Optimization and Physicochemical Characterization of Pectin Extraction from Watermelon Rind (Citrullus lanatus) with Citric Acid

Watermelon rind was used for the pectin extraction with citric acid as the extractant solvent. The effects of pH (2.0-3.0), extraction time (45-75 min) and liquid-solid ratio (1:10-1:40 g/ml) on the pectin yield, degree of esterification, methoxyl content and anhydrouronic acid content were investigated using BoxBehnken surface response experimental design. The pH was the most significant variable for the pectin yield and properties. The responses optimized separately showed different optimal conditions for each of the variables studied in this work. Therefore, the desirability function was used to determine the sole theoretical optimum for the highest pectin yield and highest anhydrouronic acid content, which was found to be: pH of 2.0, extraction time of 62.31 min and liquid-solid ratio of 35.07 g/mL. Under this optimal condition, the pectin yield, degree of esterification, methoxyl content and anhydrouronic acid content were 24.30%, 73.30%, 10.45% and 81.33%, respectively. At optimal conditions, watermelon rind pectin can be classified as high methoxyl and rapid-set pectin with high quality and high-purity. INTRODUCTION Watermelon (Citrullus lanatus) is a Curcubitacea, creeping herbal plant or climbing plant characterized as a large and juicy fruits. Watermelon fruit is composed of flesh (edible part), seeds distributed throughout the flesh and the rind representing 30-40% of the total weight1,2. Health benefits such as prevention against cardiovascular diseases are attributed to the fruit3,4. The literature have determined that the watermelon seeds5 and watermelon rinds6 contain antioxidant properties. Watermelon is mainly used for local production of juices, which generate large amounts of waste without a proper disposal treatment. These watermelon residues have great potential to produce pectin and other value-added products. Pectin is a linear polysaccharide mainly composed of anhydro galacturonic acid units with α(1→4) bonds, and the carboxyl acid groups are esterified with methyl groups partially7. Its structure includes other components such as neutral sugars: arabinose, galactose, rhamnose and xylose7,8, which can react with methanol to form methyl esters or neutralized by a base9. Pectin has gelling properties, that are used for the aqueous gels formation in the food industry10,11. Some of the pharmaceutical applications of pectin are to diminish lipid digestion12, to improve lipid hepatic accumulation13, glucose tolerance for anti-diabetic effects14 and its anti-inflammatory activity in high methoxyl pectins15. The variables that affect the extraction yield are pH, temperature, time and liquid-solid ratio (LSR). The quality parameters of pectin include extraction yield and properties such as ash content16 free acidity17, methoxyl content18, degree of esterification19, anhydro galacturonic acid content20,21 and Fourier transform infrared spectroscopy (FTIR) analysis22,23. Pectin is classified through its degree of esterification (DE) due to define its gelling properties. High methoxyl pectins, characterized by forming gels in high soluble solids and acidic systems, which have galacturonic acid units that are more than 50% (DE > 50%) esterified with methyl groups. In low methoxyl pectins (DE < 50%), the gelation occurs on a widespread pH range than high methoxyl pectins, and require the presence of divalent cations7,8,24. The primary sources of industrial pectin extraction are apple pomace and citrus peels24,25. However, pectin from agro-industrial processes wastes is generating a great interest because their use can reduce environmental impacts provoked by themselves and give added value to the agro-industrial production chain. Many biological residues such as banana peels26, mango peels27,28 and melon peels11 have been utilized to obtain pectin. Hence, watermelon rind can be used to produce pectin, while adding value to the watermelon's agribusiness. Several works have reported the extraction of pectin from watermelon wastes by using different methods such as acid hydrolysis29,30 and assisted microwave extraction31,32. However, the acid hydrolysis is the most employed method for the pectin extraction from food waste. Therefore, the objectives of this study are to optimize and characterize the pectin extraction from watermelon rind using citric acid as solvent for extraction through the response surface methodology (RSM) and determine optimal conditions for the highest pectin yield