Assessment of Water Quality in the Panama Canal Watershed Using Multivariate Analysis of Physicochemical and Biological Parameters

Abstract

In the hydrometeorological context of watersheds, water quality is strongly related to its physical, chemical, and biological characteristics. In this regard, the joint analysis of these parameters at the watershed level is highly important. The objective of this study was to analyze a total of twenty-three (23) physicochemical and biological water parameters in the Panama Canal watershed with the aim of determining the interrelationships among them, explaining their clustering and simultaneously identifying homogeneous hydrological stations. Multivariate statistical techniques were used for data analysis. The principal component analysis revealed that physicochemical and biological water parameters can be grouped into two dimensions, suggesting potential temporal or spatial patterns in water quality. Furthermore, these parameters were not homogeneous across the various stations of the reservoir. The cluster analysis grouped the fourteen (14) sampling stations with similar characteristics into three groups or clusters. In the context of future research, this study established a precedent for the interpretation of the complex patterns of water quality in river basins. Finally, this research is of great significance for those responsible for environmental management, as its results have a direct impact on the management of watershed areas.

1. Introduction

Water is a critical resource for sustaining livelihoods, growing food, producing energy, promoting industrial development as well as ensuring the integrity of ecosystems and the services they provide [1,2,3]. Managing and preserving the quality of potable water resources is a challenge [4] that many countries must face in order to improve their economic development levels [5]. Similarly, in reservoirs, water is a key element that integrates other components of watersheds, enabling the development of life [6]. Potable water resources within watersheds serve various water users such as industry, agriculture, maritime transport, and nature conservation [7], among others. However, the functioning of freshwater ecosystems is threatened by the impact of invasive species, climate change, and other emerging hazards [8,9].

Point-source pollutants [10] such as wastewater effluents, and diffuse pollution from agricultural areas can threaten the water quality. Additionally, the hydrological system itself is complex and subject to multiple regulatory frameworks, all of which add to the complexity of addressing water quality issues [11]. In the hydrometeorological context of watersheds, water quality is strongly related to its spatiotemporal evolution [12,13] and its physical, chemical, and biological characteristics [14]. Therefore, the establishment of water quality monitoring programs at the watershed level is crucial. In this regard, the Water Quality Surveillance and Monitoring Program (PVSCA) at the Panama Canal watershed is the management and monitoring framework for this essential resource. The PVSCA includes 40 sampling stations, located in reservoirs, rivers, and sub-basins of the rivers [15]. One of the most representative reservoirs, where water is used for various purposes (human consumption, navigation, agricultural activities, tourism), is Gatún. The Gatún Reservoir is the key water reserve for the operation of the Panama Canal as it stores the essential water supply needed for the Canal’s operation. Consequently, this research focused on the statistical analysis of water quality parameters from the fourteen stations within the Gatún Reservoir in the Panama Canal watershed.

A watershed is the land area where runoff water converges toward a common drainage point [16]. According to a previous study [17], the water quality depends on the characteristics of the watershed; therefore, it is highly useful for planning and managing water resources. The various factors influencing water quality in a watershed make it susceptible to changes in its physicochemical and biological characteristics [18]. According to previous research [19], variations in these factors affect the suitability of water for both natural ecosystems and various human uses. On the other hand, human activities such as urban development, deforestation, plantations, water extraction for human consumption, wastewater discharge, and road construction, among others, alter the water quality [20].

Therefore, water quality monitoring is a fundamental aspect for understanding the environmental status of reservoirs [21], promoting their conservation, and preventing and controlling eutrophication [22]. However, this monitoring generates a series of multivariate data that must be organized in such a way that complex relationships between multiple variables can be explored. To this end, the use of multivariate statistical techniques is a valuable tool [23] for the reliable management of water quality parameters. These statistical models allow for the simultaneous evaluation of the multiple relationships that may exist between water parameters and their evolution on a spatiotemporal scale [24,25].

