Abstract

Water resources scarcity in Algeria, their fragility and their unequal distribution have resulted in a serious shortage, which, in spite of all the efforts, seems inevitable. This study consists of evaluating the impact of human activity on the water quality of Aïnzeda lake (NE Algeria), a typical case study of the difficulties posed by the problem of surface water quality in semi-arid regions. Principal Component Analysis (PCA) and the trend method were applied to interpret the physico-chemical data of monthly analyzed samples, over a 25-year period (1988–2012). The trend method results show that most chemical elements have a direct relationship with urbanization and agricultural practices in the area. The change in the watershed climatic conditions (increase of 9% in air temperature, 7% in the lake water temperature, and decrease of 8% in precipitation) is also responsible for the degradation of the water quality. The PCA shows that salinization (51.73%), and anthropogenic and agricultural pollution (13.49%) are the most significant degradation factors. These two approaches have enabled us to prove that aridity and anthropogenic or agricultural activities have a negative impact on the lake's surface water quality.

HIGHLIGHTS

  • The study consists of evaluating the impact of human activity on the water quality of Aïnzeda lake (NE Algeria).

  • The Principal Component Analysis and the trend method were applied to interpret the physico-chemical data of monthly analyzed samples, over a 25-year period (1988–2012).

  • The results show that anthropogenic and agricultural pollution are the significant degradation factors for the lake's surface water quality.

INTRODUCTION

Given the aridity of most Algerian territory, access to water is still a lingering problem. It can be seen as a real obstacle to the country's development. Currently, Algeria has 75 dams with a total capacity of 6.5 billion m3. Nevertheless, taking into account such factors as population growth, urbanization, industrialization and agricultural sector development, it is mandatory to continue investing in these hydraulic infrastructures since the renewable freshwater resources are on the decrease (these have dropped from 962 m3/inhabitant in 1962 to less than 290 m3/inhabitant in 2014). To face this situation, Algerian authorities are planning to increase the total number of dams to 140 by 2030, aiming to double the storage capacity by nearly 12 billion m3 throughout the country. However, if the construction of these hydraulic infrastructures is indeed a necessity to guarantee the essential water supply to the country throughout the year, it is still urgent to control and safeguard the dams' water quality.

Several dam lakes have experienced water quality degradation problems over the past few decades (Belhadj et al. 2011; Bouzid-Lagha & Djelita 2012; Boudoukha & Boulaarak 2013; Hadji et al. 2013; Demdoum et al. 2015; Mokadem et al. 2016; Hamed et al. 2017; Hamad et al. 2018). This is mainly due to different pollution sources. Added to that, the slow natural reservoir restocking process and the impact of climate change have to be highlighted as other major difficulties.

Data series can be analyzed using several numerical models (Regnie 1965; David et al. 1983). For example, the use of the linear trend technique makes it possible to emphasize the possible effects of man-made pollution and detect breaks in the time series (Etchanchu & Probst 1988). However, the disparity of the data imposes our immediate resort to multidimensional data analysis methods such as Principal Component Analysis (PCA). The PCA method has been widely used in the analysis of water quality data to obtain useful information such as a reduced number of latent factors/sources of pollution and assessment of temporal and spatial variations (Banerjee et al. 2019).

Aïnzeda lake is the main source that supplies the various urban centers of the highlands region with drinking and industrial water. This dam has experienced water quality deterioration due mainly to the various contaminants added to the natural processes of erosion and leaching.

In the last few decades, some authors have reported on the assessment of surface water quality of dams situated in the north-east of Algeria by using the multivariate statistical technique (PCA) or trend method (Boudoukha & Boulaarak 2013; Guerraiche et al. 2016; Bouguerne et al. 2017) but over a short period (not exceeding 10 years) or using a limited number of physico-chemical data. However, in this research, the PCA technique and the trend method were used jointly to analyze the recent physico-chemical data over a long period (25 years).

This study deals with the chemical composition fluctuations of Aïnzeda dam water between 1988 and 2012, and explains the changes in the concentrations of the main dissolved components.

Study area

Aïnzeda lake is located 25 km west of Setif city. The coordinates of its location are 36° 10′ 28″ North latitude and 5° 08′ 58″ East longitude. The capacity of this dam is 125 M·m3, of which 110 M·m3 are useful. The reservoir area is about 1,140 hectares, with a maximum depth of 26 m. The dam's waters ensure the supply for a population of 2 million inhabitants with water for both drinking and industrial purposes. The lake is at the outlet zone of the Boussellam-upstream watershed, the area of which is 1,785 km2 (Figure 1(a)). The watershed is bounded by a ridge whose peaks reach between 1,050 m and 1,737 m, while plains range between 870 m to the south and center to 1,050 m to the north, with a slope of about 2.7% (Mebarkia 2011; Hadji et al. 2014).

Figure 1

(a) Geographical location of the study area, (b) lithological map of the Boussellem sub-watershed: [1 and 2: Quaternary formations (1. alluvions current and recent, 2. old alluvium, limestone soils crust and fluvial gravels), 3 and 4: silts and conglomerates (3. red conglomerates, pudding stones, clays, gypsum of the Mio-Pliocene, 4. conglomerates, yellow oyster marl, Miocene), 5: alternating coarse sandstones and clays of the web of Numidian flysch, 6 and 7: marl and calcareous marl of tablecloth Tellians (Eocene), 8: bedded limestone in alternation with the marly benches (Cretaceous), 9: limestone and dolomitic limestone (Jurassic-Cretaceous), 10: formations dominated by gypsum (Trias)].

Figure 1

(a) Geographical location of the study area, (b) lithological map of the Boussellem sub-watershed: [1 and 2: Quaternary formations (1. alluvions current and recent, 2. old alluvium, limestone soils crust and fluvial gravels), 3 and 4: silts and conglomerates (3. red conglomerates, pudding stones, clays, gypsum of the Mio-Pliocene, 4. conglomerates, yellow oyster marl, Miocene), 5: alternating coarse sandstones and clays of the web of Numidian flysch, 6 and 7: marl and calcareous marl of tablecloth Tellians (Eocene), 8: bedded limestone in alternation with the marly benches (Cretaceous), 9: limestone and dolomitic limestone (Jurassic-Cretaceous), 10: formations dominated by gypsum (Trias)].

