Application of central composite design approach for optimization nitrate removal from aqueous solution by immobilized Pseudomonas putida

Nitrate is considered as one of the major nutrients that many living organisms require for physiological processes. High concentrations of nitrate are a great problem to the environment as well as to public health. Surface water may contain high levels of nitrate due to excessive and random use of fertilizers in agriculture, which is used to increase the quality of soil and results in leakage of nitrate from soil to surface and ground water. Nitrate pollution may also be due to other sources like industrial effluent, which has a high concentration of nitrate, domestic wastewater, municipal sewage, canals septic tanks, sewage dumping grounds and animal feedlots (Wakida & Lerner 2006; Krishnaswamy et al. 2009; Roy et al. 2012; De Filippis et al. 2013). High levels of nitrate were detected in groundwater as reported by Filintas et al. (2008). The presence of high concentrations of nitrate in drinking water can cause public health problems like stomach cancer in infants (Jin et al. 2004).

Cell immobilization is a technique that allows microorganisms to grow in a solid matrix and successfully limits the mobility of the cells using synthetic or natural polymers, which can be used as biocatalysts instead of free cells (Bouabidi et al. 2019). Immobilization gives an easier purification: enhance cell residence time, substrate loading rate and separation procedures of solids from liquids (Stolarzewicz et al. 2011; Kunapongkiti et al. 2019).

Immobilized living cells have been reported to be very effective in bioremediation (Abdullateef et al. 2020). Kumar & Saramma (2012) reported that some studies have increasingly focused on the use of immobilized microorganisms as a valid method that avoids the harvesting problem. From previous studies, the heterotrophic nitrification and aerobic denitrification were identified for nitrogen removal process via microbial entrapment with sodium alginate gel. The immobilized cells gave a higher removal percentage than free living cells (Yan et al. 2019).

Maximum nitrate removal percentage is influenced by several factors including pH, incubation time, nitrate concentration and bacterial inoculum. Several statistical experimental designs may be used as useful techniques to improve optimizing the process variables. Screening designs, such as fractional factorial and Plackett–Burman, may be used for pre-formulation evaluations.

Response surface methodology (RSM) is used only when a few significant factors are involved in optimization process. RSM, a special type of experimental design, can be described as a collection of both mathematical and statistical techniques for designing, improving, optimizing, and formulating product designs where several variables potentially affect characteristics of products designs. RSM represents the most useful method for analyzing data and is an effective statistical technique for modelling the optimization of multiple parameters that have an effect on response surface (Baş & Boyacı 2007; Mansouri et al. 2012; Tripathy & Murthy 2012). Useful practices in response surface designs include central composite design (CCD), Box–Behnken design, and two- and three-level factorial design (Palamakula et al. 2004).

A comparison study by Rakić et al.(2014) of four experimental design types in response surface designs (two-level full factorial design, CCD, Box–Behnken design and three-level factorial design) were reported for the determination of fluconazole and its impurities. The study reported that both central composite and three-level full factorial design models provided satisfactory statistical parameters. A study by Acharya et al. (2018) represented that the CCD and Box–Behnken are the most important surface-response methods for designing the test. Another studies stated that CCD is the most common experimental design used in process optimization studies under RSM and is widely used to create second-order surface-response models (Chen et al. 2012; Moradi et al. 2016; Asadzadeh et al. 2018).

In this study, the feasibility of using immobilized Pseudomonas putida for nitrate removal from aqueous solution was estimated using the CCD in RSM for optimizing different important variables for removal efficiency. Samples were collected from Sabal Drain and also from fresh effluent fish farm at Al-Menoufiya and were collected during October 2019 before being sent to the laboratory for analysis.

Wastewater samples were collected in triplicates from Sabal Drain, which is the largest drain in Al-Menoufiya, Egypt. It extends for more than 62 km and discharges its contents into the Rosetta branch of the River Nile (Abdel-Khalek et al. 2018), as shown in Figure 1, in special containers for the testing of parameters according to Standard Methods for the Examination of Water and Wastewater (APHA 2012).

