is the major nitrogen species in the oxic zone, while ammonium dominates in the anoxic zone. The nitrification is a kinetic reaction and is dependent on several water conditions such as salinity, pH, and Eh [36,37]. However, the coexistence of nitrate and ammonium can trigger or slow down nitrogen conversion like nitrification or denitrification [38]. The coexistence of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> − may result from poor mixing in groundwater, especially in locations where both nitrogen species are released from active pollution sources [39,40]. The GoA waters are well oxygenated with redox indicators of oxidizing conditions [9]. Manasreh et al. [23] observed that the GoA water is well-ventilated due to the annual water mixing with complete saturation (100%).

This suggests that the presence of high NH<sup>4</sup> <sup>+</sup> levels could be associated with leaks from sewer system and/or because of water discharged from fish farm or fertilizer plume. Phosphate varied from 0.16–0.29 mg/L, with an average of 0.21 mg/L (Table 1 and Figure 3). Relatively higher concentrations were observed in the northern GoA, in close proximity to the phosphate terminal, where deposition of dust containing phosphate during loading/unloading activities may contribute phosphates to seawater. While phosphate showed low levels, it is the limiting nutrient for phytoplankton growth. Weak correlations between phosphate and both ammonium and nitrate were observed with r = −0.02 and 0.18, respectively (Table 2).

Aeolian deposits can provide important nutrients which stimulate the primary productivity in marine ecosystems [41–44], especially in oligotrophic water [45], like the GoA [10,20]. However, they can also deliver various contaminants that negatively impact the aquatic biodiversity.

Similar observations were reported by Badran [46], where phosphate and nitrate levels in surface water varied seasonally, with the lowest in summer and the highest in winter. The sampling period was concurrent with water stratification, and the concentrations of nutrients in surface water layer were low. In the winter season, winds drive convective mixing of deep (nutrient-rich) and surface waters, where nitrate and other nutrients are injected into the euphotic zone, resulting in seasonal plankton blooms [47]. The highest productivity (chlorophyll-a) is expected during the winter season, which declines to minimum levels in summertime [23].

Water stratification and high sunlight irradiation during summertime further draw down the inorganic nutrients in the surface water by enhancing primary productivity at the subsurface water layer (50–75 m) [34]. During photosynthesis, phytoplankton assimilate nutrients, and it is the availability of inorganic nitrogen that often limits the rate of primary production in the sea [32]. Nutrients uptake within the euphotic zone in oligotrophic water body results in a considerable depletion of their levels. Phytoplankton communities in oligotrophic waters are likely to survive by utilizing recycled nutrients [48,49].

Nutrient levels in the southern Red Sea are greater than its northern and central regions. The inflow of surface water from the Red Sea to GoA (to compensate for the high evaporation loss) is a contributing factor to lower levels of nutrients observed in GoA seawater (oligotrophic water). In late summer, an increase of 25% in nutrient levels is observed in the southern Red Sea compared to the central region, due to the inflow of nutrient-rich waters from the Gulf of Aden to the southern area of the Red Sea [50]. The highest levels of phosphate in the southern Red Sea are usually observed in October, following upwellings in the Arabian Sea.

In addition to water mixing, it is likely that nitrate and other nutrients are associated with atmospheric deposition, as this area experiences frequent dust storms. Rare flash floods carrying terrigenous sediments can be a minor contributor to nitrate and phosphate in coastal water. Dust from the phosphate terminal in the GoA provides further evidence of contribution of aeolian dust to coastal water. Fish farming and wastewater discharges may be important sources of nitrate and phosphate, as water samples were collected adjacent to the coastlines closer to touristic, industrial and other human facilities.

Sr content in seawater ranged between 9.17 and 10.42 mg/L (Table 1). The concentrations of Ca2+, Mg2+, Na<sup>+</sup> and K<sup>+</sup> in surface seawater layer varied from 423.32–486.99, 2246.2–2355.9, 9541.5–12,646.9 and 512.84–712.91 mg/L, respectively (Table 1). High temperature and evaporation rates are main

contributors to high levels of ions, among others. The spatial distributions of Ca2+, Mg2<sup>+</sup> and Sr in surface water exhibited relatively similar patterns with no trends. Ca2<sup>+</sup> was positively correlated with Mg2+, Sr, and Na<sup>+</sup> with correlation coefficients of 0.97, 0.85, and 0.52, respectively. The correlation coefficients between Mg2<sup>+</sup> and Sr was 0.87, and between Mg2<sup>+</sup> and Na<sup>+</sup> is 0.59. The K<sup>+</sup> levels in seawater was not significantly correlated with any cation tested.

Cl− exhibited spatial changes in surface water, with concentrations ranging from 22,172.88–25,991.94 mg/L. The SO<sup>4</sup> -2 values in seawater samples varied between 316.99 and 407.45 mg/L. Cl was correlated with Ca2+, Mg2<sup>+</sup> and Na<sup>+</sup> with correlation coefficients of 0.54, 0.60 and 0.74, respectively. Similar to K+, the SO<sup>4</sup> <sup>2</sup><sup>−</sup> content showed no significant correlations with any other ions of seawater. TDS values were well correlated with Cl−, Ca2+, Mg2+, Na<sup>+</sup> and Sr, with r = 0.70, 0.86, 0.88, 0.81 and 0.81, respectively. Whereas salinity was neither correlated with SO<sup>4</sup> <sup>2</sup><sup>−</sup> nor with K+. Dust deposition to GoA is also an important contributor to TDS and other ions.