and anhydrouronic acid content (AUA) simultaneously. RESULTS AND DISCUSSION Experimental results and quality parameters Table 1 presents the experimental values obtained for pectin yield, methoxyl content, degree of esterification, and anhydrouronic acid content at each Box Behnken design (BBD) point. The pectin yield range (4.19-27.86%) was within the typical values reported of 15-20% and 30-35% for dried apple pomace and citric peels, respectively24. The results indicate that the watermelon rind pectin can be classified as high methoxyl pectin25 due to its DE and methoxyl content (MeO) were higher than the reference values of 50% and 6.7%33, respectively. Furthermore, according to the Food Agriculture Organization (FAO), and the World Health Organization (WHO), the AUA suggests that a high-quality pectin (AUA > 65% FAO/WHO) was obtained34, except for runs carried out at pH 3.0. Run Independent variables Experimental responses X1 X2 (min) X3 (g/mL) Yield (%) DE (%) MeO (%) AUA (%) 1 2.0 45 25 17.02 71.13 9.91 79.11 2 3.0 45 25 5.31 67.71 7.45 62.49 3 2.0 75 25 21.23 66.83 8.36 71.02 4 3.0 75 25 5.95 66.67 6.75 57.52 5 2.0 60 10 14.59 71.42 9.39 74.68 6 3.0 60 10 4.19 71.53 6.82 54.11 7 2.0 60 40 27.16 73.33 10.66 82.50 8 3.0 60 40 5.99 75.44 7.21 54.23 9 2.5 45 10 6.77 73.43 9.40 72.69 10 2.5 75 10 8.80 68.22 7.89 65.69 11 2.5 45 40 6.12 68.42 7.71 64.00 12 2.5 75 40 4.47 79.17 10.16 72.83 13 2.5 60 25 11.48 73.47 10.51 81.24 14 2.5 60 25 11.50 74.61 10.44 79.46 15 2.5 60 25 13.00 72.93 10.03 78.08 Table 1. Experimental conditions of pectin extraction with citric acid, and responses for the Box Behnken experimental design. a X1 (pH with three levels: 2.0, 2.5 and 3.0), b X2 (extraction time with three levels: 45, 60 and 75 minutes) and c X3 (LSR with three levels: 10, 25 and 40 g/mL). Statistical analysis The analysis of variance (ANOVA) of the results was used to determine the effects of pH (X 1), extraction time (X2) and liquid-solid ratio (X3) on each of the responses of watermelon rind pectin (Table 2). ANOVA revealed that the linear effect (X1) was significant (p < 0.05) for pectin yield, MeO and AUA. Interactions (X2:X3) had significant for DE and MeO. Furthermore, square effects X1 and X2; X1 and X3 and X2 were also significant for MeO, AUA and DE, respectively. Pectin yield (%) DE (%) Source Df Sum sq Mean sq F value Pr (>F) Df Sum sq Mean sq F value Pr (>F) Model 9 584.37 64.93 6.08 0.0306* 9 156.22 17.36 4.24 0.0631 Linear 3 443.10 147.70 13.82 0.0074** 3 17.52 5.84 1.43 0.3911 Interaction 3 35.57 11.86 1.11 0.4274 3 67.34 22.45 5.48 0.0488* Square 3 105.70 35.23 3.30 0.1159 3 71.36 23.79 5.81 0.0438* Residuals 5 53.44 10.69 5 20.48 4.10 Lack of fit 3 51.92 17.31 22.77 0.0424 3 19.00 6.33 8.61 0.1058 Pure error 2 1.52 0.76 2 1.47 0.74 Total 28 1275.62 28 353.39 MeO (%) AUA (%) Df Sum sq Mean sq F value Pr (>F) Df Sum sq Mean sq F value Pr (>F) Model 9 27.00 3.00 7.40 0.0201* 9 1238.65 137.63 7.84 0.0177* Linear 3 13.57 4.52 11.15 0.0118* 3 800.20 266.73 15.19 0.0060** Interaction 3 4.29 1.43 3.53 0.1041 3 79.90 26.63 1.52 0.3184 Square 3 9.14 3.05 7.51 0.0267* 3 358.55 119.52 6.81 0.0323* Residuals 5 2.03 0.41 5 87.80 17.56 Lack of fit 3 1.89 0.63 9.39 0.0978 3 82.78 27.59 10.99 0.0845 Pure error 2 0.13 0.07 2 5.02 2.51 Total 28 58.06 28 2652.90 (Significant codes: 0 ‘***’, 0.01 ‘**’, 0.05 ‘*’) Table 2. Results of ANOVA for pectin yield, DE, MeO and AUA. By using multiple regression analysis on the experimental responses, the second-order model was utilized to fit experimental data and predict the effects of dependent variables. The validity of the model could be confirmed because the lack-of-fit tests were not significant, and the following determination coefficients (R) 0.9162, 0.8841, 0.9338 and 0.9301 were obtained for pectin yield, DE, MeO and AUA, respectively. These results show that the calculated responses by the model were reliable and adequate. The second-order equations (coded factors) for pectin yield (Eq. 1), DE (Eq. 2), MeO (Eq. 3) and AUA (Eq. 4) were given as follows Pectin yield (%)=11.99−7.32 X1+0.65 X2+1.17 X3−0.89 X1 X2 −2.69 X 1 X3−0.92 X2 X3+3.41 X1 −3.03 X2 −2.42 X 3 2 (1) DE (%)=73.67−0.17 X 1+0.025 X 2+1.47 X3+0.82 X1 X2 +0.50 X1 X3+3.99 X2 X3−2.48 X1 −3.10 X2 +1.74 X3 2 (2) MeO(%)=10.33−1.26 X 1−0.16 X2+0.28 X 3+0.21X 1 X 2 −0.22 X1 X3+0.99 X2 X3−1.24 X1 −0.97 X 2 −0.57 X 3 2 (3) AUA(%)=79.59−9.87 X 1−1.40 X2+0.80 X3+0.78 X 1 X 2 −1.93 X1 X3+3.96 X2 X3−7.24 X1 −4.81 X2 −5.97 X3 2 (4) Effect of independent variables on pectin yield ANOVA showed that pH (linear effect) was the only variable