Geologically speaking (Figure 1(b)), the plain is largely covered by heterogeneous Mio-Plio-Quaternary soils formed by clays, silts, alluviums and calcareous crusts (Boudoukha et al. 1997). Northern and southern borders are formed by carbonate lands belonging, respectively, to the Tell lands and the allochthonous South-Setifian complex (Vila 1980). Most of the watershed area consists of low to medium permeability soils represented by sandy clays and lacustrine limestones; the rest is occupied by clays and impermeable marls. The watershed vegetation cover is very weak, which allows heavy rainfall to displace huge amounts of suspended matter in addition to other soluble elements, to the reservoir.

The region is characterized by a semi-arid climate, with an average monthly temperature of 14.8 °C during the period (1988–2012). The average inter-annual rainfall is around 360 mm. Evaporation takes up to about 94% of the precipitation, while the remaining 6% will either flow with the runoff or be infiltrated. A 20 mm annual rainfall excess enables a permanent flow with an average of 1.14 m3/s during the rainy period (January and February). For the rest of the year, the region is subjected to an irregular hydrological regime allowing the leaching of the soluble chemical elements, especially fertilizers and the agglomerations' sewage water, with an average annual flow of 0.42 m3/s measured at the Farmatou station. This causes the reservoir to receive a large portion of the annual liquid inputs in a few weeks and to subsequently become a stagnant water mass where evaporation is the dominant factor with an annual average of 11.44 Mm3/year.

METHODS

In order to study the temporal variation of water chemical parameters and to understand the pollution mechanisms, we have collected the results of sampling and analysis series undertaken by the laboratory of the National Hydraulic Resources Agency (ANRH) which has adopted a sampling frequency of one sample per month during a 25-year observation period (1988–2012), completed by our own sampling for two years with the same frequency. This adopted strategy allowed us to collect 300 water analyses.

Temperaure, pH, electrical conductivity (EC) and dissolved oxygen (O2dis) were measured in situ using a WTW multiparameter field device. Concentrations of calcium (Ca2+), magnesium (Mg2+), sodium (Na+), potassium (K+), chloride (Cl), sulphates (SO42−), bicarbonates (HCO3), nitrates (NO3), nitrites (NO2), ammonium (NH4+), orthophosphate (PO43−), organic matter (OM), chemical oxygen demand (COD) and biochemical oxygen demand after 5 days (BOD5), total suspended solids (TSS) and turbidity were determined according to standard technical analyses (Rodier 2009).

The trends method was used for the results analysis treatment. It consists of adjusting a cloud of points (t, Yt) by a line passing through two calculated points. The series is subdivided into two subsets of the same size. For each subset, the averages of the ‘t’ and ‘Yt’ are calculated.

Two points, (t1, Y1), (t2, Y2), called average points, are therefore obtained through which the trend line passes. Instead of estimating the average points, we can rather calculate the median, which enables us to limit the outliers' effect. Based on the linear least squares method, these models allow adjustment of a time series (Yt) with the function Ct = at + b. We determine the least squares (trend) straight line (Ct = at + b) of the point cloud (t, Yt), which minimizes the distance Σ (Yt − (at + b))2. This method enables us to get a better fit for the trend line, whose effectiveness is limited where an outlier persists in the series. If the line slope is ascending, it reflects an increase in the subject matter parameter, and we can deduce a positive trend; otherwise, we may reveal a negative tendency.

The long-run trends in Aïnzeda lake chemistry were quantified over the period 1988–2012 using linear regressions between instantaneous monthly concentrations and time (Table 1).

Table 1

Statistical characteristics of climatic factors and physico-chemical data of Aїnzeda dam water from 1988 to 2012

