Comparative Validation of Vanadium(III) Chloride Reduction and Standard Methods for Nitrate Quantification in Organic-Rich Tropical Waters

Abstract

Natural organic matter (NOM) can interfere with the determination of nitrate (NO₃⁻) in tropical waters. This study compared three standard methods (APHA 4500-NO₃⁻ B, E, and J) with an optimized vanadium(III) chloride (VCl₃) reduction method coupled with the Griess reaction. The VCl₃ reaction was optimized at 1.0 M HCl in the VCl₃ reagent, 60 °C, and 45 min. At dissolved organic carbon (DOC) above 3.0 mg C/L, direct ultraviolet (UV) spectrophotometry (Method B) overestimated nitrate, and its 275 nm correction exceeded the 10% rejection criterion of the method. Spike recovery with enzymatic reduction (Method J) decreased to 54.1 ± 4.2% at 20.0 mg C/L DOC, whereas cadmium (Method E) and VCl₃ reduction gave 98.5 ± 2.2% and 101.2 ± 1.8%, respectively. For the VCl₃ method, the limit of quantification (LOQ) was 0.010 mg N-NO₃⁻/L (relative standard deviation, RSD, 3.8%; recovery 98.2% at this level), repeatability was 1.0–2.2% RSD, and inter-analyst intermediate precision was 2.2–3.8% RSD (cadmium reduction: 4.5–6.2%). In 12 natural water samples, the VCl₃ and cadmium results differed by less than 2%. The VCl₃ method avoids packed columns and cadmium waste.

1. Introduction

Nitrate (NO₃⁻) contamination of surface water and groundwater is a widespread drinking water quality problem [1,2]. Agricultural intensification, nitrogen fertilizer application, untreated domestic wastewater discharges, and inadequate sanitation increase nitrate concentrations in water resources [3,4]. Ingestion of water with elevated nitrate concentrations is associated with infantile methemoglobinemia [5,6] and, in epidemiological studies, with the endogenous formation of N-nitroso compounds (e.g., nitrosamines) and increased cancer risk [7,8]. The World Health Organization (WHO) guideline value is 50 mg NO₃⁻/L (approximately 11 mg N-NO₃⁻/L) [9], and the US Environmental Protection Agency (US EPA) maximum contaminant level is 10 mg N-NO₃⁻/L [10].

Nitrate contamination of groundwater has been documented worldwide [11,12,13,14,15,16], including the tropical karst aquifer of the Yucatan Peninsula in Mexico [13,14] and public-supply wells in Brazil [16]. In Colombia, rural groundwater and surface waters from páramo ecosystems are affected by agricultural activities [17,18]. Natural waters also contain natural organic matter (NOM), composed mainly of humic and fulvic substances, whose concentration and aromaticity vary widely among sources [19,20]. NOM-rich waters complicate the analytical determination of nitrate.

Direct ultraviolet (UV) spectrophotometry at 220 nm, with a correction at 275 nm for dissolved organic matter, is rapid and requires no reagents [21]. However, NOM, surfactants, and some metal ions also absorb in this region, and the standard method should not be applied when the correction exceeds 10% of the reading at 220 nm [22]. Ion chromatography and capillary electrophoresis provide higher selectivity but require more costly equipment and consumables [23]. Ion-selective electrodes are portable but have higher limits of detection and are subject to interference from other anions [24]. Reduction of nitrate to nitrite followed by the Griess reaction moves the measurement to the visible region (λ = 540 nm), where NOM absorbs much less [25,26,27].

Three reduction systems are commonly coupled with the Griess reaction: cadmium (Cd) reduction, vanadium(III) chloride (VCl₃) reduction, and enzymatic reduction with nitrate reductase (NaR) [28,29,30,31,32,33,34,35,36,37]. Cadmium reduction is the basis of standard methods such as those of the American Public Health Association (APHA 4500-NO₃⁻ E) and US EPA Method 353.2 [28,29], but cadmium is toxic and carcinogenic [30], and packed columns require regular activation and maintenance. Homogeneous reduction with acidic VCl₃ avoids columns and has been adapted to small volumes, microplates, and automated systems [31,32,33,34]. Enzymatic reduction with NaR and β-NADH avoids toxic metals [35,36,37], but the enzyme is costly and sensitive to storage conditions [38].