that affected pectin yield significantly. The determination coefficient indicates that 91.62% of the total variation can be explained by the quadratic model (Eq. 1), which suggest that the model is suitable to predict the yield pectin from watermelon rind under the experimental conditions evaluated in this work. Fig. 1a shows that the pectin yield was favored at lower pH (negative regression term) and, the higher pectin yield was obtained at the lowest pH of 2.0. The carboxyl groups present in pectin are hydrated due to the acidified extraction solvent at lower pH; the loss of charges in carboxyl groups tends to reduce the repulsive forces, promoting the pectin precipitation. Indeed, more pectin dissolution by hydrolysis of non-soluble pectin and increasing of pectin mass transfer from the plant source at lowering pH have shown by the literature. These conditions were shown to have led to increased the pectin release and recovery. These results were similar to other reported pectins, which showed a significant rise of pectin yield at lower pH for extracted pectin from mango peels and watermelon rind. Fig. 1a illustrates that at high points of extraction time (X2) and low values of LSR (X3) the pectin yield decreased; this behavior is in agreement with results reported in the pectin extraction from seed watermelon peel and pectin extraction from carrot pomace. This is likely because of extended extraction time; the citric acid causes the breakdown of glycoside bonds and ester bonds of pectin. On the other hand, excessive or insufficient amounts of LSR influenced the pectin mass transfer due to the solution saturation or low dissolution capacity. Therefore, the pectin yields were smaller at extreme LSR conditions. Effect of independent variables on DE As can be seen, in Table 2 and Fig. 1b, the extracted pectin is considered like a high methoxyl pectin with DE ranged from 66.67% to 75.44%. According to ANOVA, linear effects were not significant, and the interaction between time extraction and LSR (X2:X3) and the quadratic effect of the time extraction (X2) affected the DE (Table 2). High methoxyl pectin (DE > 50%) has been obtained from pomelo peel, pomelo albedo and watermelon rind. Indicating that our findings are in agreement with these previous works. It was noticed that the DE increased for more extensive extraction time. These results are in accordance with the reported values given in pectin from cocoa husks. Under the conditions tested in this study, the DE showed that the obtained watermelon rind pectin was characterized with a low degree of de-esterification of polygalacturonic chains. Effect of independent variables on MeO The methoxyl content listed in Table 2, varied from 6.75 to 10.66%. Based on the Regression analysis (Eq. 3), it can be suggested that pH (linear and quadratic coefficients) negatively influenced the MeO (Fig.1c). In the other word, decreasing pH led to get higher the methoxyl content. Other significant factors were interaction coefficients terms (X2:X3, Fig. 1c) and quadratic coefficient of the time extraction (X2). In addition, the R squared pointed out that 6.62% of data variability could not be explained by the Eq. (3). The acidic extraction (citric acid) was favorable towards the methoxylation of side acid groups of polygalacturonic chains. This is probably due to the methoxylation was catalyzed by acidity medium (low pH), which increases the reactivity of the acid groups by shifting the equilibrium forward to methyl ester formation. The methoxyl content of citrus peel pectin and apple pectin40 were comparable with obtained results. Effect of independent variables on AUA Similar to the pectin yield, pH was the most significant variable for the anhydrouronic acid content. Unlike the yield, the quadratic coefficient term of LSR (X3) was also significant. In Fig. 1d observes that AUA (purity pectin) increased at lower levels of pH and low-intermediate levels of LSR. However, the highest value of AUA (82.50%) was obtained at the lowest pH (2.0), greatest LSR (40 g/mL) and time extraction of 60 minutes. Several works from different plant sources like orange peels37, ponkan peels20 and banana peels26 have shown that low pH and LSR between 20 and 25 g/mL increased the AUA16,20,37. By contrast, it has been previously shown that longer extraction time improved pectin purity16,21,26, which did not agree with the results in the present study. Figure 1. Response surface showing the effect of dependent variables (pH, time extraction and LSR) on the pectin yield (a), DE (b), MeO (c) and AUA (d). Pectin