Academics and researchers from various countries have employed multivariate methods to study water quality. For example, in Argentina, principal component analysis and multivariate analysis of variance have been used [26]; in Peru [27], Honduras [28], Saudi Arabia [29], and Bangladesh [30], factor analysis and cluster analysis were applied; while in China [31] and Mexico [32], analysis of variance was implemented. In any case, multivariate methods have been widely used to provide a clear interpretation of the water quality in both surface water and groundwater. However, the study of water quality parameters in the Gatún Reservoir from a multivariate perspective represents a relatively unexplored research area. Therefore, in this research, 23 physicochemical and biological parameters from the 14 stations within the Gatún Reservoir were analyzed using multivariate techniques, specifically principal component analysis, cluster, and HJ-biplot.

The objectives of this study were to: (1) determine the physicochemical and biological parameters that best explain the variability in water quality in Gatún Reservoir; (2) analyze the spatial variability of water quality in Gatún Reservoir and determine the presence of homogeneous zones based on the sampling stations. This study was based on the following hypotheses: (1) some parameters have a greater influence on the variability of water quality in Gatún Reservoir; and (2) there is variation in the water quality between the different sampling sites.

2. Materials and Methods

In this study, an analysis of the physicochemical and biological parameters of water quality was conducted for the Gatun Reservoir in the Panama Canal watershed (https://pancanal.com/cuenca-hidrografica accessed on 23 February 2025).

The Gatún Reservoir, located in the Central Mountain Range of Panama (Figure 1), covers an area of 436 km² at an elevation of 27 m above sea level, with its geographical location defined by the coordinates 9°12′ N and 79°54′ W.

The data corresponding to the year 2022 were obtained from the fourteen (14) permanent stations that make up the Gatún Reservoir, following the protocols of the Environmental Quality Analysis Team of the Panama Canal Authority. The methodology used for laboratory analyses is the one recommended by the Standard Methods for the Examination of Water and Wastewater, 23rd Edition [33], adopted by the Environmental Quality Analysis Team of the Panama Canal Authority.

A total of twenty-three (23) physicochemical and biological indicators (see

Table 1. Statistical summary of the physicochemical and biological water quality parameters.

Type Parameter	Parameter	Units	Minimum	Maximum	Mean	Standard Deviation
Bacteriological parameters	Total coliforms (TC)	NMP/100 mL	63	120,000	2566	9521.8
	Escherichia coli (E. coli)	NMP/100 mL	10	1700	51	146.28
Biological parameter	Chlorophyll a (CHL_A)	µg/L	1.8	40.20	7.20	5.19
Solids	Total dissolved solids (TDS)	mg/L	7	286	147	67.86
	Total suspended solids (TSS)	mg/L	10	15	10	0.48
	Turbidity (Turb)	NTU	0.6	43.3	3.1	4.73
Nutrients	Phosphorus as phosphate (P_PO4)	mg/L	0.020	0.035	0.020	0.0012
	Nitrogen as nitrate (N_NO3)	mg/L	0.010	0.284	0.032	0.042
Major anions	Total alkalinity (T_Alc)	mg/L	14	60	36	12.53
	Sulfate (SO4)	mg/L	0.6	20.6	8.5	4.28
	Chlorides (Cl)	mg/L	3.9	116.8	48.8	31.86
Major cations	Sodium (Na)	mg/L	2.68	64.61	28.21	17.34
	Calcium (Ca)	mg/L	1.13	16.30	5.92	2.69
	Magnesium (Mg)	mg/L	1.32	9.77	5.29	2.1
	Potassium (K)	mg/L	0.65	3.12	1.8	0.603
	Water hardness (Hardness)	mg/L	8.5	72.4	36.5	13.89
In Situ	Conductivity (Cond)	µS/cm	42	514	251	124.7
	Salinity (S)	UPS	0.10	0.25	0.14	0.041
	Hydrogen ion potential (pH)	pH units	6.38	8.28	7.39	0.391
	Dissolved oxygen (DO)	mg/L	2.64	8.35	6.74	0.96
	Oxygen saturation percentage (SatO2)	%	33	108	88	12.81
	Transparency (Transp)	m	0.33	5.0	2.5	1.13
	Temperature (T)	°C	26.6	30.9	29.2	0.72

) were monitored, with the sampling frequency conducted monthly at the surface of each station in the reservoir.