ParametersTrend parameters
Statistical parameters
Norms
Relation: Ct/tCt0CtfTrends (%)MinMaxMeanMedianSDCVWHO.N.A.N.
VL (Mm3Ct = −0.0007t + 115.62 95 88.5 −7 19.7 125.6 91.17 96.6 27.02 0.007 
P (mm) Ct = −0.0004t + 43.67 32 29.5 −8 137.9 30.7 27.02 25.04 0.003 
Air T (°C) max Ct = 0.0038t + 19.995 5.1 5.9 +15 6.3 36.4 20.56 19.25 8.54 0.0015 
mean Ct = 0.0038t + 14.229 14.2 15.4 +9 2.2 28.9 14.79 14 7.54 0.002 
min Ct = 0.0026t + 5.1018 20 21.1 +6 −6.3 18.6 5.5 4.25 6.34 0.0013 
Water T (°C) Ct = 0.0001t + 12.80 16.5 17.7 +7 30 17.1 17 6.63 0.002 <25 25 
EC (μS/cm) Ct = 0.056t–1011.40 785 1,295 +65 600 1,900 1,038.2 1,000 211.5 0.48 1,500 2,800 
TDS (mg/L) Ct = 0.0169t + 102.52 640 800 +25 386 1,020 721.43 710 110 0.16 2,000 2,000 
pH Ct = −0.0007t + 115.62 8.42 7.75 −8 6.83 10 8.11 8.2 0.34 0.006 6.5–8 6.5–8.5 
O2dis (%) Ct = −9E-05t + 90.98 88.1 87.2 −1 43.1 146.34 87.68 89.4 17.25 0.002 80–120  
TSS (mg/L) Ct = 0.005t–113.59 48 90 +87 414 69.46 60 60 0.04 30  
Turbidity (NTU) Ct = 0.0003t–7.83 4.4 +120 0.3 16.8 3.14 2.1 2.88 0.063 
Ca2+ (mg/ L) Ct = 0.0012t + 24.193 64 74.5 +16 26 150 69.33 68 15.89 0.034 200 200 
Mg2+ (mg/L) Ct = 0.0018t–32.508 24.2 38.9 +60 140 31.62 30 16.47 0.07 150 150 
Na+ (mg/L) Ct = 0.0057t–103.34 79 126 +60 18 200.2 102.53 100 29.81 0.2 200 200 
K+ (mg/L) Ct = −0.001t + 42.822 11.1 −72 30 7.05 5.40 0.2 12 20 
Cl (mg/L) Ct = 0.0111t–248.54 109 200 +83 30 315 153.9 150 51.5 0.3 250 500 
SO42− (mg/L) Ct = 0.0059t–54.87 134 182 +35 48 380 157.9 156 43.69 0.102 250 400 
HCO3 (mg/L) Ct = 0.0015t + 107.47 155 167 +7 50 237.9 160.92 162 32.94 0.011 
NO3 (mg/L) Ct = −6E-05t + 5.12 3.3 2.8 −15 24 3.65 0.002 50 50 
NO2 (mg/L) Ct = 1E-05t–0.32 0.01 0.08 +700 0.95 0.05 0.01 0.11 0.06 0.01–0.1 0.1 
NH4+ (mg/L) Ct = 3E-05t–0.88 0.05 0.25 +400 2.04 0.15 0.05 0.277 0.07 0.5 0.5 
PO43− (mg/L) Ct = 8E-06t–0.14 0.13 0.19 +46 1.1 0.16 0.1 0.18 0.01 0.1–0.5 0.5 
OM (mg/L) Ct = 0.0004t–6.34 7.9 12.1 +53 1.6 18.8 10 9.9 0.15 
COD (mg/L) Ct = 0.003t–62.98 61 33.5 +82 10 109 47.23 46.5 20.97 0.14 20–25 30 
BOD5 (mg/L) Ct = 0.0003t–9.01 +150 0.7 14 3.45 2.39 0.14 3–5 
COD/BOD5 Ct = −0.0011t + 60.18 25 16 −36 1.16 111.4 20.5 14.66 17.76 0.03 
N/P Ct = −0.0041t + 192.8 60 24 −60 0.1 150 43 21 44.6 0.06 
ParametersTrend parameters
Statistical parameters
Norms
Relation: Ct/tCt0CtfTrends (%)MinMaxMeanMedianSDCVWHO.N.A.N.
VL (Mm3Ct = −0.0007t + 115.62 95 88.5 −7 19.7 125.6 91.17 96.6 27.02 0.007 
P (mm) Ct = −0.0004t + 43.67 32 29.5 −8 137.9 30.7 27.02 25.04 0.003 
Air T (°C) max Ct = 0.0038t + 19.995 5.1 5.9 +15 6.3 36.4 20.56 19.25 8.54 0.0015 
mean Ct = 0.0038t + 14.229 14.2 15.4 +9 2.2 28.9 14.79 14 7.54 0.002 
min Ct = 0.0026t + 5.1018 20 21.1 +6 −6.3 18.6 5.5 4.25 6.34 0.0013 
Water T (°C) Ct = 0.0001t + 12.80 16.5 17.7 +7 30 17.1 17 6.63 0.002 <25 25 
EC (μS/cm) Ct = 0.056t–1011.40 785 1,295 +65 600 1,900 1,038.2 1,000 211.5 0.48 1,500 2,800 
TDS (mg/L) Ct = 0.0169t + 102.52 640 800 +25 386 1,020 721.43 710 110 0.16 2,000 2,000 
pH Ct = −0.0007t + 115.62 8.42 7.75 −8 6.83 10 8.11 8.2 0.34 0.006 6.5–8 6.5–8.5 
O2dis (%) Ct = −9E-05t + 90.98 88.1 87.2 −1 43.1 146.34 87.68 89.4 17.25 0.002 80–120  
TSS (mg/L) Ct = 0.005t–113.59 48 90 +87 414 69.46 60 60 0.04 30  
Turbidity (NTU) Ct = 0.0003t–7.83 4.4 +120 0.3 16.8 3.14 2.1 2.88 0.063 
Ca2+ (mg/ L) Ct = 0.0012t + 24.193 64 74.5 +16 26 150 69.33 68 15.89 0.034 200 200 
Mg2+ (mg/L) Ct = 0.0018t–32.508 24.2 38.9 +60 140 31.62 30 16.47 0.07 150 150 
Na+ (mg/L) Ct = 0.0057t–103.34 79 126 +60 18 200.2 102.53 100 29.81 0.2 200 200 
K+ (mg/L) Ct = −0.001t + 42.822 11.1 −72 30 7.05 5.40 0.2 12 20 
Cl (mg/L) Ct = 0.0111t–248.54 109 200 +83 30 315 153.9 150 51.5 0.3 250 500 
SO42− (mg/L) Ct = 0.0059t–54.87 134 182 +35 48 380 157.9 156 43.69 0.102 250 400 
HCO3 (mg/L) Ct = 0.0015t + 107.47 155 167 +7 50 237.9 160.92 162 32.94 0.011 
NO3 (mg/L) Ct = −6E-05t + 5.12 3.3 2.8 −15 24 3.65 0.002 50 50 
NO2 (mg/L) Ct = 1E-05t–0.32 0.01 0.08 +700 0.95 0.05 0.01 0.11 0.06 0.01–0.1 0.1 
NH4+ (mg/L) Ct = 3E-05t–0.88 0.05 0.25 +400 2.04 0.15 0.05 0.277 0.07 0.5 0.5 
PO43− (mg/L) Ct = 8E-06t–0.14 0.13 0.19 +46 1.1 0.16 0.1 0.18 0.01 0.1–0.5 0.5 
OM (mg/L) Ct = 0.0004t–6.34 7.9 12.1 +53 1.6 18.8 10 9.9 0.15 
COD (mg/L) Ct = 0.003t–62.98 61 33.5 +82 10 109 47.23 46.5 20.97 0.14 20–25 30 
BOD5 (mg/L) Ct = 0.0003t–9.01 +150 0.7 14 3.45 2.39 0.14 3–5 
COD/BOD5 Ct = −0.0011t + 60.18 25 16 −36 1.16 111.4 20.5 14.66 17.76 0.03 
N/P Ct = −0.0041t + 192.8 60 24 −60 0.1 150 43 21 44.6 0.06 

Ct0 and Ctf: initial and final values calculated using the regression at time t0 (January 1988) and tf (December 2012), Trend (%) =](Ctf – Ct0)/Ct0[×100, Relation Ct/t: Relation: Dependent variable/time; where Ct = at + b, SD: Standard Deviation, CV: Coefficient of Variation, VL: Storage capacity of like, WHO.N.: World Health Organization Norm, A.N.: Algerian Norm.