optimization Based on the polynomial models fitted for each of the responses, a separate optimization process was carried out to find out the optimal conditions that maximized pectin yield, DE, MeO and AUA independently. Different optimal conditions were obtained (Table 3). These optimal predictions confirmed the strong influence of pH (low values) on the pectin yield, MeO content and AUA content as well as previously shown. It is noteworthy that for all responses except DE, the optimal conditions were at low pH, intermediate values of time extraction (nearly 60 min) and midpoints to high levels of LSR for the study regions tested in this work. In order to unify to one optimal condition (theoretical optimum), an analysis of the multiple responses through the desirability discontinuous functions41 using R-programming software42 was performed. The conditions obtained for the highest pectin yield and highest AUA simultaneously were found at pH of 2.0, extraction time of 62.31 min and LSR of 35.07 g/mL (Table 3). Theoretical optimum was quite similar to the optimal conditions predicted for pectin yield model in Eq. (1) as is shown in Table 3. In fact, the theoretical optimum pectin yield (24.30%) can be considered equivalent to highest predicted by the Eq. (1), which is slightly lower than the experimental value, 27.16%. In addition, regarding the other responses, the theoretical predictions are very good, similar to the experimental values (Table 1). According to these findings, the models were well fitted to predict the effect of independent variables and their responses. Dependent variable X1 X2 (min) X3 (g/mL) Optimal value† Theoretical predicted‡ Pectin yield (%) 2.00 62.07 36.57 24.32 24.30 DE (%) 2.59 70.05 40.00 78.26 73.00 MeO (%) 2.23 60.97 31.12 10.72 10.45 AUA (%) 2.15 57.77 26.97 83.24 81.33 Table 3. Optimized responses and theoretical optimum.† Optimum calculated by each model separately ‡ Responses at pH = 2.0, extraction time = 62.31 min and LSR = 35.07 g/mL. The optimal pectin yield (24.30%) was comparable with other studies that also used the desirability function to optimize simultaneous responses. The pectin extraction from melon peels11, mango peels28 and ponkan peels20 reported maximum pectin yields of 29.48%, 30.0% and 25.6%, respectively. Previous works from watermelon rind have reported pectin yields of approximately 19%31 and 25%30,32. It has shown that citric acid was a better solvent extractant than hydrochloric acid in the pectin extraction from watermelon rind29. Nevertheless, Their maximum pectin yield (8.38%) obtained was very low compared to the reported values by this study and previous works, despite their optimal conditions (pH of 2.0, extraction time of 180 min, LSR of 25 and temperature of 80 °C) were pretty similar to ours. This might be due to the Colombian variety of watermelon and using methanol as a precipitating agent, which has shown to be a better precipitating agent than ethanol in the extraction of pectin from pomelo albedo17. For this reason, the highest properties of pectin were obtained from watermelon rind. Characterization of pectin The characteristics of extracted pectin from watermelon rind using citric acid under the highest yield experimental conditions (pH of 2.0, time of 60 min and LSR of 40 g/mL) were comparable with commercial pectin (Merck). Table 4 shows the remarkable properties of the pectin obtained. The ash content and alkalinity of ash were 1.24% and 0.39%, respectively, which were lower than the accepted levels for the standard pectin. The ash content is a measure of the degree of purity40 and quality43 of pectin. Based on these results, the extracted pectin presented low amounts of soluble solids and high purity. Free acid registered a high value compared to standard pectin; this might be due to the chemical nature of carboxyl groups, which in harsh pH conditions are hydrolyzed, increasing pectin acidity. High methoxyl and rapid-set pectin was obtained from watermelon rind because of the MeO (10.66%), and DE (73.33%) values are ranged for this pectin type (Table 4). Rapid-Set pectin is used in high sugar jams, jellies, and marmalades7,8, indicating that pectin from watermelon rind could be employed in food applications. High methoxyl pectin from watermelon rind has been previously reported29–32, and these findings are in agreement with obtained results. From the results of AUA, the extracted pectin had a value of 82.5%, which was higher than those reported by similar works with watermelon rind29–31 and as well as the percent requirements for commercial