To perform the data analysis, multivariate statistical techniques were employed including principal component analysis, correlation analysis, cluster analysis, and the HJ-biplot technique [34,35,36,37,38]. Additionally, two statistical tests were employed: the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity to determine the validity of the data. These analyses were performed using computational routines in R software, version 4.2.1.

The data analyzed in this study were previously standardized using the z-scale transformation to prevent or correct errors arising from differences in the magnitudes or measurement units of the parameters and mitigate potential statistical biases.

3. Results and Discussion

3.1. Description of the Physicochemical and Biological Water Parameters

To describe the variation of each parameter, Table 1 presents the minimum, maximum, average values, and standard deviations of all water samples collected from the different stations of the Gatún Reservoir in the Panama Canal watershed.

The results regarding the physicochemical and biological variables are presented in Table 1. These data suggest that during the analyzed period, the water temperature remained within a relatively narrow range (26.6–30.9 °C, average 29.2 ± 0.72 °C), which is typical of tropical or subtropical water bodies [39]. Parameters related to water mineralization showed moderate values: the conductivity exhibited a wide range (42–514 μS/cm, average 251 ± 124.7 μS/cm), while the total dissolved solids (TDS) ranged from 7 to 286 mg/L (average 147 ± 67.86 mg/L). These values suggest a freshwater body with moderate mineralization [40]. The oxygenation status of the system appears to be adequate, with the dissolved oxygen concentrations ranging from 2.64 to 8.35 mg/L (average 6.74 ± 0.96 mg/L) and the saturation percentages reaching up to 108%. The pH remained within the typical range for natural waters (6.38–8.28, average 7.39 ± 0.391).

Regarding nutrients, the nitrate concentrations (0.010–0.284 mg/L N-NO₃) and phosphate concentrations (0.020–0.035 mg/L P-PO₄) were relatively low, which could indicate a system with limited eutrophication. Chlorophyll a, an indicator of phytoplankton biomass, ranged from 1.8 to 40.20 μg/L (average 7.20 ± 5.19 μg/L), suggesting significant variations in primary productivity.

The microbiological indicators revealed the presence of fecal contamination, with the total coliforms varying widely (63–120,000 CFU/100 mL) and the presence of E. coli (10–1700 CFU/100 mL), which could indicate an anthropogenic influence on the system [41]. Water transparency (0.33–5.0 m) and turbidity (0.6–43.3 NTU) showed considerable variations, suggesting significant dynamics in the amount of suspended material and the optical conditions of the water.

3.2. Principal Component Analysis (PCA)

PCA was performed to facilitate the dimensionality reduction of the data, synthesizing the 23 water quality parameters into a manageable number of principal components. These components identify the elements involved in the formation of factors that explain the correlations between parameters.

3.2.1. Bartlett’s Test of Sphericity

Bartlett’s test of sphericity (

Table 2. Conditions for the application of principal component analysis (PCA).

Barlett’s Test of Sphericity		Kaiser–Meyer–Olkin Measure of Sampling Adequacy (KMO)
Approximate Chi-Square	8173.72	0.84
Degrees of Freedom	253
Significance	0.00

) was used to validate whether the water parameters were sufficiently correlated to perform a principal component analysis. In this case, the hypotheses being tested were H₀: the variables are not correlated, versus the alternative hypothesis H₁: the variables are correlated.

3.2.2. Kaiser–Meyer–Olkin (KMO) Test

The Kaiser–Meyer–Olkin (KMO) test is used to assess the adequacy of the sample for conducting factor analysis or principal component analysis. It evaluates whether the partial correlations between variables are small enough to justify the use of these techniques. A KMO value closer to 1 indicates that the data are suitable for factor analysis, while values closer to 0 suggest that the data may not be suitable.

In Table 2, the result of Bartlett’s test (significance = 0) suggests that there was insufficient evidence to accept the null hypothesis, implying that the water quality parameters are correlated. Additionally, Table 2 presents a value of 0.84 for the Kaiser–Meyer–Olkin statistic, indicating that it is appropriate to perform the principal component analysis.