Side-by-side comparisons of these reduction methods and direct UV spectrophotometry in NOM-rich drinking water sources are scarce. The objectives of this study were to (i) optimize the HCl concentration (0.1–3.0 M), reaction temperature (20–80 °C), and incubation time (15–90 min) of the VCl₃ method; (ii) compare the working range, limit of quantification (LOQ), repeatability, intermediate precision (between-day and inter-analyst), and trueness of the VCl₃ method with those of APHA 4500-NO₃⁻ B, E, and J; (iii) evaluate the effect of NOM on each method by spike recovery over a dissolved organic carbon (DOC) gradient of 0–20 mg C/L; and (iv) apply the four methods to 12 natural water samples from Colombian catchments.

2. Materials and Methods

The study comprised four parts: (i) single-factor optimization of the VCl₃ method (HCl concentration in the VCl₃ reagent, incubation temperature, and incubation time); (ii) validation of the VCl₃ method and of three standard methods (APHA 4500-NO₃⁻ B, E, and J) by working range, LOQ, repeatability, intermediate precision, and trueness; (iii) evaluation of the NOM effect on each method by spike recovery at added DOC levels of 0–20 mg C/L; and (iv) determination of nitrate in 12 natural water samples by the four methods.

2.1. Reagents and Analytical Standards

All reagents were of analytical grade and were used without further purification. Ultrapure water (Millipore, Bedford, MA, USA) was used to prepare reagents, dilutions, and blanks. A certified nitrate standard solution (1000 mg N-NO₃⁻/L, Certipur^®, Merck, Darmstadt, Germany) was used to prepare calibration standards (0.005–10.0 mg N-NO₃⁻/L). The Griess reagents were a sulfanilamide reagent (1.0% w/v sulfanilamide in 1.2 M HCl, prepared by diluting concentrated HCl, 37% w/w, approximately 12 M) and an N-(1-naphthyl)ethylenediamine dihydrochloride (NEDD) reagent (0.1% w/v in ultrapure water, stored in amber glass bottles at 4 °C).

For the NOM interference tests, Suwannee River NOM (SRNOM 2R101N, International Humic Substances Society, Denver, CO, United States) and humic acid sodium salt (Sigma-Aldrich, Taufkirchen, Germany) were used to fortify samples with up to 20.0 mg C/L of added DOC.

2.2. Sampling, Preservation, and Characterization of Natural Water Samples

Twelve surface water and groundwater samples ere collected in agricultural catchments of Colombia (Risaralda, Cauca Valley, and Andean plateau regions) following ISO 5667-3. Samples were collected in high-density polyethylene containers, kept at 4 °C, and filtered through 0.45 μm polyethersulfone membranes within 6 h of collection. Because unacidified samples can be stored for up to 48 h below 6 °C before nitrate determination [22], all natural samples were stored at 4 °C and analyzed within 48 h of collection. pH and electrical conductivity (HQ40d, Hach, Loveland, CO, USA), DOC by catalytic combustion oxidation (TOC-L, Shimadzu, Kyoto, Japan), and specific UV absorbance at 254 nm (SUVA₂₅₄; Genesys 50, Thermo Fisher Scientific, Waltham, MA, USA) were determined.

2.3. Analytical Methodologies and Optimization Protocols

Absorbance was measured with a double-beam UV-Vis spectrophotometer (Genesys 50, Thermo Fisher Scientific, Waltham, MA, USA) and 1 cm quartz cells; a second UV-Vis spectrophotometer (Evolution 201, Thermo Fisher Scientific, Waltham, MA, USA) was used for intermediate inter-instrument precision. Measurements were made at λ = 540 nm for the Griess-based methods and at 220 and 275 nm for direct UV spectrophotometry. Before reduction, all natural samples were analyzed for nitrite with the Griess reaction without reductant; nitrite was below the LOQ (<0.010 mg N-NO₂⁻/L) in all samples, so the absorbance measured after reduction was attributed to nitrate. Standards and samples with an absorbance of 2.0 or higher were diluted with ultrapure water before measurement, and the dilution factor was applied in the calculations.

2.3.1. Standard Methods

The three reference methods were applied as described in Standard Methods for the Examination of Water and Wastewater [22,28,37]:

-

Direct UV spectrophotometry (APHA 4500-NO₃⁻ B): Absorbance was measured at 220 nm and corrected for dissolved organic matter using the reading at 275 nm.