food or pharmaceutical purpose33,34. It should be pointed out that the AUA is a determining quality parameter and gelling properties of pectin19,33. Therefore, the results related to AUA also indicated that high-quality pectin was extracted from watermelon rind by using citric acid. Properties Standard pectin Extracted pectin Ash (%) 3.77 ± 3.39 1.24 Alkalinity of ash (Calcium carbonate ,%) 2.34 ± 2.90 0.39 Free acid (meq/g) 0.78 ± 0.46 1.25 MeO (%) ≥ 6.70 (USP) 10.66 DE (%) ≥ 50 high methoxyl pectin 71-74 rapid set 73.33 AUA (%) ≥ 74 (USP) ≥ 65 (FAO/WHO) 82.50 Table 4. Chemical properties pectin obtained from watermelon rind compared with standard pectin. FTIR analysis The information on the functional groups presents in the extracted pectin from watermelon rind was evaluated by FTIR analysis at highest yield experimental conditions (pH of 2.0, time extraction of 60 min and LSR of 40 g/mL). The FTIR spectra of watermelon rind pectin is illustrated in Figure 2. The peaks related to O-H and C-H stretching vibrations were between 3500-3250 cm and 3000-2700 cm, respectively. Stretching vibrations (=CO) of esterified carboxyl groups and free carboxyl groups (18001600 cm) were observed in the fingerprint regions of the pectin spectrum10,44–46. The peak of CH bending vibrations for pyranose ring (approximately 1338 cm) and the peak of COO stretching vibration for ester groups (1330-1210 cm) can be assigned in Fig. 244. Furthermore, the characteristic overlapped peaks of glycoside bonds and pyranose cycles around 1000 cm -1 and weak peaks (830-500 cm) associated with αand β-configurations were identified, indicating that the pectin is the principal component44,47. The bands of FTIR spectra of the obtained pectin were similar to reported with analogous work of pectin extraction from watermelon rind29. Figure 2. FTIR spectra of extracted pectin from watermelon rind. CONCLUSIONS In this study, the pectin extracted from watermelon rind with citric acid is considered as a high methoxyl pectin (DE > 50% and MeO > 6.7%) and has high quality (AUA > 65%, FAO/WHO, except for runs at pH of 3.0). Based on the results of BBD, pH was the most significant variable on the pectin yield and its properties. The simultaneous optimization to obtain the highest pectin yield and highest AUA showed that the optimal conditions were found at pH of 2.0 (lowest level), extraction time of 62.31 min (intermediate points) and 35.07 g/mL (highmidpoints), which were close to optimal experimental conditions (pH of 2.0, time of 60 min and LSR of 40 g/mL) for the highest yield pectin. Under this optimum the yield pectin, DE, MeO and AUA were 24.32%, 73.33% (rapid-set pectin), 10.66 and 82.50% (AUA > 74% USP, high quality), respectively. In addition, ash content and alkalinity of ash revealed high purity of pectin at the optimal condition. MATERIALS AND METHODS Samples preparation Watermelon rinds were obtained from local market of watermelon juices located in the Barranquilla center and around the Universidad del Atlántico, Colombia. They were washed, manually cut, and heated in distilled water until boiling to denature enzymes and inactive microorganisms.48 The collected material was milled for homogenization and was dried in a convection oven at a temperature no higher than 80 °C for 1440 minutes. Finally, the treated watermelon rind was packed and stored in a desiccator for later use. Chemicals and solvents All chemicals and solvents such as citric acid, ethanol and methanol used were of analytical grade. Pectin extraction The pectin extraction was carried out through acid hydrolysis with citric acid, according to the methodology with some modifications.49 The dried watermelon rind was stirred in citric acid solutions with the following defined conditions for all runs of the Box-Behnken design: pH (2.0, 2.5 and 3.0); extraction time (45, 60, 90 min) and liquid-solid ratio (1:10, 1:25 and 1:40 g/mL) at a constant temperature of 80 °C. The resultant slurry was vacuum filtered with a micro cloth using vacuum pressure. The residual liquid was precipitated with methanol in a 60% volume solution. The obtained precipitate was washed three times with 70% ethanol and subsequently was rewashed with 96% ethanol. The collected precipitate was dried in a convection oven until constant weight at 50 °C for 12 hours. Pectin yield The pectin yield of watermelon rind on dry weight basis was determined as follow (Eq. 5)10,43 Pectin yield (%)= weight of dried pectin(g) weight of driedwatermelonrind (g) (5) Where the dried pectin was obtained after the treatment of filtration, precipitation and drying; the dried watermelon rind was collected as result of sample preparation. Pectin characterization Determination of degree of esterification and anhydrouronic acid content The degree of esterification and the anhydrouronic acid content were determined by titration relating the methoxyl content with the equivalent weight. 