3.2.3. Analysis of the Correlation Matrix

The correlation matrix was used to determine the Pearson correlation coefficients (

Table 3. Pearson correlation matrix of the physical, chemical, and biological parameters.

Parameter	T_Alc	Ca	CHL_A	Cl	Cond	TC	Hardness	E. coli	K	Mg	N_NO3	Na	DO	SatO2	P_PO4	pH	S	SO4	TDS	TSS	T	Transp	Turb
T_Alc	1.00
Ca	0.86	1.00
CHL_A	−0.08	−0.07	1.00
Cl	0.18	0.35	−0.28	1.00
Cond	0.38	0.53	−0.26	0.97	1.00
TC	0.08	0.10	−0.02	−0.15	−0.11	1.00
Hardness	0.73	0.88	−0.17	0.73	0.85	0.00	1.00
E. coli	0.11	0.08	−0.03	−0.20	−0.16	0.88	−0.04	1.00
K	0.07	0.27	−0.29	0.97	0.92	−0.14	0.65	−0.20	1.00
Mg	0.51	0.63	−0.23	0.90	0.96	−0.08	0.93	−0.12	0.83	1.00
N_NO3	0.43	0.47	0.25	−0.24	−0.11	0.29	0.22	0.40	−0.28	−0.01	1.00
Na	0.23	0.39	−0.27	0.99	0.97	−0.15	0.75	−0.19	0.97	0.91	−0.20	1.00
DO	−0.07	0.05	−0.09	0.61	0.55	−0.37	0.30	−0.49	0.61	0.45	−0.44	0.61	1.00
SatO2	−0.09	0.03	−0.10	0.61	0.54	−0.38	0.29	−0.49	0.62	0.45	−0.50	0.61	1	1.00
P_PO4	−0.08	−0.07	−0.02	−0.05	−0.06	−0.02	−0.09	−0.02	−0.03	−0.09	−0.04	−0.05	0.09	0.09	1.00
pH	0.35	0.44	−0.25	0.69	0.70	−0.17	0.63	−0.24	0.64	0.67	−0.17	0.71	0.74	0.73	0.03	1.00
S	0.22	0.41	−0.24	0.92	0.93	−0.09	0.74	−0.14	0.90	0.87	−0.10	0.92	0.50	0.50	−0.07	0.59	1.00
SO4	0.48	0.62	−0.11	0.81	0.86	−0.05	0.84	−0.09	0.76	0.87	0.09	0.82	0.44	0.42	−0.03	0.63	0.79	1.00
TDS	0.41	0.54	−0.21	0.89	0.93	−0.07	0.83	−0.12	0.84	0.92	−0.08	0.90	0.50	0.45	−0.06	0.65	0.84	0.82	1.00
TSS	0.19	0.13	0.00	−0.03	0.02	0.47	0.10	0.44	−0.04	0.05	0.28	−0.01	−0.23	−0.24	−0.01	−0.14	0.02	0.11	0.08	1.00
T	−0.30	−0.20	−0.12	0.21	0.14	−0.33	−0.04	−0.36	0.27	0.10	−0.50	0.19	0.36	0.43	0.02	0.13	0.10	−0.03	0.10	−0.30	1.00
Transp	−0.05	−0.01	−0.39	0.61	0.54	−0.22	0.28	−0.27	0.62	0.46	−0.50	0.60	0.62	0.63	0.00	0.55	0.50	0.38	0.50	−0.24	0.31	1.00
Turb	0.28	0.23	0.20	−0.37	−0.27	0.68	−0.01	0.77	−0.39	−0.20	0.74	−0.34	−0.61	−0.63	−0.03	−0.38	−0.22	−0.10	−0.22	0.56	−0.54	−0.58	1.00

Note: The values in bold indicate a correlation of 0.5 or higher between the water quality parameters.

) of the water quality parameters. The significance values (p-values) associated with Pearson’s correlation coefficients are presented in

Table 4. Significance values (p-value) of the Pearson’s correlation coefficients.