To isolate bacteria capable of reducing nitrate (NO₃⁻) in aqueous solutions, about 10 mL of collected wastewater sample was first individually grown in a 250 mL Erlenmeyer flask containing sterilized 100 mL of a modified minimum denitrification medium (DM) according to Kesserű et al. (2003): 1.8 g K₂HPO₄, 1.08 g KH₂PO₄, 0.1 g MgSO₄ 7H₂O, 0.5 g NH₄Cl, 0.1 g KCl, 0.5 g/L NaNO₃, and 0.05 CaCl₂ 2H₂O g/L. The enrichment was incubated for 24 h at 35 °C to isolate bacteria that can survive in high concentrations of nitrate. The inoculum was used for further removal studies. The obtained bacterial consortia survived harmful conditions in the presence of high levels of nitrate in the solutions and were used as inoculum for further bio-removal studies. One mL of consortium was spread on DM agar plates and amended with 0.5 g/L of nitrate concentration before incubating for 24 h at 35 °C. The predominated bacterial colonies were selected for the rest of the research. The strains were maintained on slants of nutrient agar.

Nitrate tests from Angar et al. (2016) were followed to select the bacterial strain with the nitrate reductase enzyme, which reduces nitrate (NO₃⁻) to harmless nitrogen gas (N₂) and then becomes denitrifying bacteria. A heavy inoculum of test organism was inoculated in sterilized nitrate broth tubes fitted with Durham tubes containing peptone 5 g/L, meat extract 3 g/L, potassium nitrate 1 g/L and with a final pH of 7.0 ± 0.2. It was incubated at 35 °C for 24 h. After incubation, nitrogen was checked in the Durham tubes before adding reagents. If there was no nitrogen gas, two reagents were added: three drops of 1% of nitrate reagent A (sulfanilic acid) and one drop of 0.02% nitrate reagent B (alpha-naphthylamine). The development of a red or purple colour within 5 min was considered a positive reaction for nitrate reduction to nitrite. Negative tests were further confirmed with the addition of zinc dust, where zinc can reduce the nitrate to nitrite. Failure of colour development was recorded as a positive test for nitrate reduction beyond nitrite to gaseous or non-gaseous nitrogen compounds. The development of the red colour after the addition of zinc particles indicated the presence of nitrate in the culture medium.

The immobilization technique for isolated nitrate-reducing bacterial strains was used. First, about 2% of sodium alginate was dissolved in saline 0.85% at pH 7 and then autoclaved for 5 min. About 5 mL of bacterial culture cotaining 10⁶ colony-forming units (CFU) per millilitre was centrifuged at 2,000 rpm for 10 min. The supernatant was discarded and the cells were re-suspended in 5 mL of saline solution (0.85%). About 5 mL of sodium alginate was mixed with 5 mL of bacterial suspension and was shaken manually for 20 min. Round beads, approximately 3 mm in diameter, were obtained by dropping the alginate bacterial mixture through a sterilized syringe into a sterile 250 mL Erlenmeyer flask containing 50 mL of cold sterile 2% calcium chloride, at room temperature in sterile conditions. The formed beads were allowed to harden in calcium chloride for 1 h. The beads were washed several times in sterile saline solutions (0.85%) to remove any remains of calcium chloride and were later stored at 4 °C in autoclaved distilled water (Kumar & Saramma 2012).