Principal component analysis (PCA) was also used. This consists of representing graphically the maximum of the information contained in a data table (Diday et al. 1982). The richness of Aïnzeda lake water physicochemical analyses justifies the application of this method. However, the concentration range's disparity imposes the use of reduced relevant data. The purpose of using the PCA for hydrochemical data in the study area was to characterize the water chemistry during the observation period and give a preliminary idea of the pollution elements and sites. The analysis interpretation was achieved using Origin Pro 9.0 software, which contains a module dedicated to statistical processing.

RESULTS AND DISCUSSION

The linear trend was determined using instantaneous concentrations as a function of time (Table 1). The initial and final average values for the considered period allowed us to calculate these variations according to the following formula (Equation (1)): 
formula
(1)

However, Ct0 and Ctf, initial and final values were calculated using the regression at time t0 (January 1988) and tf (December 2012).

In addition, the moving average values are calculated from the mean of the complete data series, using the moving average analysis tool in Excel (Figures 27).

Figure 2

(a) Monthly evolution of the water's volume of the like (VL) and precipitation (P) during 1988–2012 in Aïnzeda meteorological station, (b) Variations of air (min- mean- max) temperatures (air T) with relevance to time during the period 1988–2012, (c) Variations of water temperature of the like (water T) with relevance to time during the period 1988–2012, (d) Changes in total suspended solids (TSS) of the dam's water in function of water volume (1988–2010).

Figure 2

(a) Monthly evolution of the water's volume of the like (VL) and precipitation (P) during 1988–2012 in Aïnzeda meteorological station, (b) Variations of air (min- mean- max) temperatures (air T) with relevance to time during the period 1988–2012, (c) Variations of water temperature of the like (water T) with relevance to time during the period 1988–2012, (d) Changes in total suspended solids (TSS) of the dam's water in function of water volume (1988–2010).

Figure 3

(a) Changes in turbidity of the dam's water in function of water volume (1988–2010), (b) Changes in salinity (electrical conductivity – EC) of the dam's water in function of water volume (1988–2012), (c) variation of dissolved oxygen (O2dis) of the dam water (1988–2012) in function of water volume, (d) variation of the pH of the dam water (1988–2012) in function of water volume.

Figure 3

(a) Changes in turbidity of the dam's water in function of water volume (1988–2010), (b) Changes in salinity (electrical conductivity – EC) of the dam's water in function of water volume (1988–2012), (c) variation of dissolved oxygen (O2dis) of the dam water (1988–2012) in function of water volume, (d) variation of the pH of the dam water (1988–2012) in function of water volume.

Figure 4

Variation of the carbonated elements ((a) Ca2+, (b) Mg2+ and (c) HCO3) and the water volume of the dam during the period 1988–2010.

Figure 4

Variation of the carbonated elements ((a) Ca2+, (b) Mg2+ and (c) HCO3) and the water volume of the dam during the period 1988–2010.

Figure 5

Variation of the salt elements ((a) Na+, (b) Cl, (c) SO42− and (d) K+) and the water volume of the dam during the period 1988–2010.

Figure 5

Variation of the salt elements ((a) Na+, (b) Cl, (c) SO42− and (d) K+) and the water volume of the dam during the period 1988–2010.

Figure 6

(a) Variations of organic matter, (b) COD, (c) BOD5, in the function of water volume and (d) COD/BOD5 report.

Figure 6

(a) Variations of organic matter, (b) COD, (c) BOD5, in the function of water volume and (d) COD/BOD5 report.

Figure 7

Nutrients variation over time ((a) NO3, (b) NO2, (c) NH4+ and (d) PO43−) in function of water volume.

Figure 7

Nutrients variation over time ((a) NO3, (b) NO2, (c) NH4+ and (d) PO43−) in function of water volume.

Impact of weather conditions on the watershed and its water quality

The intense and persistent drought, observed in the region during the study period (1988–2012), is characterized by a rainfall decrease of −8% accompanied by a decrease of the dam's stored water volume of −7%, of which more than 50% occurred during the year 2001–2002. This was the driest year, during which the dam's water volume reached its minimum (19.7 M·m3 in October 2002), but the heavy rains of the years 2002 (633 mm) and 2003 (513 mm) caused the volume to jump about 125.6 M·m3 in a few days (Figure 2(a) and 2(b)).

The average air temperatures recorded positive trends of respectively +6% for the minimum temperature, +9% for the average and +15% for the maximum, which had an effect on the water average temperature that has also recorded an increase of +7% (Figure 2(c)) causing the water salinity to increase due to evaporation.

The changes affecting the watershed due to siltation are estimated to be 189,400 m3/year, due to the nature and morphology of the sloping land, the fragility of the vegetation cover, the lack of afforestation and the urbanization upstream of the dam generating strong erosion that reduces the dam's storage capacity by 0.2%/year.

During the dry periods, the flow depends on discharges from the uncontrolled urban and industrial centers of the region. The self-purification of the various rejects then becomes insufficient, which leads to an enrichment of the watershed in more and more important quantities of soluble and insoluble materials. The agricultural lands cover almost 70% of the watershed area. The vegetation cover is very weak, which allows the mobilization of huge quantities of total suspended solids (TSS) during heavy rainfall periods (reaching their maximum 414 mg/L in February 1988 and an average of 70 mg/L) (Figure 2(d)). In addition, the other soluble elements, which will be driven to the lake, have a high trend of +87% during this period, accompanied by an increase in turbidity (Figure 3(a)) of +120%.

Statistical characterization of the reservoir water physicochemical parameters

The chemical elements statistical analysis (Table 1) shows that the mean and median values are very close, which proves the representativity of the mean and the symmetrical distribution of the samples. The examination of the standard deviation and the variation coefficient shows that the majority of the analyzed parameters have a low to medium variation around the calculated average.