-

Cadmium reduction (APHA 4500-NO₃⁻ E): Samples (25 mL) buffered with NH₄Cl–EDTA (pH 8.5) were passed through a glass column packed with copperized cadmium granules at 7.0–10.0 mL/min, and the eluate was reacted with sulfanilamide and NEDD before measurement at 540 nm.

-

Enzymatic reduction (APHA 4500-NO₃⁻ J): Nitrate was reduced with eukaryotic nitrate reductase (AtNaR2) in 50 mM potassium phosphate buffer (pH 7.2) containing 100 μM β-NADH at 25 °C for 30 min, followed by color development with sulfanilamide and NEDD and measurement at 540 nm.

2.3.2. Vanadium(III) Chloride (VCl₃) Method and Optimization

The VCl₃ method was optimized by single-factor experiments with a 1.0 mg N-NO₃⁻/L standard (n = 3), varying one parameter at a time while the other two were kept at their optimized values:

HCl concentration in the VCl₃ reagent: 0.1, 0.5, 0.8, 1.0, 1.2, 2.0, and 3.0 M;

Incubation temperature: 20, 37, 50, 60, and 80 °C (thermostated water bath);

Incubation time: 15, 30, 45, 60, and 90 min.

Under the optimized conditions, a combined reagent was prepared by mixing the acidic VCl₃ solution (0.8% w/v VCl₃; 400 mg in 50 mL of 1.0 M HCl), the sulfanilamide reagent, and the NEDD reagent (2:1:1, v/v/v). Then, 1.00 mL of combined reagent was added to 1.00 mL of sample in a capped vial (final volume 2.00 mL). The final concentrations in the reaction mixture were 0.20% w/v VCl₃ (12.7 mM), 0.40 M HCl, 0.125% w/v sulfanilamide (7.26 mM), and 0.0125% w/v NEDD (0.48 mM). The vials were vortexed, incubated at 60 °C for 45 min, cooled to 20 °C for 5 min, and measured at 540 nm.

2.4. Method Validation

Validation followed the Eurachem guide [39], and terminology followed the International Vocabulary of Metrology (VIM) [40]. The LOQ of each method was first estimated as $LOQ={\displaystyle \raisebox{1ex}{$10\ast \sigma b$}\!\left/ \!\raisebox{-1ex}{$S$}\right.}$, where $\sigma$b is the standard deviation of 10 independent blank measurements, and S is the slope of the calibration function. The estimated LOQ was then verified experimentally for each method by analysing blank samples spiked at that concentration (n = 10); the LOQ was accepted when the RSD was ≤20% and the recovery was 80–120%. Precision was evaluated at three levels, 0.1 (low), 1.0 (medium), and 5.0 (high) mg N-NO₃⁻/L, in synthetic and natural water matrices and expressed as relative standard deviation (RSD):

Repeatability: six consecutive replicates (n = 6) analysed by the same analyst with the same instrument on the same day.

Intermediate precision (between-day): five determinations (n = 5) at each level (low, medium, and high) over two consecutive days, with freshly prepared reagents and an independent calibration on each day. This two-day design matches the 48 h holding time of unacidified natural samples [22]; it does not represent long-term precision.

Intermediate precision (inter-analyst and inter-instrument): independent runs by three analysts using two spectrophotometers (Genesys 50 and Evolution 201) in the same laboratory. For inter-analyst/equipment intermediate precision, Analysts 1 and 2 operated the Genesys 50 UV-Vis spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), while Analyst 3 operated the Evolution 201 UV-Vis spectrophotometer(Thermo Fisher Scientific, Waltham, MA, USA), with each analyst analysing independent replicate sets. Reported intermediate precision RSD values thus reflect combined variance across both different operators and spectrophotometers.

NOM effect: natural water samples were spiked with nitrate and fortified with the NOM materials described in Section 2.1 to obtain added DOC levels from 0 (unfortified) to 20 mg C/L. Trueness was estimated from spike recovery (%R), which reflects the bias for the added analyte only.

2.5. Statistical Analysis

Results are reported as mean ± standard deviation. Optimization experiments and natural samples were analyzed in triplicate (n = 3); the number of replicates for each validation parameter is given in Section 2.4. Differences among the four methods were evaluated by one-way analysis of variance (ANOVA) (α = 0.05); for the natural samples, ANOVA was followed by Tukey’s honestly significant difference (HSD) test. Data were processed with Excel and SPSS v24.