For identification of the bacterial strain, the genomic DNA was extracted from bacterial culture grown in nutrient broth and was centrifuged at 3,000 rpm for 15 min. Extraction of genomic DNA was carried out using the PrepMan™ ultra sample preparation reagent (PN 4322547). The 16S rRNA genes were amplified by polymerase chain reaction (PCR) using the universal bacterial specific primer performed in a 20 μL reaction containing 10 μL Premix Taq, 1.5 μL of DNA template, 0.5 μL forward primer, 0.5 μL reverse primers and 7.5 μL double-distilled water. The PCR programming used for the initial step was 95 °C for 10 min, 30 cycles at 95 °C for 30 s, 60 °C for 30 s and 72 °C for 45 s and the final step was at 72 °C for 10 min. The amplified PCR product was detected in the samples by running on 1.5% agarose gel and sequencing as the 16S rRNA amplicon. Resulting gene sequences were subjected to BLAST search tool (Altschul et al. 1997) in the NCBI gene database and submitted to GenBank to obtain the accession number. Alignment with the similar sequences was carried out using ClustalW (Thomas et al. 2013). Finally, MEGA X software (https://www.megasoftware.net) was used to build the phylogenetic tree by the neighbour-joining method, with 1,000 bootstrap replicates and the maximum composite likelihood model (Kumar et al. 2008; Kumari et al. 2016).

RSM based on CCD is a mathematical and statistical technique which is useful for designing experiments, analyzing the interaction between independent variables and creating the optimum synergy of this response before checking the response patterns (Khuri & Mukhopadhyay 2010).

Five independent variables factors, i.e. pH, temperature, nitrate concentration, bacterial inoculum concentration and sodium alginate concentration, were studied at five levels: −α(minimum), −1(low), 0(central), +1(high) and +α(maximum), where these levels were selected based on preliminary study results, as well as previous studies (Tong et al. 2014; Asadzadeh et al. 2018). The factors and levels are shown in Table 1.

After statistical modelling and regression analysis, analysis of variance (ANOVA) was used with the significance level P-value (P ≤ 0.05), F-value and the coefficient R² to define the significance of each factor in the fitted equations and to evaluate the goodness of fit in each case (Asadzadeh et al. 2018). Minitab was used to produce three-dimensional (3D) response surface plots as a graphical representation for the regression equation for the optimization of reaction conditions. It was considered as the most useful approach in revealing the conditions of the reaction system. They were plotted based on the simultaneous interaction effect of the two-factor levels on the responses by plotting the removal percentage (response) on the Z-axis and two independent variables on the X-axis and Y-axis, and keeping the other variables at the coded zero point (Sarrai et al. 2016).

Fresh effluent samples from an open fish farm located at Al-Menoufiya governorate were collected. The collected raw effluent samples were immediately transported to the environmental laboratory and analyzed for water quality parameters.

Collected samples were treated by immobilized denitrifying strain, using the suggested optimum condition by RSM based on CCD. Nitrate concentration was measured after biotreatment in the collected samples, and the nitrate bioremediation percentage was calculated using Equation (1). When the raw effluent served as the control, all experiments were performed in triplicate.

The free and immobilized strain was stored at 4 °C in saline solution for 2 months and used for treating the collected fish farm samples. The storage, reusability activity and bioremediation percentage was determined every week.

The consortium from the collected wastewater sample was spread after enrichment separately on DM agar amended with 500 mg/L of nitrate. On the basis of morphology, nine different colonies were obtained. Further selection for these colonies was made on the basis of nitrate reduction test. Nitrate test with isolated strains is shown in Figure 2. The strong nitrate reduction ability appeared with isolate named (S6), which reduced nitrate to nitrogen gas, and was therefore selected for their nitrate removal efficiency for further studies. Isolate with name (S2a) reduced nitrate to nitrite, and the solution changed to a pink colour, as shown in Figure 2. The other strains gave no change in colour when reagents A, B or the zinc powder were added, suggesting that nitrate reduction did not occurred. Positive results appear red, representing nitrate reduction to nitrite (NO₃ → NO₂). Negative results showed no colour change and the test needed to be completed by adding zinc dust. After the addition of zinc dust, a positive result showed appear no colour, suggesting nitrate reduction nitrite to gaseous or non-gaseous nitrogen compounds. A negative results shows red colour and suggests no nitrate reduction (Marietou et al. 2009).