To the parameters Twater, pH, O2xlis and NO3 is assigned a very small variation (<1%) around the mean. Turbidity, TSS, Ca2+, Mg2+, HCO3, NO2, NH4+, and PO43− also show a small variation (<10%) around the mean. For the rest of the elements (Na+, K+, SO42−, TDS (total dissolved solids), OM, COD, and BOD5), there is an average variation between 10% and 30% around the average. On the other hand, the EC and Cl show a strong variation around the average (>30%). The strong variation averages mainly characterize the parameters of the pollution resulting from effluents and the ground leaching following torrential precipitations spread over time.

The use of the moving average shows that the monthly variability is higher than the annual one. This follows a multi-year cycle of 2–4 years in relation to the torrential precipitations, the driving force of this hydro-physico-chemical typology. The examination of these data (Table 1) proves that the average concentrations of certain chemical elements such as organic matter (OM), COD, total suspended solids (TSS), pH and temperature in the waters exceed the WHO standards. In fact, 13.3% of the recorded data exceed the WHO standards during the summer period.

Evolution of the reservoir water chemistry

For most of the chemical parameters, the slope can be seen to differ significantly from zero, except for the parameters K+, NO3, pH, and O2dis where the values of the slope are negative. They are −72%, −15%, −8% and −1%, respectively. However, all the other parameters have a positive slope that varies between +7% and +700%.

Physicochemical parameters

The water salinity represented by the electrical conductivity shows an increase of 65% (Figure 3(b)) following a 7% decrease of the water volume in the dam. This phenomenon is emphasized by the intense evaporation at the dam, where the average air temperature is 9% higher during this period and reaches very high values (up to 36.4 °C) (Mebarkia 2011) accompanied by a 7% increase in the average temperature of the water (Figure 2(c)).

The dissolved oxygen in the water of the reservoir showed a negative trend of −1% (Figure 3(c)), which is consistent with the decrease in alkaline pH (Figure 3(d)) by −8%. This might be tied to the organic material oxidation according to reaction (Equation (2)) or the increase in CO2 pressure, which leads to a decrease in pH according to the formula (Equation (3)) (Kempe 1982). The large area of the water body and the reservoir's shallow depth make it possible to compensate for the oxygen consumption during the organic matter oxidation, which explains the low dissolved oxygen trend of −1%. 
formula
(2)
 
formula
(3)

The major component concentrations are below the WHO standards but show positive trends for Ca2+ (16%), Mg2+ (60%), Na+ (60%), HCO3 (7%), Cl (83%), SO42− (35%), in agreement with the increase of the EC trend except for K+, which records a negative trend of −72%. The increase of Ca2+, Mg2+ and slightly the HCO3 elements (Figure 4(a)–4(c)) may result from the dissolution of the emerging carbonate formations in the watershed.

However, the more dominant elements Cl and SO42− (Figure 5(b) and 5(c)) are related to the wastewater discharge and the alteration of both gypsum and salt formations as well as sodium-rich formations. On the other hand, the enrichment in Na+ and Cl elements, with maximum values of 200.2 mg/L and 315 mg/L, respectively, is due to the contamination by the Triassic outcrops, particularly that of the Guelal zone diapir (the center of the watershed). These elements recorded very high concentrations during the driest period of 2001–2002, but decreased during the highly wet periods owing to the water dilution phenomenon. The binding of the ratio Ca2+ + Mg2+ + HCO3/Na+ + Cl + SO42− vs EC shows that this salinity is linked to the Na+, Cl and SO42− elements rather than the Ca2+, Mg2+ and HCO3 elements with a ratio (Equation (4)): 
formula
(4)

Organic parameters

The examination of the change in organic matter values recorded in the reservoir water (Figure 6) during the 1988–2012 period depicts a positive trend of 53% which is in line with the COD and BOD5 positive trends of 82% and 150%, respectively. The organic matter (OM) pollution is the result of wastewater and some highly polluting food industries' discharges: slaughterhouses, cheese factories, dairies and sweets (Kebiche et al. 1999). In fact, the region is densely populated and witnesses an industrial expansion, especially that of food, which could justify the high organic matter values. The average COD/BOD5 ratio is 20>4, indicating the presence of a difficult-to-biodegrade organic material. The COD values (10–109 mg/l) and BOD5 (0.7–14 mg/l) can be explained by the introduction of organic matter (OM) degradation conditions by microorganisms. This degradation was accompanied by a decrease in dissolved oxygen of −1%.

Nitrogen and phosphate parameters

Nitrogen pollution (NO3, NO2 and NH4+) depends on the supply of agricultural land with nutrients (land application, livestock releases and fertilizers) and the discharge of wastewater. Domestic and industrial discharges and livestock manure inputs are important ammonia nitrogen sources. The most used fertilizers in the Bousselem watershed are phosphorus and potassium urea, ammonium nitrate, superphosphates, potassium chloride and to a lesser extent ammonium sulphate, sodium sulphate, calcium nitrate and potassium sulphate (Boudoukha & Boulaarak 2013). This nitrogen category is the largest in mass and the most difficult to evaluate. The trends of the different nitrogen forms show that the highest increases were registered for nitrites with +700% and ammonium with +400% (Figure 7) during the 1988–2012 period. This increase in nitrite results either from the ammonium oxidation according to the reaction (Equation (5)) (Martin 1979), or the reduction of nitrates by denitrification. 
formula
(5)
The increase in ammonium, in the watershed waters, is due to (i) the quantity of the discharged animal and human organic matter, (ii) the vegetal matter in the watercourses, and (iii) the industrial and agricultural discharges. This is in line with the decrease in nitrates, which recorded a negative trend of −15%. This decrease is probably due to its consumption by algae followed by a decrease in dissolved oxygen of −1% to turn into nitrites (NO2). This is therefore a natural denitrification phenomenon due to the consumption of dissolved O2 by the microorganisms present in the dam water according to Equation (6). 
formula
(6)

The relationship between nitrates, nitrite and ammonium concentrations in the reservoir water during the period 1988–2012 (Figure 8(a)) shows an inverse evolution of NO3 compared to NO2 and NH4+. The nitrates represent the oxidized form of nitrogen, whereas the other two forms (nitrites and ammonium) represent their reduced forms. This transformation is achieved by anaerobic bacteria that promote the nitrates' denitrification according to Equation (6). This change is marked by the decrease of nitrates in the water, indicating a change of form. These reactions are accompanied by a decrease in oxygen in the aquatic environment.