3. Results

3.1. Characterization of Natural Water Samples

The DOC of the natural samples ranged from 1.8 to 18.2 mg C/L, and SUVA₂₅₄ ranged from 1.8 to 4.6 L/(mg·m) (Table 3). The pH was 6.4–7.2, and the electrical conductivity was 115–480 μS/cm.

3.2. Optimization of the Vanadium(III) Chloride Method

The effects of the HCl concentration in the VCl₃ reagent, incubation temperature, and incubation time on the absorbance at λ = 540 nm are summarized in Table 1. Because the conversion of nitrate to nitrite was not measured independently, the response is expressed relative to the maximum mean absorbance (relative color response).

Table 1. Single-factor optimization of the vanadium(III) chloride (VCl₃) method (1.0 mg N-NO₃⁻/L).

Optimization Parameter	Experimental Level	Mean Absorbance (λ = 540 nm)	Relative Color Response (%)	Observations *
HCl in VCl3 reagent (60 °C, 45 min)	0.1 M	0.112 ± 0.008	32.0 ± 2.3	Incomplete VCl3 dissolution; lower absorbance.
	0.5 M	0.280 ± 0.011	80.0 ± 3.1	Lower absorbance.
	0.8 M	0.346 ± 0.006	98.9 ± 1.7	High conversion; near maximum response
	1.0 M (Optimal)	0.350 ± 0.005	100.0 ± 1.4	Maximum absorbance.
	1.2 M	0.347 ± 0.005	99.1 ± 1.4	High conversion; comparable to optimal plateau
	2.0 M	0.298 ± 0.012	85.1 ± 3.4	Decrease in absorbance.
	3.0 M	0.263 ± 0.015	75.1 ± 4.3	Decrease in absorbance; color fading.
Incubation temperature (1.0 M HCl, 45 min)	20 °C	0.105 ± 0.007	30.0 ± 2.0	Slow color development.
	37 °C	0.203 ± 0.009	58.0 ± 2.6	Incomplete color development within 45 min.
	50 °C	0.315 ± 0.008	90.0 ± 2.3	Below maximum response.
	60 °C (Optimal)	0.350 ± 0.005	100.0 ± 1.4	Maximum absorbance.
	80 °C	0.297 ± 0.014	84.9 ± 4.0	Decrease in absorbance.
Incubation time (1.0 M HCl, 60 °C)	15 min	0.210 ± 0.010	60.0 ± 2.8	Incomplete color development.
	30 min	0.312 ± 0.008	89.1 ± 2.3	Approaching plateau.
	45 min (Optimal)	0.350 ± 0.005	100.0 ± 1.4	Plateau reached.
	60 min	0.349 ± 0.006	99.7 ± 1.7	Plateau maintained.
	90 min	0.342 ± 0.009	97.7 ± 2.5	No significant change (p > 0.05).

* Mean ± standard deviation (n = 3). The relative color response was calculated with respect to the maximum mean absorbance (1.0 M HCl in the VCl₃ reagent, 60 °C, 45 min). The HCl concentration refers to the VCl₃ reagent; under the optimized conditions, the final HCl concentration in the reaction mixture was 0.40 M.

3.2.1. Effect of Acidity

The absorbance increased from 0.112 ± 0.008 at 0.1 M HCl to a maximum of 0.350 ± 0.005 at 1.0 M HCl (Table 1), and a similar maximum response was obtained between 0.8 and 1.2 M HCl At 0.1 M HCl, VCl₃ did not dissolve completely. At 2.0 and 3.0 M HCl, the relative color response decreased to 85.1% and 75.1%, respectively, and color fading was observed. A possible explanation, not tested in this study, is that excess acidity slows the coupling of the diazonium ion with NEDD. Therefore, 1.0 M HCl in the VCl₃ reagent was selected.

3.2.2. Effect of Temperature and Incubation Time

At 45 min, the relative color response was 30.0% at 20 °C, 58.0% at 37 °C, 90.0% at 50 °C, and 100.0% at 60 °C. At 80 °C, the absorbance decreased by 15.1% (0.297 ± 0.014), possibly because of thermal degradation of the azo dye. At 60 °C, the response reached 89.1% at 30 min and plateaued at 45 min (100.0%); no significant change was observed between 45 and 90 min (one-way ANOVA, p > 0.05). Therefore, 60 °C and 45 min were selected.