The final bacterial beads appeared after they were washed several times in distilled water to remove any calcium chloride residue, as shown in Figure 3.

Partial sequence of isolate S6 based on 16S rRNA gene sequence with (1,527 bp) was obtained and submitted to the GenBank. The accession number MT364822.1 was obtained. Analysis using BLAST indicated that strain P. putida MT364822.1 (http://www.ncbi.nlm.nih.gov/MT364822.1), with excellent nitrate removal ability isolate during the study, was closely related to Pseudomonas sp. There are several studies about denitrifying species such as Pseudomonas sp., which can be isolated from canals, ponds, soils and activated sludge, that have the ability to utilize nitrate (Kesserű et al. 2003; Wu et al. 2012b). A phylogenetic tree was constructed on the basis of the gene sequences 16S rRNA of the isolate, with other reference sequences shown in Figure 4, and the result showed that P. putida MT364822.1 (S6) belongs to P. putida species.

RSM based on CCD was employed for studying interaction effects between five independent variables to optimize the nitrate bioremediation process. The variables were tested at five different levels −α(minimum), −1(low), 0(central), +1(high) and +α(maximum) acquired from preliminary experiments in this study, as shown in Table 1.

According to the experimental CCD design, a total of 32 runs were carried out, and the results of the statistical plan and response procedures are summarized in Table 2. Results showed that the efficacy of nitrate removal in five different levels of the five variables was different. According to the results of the experimental CCD study, the highest and lowest percentages of nitrate removal by immobilized P. putida MT364822.1 strain, were 90.94% and 57.81%, respectively, and were obtained in experiment run no. 26 and 9, respectively. This may be attributed to the change in variables, which results in different interactions that affect the removal efficiency, allowing them to become an effective tool for optimizing the removal process. In addition, high accuracy of the model was estimated from a small difference between the actual and predicted efficiency in estimating nitrate removal efficiency (response variable).

In the above equation, A is the immobilized bacterial beads (mg), B is sodium alginate concentration (%), C is nitrate concentration (mg/L), D is the pH and E is the temperature (°C).

The second-order polynomial equation is widely used in RSM because it is very flexible, it can take on a wide variety of functional forms, so it will work well as an approximation to the true response surface; it is easy to estimate the parameters (β) in the second-order model and there is evidence that second-order models work well in solving real response surface problems (Carley et al. 2004).

The results of experiments were subjected to variance analysis (ANOVA) to validate the model and variance analysis. The significance of each coefficient in the model was determined using the F-test and probability of error (P-values) for ANOVA at a 95% confidence. Experimental design was significant when P-value <0.05 while P-values >0.1 indicated that the model terms were not significant (Moradi et al. 2016). The resulting ANOVA on this model are listed in Table 3. Based on results of P-values, the model was highly significant, as shown in the (0.000) P-value in the regression. Moreover, first-order effects of all five variables (A, B, C, D and E) were significant for responses process. The P-values were larger than (0.05) for the model interaction, B*B (0.825), A*B (0.316), C*D (0.057) and C*E (0.925), which represented the insignificance of these terms in nitrate removal efficiency by immobilized P. putida MT364822.1.