Parameter	T_Alc	Ca	CHL_A	Cl	Cond	TC	Hardness	E. coli	K	Mg	N_NO3	Na	DO	SatO2	P_PO4	pH	S	SO4	TDS	TSS	T	Transp	Turb
T_Alc
Ca	0.00
CHL_A	0.30	0.39
Cl	0.00	0.00	0.00
Cond	0.00	0.00	0.00	0.00
TC	0.30	0.20	0.78	0.05	0.14
Hardness	0.00	0.00	0.03	0.00	0.00	0.95
E. coli	0.17	0.32	0.67	0.01	0.04	0.00	0.65
K	0.34	0.00	0.00	0.00	0.00	0.06	0.00	0.01
Mg	0.00	0.00	0.00	0.00	0.00	0.28	0.00	0.13	0.00
N_NO3	0.00	0.00	0.00	0.00	0.14	0.00	0.00	0.00	0.00	0.87
Na	0.00	0.00	0.00	0.00	0.00	0.06	0.00	0.01	0.00	0.00	0.01
DO	0.39	0.52	0.25	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
SatO2	0.26	0.67	0.18	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.0
P_PO4	0.32	0.40	0.80	0.51	0.41	0.81	0.25	0.78	0.66	0.24	0.60	0.49	0.25	0.24
pH	0.00	0.00	0.00	0.00	0.00	0.02	0.00	0.00	0.00	0.00	0.02	0.00	0.00	0.00	0.70
S	0.00	0.00	0.00	0.00	0.00	0.23	0.00	0.08	0.00	0.00	0.18	0.00	0.00	0.00	0.38	0.00
SO4	0.00	0.00	0.14	0.00	0.00	0.52	0.00	0.23	0.00	0.00	0.26	0.00	0.00	0.00	0.66	0.00	0.00
TDS	0.00	0.00	0.01	0.00	0.00	0.35	0.00	0.11	0.00	0.00	0.28	0.00	0.00	0.00	0.43	0.00	0.00	0.00
TSS	0.00	0.09	0.99	0.69	0.79	0.00	0.21	0.00	0.63	0.48	0.00	0.95	0.00	0.00	0.89	0.07	0.78	0.17	0.29
T	0.00	0.0	0.13	0.01	0.07	0.00	0.64	0.00	0.00	0.21	0.00	0.01	0.00	0.00	0.82	0.10	0.20	0.68	0.18	0.00
Transp	0.49	0.90	0.00	0.00	0.00	0.00	0.0	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.98	0.00	0.00	0.00	0.00	0.00	0.00
Turb	0.00	0.0	0.01	0.00	0.00	0.00	0.86	0.00	0.00	0.01	0.00	0.00	0.00	0.00	0.71	0.00	0.00	0.20	0.00	0.00	0.00	0.0

Note: The values in bold correspond to statistically significant parameters (p < 0.05).

. Several groups of significant correlations were observed, revealing important patterns in the dynamics of the aquatic system.

Conductivity showed strong positive correlations with several ions including chlorides (r > 0.8), sodium, and total dissolved solids (TDS). This relationship is expected, as conductivity is a direct measure of the water’s ionic content [42]. The TDS positively correlated with chlorides, sodium, and conductivity, confirming the consistency of the parameters describing the water’s mineralization.

On the other hand, chlorophyll a showed a negative correlation with transparency, which is typical in aquatic systems where phytoplankton affects light penetration [43].

Phosphorus as phosphate did not show strong correlations with other parameters, suggesting that its dynamics are controlled by multiple factors.

Calcium and magnesium showed positive correlations with each other and with the total hardness, which is consistent with the chemical definition of water hardness [44]. Sodium and chlorides presented a strong positive correlation, suggesting a common origin, possibly related to geological or anthropogenic influences.

Dissolved oxygen and the oxygen saturation percentage showed a very strong positive correlation, as expected due to their direct relationship. Temperature showed moderate negative correlations with dissolved oxygen, which is consistent with the physical relationship between these parameters [45]. The total coliforms and E. coli showed a positive correlation with each other, indicating common sources of fecal contamination [46]. Interestingly, they did not show strong correlations with other parameters, suggesting that microbiological contamination might be more related to specific events rather than the overall physicochemical conditions of the system.

The correlation matrix revealed an aquatic system where the main processes were related to the mineralization of water and biological productivity. The lack of strong correlations between some parameters suggests the complexity of the system and the possible influence of multiple environmental and anthropogenic factors on the water quality.