0.50g of watermelon rind pectin was dissolved into ethanol/water solution (1:20 v/v); 5 drops of phenol red indicator were added, and the sample was titrated with 0.1N sodium hydroxide (V1, mL) until the indicator changed. Then, 25 ml of 0.25N NaOH was added, and the sample was heated and stirred vigorously. Five drops of phenol red and 25 mL of 0.25N HCl were added again and, it was titrated with 0.1N NaOH (V2, mL) until the color change from yellow to faint pink endpoint. The MeO and the DE were then calculated according to Eq. (6) and Eq. (7), respectively. MeO(% )= V 2(mL)0.1(N )31 weight of sample (mg) x 100 (6) DE (%)= V 2(mL) V 1(mL)+V 2(mL) x100 (7) The Eq. (8) was used to calculate the anhydrouronic acid content18 AUA (%)= 0.1(N )(V 1(mL)+V 2(mL ))176 Weight of sample(mg) x100 (8) Where176 is the molecular weight of anhydrouronic acid expressed as mg/meq; V1 and V2 were the volumes used for first and second titration, respectively. Fourier transform infrared spectroscopy Fourier transform infrared (FTIR) spectra of the extracted pectin was used to evaluate its structural chemical properties. FTIR analysis was performed in the SHIMADZU (IRAffinity-1) spectrometer with a resolution of 4cm-1 and 130 scans of wavelengths ranging from 7000 to 400 cm-1. Experimental design The Box-Behnken design response surface design was used to evaluate the effect of three independent variables such as pH (X1: 2.0-3.0), the extraction time (X2: 45-75 min) and the liquid-solid ratio, LSR (X3: 10-40 g/ml) on simultaneous responses (pectin yield, the MeO, the DE and AUA). Table 5 summarizes the levels and code of three independent variables. Dependent variables Real values of coded levels -1 0 1 X1: pH 2.0 2.5 3.0 X2: Time (min) 45 60 75 X3: Liquid-solid ratio, LSR (g/mL) 1:10 1:25 1:40 Table 5. Levels and code of variables chosen for the Box-Behnken design. The experimental design and variance analysis (ANOVA) were performed using R-programming software 4.0.2. The responses variables were fitted to the second-order polynomial model as given for the following Eq. (9). Y=B0+∑ i=1 3 Bi X i+∑ i=1 3 Bi iX i +∑ i=1 2 ∑ j=i+1 3 Bi j X i X j (9) Where: Y is the response variable; B0, Bi, Bii and Bij are the coefficients terms for the regression model; Xi and Xj are the levels of independent variables. The responses were initially optimized separately to predict the 3D surface and the maximum values. Latterly, an optimization of simultaneous responses using the desirability function41 was performed. References 1. Durán, R., Villegas, M. E. & Nieves, I. Caracterización y extracción de citrulina de la corteza de la sandía (Citrullus lanatus “thunb”) consumida en Valledupar. Temas Agrar. 22, 61 (2017). 2. Rico, X., Gullón, B., Alonso, J. L. & Yáñez, R. Recovery of high value-added compounds from pineapple, melon, watermelon and pumpkin processing by-products: An overview. Food Res. Int. 132, 109086 (2020). 3. Costa-Rodrigues, J., Pinho, O. & Monteiro, P. R. R. Can lycopene be considered an effective protection against cardiovascular disease? Food Chem. 245, 1148–1153 (2018). 4. Shanely, R. A. et al. Daily watermelon consumption decreases plasma sVCAM-1 levels in overweight and obese postmenopausal women. Nutr. Res. 76, 9–19 (2020). 5. Wen, C. et al. Purification and identification of novel antioxidant peptides from watermelon seed protein hydrolysates and their cytoprotective effects on H2O2-induced oxidative stress. Food Chem. 327, 127059 (2020). 6. Al-Sayed, H. M. A. & Ahmed, A. R. Utilization of watermelon rinds and sharlyn melon peels as a natural source of dietary fiber and antioxidants in cake. Ann. Agric. Sci. 58, 83–95 (2013). 7. Brejnholt, S. M. 13 Pectin. in Food Stabilisers, Thickeners and Gelling Agents (2009). 8. Endreß, H.-U. & Christensen, S. H. 12 Pectins BT Handbook of Hydrocolloids (Second edition). in Woodhead Publishing Series in Food Science, Technology and Nutrition (2009). doi:http://dx.doi.org/10.1533/9781845695873.274. 9. Zegada, V. Y. Pectin Extraction From Orange Peels Waste By Microwave Assisted Acid. Cent. Investig. Procesos Ind. 1, 65–76 (2015). 10. Muthukumaran, C. et al. Pectin from muskmelon (Cucumis melo var. reticulatus) peels: extraction optimization and physicochemical properties. 