Lack-of-Fit test coefficient was created to determine whether the selected model was sufficient to describe the observed data or whether a more complicated model should be used. The test would be significant if the model was not well-fitted (Wu et al. 2012a). The high P-value for lack-of-Fit in this model (0.762) was not significant, thus implying that the model was valid and significant to demonstrate the correlation between the factor variables and response process (Jafari et al. 2014). The F-value of the model was 13.69, which indicated the significance of the model. Also, the fitting of selected models to the experimental data was tested by calculating the determination regression coefficient (R²). The high value of determination regression coefficient (R² = 0.9614) represented a good correlation between the experimental values and predicted response, as shown in Figure 2. The model will be stronger when the determination regression coefficient (R²) becomes nearer to one (Jaafari & Yaghmaeian 2019). The significance of the model was also confirmed by the adjusted R² and predicted R². The adjusted regression coefficient (R²) for the removal efficiency model (0.8912) was in reasonable agreement with determination regression coefficient R² value. The predicted regression coefficient (R²) was estimated 0.7831 and revealed that the predicted response could be well calculated by the model. Therefore, there was a satisfactory match between the adjusted regression coefficient of the experiment and the predicted regression coefficient of the model in this study.

Normal probability plot of the residuals is a graphical plotting technique for assessing the normal distribution of data (Chambers et al. 1983). If the points on the plot fall fairly close to the straight line, then the data are normally distributed, and some scatter is to be expected (Pokhrel & Viraraghavan 2006). Figure 5(a) plots the residuals values against the predicted response. Residuals appeared scattered randomly around zero, suggesting that errors have a constant variance. Predicted versus experimental values were plotted and are shown in Figure 5(b) where shows a good correlation and agreement between the experimental values and predicted response in which all the points are close to the diagonal line. Figure 5(c) shows normal probability plot of residual values. Experimental points appear arranged reasonably, suggesting normal distribution. In Figure 5(d) the residuals are plotted against the order of runs used in the design. Residuals seem to be scattered randomly around zero. Except for two point (runs number 4 and 22 having a residual value of 4.9 and −5 respectively), all other points were found to fall in the critical alpha value ranges of +2 to −2. A histogram of the residuals plots used to demonstrate the distribution of the residuals for all observations are shown in Figure 5(e). Errors were distributed normally with zero and this shows a symmetrical histogram (Ranjan et al. 2009). A Pareto chart ranks the effect of effective variables on nitrate removal process (response) as in Figure 5(f). As shown in Figure 5(f), sodium alginate concentration and nitrate concentration are the most important parameters for nitrate removal efficiency, followed by pH, temperature and finally the bacterial inoculum, where A is the bacterial beads volume, B is the sodium alginate concentration, C is the nitrate concentration, D is the pH and E is the temperature.

To demonstrate the interaction effects between variables gained by solving second-order polynomial equation, three-dimensional (3D) plots were created (Asadzadeh et al. 2018) and shown in Figure (6), where (a) bacterial beads volume and sodium alginate; (b) bacterial inoculum and nitrate concentration; (c) pH and bacterial concentration; (d) temperature and bacterial inoculum concentration; (e) sodium alginate concentration and nitrate concentration; (f) pH and sodium alginate concentration; (g) sodium alginate and temperature; (h) nitrate concentration and pH; (i) temperature and nitrate concentration; and (j) pH and temperature of 35° in efficacy of nitrate removal by immobilized P. putida MT364822.1.

The independent variable interactions are illustrated by 3D surface plots, derived from the quadratic polynomial model, as shown in Figure 6(a)–6(j).

The study was performed using different volumes of immobilized bacterial cells, the results of which are shown in Figure 6(a)–6(c). Figure 6(a) shows the interaction between bacterial inoculum and sodium alginate concentration on the nitrate removal rate. As expected, there is an increase in the nitrate removal rate with an increase of bacterial beads. The maximum nitrate removal rate was achieved by increasing the bacterial beads range to 3 g. These results are in agreement with previous studies that have reported that the removal rate was dependent on bacterial beads volume and maximum nitrate removal, obtained by increasing bacterial concentration (Shawky et al. 2015). Also, it was observed that higher nitrate removal was achieved when sodium alginate concentration was 2%. Figure 6(b) illustrates the interaction of bacterial beads volume and nitrate concentration on the removal process. As shown in the 3D plots, increasing nitrate concentration up to 600 mg/L gradually decreases the removal process. 3D plots in Figure 6(c) show the combined effect of pH and bacterial concentration. The removal rate was about 90.9% when increasing pH to 7.0 and increasing the bacterial concentration balls to 3 g.