3.2.4. Number of Components to Retain

To determine the number of principal components to retain, the scree plot was used. Figure 2 shows the components generated along with their percentage of explained variance.

Figure 2 allows for the visual appreciation of the optimal number of components to retain in order to reduce the water quality parameters. It was observed that the first component explained 45.20% of the variance, the second component 20.70%, and the third 7.70%; thus, the first three components accounted for 73.60% of the data variability.

The subsequent components, such as PC4 with 4.9%, PC5 with 4.6%, and PC6 with 3.7%, contributed minimal variance and did not significantly aid in understanding the data. It can thus be concluded that by using two principal components, we can largely capture the variance of the original data.

3.2.5. Matrix of Principal Component Loadings

Below, the selected principal components are represented in matrix form (

Table 5. Factor loadings of water parameters for the first two components.

Parameter	Dim1	Dim2
T_Alc	0.328	0.663
Ca	0.482	0.664
CHL_A	−0.287	0.057
Cl	0.959	0.021
Cond	0.966	0.177
TC	−0.243	0.608
Hardness	0.809	0.502
E. coli	−0.310	0.656
K	0.922	−0.049
Mg	0.927	0.293
N_NO3	−0.232	0.747
Na	0.963	0.055
DO	0.701	−0.474
SatO2	0.699	−0.500
P_PO4	−0.038	−0.107
pH	0.803	−0.042
S	0.882	0.158
SO4	0.844	0.339
TDS	0.903	0.232
TSS	−0.090	0.555
T	0.239	−0.581
Transp	0.647	−0.389
Turb	−0.441	0.812

). The columns display the two chosen principal components, and the rows represent the water quality parameters. Therefore, the coefficients in this matrix characterize the degree of correlation between each parameter and the corresponding principal component.

The first dimension was strongly characterized by parameters related to the mineralization of water, showing high factor loadings (>0.8) for conductivity (0.966), sodium (0.963), chlorides (0.959), magnesium (0.927), potassium (0.922), dissolved solids (0.903), sulfates (0.844), hardness (0.809), and pH (0.803). Considering these parameters, we could label this dimension as the “mineralization and ionic status of water.” This designation is consistent with previous studies where mineralization was identified as the main factor of variation in aquatic systems [27,47].

In the second dimension, the significant factor loadings (>0.5) were turbidity (0.812), N-NO₃ (0.747), total alkalinity (0.663), calcium (0.664), E. coli (0.656), total coliforms (0.608), total suspended solids (0.555), and temperature (−0.581).

Based on these parameters, we could label this dimension as “runoff and particle load”. This grouping suggests processes related to the input of particulate material and pollution, possibly of anthropogenic origin, as it combined indicators of fecal contamination (E. coli, coliforms) with turbidity and nutrients [48].

It is interesting to note that temperature showed a negative correlation with this component, which could indicate a seasonal relationship with pollution processes.

The remaining dimensions showed less pronounced factorial loadings and were more difficult to interpret, suggesting that the main processes controlling the variability of the system were well-captured by the first two dimensions.

Additionally, to better visualize the relationships described, Figure 3 graphically presents the previous results through the correlation matrix between the components and the water quality parameters.

Figure 4 provides a clearer view of the parameters that had the greatest weight on the first and second components. Positive correlations are shown in blue tones and negative correlations in red tones. The size of the circle represents the degree of correlation; the larger the area of the circle, the greater the correlation between the parameter and the dimension.

It is possible that certain variables are better represented in another dimension, suggesting an independent process that is important in the system but explains less of the total variance than the first two components. This low representation in the dominant principal components does not mean that these variables are less important; rather, it indicates that they likely respond to more complex and specific environmental controls that would require further detailed analysis.

3.2.6. HJ-Biplot Analysis

The HJ-biplot (Figure 4) provides the most effective multivariate representation of the water quality parameters in a reduced dimension. A significant dispersion of points was observed along both axes, indicating that the samples exhibited substantial variability in both dimensions. The greatest dispersion occurred along the first dimension (horizontal axis), aligning with the highest variance explained by this dimension.