3.3. Validation Parameters

Calibration curves were constructed with six standards distributed over the working range of each method (Table 2), and the coefficients of determination (R²) ranged from 0.9981 to 0.9996 (Table 2). The calibration slopes of the Griess-based methods (0.3245–0.3512 AU·L/mg N) were about four times that of direct UV spectrophotometry (0.0842 AU·L/mg N); the calibration functions are shown in Figure S1. The LOQ was 0.010 mg N-NO₃⁻/L for the cadmium and VCl₃ methods, 0.050 mg N-NO₃⁻/L for the enzymatic method, and 0.100 mg N-NO₃⁻/L for direct UV spectrophotometry in low-NOM matrices; for direct UV spectrophotometry, no reliable LOQ could be established in high-NOM samples. For the VCl₃ method, the analysis of 10 blank samples spiked at 0.010 mg N-NO₃⁻/L gave an RSD of 3.8% and a recovery of 98.2 ± 3.7%, which met the acceptance criteria.

Table 2. Validation parameters of the four methods for nitrate determination.

Parameter	Direct UV (B)	Cadmium (E)	Enzymatic (J)	VCl3	ANOVA
Working range (mg N-NO3−/L)	0.100–10.0	0.010–10.0	0.050–10.0	0.010–10.0	—
Coefficient of determination (R2)	0.9981	0.9994	0.9987	0.9996	—
Slope (AU·L/mg N)	0.0842 ± 0.0011	0.3512 ± 0.0024	0.3245 ± 0.0031	0.3498 ± 0.0018	—
Intercept (AU)	0.0125 ± 0.0018	0.0084 ± 0.0012	0.0112 ± 0.0015	0.0041 ± 0.0009	—
LOQ (mg N-NO3−/L)	0.10 a	0.010 a	0.050 a	0.010 a,b	Verified experimentally b
Repeatability (RSD, %; n = 6)	3.5–6.8	2.0–3.1	2.1–4.2	1.0–2.2	F(3,20) = 14.32; p < 0.001
Intermediate precision, between-day (RSD, %; 2 days, n = 5 per level)	4.2–7.5	2.5–5.5	3.2–5.0	1.5–4.5	F(3,16) = 8.64; p = 0.001
Intermediate precision, inter-analyst (RSD, %; 3 analysts)	5.8–8.9	4.5–6.2	4.0–5.8	2.2–3.8	F(3,8) = 18.91; p < 0.001
Spike recovery at DOC = 20 mg C/L (%)	>150 (unreliable)	98.5 ± 2.2	54.1 ± 4.2	101.2 ± 1.8	F(3,20) = 142.5; p < 0.001

^a Estimated as 10σb/S. ^b Verified experimentally at LOQ (n = 10 matrix blank spikes): Direct UV (0.100 mg N/L): RSD 6.5%, recovery 95.8 ± 6.2%; Cadmium (0.010 mg N/L): RSD 4.2%, recovery 97.4 ± 4.1%; Enzymatic (0.050 mg N/L): RSD 4.8%, recovery 96.5 ± 4.6%; VCl3 (0.010 mg N/L): RSD 3.8%, recovery 98.2 ± 3.7%.

RSD ranges cover the three concentration levels (0.1, 1.0, and 5.0 mg N-NO₃⁻/L). Slope and intercept: value ± standard error (six calibration standards). ANOVA: one-way ANOVA among the four methods (α = 0.05), with degrees of freedom in parentheses. ^a Estimated as 10σ_b/S and verified experimentally at this concentration. ^b 10 blank samples spiked at 0.010 mg N-NO₃⁻/L: RSD 3.8%; recovery 98.2 ± 3.7%.

Repeatability was 1.0–2.2% RSD for the VCl₃ method, 2.0–3.1% for cadmium reduction, 2.1–4.2% for the enzymatic method, and 3.5–6.8% for direct UV spectrophotometry (F(3,20) = 14.32, p < 0.001; Table 2).

Between-day intermediate precision, evaluated over two consecutive days, was 1.5–4.5% RSD for the VCl₃ method, 2.5–5.5% for cadmium reduction, 3.2–5.0% for the enzymatic method, and 4.2–7.5% for direct UV spectrophotometry (F(3,16) = 8.64, p = 0.001). Inter-analyst intermediate precision was 2.2–3.8% RSD for the VCl₃ method, 4.0–5.8% for the enzymatic method, 4.5–6.2% for cadmium reduction, and 5.8–8.9% for direct UV spectrophotometry (F(3,8) = 18.91, p < 0.001).