Table 3 compares seawater chemistry of the GoA (northernmost Red Sea) analyzed in this study relative to other regions of the Red Sea. Relatively elevated levels of pH, TA, Cl−, NO<sup>3</sup> <sup>−</sup>, PO<sup>4</sup> <sup>3</sup>−, NH<sup>4</sup> +, Mg2<sup>+</sup> and K were observed for GoA, compared to the central and northern Red Sea. Whereas SO<sup>4</sup> <sup>2</sup>−, TDS, Ca, and Na showed values that are comparable to or lower than those for other parts of the Red Sea water.


**Table 3.** Comparison of water quality parameters in different regions of the Red Sea and GoA.

The concentrations of metals in surface seawater layer are shown in Table 1 and presented in Figure 5. Zn showed the highest concentration, with an average concentration of 6.05 µg/L. Other metals that followed were in the order Cr > Fe > Cu > Mn > Cd > Pb = Co (Table 1, Figure 5).

Spatial variability of metals contents in seawater samples exhibited increasing trends toward the south (Figure 5), where the industrial complex is located. Potential impacts from heavy metals are commonly confined to areas in the vicinity of urban or industrialized regions on the coastal edge. However, these levels of metals also suggest that they have probably been derived from multiple sources, including a geogenic origin.

Al-Taani et al. [9] reported high levels of dissolved oxygen, with redox values indicating oxidizing conditions in the coastal water GoA, which may favor immobilization of some metals with relatively low levels in seawater samples. Zn in seawater varied from 3.63–8.28 µg/L with an average of 6.05 µg/L (Table 1 and Figure 5). These values are higher than those reported for the Saudi GoA [9], the offshore surface seawater of Red Sea [53], the average oceanic concentration [54], and the Mediterranean surface seawater [55] (Table 4).

In addition, atmospheric dry deposition is the primary external source of trace metals to GoA [14]. Aeolian dry fluxes of certain trace elements (e.g., Cd, Pb, Cu and Zn) to the ocean water may surpass those of riverine sources [56,57]. Aeolian dust of Zn to GoA ranges between 1.68 mg/m<sup>2</sup> /year (in Eilat city at the northwestern corner; [31]) and 4.02 mg/m<sup>2</sup> /year (in Aqaba city in northeastern region; [8]). High concentrations of Zn were observed GoA seawater ranging from 5.71–11.55 µg/L [58] in the vicinity of Industrial Complex.


**Table 4.** Comparison of selected metals (µg/L) in surface seawater of GoA relative to other regions.

a : [59,60]. <sup>b</sup> : [61]. <sup>c</sup> : [62]. <sup>d</sup> : [63,64]. <sup>e</sup> : [65,66]. <sup>f</sup> : [67]. <sup>g</sup> : [62,65].

Fe contents of seawater varied from 0.75–1.94 µg/L with a mean value of 1.29 µg/L. These concentrations of Fe showed spatial variability in surface water layer with generally greater values in the southern GoA (Figure 5). During the stratified summer, surface water becomes enriched in Fe [13], but the winter mixing of surface and deep water layers, decreases these Fe levels. The average concentration of Fe measured in the present study is relatively comparable to that for the Red sea offshore seawater [53], but higher than the averages for oceanic concentration [54] and the Mediterranean seawater [68] (Table 4). Higher concentrations of Fe were observed in the Saudi GoA with about 15.25 µg/L, suggesting that atmospheric dry deposition in this area is more intense. It is believed that Fe is probably derived from crustal sources. The average dry flux of Fe to the GoA waters varied from about 216 mg/m<sup>2</sup> /year [31] to 440 mg/m<sup>2</sup> /year [8].

Cr levels varied from 0.96–1.91 µg/L with average value about 1.44 µg/L (Table 1). These values are higher than those observed in the Saudi GoA seawater [9], the mean oceanic level [54], but less than those for the Mediterranean Sea [63,64] (Table 3). The spatial pattern of Cr suggests that these high levels of Cr at the southern end of GoA are most likely related to discharge of brine water from desalination plant [68]. In addition to the industrial wastewater, mineral dust from fertilizer and cement factories remain potential sources of Cr to GoA seawater. atmospheric aerosol deposition to GoA fluctuated between 0.96 mg/m<sup>2</sup> /year in Eilat city [31], and about 1.42 mg/m<sup>2</sup> /year in Aqaba city [8].