Figure 6(e) shows the interaction effect of the sodium alginate concentration and nitrate concentration on the removal rate. As the sodium alginate increases from 1.5 to 2%, the beads showed a better transparency and hardness, and so the removal efficiency of bacterial isolate increased, with the highest removal efficiency at 87.86% and 90.7%. A study by Dong et al. (2017) demonstrated that as sodium alginate increased from 0.8% to 1.5%, the removal efficiency increased because the immobilized balls were transparent and soft, which affected positively on the removal efficiency. However, when the concentrations were higher than 2%, the immobilized balls were opaque, hard, and trailing and negatively affected the removal efficiency. Results and studies suggest that the concentration of sodium alginate affects the hardness of immobilized cells. The strength of immobilized cells increases as the concentration of sodium alginate increases. However if the concentration was too high, immobilized cells would not grow, and if the concentration was too low, immobilized cells would break.

Figure 6(g) depicts the effects of sodium alginate and temperature. From the curve, it could be found that the pattern of response increased with a sodium alginate increase from 1.0 to 2.0% and temperature increase from 25 to 35 °C.

The denitrification abilities of microorganisms can be enhanced by gradually exposing them to higher concentrations of nitrate. Adaptation of a microbial community towards toxic or recalcitrant compounds is useful in improving the rate of bioremediation process (Dafale et al. 2010). As shown in Figure 6(h), nitrate concentration of 600 mg/L showed gradual decrease denitification ability of P. putida MT364822.1, with a nitrate removal of 56%. There was also an increase in the response process with an increase in initial nitrate concentration from 200 to 400 mg/L. Similar results were reported by Chen et al. (2015), which reported that the denitrification rate increased with initial nitrate concentration (100–350) mg/L and decreased at high nitrate loading (550–600) mg/L. Finally, nitrate removal efficiency was found to be inhibited at high nitrate loading (800–1,000 mg/L). In addition, Figure 6(i) shows the mutual effect of temperature and nitrate concentration on nitrate removal, with the highest efficiency at 400 mg/L and temperature of 35 °C.

Temperature is another influential important environmental factor for bacterial growth and their survival (He et al. 2018). The effect of simultaneous variations of pH and temperature on nitrate removal was studied, with the highest efficiency at pH 7 and temperature of 35 °C, as shown in Figure 3(j). It can be concluded from the 3D graphs in Figure 6(d) that increasing temperature leads to a linear increase in nitrate removal process in the model. A study by Munna et al. (2015) reported that P. putida is able to grow and conduct aerobic denitrification, with a nitrate removal of 82% at 35 °C, while growth declined at lower temperatures (25 °C). However, Wu et al. (2012b) evaluated that at 45 °C, the nitrate removal was almost completely reduced. At a temperature of 20 °C, the cell activity for nitrate removal was inhibited. Ji et al. (2015) reported that the optimum temperature for denitrification process was ranged from 25 to 37 °C, which is generally the optimum temperature for bacterial activity.

pH had a great effect on nitrate removal and also on bacterial growth, and is one of the leading factors that can influence bacterial ability in reducing efficiency (Zhang et al. 2011). Figure 6(f) shows the changing parameter of pH and sodium alginate concentration on nitrate removal yield. The optimum pH value is 7 and nitrate removal rate was gradually decreased when the pH increased to 9.0, with the removal percentage about 60.0%. These findings are confirmed by Cai et al. (2015) who concluded that neutral conditions enhanced denitrification rate. Glass & Silverstein (1998) also reported that optimum pH for denitrification was close to 8.0, while acidic environments (pH <5) inhibited denitrification and prevented the denitrification chain (Brady & Weil 2002). Burghate & Ingole (2017), however, reported that there was no significant effect of pH between 7.0 and 8.0 on the denitrification rate.