3 Biotech 7, (2017). 11. Raji, Z., Khodaiyan, F., Rezaei, K., Kiani, H. & Hosseini, S. S. Extraction optimization and physicochemical properties of pectin from melon peel. Int. J. Biol. Macromol. 98, 709–716 (2017). 12. Espinal-Ruiz, M., Restrepo-Sánchez, L. P., Narváez-Cuenca, C. E. & McClements, D. J. Impact of pectin properties on lipid digestion under simulated gastrointestinal conditions: Comparison of citrus and banana passion fruit (Passiflora tripartita var. mollissima) pectins. Food Hydrocoll. 52, 329–342 (2016). 13. Zhu, R. G. et al. Comparative effects of hawthorn (Crataegus pinnatifida Bunge) pectin and pectin hydrolyzates on the cholesterol homeostasis of hamsters fed high-cholesterol diets. Chem. Biol. Interact. 238, 42–47 (2015). 14. Liu, Y., Dong, M., Yang, Z. & Pan, S. Anti-diabetic effect of citrus pectin in diabetic rats and potential mechanism via PI3K/Akt signaling pathway. Int. J. Biol. Macromol. 89, 484–488 (2016). 15. Tummalapalli, M. et al. Drug loaded composite oxidized pectin and gelatin networks for accelerated wound healing. Int. J. Pharm. 505, 234–245 (2016). 16. Chan, S. Y. & Choo, W. S. Effect of extraction conditions on the yield and chemical properties of pectin from cocoa husks. Food Chem. 141, 3752–3758 (2013). 17. de la Hoz, S., Vazquez, B., Perez, J., Albis, A. & Quiñones, L. Effect of precipitating solvents on the extraction of pectin from the pomelo albedo by acid hydrolysis. Contemp. Eng. Sci. 11, 3849–3855 (2018). 18. Ismail, N. S. M., Ramli, N., Hani, N. M. & Meon, Z. Extraction and characterization of pectin from dragon fruit (Hylocereus polyrhizus) using various extraction conditions. Sains Malaysiana 41, 41–45 (2012). 19. Jafari, F., Khodaiyan, F., Kiani, H. & Hosseini, S. S. Pectin from carrot pomace: Optimization of extraction and physicochemical properties. Carbohydr. Polym. 157, 1315–1322 (2017). 20. Colodel, C., Vriesmann, L. C., Teófilo, R. F. & de Oliveira Petkowicz, C. L. Extraction of pectin from ponkan (Citrus reticulata Blanco cv. Ponkan) peel: Optimization and structural characterization. Int. J. Biol. Macromol. 117, 385–391 (2018). 21. Peng, X. et al. Box–Behnken design based statistical modeling for the extraction and physicochemical properties of pectin from sunflower heads and the comparison with commercial low-methoxyl pectin. Sci. Rep. 10, 1–10 (2020). 22. Kalapathy, U. & Proctor, A. Effect of acid extraction and alcohol precipitation conditions on the yield and purity of soy hull pectin. Food Chem. 73, 393–396 (2001). 23. Wang, X., Chen, Q. & Lü, X. Pectin extracted from apple pomace and citrus peel by subcritical water. Food Hydrocoll. 38, 129–137 (2014). 24. Lopes da Silva, J. A. & Rao, M. A. Pectins: Structure, Functionality, and Uses. in Food Polysaccharides and Their Applications (2006). doi:doi:10.1201/9781420015164.ch6\ r10.1201/9781420015164.ch6. 25. Pilgrim, G. W., Walter, R. H. & Oakenfull, D. G. Jams, Jellies, and Preserves. in The Chemistry and Technology of Pectin (1991). doi:10.1016/b978-0-08-092644-5.50007-8. 26. Oliveira, T. Í. S. et al. Optimization of pectin extraction from banana peels with citric acid by using response surface methodology. Food Chem. 198, 113–118 (2016). 27. Barreto, G. E., Púa, A. L., De Alba, D. D. & Pión, M. M. Extracción y caracterización de pectina de mango de azúcar (Mangifera indica L.). Temas Agrar. 22, 78 (2017). 28. do Nascimento Oliveira, A. et al. Optimization of pectin extraction from Ubá mango peel through surface response methodology. Int. J. Biol. Macromol. 113, 395–402 (2018). 29. Lee, K. Y. & Choo, W. S. Extraction Optimization and Physicochemical Properties of Pectin from Watermelon (Citrullus lanatus) Rind: Comparison of Hydrochloric and Citric acid Extraction. J. Nutraceuticals Food Sci. 5, 1–8 (2020). 30. Petkowicz, C. L. O., Vriesmann, L. C. & Williams, P. A. Pectins from food waste: Extraction, characterization and properties of watermelon rind pectin. Food Hydrocoll. 65, 57–67 (2017). 31. Prakash Maran, J., Sivakumar, V., Thirugnanasambandham, K. & Sridhar, R. Microwave assisted extraction of pectin from waste Citrullus lanatus fruit rinds. Carbohydr. Polym. 101, 786–791 (2014). 32. Sari, A. M., Ishartani, D. & Dewanty, P. S. Effects of microwave power and irradiation time on pectin extraction from watermelon rinds (Citrullus lanatus) with acetic acid using microwave assisted extraction method. IOP Conf. Ser. Earth Environ. Sci. 102, (2018). 33. USP. USP Monographs: Pectin. Pharmacopeial Forum 783 (2007). 34. JECFA. Compendium of Food Additive Specifications. Joint FAO/WHO Expert Committee of Food Additives. FAO JECFA Monofraphs (2016). 