Spike recoveries in the presence of NOM are summarized in Table 2. For direct UV spectrophotometry, the 275 nm correction exceeded 10% of the 220 nm reading when the DOC was above 3.0 mg C/L, which is the rejection criterion of the standard method [22]; recoveries in this range were 135–185%, or negative values were obtained after correction.

For the enzymatic method, no NOM effect was observed at DOC < 2.0 mg C/L, but recovery decreased at higher DOC (one-way ANOVA, p < 0.01), to 72.4 ± 3.1% at 10.0 mg C/L and 54.1 ± 4.2% at 20.0 mg C/L.

The recoveries of the cadmium and VCl₃ methods at a DOC of 20.0 mg C/L were 98.5 ± 2.2% and 101.2 ± 1.8%, respectively; for both chemical reduction methods, recoveries remained close to 100% over the DOC range tested.

3.4. Analysis of Natural Water Samples

The nitrate concentrations determined in the 12 natural samples are shown in Table 3. The cadmium and VCl₃ results ranged from 0.65 to 9.12 and from 0.64 to 9.20 mg N-NO₃⁻/L, respectively, differed by −1.5% to +1.6%, and belonged to the same Tukey group in all samples. For S01 and S02, no significant differences were found among the four methods (p > 0.05). For S03–S12, direct UV spectrophotometry gave significantly higher results than the other three methods (116–305% of the cadmium result). The enzymatic results were significantly lower than the cadmium and VCl₃ results in S05–S07 and S09–S12 (53–85% of the cadmium result); in S04 and S08 (91% and 65%), the difference was not significant.

Table 3. Nitrate concentrations (mg N-NO₃⁻/L) in 12 natural water samples from Colombian catchments determined by the four methods.

ID	Source	DOC (mg C/L)	SUVA254 (L/(mg·m))	Direct UV (B)	Cadmium (E)	Enzymatic (J)	VCl3	p
S01	Groundwater	1.8	1.8	2.45 ± 0.12	2.41 ± 0.05	2.38 ± 0.08	2.42 ± 0.04	0.756
S02	Groundwater	2.2	2.1	5.12 ± 0.18	4.98 ± 0.09	4.92 ± 0.11	5.01 ± 0.06	0.287
S03	Upland river	2.8	2.3	0.88 ± 0.08 a	0.76 ± 0.02 b	0.74 ± 0.03 b	0.75 ± 0.02 b	0.016
S04	River intake	4.5	3.1	1.85 ± 0.15 a	1.22 ± 0.03 b	1.11 ± 0.05 b	1.24 ± 0.02 b	<0.001
S05	River intake	6.2	3.4	3.42 ± 0.22 a	2.15 ± 0.04 b	1.82 ± 0.07 c	2.18 ± 0.03 b	<0.001
S06	Agricultural runoff	8.5	3.8	8.95 ± 0.45 a	5.40 ± 0.11 b	4.21 ± 0.18 c	5.45 ± 0.08 b	<0.001
S07	Reservoir	10.1	3.9	2.10 ± 0.19 a	0.98 ± 0.02 b	0.71 ± 0.04 c	0.97 ± 0.02 b	<0.001
S08	Wetland outlet	12.4	4.1	1.98 ± 0.21 a	0.65 ± 0.02 b	0.42 ± 0.03 b	0.64 ± 0.01 b	<0.001
S09	Agricultural canal	14.2	4.2	12.10 ± 0.72 a	6.82 ± 0.12 b	4.15 ± 0.22 c	6.88 ± 0.09 b	<0.001
S10	Drainage ditch	15.8	4.4	16.50 ± 0.88 a	9.12 ± 0.18 b	5.20 ± 0.29 c	9.20 ± 0.14 b	<0.001
S11	Lagoon	17.1	4.5	4.85 ± 0.35 a	1.85 ± 0.04 b	0.99 ± 0.06 c	1.87 ± 0.03 b	<0.001
S12	Raw stream	18.2	4.6	7.20 ± 0.52 a	3.10 ± 0.07 b	1.65 ± 0.11 c	3.14 ± 0.05 b	<0.001

Mean ± standard deviation (n = 3). p: one-way ANOVA among the four methods. Different superscript letters within a row indicate significant differences (Tukey HSD, α = 0.05). ANOVA and Tukey HSD were calculated from the means and standard deviations shown (n = 3 per method).