Figure 8

(a) Evolution of the levels of nitrates, nitrites and ammonium in the water of the dam (1988–2012), (b) variation of NT/PT ratio in Aïnzeada dam water over the period 1988–2012.

Figure 8

(a) Evolution of the levels of nitrates, nitrites and ammonium in the water of the dam (1988–2012), (b) variation of NT/PT ratio in Aïnzeada dam water over the period 1988–2012.

The orthophosphate PO43−, which has recorded an increase of +46%, does not follow the same negative trend as nitrates. It rather consists of low concentrations (average of 0.16 mg/L). This situation can be explained by the fact that phosphorus is easily adsorbed by the colloids. The phosphorus concentrations in the wadi Boussellam water are regulated by several biogeochemical processes such as apatite precipitation (PO4)3(F,Cl,OH)Ca5 (Golterman & Meyer 1985) and the consumption by aquatic plants (Johnson et al. 1976; Probst 1985; Pilleboue & Dorioz 1986; Kattan et al. 1987).

Total nitrogen/total phosphorus ratio (NT/PT)

The NT/PT rapport values, a water eutrophication status indicator (Siep 1994), provide information on the probable presence of atmospheric nitrogen-fixing algae at ratio values below 29 (Downing & McCauley 1992). The N/P ratio (Figure 8(b)) shows a negative trend of −60%, with values ranging from 0.1 to 150 during the observation period. The ecosystem characteristics (strong evaporation, wind erosion) are responsible for the reduction of nitrates (which are either consumed by algae or lost by denitrification). 53% of the values of the NT/PT ratio are less than 29, which allows the appearance of Cyanophycea in Aïnzeda dam reservoir.

Phosphorus acts as a limiting factor in the waters, which is confirmed by the phosphate intake by leaching at the end of the winter and spring. Therefore, the microorganisms develop rapidly and can lead to the appearance of scums on the surface. Many Cyanophyceae strains can produce toxins that are dangerous for drinking water supply or bathing (Duchemin 2010).

Principal component analysis application to Aïnzeda dam water chemical data

The objectives of this part are to apply the PCA to a set of water quality data to identify the water quality characteristics due to natural or anthropogenic influences.

This study was carried out on the physicochemical elements of the samples taken from the dam water for the period 1988–2012. The analyzed parameters (Table 2) are: Ca2+, Mg2+, Na+, K+, Cl, SO42−, HCO3, NO3, NH4+, NO2, PO43−, EC, TDS, COD and BOD5. The analysis interpretation was achieved using Origin Pro 9.0 software, which has a module dedicated to statistical processing.

Table 2

Correlation matrix for the period 1988–2012 of the waters of the Aïnzeda dam

CaMgNaKClSO4HCO3NO3NH4NO2PO4ECTDSCODBOD5
Ca               
Mg −0.249              
Na 0.201 0.092             
0.074 −0.010 −0.167            
Cl 0.307 0.178 0.810 −0.101           
SO4 0.644 0.144 0.700 −0.169 0.201          
HCO3 0.022 0.132 0.120 0.042 0.130 0.013         
NO3 0.020 0.042 −0.174 −0.012 −0.177 0.069 0.076        
NH4 0.094 −0.021 0.294 −0.099 0.269 0.068 0.141 0.010       
NO2 0.125 −0.061 0.050 −0.110 0.061 0.053 0.022 0.014 0.191      
PO4 0.186 −0.014 0.069 0.003 0.154 0.008 0.221 0.169 −0.023 0.028     
EC 0.380 0.267 0.802 −0.261 0.794 0.413 0.259 −0.107 0.261 0.079 0.139    
TDS 0.289 0.103 0.670 −0.167 0.612 0.348 0.091 −0.235 0.223 0.022 −0.043 0.718   
COD 0.138 −0.035 0.380 −0.021 0.407 0.104 0.086 −0.069 0.149 −0.075 0.1184 0.374 0.230  
BOD5 0.225 −0.056 0.222 −0.084 0.288 0.143 0.013 −0.077 0.127 0.106 0.054 0.143 0.143 0.111 
CaMgNaKClSO4HCO3NO3NH4NO2PO4ECTDSCODBOD5
Ca               
Mg −0.249              
Na 0.201 0.092             
0.074 −0.010 −0.167            
Cl 0.307 0.178 0.810 −0.101           
SO4 0.644 0.144 0.700 −0.169 0.201          
HCO3 0.022 0.132 0.120 0.042 0.130 0.013         
NO3 0.020 0.042 −0.174 −0.012 −0.177 0.069 0.076        
NH4 0.094 −0.021 0.294 −0.099 0.269 0.068 0.141 0.010       
NO2 0.125 −0.061 0.050 −0.110 0.061 0.053 0.022 0.014 0.191      
PO4 0.186 −0.014 0.069 0.003 0.154 0.008 0.221 0.169 −0.023 0.028     
EC 0.380 0.267 0.802 −0.261 0.794 0.413 0.259 −0.107 0.261 0.079 0.139    
TDS 0.289 0.103 0.670 −0.167 0.612 0.348 0.091 −0.235 0.223 0.022 −0.043 0.718   
COD 0.138 −0.035 0.380 −0.021 0.407 0.104 0.086 −0.069 0.149 −0.075 0.1184 0.374 0.230  
BOD5 0.225 −0.056 0.222 −0.084 0.288 0.143 0.013 −0.077 0.127 0.106 0.054 0.143 0.143 0.111 

Correlation matrix analysis

The correlation coefficient is commonly used to estimate the relationship between two variables. It is a simple statistical tool to show the degree of connection between the various variables.