Thus, vectors pointing in similar directions indicate positive correlations, while vectors in opposite directions suggest negative correlations. The length of the vectors represents the magnitude of their contribution to the total variability. Parameters with similar characteristics can be identified, suggesting possible temporal or spatial patterns in water quality.

The HJ-biplot shows a correlation circle where the distance of the variables from the center and their proximity to the outer circle indicate how well they are represented in the plane formed by dimensions 1 and 2. In this case, both chlorophyll a and P-PO₄ were located near the center of the circle, indicating a poor representation in these dimensions. This situation can be explained by several factors including dynamics independent of the main processes. For example, while dimensions 1 and 2 primarily represent processes of mineralization and runoff-related contamination, chlorophyll a and P-PO₄ may be responding to more complex and specific ecological processes [49].

Chlorophyll a production is controlled by multiple factors such as light, temperature, nutrients, and biological processes [50], which can vary independently of the mineralization or general contamination of the system. Temporal variability is possible, as primary production cycles (reflected in chlorophyll a) and phosphorus dynamics often operate on different timescales than the general physicochemical processes [50,51].

Phosphorus may be subject to rapid absorption–release cycles by plankton and sediments, resulting in variation patterns that do not align with the main environmental gradients. This is a parameter controlled by multiple factors; that is, the concentration of P-PO₄ in aquatic systems is strongly regulated by biogeochemical processes such as adsorption–desorption with particles, chemical precipitation, rapid biological absorption, and release from sediments, among others. These processes can occur simultaneously, making its variation more complex and less correlated with other parameters [49].

The left side of the first axis (Dim1) was characterized by vectors representing major ions (Na, Cl, K) and parameters related to mineralization (conductivity, TDS). In the upper quadrant, variables associated with contamination and turbidity (Turbidity, E. coli, Total Coliforms) were grouped together.

Variables such as DO and SatO₂ were projected in the opposite direction to the contamination indicators, suggesting an inverse relationship with water quality. Additionally, an inverse correlation was evident between the water transparency and chlorophyll a concentration, consistent with the findings reported by [47].

The main gradient (Dim1) represented an increase in mineralization from left to right, typical of systems influenced by natural geochemical processes or saline intrusion [52]. The second gradient (Dim2) appeared to be related to contamination processes and the influx of particulate matter, possibly associated with runoff or anthropogenic activities [53,54,55].

The clear separation of the samples suggests that the system experiences significant variations in its water quality. The association of microbiological variables with turbidity could indicate contamination events related to runoff or discharge. The independence between mineralization and contamination gradients suggests that these processes operate relatively independently.

The inverse correlations observed in the water analysis suggest how physical, chemical, and biological factors interact with each other and influence the water quality in a complex manner. As the temperature increases, organic matter decomposition processes accelerate, which reduces the concentration of parameters such as the total carbon, nitrates, and turbidity, indicating higher biological and chemical activity. On the other hand, the presence of suspended solids, organic matter, and microbiological contaminants like E. coli tends to reduce the water transparency and decrease the dissolved oxygen levels.

3.3. Cluster Analysis

In this research, a cluster analysis was conducted to group the sampling stations and identify the similarity between the fourteen sampling stations of the Gatun Reservoir.

For the analysis, the Euclidean distance was used as a similarity measure, and the Ward method was applied for hierarchical clustering on the normalized data. As a result, the dendrogram shown in Figure 5 was constructed.

The cluster analysis revealed a spatial organization of the sampling stations, which were grouped into three main categories:

Cluster 1, composed of the following stations Batería 35 (BAT), Escobal (ESC), Arenosa (ARN), Las Raíces (RAI), Toma de Agua Sabanitas (TAS), Toma de Agua Monte Esperanza (TMH), Barro Colorado (BCI), and Monte Lirio (MLR), consisted of stations that shared similar environmental conditions and tended to exhibit comparable water quality profiles. For example, the water circulation in the Gatun Reservoir may be uniform in certain areas, resulting in similar hydrodynamic conditions between the stations. This can influence parameters such as water temperature, dissolved oxygen concentration, and nutrient distribution. Most of these stations are located near the Gatun and Agua Clara locks in the Atlantic, with greater depth and spatial area of the reservoir.