Based on the above findings and confirming the model's accuracy the optimum values for the independent variables and nitrate removal efficiency (responses) were obtained using the response optimizer in Minitab 18 program. Maximum efficacy of nitrate removal using immobilized P. putida MT364822.1 was predicted to be 91.1% when optimal conditions were pH 7, temperature 35 °C, nitrate concentration 400 mg/L, sodium alginate 2% and bacterial beads 3 g/L.

Water quality parameters of raw fish farm effluent are summarized in Table 4. The applicability of effective immobilized P. putida MT364822.1 and pure strain on nitrate removal from fish farm was estimated. Optimum conditions suggested by the CCD design were used. The pure strain had a faster removal rate of nitrate concentration than the immobilized P. putida MT364822.1 beads, which reached 89% and 82%, respectively, during the first 24 h of biotreatment (Figure 7).

One of the main objectives of the immobilization is to extend the storage time of free cells. Therefore, storage stability and reusability of the immobilized system is very important for successful biotreatment and for the industrial application. As shown in Figure 8, the immobilized P. putida MT364822.1 was quite stable when stored at 4 °C for 8 weeks and showed the strength of the immobilized strain in repeated use for eight cycles, with retained activity up to 82%. However, the free strain was completely active for 2 weeks, after which the activity decreased about 19–55% after storage for 3–8 weeks. This enhanced stability of immobilized cells is probably a result of autolysis prevention by immobilization so the immobilized P. putida MT364822.1 had a longer removal of nitrate pollution (Yan et al. 2019). Various reports have discussed the high operational stability of immobilized strains (Lu et al. 2012).

Majeau et al. (2010) discussed that free strains are not stable under normal storage conditions, because when at room temperature or under refrigeration they lose their activity gradually over time. This instability is because protein may be denatured over time, but this can be resolved by immobilization (Leonowicz et al. 1988). Stability of the immobilized strain may be due to the improved stabilization of its active conformation and lower flexibility of the strain in the immobilized form (Lonappan et al. 2018).

Mogharabi et al. (2012) studied the reusability of the entrapped strains and the results showed that 85% of the activity was retained after five successive cycles. Crestini et al. (2010) found that immobilized strains retained 68% of its activity for ten cycles. Reusability of the immobilized strains was suggested where the decrease in removal efficiency by immobilized strain during recycling could be due to detachment of the immobilized bacteria in the washing step (Bryjak et al. 2007; San et al. 2014).

The current study showed that the immobilization technique for certain species of bacteria was perfect for controlling nitrates in aqueous solution. The high removal efficiencies of nitrate by immobilized P. putida MT364822.1, which were isolated during the study from the Sabal Drain, represented the denitrification ability of this strain. RSM experimental design based on CCD was effective and was used to obtain the operational conditions for the maximum removal rates of nitrate by immobilized P. putida MT364822.1. In this study, the effective parameters in the CCD that supported optimal bacterial activity for nitrate removal in liquid medium included pH 7, temperature 35 °C, nitrate concentration 400 mg/L, bacterial beads volume 3 g/L and sodium alginate concentration 2%. Furthermore, the model was presented as 3D response surface plot to investigate the optimum zone of nitrate removal. Nitrate concentration removal from raw fish farm samples over 24 h reached 89% and 82.2% by bacterial solution and immobilized P. putida MT364822.1 beads, respectively. The removal efficiency of the pure strain decreased after storage for 3–8 weeks. Storage and reusability experiments showed the highest operational stability of immobilized P. putida MT364822.1 for eight cycles with retained activity up to 82.2%. It is therefore suggested that immobilized P. putida MT364822.1 is an excellent bioremoval for nitrate in wastewater.

Special thanks go to all the staff in the Central Laboratory for Environmental Quality Monitoring for their collaboration during measurements.

All relevant data are included in the paper or its Supplementary Information.