Table 3. Nitrate concentrations (mg N-NO₃⁻/L) in 12 natural water samples from Colombian catchments determined by the four methods.

ID	Source	DOC (mg C/L)	SUVA254 (L/(mg·m))	Direct UV (B)	Cadmium (E)	Enzymatic (J)	VCl3	p
S01	Groundwater	1.8	1.8	2.45 ± 0.12	2.41 ± 0.05	2.38 ± 0.08	2.42 ± 0.04	0.756
S02	Groundwater	2.2	2.1	5.12 ± 0.18	4.98 ± 0.09	4.92 ± 0.11	5.01 ± 0.06	0.287
S03	Upland river	2.8	2.3	0.88 ± 0.08 a	0.76 ± 0.02 b	0.74 ± 0.03 b	0.75 ± 0.02 b	0.016
S04	River intake	4.5	3.1	1.85 ± 0.15 a	1.22 ± 0.03 b	1.11 ± 0.05 b	1.24 ± 0.02 b	<0.001
S05	River intake	6.2	3.4	3.42 ± 0.22 a	2.15 ± 0.04 b	1.82 ± 0.07 c	2.18 ± 0.03 b	<0.001
S06	Agricultural runoff	8.5	3.8	8.95 ± 0.45 a	5.40 ± 0.11 b	4.21 ± 0.18 c	5.45 ± 0.08 b	<0.001
S07	Reservoir	10.1	3.9	2.10 ± 0.19 a	0.98 ± 0.02 b	0.71 ± 0.04 c	0.97 ± 0.02 b	<0.001
S08	Wetland outlet	12.4	4.1	1.98 ± 0.21 a	0.65 ± 0.02 b	0.42 ± 0.03 b	0.64 ± 0.01 b	<0.001
S09	Agricultural canal	14.2	4.2	12.10 ± 0.72 a	6.82 ± 0.12 b	4.15 ± 0.22 c	6.88 ± 0.09 b	<0.001
S10	Drainage ditch	15.8	4.4	16.50 ± 0.88 a	9.12 ± 0.18 b	5.20 ± 0.29 c	9.20 ± 0.14 b	<0.001
S11	Lagoon	17.1	4.5	4.85 ± 0.35 a	1.85 ± 0.04 b	0.99 ± 0.06 c	1.87 ± 0.03 b	<0.001
S12	Raw stream	18.2	4.6	7.20 ± 0.52 a	3.10 ± 0.07 b	1.65 ± 0.11 c	3.14 ± 0.05 b	<0.001

Mean ± standard deviation (n = 3). p: one-way ANOVA among the four methods. Different superscript letters within a row indicate significant differences (Tukey HSD, α = 0.05). ANOVA and Tukey HSD were calculated from the means and standard deviations shown (n = 3 per method).

4. Discussion

The main differences among the four methods were related to their susceptibility to NOM and their precision. These differences are discussed below in relation to the reaction chemistry and the operational characteristics of each method.

In acidic solution, each V³⁺ ion is oxidized to VO²⁺ (V(IV)) and releases one electron; therefore, two V³⁺ ions are required for the two-electron reduction of nitrate to nitrite:

2V³⁺ + NO₃⁻ + H₂O → 2VO²⁺ + NO₂⁻ + 2H⁺

Reduction of nitrate by VCl₃ is slow at room temperature, and published VCl₃ methods use heating or long incubation times [31,33,34]; the temperature dependence in Table 1 is consistent with these reports. The decrease in absorbance at 80 °C may result from thermal degradation of the diazonium intermediate or the azo dye; this was not investigated further.

The HCl concentration in the VCl₃ reagent affected both the dissolution of VCl₃ and color development. Under the optimized conditions, the final HCl concentration in the reaction mixture was 0.40 M, which provides the acidic medium required for the diazotization of sulfanilamide [26,27]. The mechanism of the lower response at higher acidity was not studied.