Cluster 2, composed of stations such as Laguna Alta (LAT), Toma de Agua Mendoza (TME), Humedad (HUM), and Toma de Agua Cuipo (TAC), represented peripheral areas with aquatic vegetation and lower impact from canal operations; it presents a scenario where maritime traffic has a much lesser influence. The main challenge lies in nutrient sources that may come from nearby agricultural areas, potentially contributing to moderate levels of nitrogen and phosphorus or other pollutants at certain times of the year. Additionally, turbidity may be influenced by rivers from the sub-basins of Los Hules, Tinajones, and Caño Quebrado, which discharge into the reservoir, close to these sites (LAT and TME). The aquatic vegetation in this cluster could be a positive factor that helps regulate turbidity and absorb nutrients; however, the excessive growth of aquatic plants can lead to low dissolved oxygen levels, especially in quieter and shallower areas.

Cluster 3, composed of the stations Gamboa (DC) and Toma de Agua Paraíso (TMR), consisted of stations located in the narrow part of the reservoir; these are sites where canal operations and maritime traffic could quickly influence the water characteristics in this area, especially in TMR. The stations are located near the Miraflores, Pedro Miguel, and Cocolí locks on the Pacific side. In the case of DC, the middle section of the Chagres River and Chilibre exerts a direct influence. Both sites play a vital role in the supply of drinking water in Panama, and their continuous monitoring is essential to ensure that the water extracted from the Gatun Reservoir meets the necessary standards for human consumption.

The clear zoning suggests the need for specific management strategies for each area. Monitoring should maintain representative stations for each group to adequately capture the variability of the system. Management measures should consider the unique characteristics of each zone, particularly in relation to sediment control, water quality conservation, and the management of the impact of maritime traffic in the canal area. The spatial coherence of the clustering, indicating that stations within each cluster are geographically close, reinforces the interpretation that the observed patterns respond to spatially structured processes.

4. Conclusions

This study enabled a comprehensive analysis of the physicochemical and biological parameters of the water in the Gatún Reservoir, providing a detailed understanding of the interrelationships among these parameters.

On the one hand, principal component analysis was employed to identify the parameters that contributed the most to the variation in the water quality of the Gatún Reservoir. The first component revealed that the parameters with the greatest influence on water quality in the Gatún Reservoir were the conductivity, chlorides, sodium, magnesium, potassium, phosphorus, dissolved solids, sulfates, hardness, salinity, and pH. The second component highlighted that other parameters such as turbidity, nitrogen, total alkalinity, calcium, E. coli, total coliforms, total suspended solids, and temperature also affected the water quality, but to a lesser extent. Consequently, the water quality in the watershed ecosystem is primarily controlled by two processes: the natural mineralization of the water and contamination processes along with the input of particulate matter, likely associated with anthropogenic activities and runoff. As a result, the number of parameters was reduced from twenty-three to two principal components.

On the other hand, clustering patterns were identified that reflect the characteristics of the 14 monitored stations, allowing for the recognition of three clusters sharing homogeneous conditions. The first cluster consisted of eight stations. These stations were mostly located near the Gatún and Agua Clara locks in the Atlantic, where the water reaches greater depths, and the operations of the Panama Canal have a more significant impact. The second cluster, made up of four stations, corresponded to peripheral areas with aquatic vegetation, a positive factor that helps regulate water turbidity. The third cluster, consisting of two stations, was located near the Miraflores, Pedro Miguel, and Cocolí locks, where maritime traffic can rapidly influence the water characteristics in this area. Based on this classification, homogeneous sampling stations with similar characteristics were identified. The clear zonation underscores the importance of adopting management strategies tailored to each cluster.

Finally, the authors acknowledge that this research study had certain limitations, among which its cross-sectional approach stands out. For a deeper understanding of the dynamics of the water quality parameters in the Panama Canal, it would be advisable to conduct a longitudinal study that allows for the analysis of spatiotemporal variations. Furthermore, future research could involve analyzing the data to differentiate between the dry and wet seasons more precisely.