The inter-variable correlation matrix calculation for the period 1988–2012 is presented in Table 2. This matrix allows us to have an idea about the strong correlations between the main ionic associations (couples) encountered at the dam water. The coefficients vary between the maximum value 0.810 corresponding to the couple (Na+/Cl) and the minimum value −0.075 for the couple (NO2/COD). However, most of the variables are positively correlated and the correlated items have the same origin. The correlation matrix outcome of the PCA shows that Na+ has a good significant relationship (0.670–0.810) with TDS, SO42−, EC and Cl. Cl has a significant relationship (0.612–0.794) with EC and TDS. Ca2+ influences SO42− (r = 0.644) and EC influences TDS (r = 0.718).

For example, the Na+/Cl pair is included in the chemical facies existing in the sodium chloride region. These originate from the evaporitic rocks existing in the dam watershed (Guelal gypsum deposit). Immediately, the couple (EC/TDS), which gives a good correlation, confirms the contamination of the dam water by anthropogenic pollution (uncontrolled discharges of wastewater). It has also been found that nitrates, which have no correlation with other elements, are the result of agricultural pollution (chemical fertilizer application). However, the other parameters appear weakly related to the other variables.

Correlation between variables and main axes

Table 3 displays the correlation between the variables and the main axes to determine the possible contributing factors in the hydrochemistry of water samples. It shows that the variables are better represented with the main axis F1 for the period 1988–2012. The main axis F1 has a positive strong correlation with Na+, Cl, SO42−, EC and TDS. For these parameters, F1 varies between 0.637 and 0.882.

Table 3

Correlation between variables and main axes

PCA (Period: 1988–2012)
Parameters
Ca2+Mg2+Na+K+ClSO42−HCO3NO3NH4+NO2PO43ECTDSCODBOD5
F1 0.262 0.078 0.742 −0.115 0.724 0.637 0.126 −0.30 085 0.183 0.057 0.074 0.882 0.741 0.231 0.176 
F2 −0.235 0.486 −0.153 0.161 −0.028 −0.067 0.429 0.385 0.049 0.175 0.483 −0,027 −0.195 0.017 0.108 
PCA (Period: 1988–2012)
Parameters
Ca2+Mg2+Na+K+ClSO42−HCO3NO3NH4+NO2PO43ECTDSCODBOD5
F1 0.262 0.078 0.742 −0.115 0.724 0.637 0.126 −0.30 085 0.183 0.057 0.074 0.882 0.741 0.231 0.176 
F2 −0.235 0.486 −0.153 0.161 −0.028 −0.067 0.429 0.385 0.049 0.175 0.483 −0,027 −0.195 0.017 0.108 

PCA correlation circles analysis

On the main plane (Figure 9), the first factorial axis F1 represents 51.73% of the total inertia. It groups two clouds of points. The first group includes the parameters Na+, Cl, EC and TDS, while the second involves the parameters SO42−, NH4+, COD and BOD5.

Figure 9

Graphical representation of variables on the factorial design 1–2 (period: 1988–2012).

Figure 9

Graphical representation of variables on the factorial design 1–2 (period: 1988–2012).

This first axis is linked to the minerals' dissolution (the existing gypsum formation in the watershed) and the uncontrolled wastewater discharges (released from agglomerations). The second axis, F2, represents 13.49% of the total inertia. It groups the parameters Ca2+, HCO3, NO3 and PO43− together and confirms that the pollution is either natural due to the leaching of the carbonate formations leveling the upstream part of the watershed (Ca2+ and HCO3) or agricultural, caused by the intense use of chemical fertilizers in the agricultural area of the watershed (NO3 and PO43−).

Following the PCA analysis, it can be concluded that human activities, especially domestic and industrial discharges as well as agricultural practices, can locally affect the chemical quality of Aïnzeda dam water. Furthermore, the rainfall regime plays a key role in the dam hydrochemistry by the leaching of watershed soils.

CONCLUSIONS

Like most of Eastern Algeria's surface waters, Aïnzeda dam waters seem to become more and more polluted. In the above-discussed research, 90% of the samples exceed the content of 0.01 mg/L in NO2 and 14% of the samples exceed the content of 0.5 mg/L in NH4+ recommended by the WHO standard. In order to determine the impact of human activity on Aïnzeda lake water quality (semi-arid region), a monthly monitoring of the climatic parameters of the region (precipitation, air temperature and evaporation) was carried out and the physico-chemical parameters of the lake water (water temperature, EC, TDS, pH, O2dis, TSS, turbidity, Ca2+, Mg2+, Na+, K+, Cl, SO42−, HCO3, NO3, NO2, NH4+, PO43−, OM, COD, BOD5) were investigated over a period of 25 years (1988–2012) and presented in Table 1. The results of this study show that the decrease in the reservoir water volume (7%) and the deterioration of water quality are the consequence of several factors, including: (i) the decrease in precipitation (8%), (ii) increased air temperatures (9%) and increased dam water temperature (7%) and (iii) the different anthropogenic pollution sources (urban, industrial and agricultural waste). This has led to a very high concentration of most of the chemical elements in the dam water and consequently an increase in their trends, notably the EC (65%), Cl (83%), Na+ (60%), SO42− (35%), OM (53%), COD (82%), BOD5 (150%), NO2 (700%), NH4+ (400%). The elements Ca2+ (16%), Mg2+ (60%) and HCO3 (7%) can come from the dissolution of carbonates. This deterioration in water quality is due to the decomposition of organic matter accompanied by a pH drop of 8%. The increase in ammonium is due to the large amount of organic matter and industrial waste. It is also noteworthy to mention that there is a decrease in nitrates and phosphates and an increase in nitrite. This is due to the oxidation of ammonium, accompanied by a 1% decrease in dissolved oxygen. Aïnzeda dam's water chemistry highlights all the factors involved in the mineralization acquisition process: the influence of geological formations in the watershed, domestic and industrial discharges and the use of fertilizers in association with the climate parameters of the region (high tempterature and evaporation, and low precipitation).

ACKNOWLEDGEMENTS

The authors are grateful to the authorities of Bordj Bou Arreridj, Setif and the National Hydraulic Resources Agency (ANRH) for providing data and permission to pursue and publish the present work.