The difference in inter-analyst RSD between the VCl₃ method (2.2–3.8%) and cadmium reduction (4.5–6.2%) may be related to operational differences between the two procedures. Cadmium reduction depends on column packing, flow rate (7.0–10.0 mL/min), and periodic reactivation of the copperized cadmium, and incomplete control of these factors can change the reduction efficiency [28]. The VCl₃ method is performed in a single liquid phase with fixed volumes, temperature, and time, which reduces the number of operator-dependent steps. This explanation was not tested experimentally. The homogeneous format can also be adapted to small volumes and microplates [34].

Direct UV measurement at 220 nm is based on the absorption of nitrate in the far UV, where aromatic and conjugated structures of NOM also absorb [21]. The 275 nm correction assumes a constant relationship between NOM absorbance at 220 and 275 nm, which may not hold for samples with different NOM composition [22]. In this study, the correction exceeded the 10% criterion above 3.0 mg C/L DOC, and the method gave higher results than cadmium reduction in natural samples S03–S12 (116–305%; Table 3), including S03, with a DOC of 2.8 mg C/L. Direct UV spectrophotometry is therefore not suitable for the NOM-rich waters studied here.

The decrease in recovery observed with the enzymatic method at higher DOC suggests interference with the enzymatic assay. Possible explanations include adsorption of humic substances onto the enzyme, conformational changes in the protein, and consumption of β-NADH by NOM. This study did not test these mechanisms, which require further investigation. In the natural samples, the enzymatic results were significantly lower than the cadmium results in seven of the 12 samples (53–85% of the cadmium result; Table 3).

For the cadmium and VCl₃ methods, recoveries remained close to 100% up to a DOC of 20 mg C/L. In both methods, nitrate is converted to nitrite and measured as an azo dye at 540 nm, where NOM absorbance is low [26]. In addition, the VCl₃ reaction does not depend on a protein catalyst. The agreement between the two methods in the natural samples (−1.5% to +1.6%; Table 3) supports this result.

The International Agency for Research on Cancer classifies cadmium as carcinogenic to humans (Group 1) [41], and it is toxic at low exposure levels [30]; cadmium columns therefore generate waste containing Cd²⁺ and Cu²⁺ that requires special disposal. The enzymatic method avoids toxic metals [35], but the enzyme and β-NADH require cold storage. Based on 2026 reagent prices, the direct cost per determination (including waste disposal) was approximately estimated at USD 0.12 for the VCl₃ method, USD 0.45 for cadmium reduction (including column consumables), and USD 2.80 for the enzymatic method. Vanadium compounds are also toxic and require appropriate handling and waste management, and this study did not perform a formal environmental impact assessment; therefore, the environmental advantage of the VCl₃ method is limited to avoiding cadmium waste.

This study has several limitations. First, the natural samples were collected in a limited number of Colombian catchments (DOC 1.8–18.2 mg C/L; SUVA₂₅₄ 1.8–4.6 L/(mg·m)), and NOM of different composition may behave differently. Second, between-day intermediate precision was evaluated over only two consecutive days, matching the 48 h holding time of unacidified natural samples [22]; long-term precision should be estimated from quality control data over an extended period, as recommended by Eurachem [39]. Third, interference from high concentrations of metals (e.g., Fe³⁺ or Mn²⁺) and from salinity was not evaluated. Fourth, only one nitrate reductase (AtNaR2) was tested. Finally, the VCl₃ method requires incubation at 60 °C for 45 min, and the cost estimates are approximate (2026 prices) and depend on local conditions.

5. Conclusions

In NOM-rich waters from Colombian catchments, direct UV spectrophotometry (APHA 4500-NO₃⁻ B) was not applicable above a DOC of 3.0 mg C/L, and the recovery of the enzymatic method (APHA 4500-NO₃⁻ J) decreased with increasing DOC (54.1 ± 4.2% at 20.0 mg C/L). The cadmium (APHA 4500-NO₃⁻ E) and optimized VCl₃ methods gave recoveries close to 100% and agreed within 2% in 12 natural samples. The VCl₃ method (1.0 M HCl in the reagent, 60 °C, 45 min) had an LOQ of 0.010 mg N-NO₃⁻/L and an inter-analyst RSD of 2.2–3.8% (cadmium reduction: 4.5–6.2%), and it does not require cadmium columns. The VCl₃ method is therefore a suitable alternative to cadmium reduction for routine nitrate determination in NOM-rich waters; long-term precision and other matrix interferences should be evaluated before routine implementation.