Mn content in southern water samples exhibited higher values relative to the northern part of GoA (Table 1 and Figure 5), and it is likely to originate form anthropogenic emissions [31]. The Mn contents in surface seawater ranged between 0.68 and 1.15 µg/L, with an average of 0.88 µg/L, which is three times as much as that measured for the Saudi GoA [9]. These values are also greater than those reported for the offshore water of Red Sea [53], the average oceanic concentration, and those for the Mediterranean Sea [67] (Table 4). Windblown dust of Mn to the uppermost eastern GoA (Elat city) averaged 5.28 mg/m<sup>2</sup> /year [31], whereas, in the northernmost extension, a mean value of 10.29 mg/m<sup>2</sup> /year was reported [8]. In addition, Mn is probably related to desalination plants in the neighboring cities (Eilat, Taba and Haql) [3,9], where various heavy metals, including Mn, may be released with the discharged water of thermal desalination plants, depending on the metal alloys used [68,69].

Cd concentrations in seawater ranged between 0.2 and 0.76 µg/L, with an average of 0.51 µg/L (Table 1). Similarly, elevated levels of Cd were found in the southern GoA (Figure 5), where industrial activities are concentrated. Comparable levels of Cd have been reported by Shriadah et al. [53] for the northern Red Sea (offshore seawater), but higher than those for the average oceanic levels of about 0.07 µg/L [54]. Additionally, the average Cd content in surface seawater measured in the present study is greater than those for Saudi GoA and the Mediterranean Sea (Table 4).

In addition to desalination plants, the anthropogenically derived Cd (and other metals such as Pb and Co) is likely related to the discharge of cooling water and sewage in the southern GoA [26,70,71]. The average concentration of Cd from atmospheric dust varied from 0.012 mg/m<sup>2</sup> /year in Eilat city [31] to 0.04 mg/m<sup>2</sup> /year in Aqaba city [8].

Co content in surface seawater exhibited little spatial variations, with relatively higher levels were observed in the southern portion of GoA (Figure 5). Co ranged from 0.28–0.51 µg/L, with average Co values of about 0.38 µg/L (Table 1). Lower average Co contents were reported for the northern Red Sea offshore water [53] and for the Saudi GoA [9]. Co occurs in seawater at concentrations below 0.005 µg/L [59,60]. However, higher average value has been detected in the surface water of Mediterranean Sea [62]. Potential sources of Co in GoA seawater are likely similar to those of Co [9]. Dust particles collected from GoA region showed an average of 0.1 mg/m<sup>2</sup> /year [31].

Relatively elevated levels of Cu were detected in seawater samples, varying from 0.69–1.91 µg/L, with greater values observed for the southern sampling sites (Figure 5). These values are higher than the average ocean level [54] and the offshore water of Red Sea [53] (Table 3). An average Cu value of 0.2 µg/L was reported for the Mediterranean Sea [67,68]. However, elevated levels of Cu were detected in the Saudi GoA (at the Jordan–Saudi border). The atmospheric dry deposition flux of Cu to GoA region ranged from 0.38 mg/m<sup>2</sup> /year in Eilat [31] to 0.68 mg/m<sup>2</sup> /year in Aqaba city [8]. Elevated concentrations of Cu have been reported in Jordanian GoA water, in the range of 0.74–2.28 µg/L [58], and are higher than those measured in this study.

Pb levels in seawater samples varied from 0.17–0.79 µg/L, with a spatial pattern of increasing levels in the southern part of GoA. The average Pb value of 0.38 µg/L is comparable to those measured in offshore surface water sites of Red Sea [53], but higher than those reported for the Saudi GoA [9], the average ocean, and the Mediterranean surface seawater (Table 3). Pb is of anthropogenic origin, mainly from fossil fuel burning [72]. Elevated concentrations of Pb were found in the Jordanian GoA water ranging between 0.73 and 1.43 µg/L [56]. GoA receives high dry flux of Pb varying from 0.8 mg/m<sup>2</sup> /year in Eilat [31] to about 1.42 mg/m<sup>2</sup> /year in Aqaba city [8]. All metals tested were significantly correlated (Table 2) indicating that they may have been derived from similar sources.

### **4. Conclusions**

GoA is a place for rich and diverse marine ecosystems. It is highly vulnerable to pollution, wherehuman activities in the bordering countries are intense, with high potential for water contamination. This requires cross-border collaboration to protect these naturally diverse but stressed ecosystems. This study intended to ascertain the seawater quality conditions along the eastern coast of GoA, Jordan. The sampling campaign coincided with a period of low levels of inorganic nutrients, low rates of algal growth with reduced microbial decomposition of dead algal cells. In addition to prevailing saline conditions of high temperatures and high evaporation rates, the water stratification and intense dust storms are the major contributing factors to the observed seawater chemistry. The surface distribution of water quality variables showed spatial variations with no specific patterns, except for metal contents, which exhibited southward increasing trends, closed to the industrial complex. The vast majority of these quality parameters showed relatively higher values compared to those of other regions.

**Author Contributions:** Conceptualization, A.A.A.-T. and M.R.; methodology, J.I.; software, A.A.-R.; validation, F.H., Y.N.; formal analysis, S.K.; investigation, A.A.A.-T.; resources, A.A.B.; data curation, Y.N.; writing—original draft preparation, A.A.A.-T.; writing—review and editing, F.H.; visualization, A.A.-R.; supervision, A.A.A.-T.; project administration, M.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