REFERENCES

REFERENCES
Belhadj
M. Z.
,
Boudoukha
A.
&
Mezedjri
L.
2011
Quality of surface waters and their impact on the environment in the wilaya of Skikda (North-eastern Algeria) (Natural Mercury contamination)
.
European Journal of Scientific Research
56
(
2
),
204
211
.
Boudoukha
A.
&
Boulaarak
M.
2013
Waters pollution of Hammam Grouz Dam by nutrients (East of Algeria)
.
Bulletin du Service Géologique de l'Algérie
24
(
2
),
23
34
.
Boudoukha
A.
,
Kowalsky
W. M.
&
Pharista
A.
1997
Evolution des sédiments lacustres Plio-Villafranchiens des dépressions fermées d’El Eulma. Est algérien
.
Bulletin de la Société des Sciences Naturelles du Pays Montbéliard
7
,
187
200
.
Bouguerne
A.
,
Boudoukha
A.
,
Benkhaled
A.
&
Mebarkia
A.
2017
Assessment of surface water quality of Ain Zada dam (Algeria) using multivariate statistical techniques
.
International Journal of River Basin Management
15
(
2
),
133
143
.
Bouzid-Lagha
S.
&
Djelita
B.
2012
Study of eutrophication in the Hamman Boughrara Reservoir (Wilaya de Tlemcen, Algeria)
.
Hydrological Sciences Journal
57
(
1
),
186
201
.
David
M.
,
Bonnieux
F.
&
Rainelli
P.
1983
Introduction au traitement des séries chronologiques
.
Economie rurale
157
,
49
56
.
Demdoum
A.
,
Hamed
Y.
,
Feki
M.
,
Hadji
R.
&
Djebbar
M.
2015
Multi-tracer investigation of groundwater in El Eulma Basin (Northwestern Algeria), North Africa
.
Arabian Journal of Geosciences
8
(
5
),
3321
3333
.
Diday
E.
,
Lemaire
J.
,
Pouget
J.
&
Testu
F.
1982
Eléments d'analyse de données
.
Dunod
,
Paris, France
.
Downing
J. A.
&
McCauley
E.
1992
The nitrogen: phosphorus relationship in lakes
.
Limnology and Oceanography
37
(
5
),
936
945
.
Duchemin
J.
2010
Blue-green algae, cyanophyceae, cyanobacteria
.
Water and Health
,
Normandie, France
,
22
pp.
Etchanchu
D.
&
Probst
J. L.
1988
Evolution of chemical composition of the Garonne river water during the period of 1971-1984
.
Hydrological Sciences Journal
33
(
3
),
243
256
.
Guerraiche
Z.
,
Boudoukha
A.
&
Benkadja
R.
2016
Variation of the chemical composition of Grouz dam waters, Eastern Algeria
.
Desalination and Water Treatment
57
(
11
),
4878
4887
.
Hadji
R.
,
Limani
Y.
,
Boumazbeur
A.
,
Demdoum
A.
,
Zighmi
K.
,
Zahri
F.
&
Chouabi
A.
2014
Climate change and their influence on shrinkage – swelling clays susceptibility in a semi – arid zone: a case study of Souk Ahras municipality, NE-Algeria
.
Desalination and Water Treatment
52
(
10–12
),
2057
2072
.
Hamad
A.
,
Baali
F.
,
Hadji
R.
,
Zerrouki
H.
,
Besser
H.
,
Mokadem
N.
&
Hamed
Y.
2018
Hydrogeochemical characterization of water mineralization in Tebessa-Kasserine karst system (Tuniso-Algerian transboundry basin)
.
Euro-Mediterranean Journal for Environmental Integration
3
(
1
),
7
.
Hamed
Y.
,
Redhaounia
B.
,
Ben Sâad
A.
,
Hadji
R.
,
Zahri
F.
&
Zighmi
K.
2017
Hydrothermal waters from karst aquifer: case study of the Trozza basin (Central Tunisia)
.
Journal of Tethys
5
(
1
),
33
44
.
Johnson
A. H.
,
Bouldin
D. R.
,
Goyette
E. A.
&
Hedges
A. M.
1976
Phosphorus loss by stream transport from a rural watershed: quantities, processes, and sources
.
Journal of Environmental Quality
5
(
2
),
148
157
.
Kebiche
M.
,
Benabid
C.
&
Adjal
F.
1999
Surface water pollution in the semiarid Setif region of Algeria
.
Sciences et changements planétaires / Sécheresse
10
(
2
), 137.
Kempe
S.
1982
Long-term records of CO2 pressure fluctuations in fresh waters
.
SCOPE/UNEP Sonderband
52
,
91
332
.
Martin
G.
1979
Le problem de l'azote dans les eaux
.
Technique et Documentation
.
Paris édition, France
,
279
pp.
Mebarkia
A.
2011
Étude des caracteristiques physico-chimiqes des eaux de surface; cas du barrage de ain zada wilaya de Bordj Bou-Arreridj, (NE algérien)
.
Mémoire de Magister
,
Université Badji Mokhtar d'Annaba
, p.
209
.
Mokadem
N.
,
Demdoum
A.
,
Hamed
Y.
,
Bouri
S.
,
Hadji
R.
,
Boyce
A.
,
Laouar
R.
&
Saad
A.
2016
Hydrogeochemical and stable isotope data of groundwater of a multi-aquifer system: Northern Gafsa basin e Central Tunisia
.
Journal of African Earth Sciences
114
,
174
191
.
Pilleboue
E.
&
Dorioz
J. M.
1986
Mass-balance and transfer mechanisms of phosphorus in a rural watershed of Lac Leman, France
. In
Sediments and Water Interactions
.
Springer
,
New York, NY
, pp.
91
102
.
Regnie
S.
1965
On some mathematical aspects of automatic classification problems
.
Math and Human Sciences
82
,
13
29
.
Rodier
J.
2009
L'analyse de l'eau, eaux naturelles, eaux résiduaires, eau de mer, 9ème édition. Dunod, Paris, France, 1526pp
.
Siep
K. L.
1994
Phosphorus and nitrogen limitation of algal biomass across trophic gradients
.
Aquatic Sciences
56
(
1
),
16
28
.
Vila
J. M.
1980
La chaîne alpine d'Algérie orientale et des confins algéro-tunisiens. Thèse de doctorat d'état
,
Université Pierre et Marie Curie
,
Parie VI, France
,
665
pp.