35. Garna, H. et al. Effect of extraction conditions on the yield and purity of apple pomace pectin precipitated but not washed by alcohol. J. Food Sci. 72, C001–C009 (2007). 36. Methacanon, P., Krongsin, J. & Gamonpilas, C. Pomelo (Citrus maxima) pectin: Effects of extraction parameters andits properties. Food Hydrocoll. 35, 383–391 (2014). 37. Su, D. L. et al. Efficient extraction and characterization of pectin from orange peel by a combined surfactant and microwave assisted process. Food Chem. 286, 1–7 (2019). 38. Jiang, L., Shang, J., He, L. & Dan, J. Comparisons of microwave-assisted and conventional heating extraction of pectin from seed watermelon peel. Adv. Mater. Res. 550–553, 1801–1806 (2012). 39. Zhang, X. U. E. Z., Zhang, X., Wang, Z. L. I. U. J. & Long, Y. C. D. Optimization of Pectin Extraction from Citrus Peel by Response Surface Methodology. Food Sci. 32, 128–132 (2011). 40. Khamsucharit, P., Laohaphatanalert, K., Gavinlertvatana, P., Sriroth, K. & Sangseethong, K. Characterization of pectin extracted from banana peels of different varieties. Food Sci. Biotechnol. 27, 623–629 (2018). 41. Derringer, G. & Suich, R. Simultaneous Optimization of Several Response Variables. J. Qual. Technol. 12, 214–219 (1980). 42. Kuhn, M. The desirability Package. Proc. 2012 Winter Simul. Conf. 1–17 (2012). 43. Mohamed, H. Extraction and Characterization of Pectin from Grapefruit Peels. MOJ Food Process. Technol. 2, 31–38 (2016). 44. Filippov. Practical infrared spectroscopy of pectic substances. Top. Catal. 6, 115–142 (1992). 45. Kumar, A. & Chauhan, G. S. Extraction and characterization of pectin from apple pomace and its evaluation as lipase (steapsin) inhibitor. Carbohydr. Polym. 82, 454–459 (2010). 46. Kamnev, A. A., Ptichkina, N. M., Ristic, M., Shkodina, O. G. & Ignatov, V. V. Comparative Spectroscopic Characterization of Pectins. Spectrosc. Biol. Mol. 445–446 (1995) doi:10.1007/978-94-011-0371-8_204. 47. Čopíková, J., Černá, M., Novotná, M., Kaasová, J. & Synytsya, A. Application of FT-IR spectroscopy in detection of food hydrocolloids&nbsp; in confectionery jellies and food supplements. Czech J. Food Sci. 19, 51–56 (2018). 48. Rojas, J. P. L., Perea, A. V. & Stashenko, E. E. Obtención de aceites esenciales y pectinas a partir de subproductos de jugos cítricos. Vitae 16, 110–115 (2009). 49. Wai, W. W., Alkarkhi, A. F. M. & Easa, A. M. Effect of extraction conditions on yield and degree of esterification of durian rind pectin: An experimental design. Food Bioprod. Process. 88, 209– 214 (2010). 50. Owens, H. S. et al. Methods Used at Western Regional Research Laboratory for Extraction and Analysis of Pectic Materials. Bureau of Agricultural and Industrial Chemistry (1952). doi:10.1017/CBO9781107415324.004. 51. Norziah, M. H., Fang, E. O. & Karim, A. A. Extraction and Characterisation of Pectin From Pomelo Fruit Peels. Gums Stabilisers Food Ind. 10 27–36 (2000) doi:10.1533/9781845698355.1.27. 52. Pasandide, B., Khodaiyan, F., Mousavi, Z. E. & Hosseini, S. S. Optimization of aqueous pectin extraction from Citrus medica peel. Carbohydr. Polym. 178, 27–33 (2017). Author contributions J. P. designed and supervised the work and wrote-edited the manuscript. K. G. and L. V. carried out the experiments and chemical analysis. K. G. edited the main manuscript. All authors did the statistical and data analysis. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to J. P.


INTRODUCTION
Watermelon (Citrullus lanatus) is a Curcubitacea, creeping herbal plant or climbing plant characterized as a large and juicy fruits. Watermelon fruit is composed of flesh (edible part), seeds distributed throughout the flesh and the rind representing 30-40% of the total weight 1,2 . Health benefits such as prevention against cardiovascular diseases are attributed to the fruit 3,4 . The literature have determined that the watermelon seeds 5 and watermelon rinds 6 contain antioxidant properties. Watermelon is mainly used for local production of juices, which generate large amounts of waste without a proper disposal treatment. These watermelon residues have great potential to produce pectin and other value-added products.
Pectin is a linear polysaccharide mainly composed of anhydro galacturonic acid units with α(1→4) bonds, and the carboxyl acid groups are esterified with methyl groups partially 7 . Its structure includes other components such as neutral sugars: arabinose, galactose, rhamnose and xylose 7,8 , which can react with methanol to form methyl esters or neutralized by a base 9 . Pectin has gelling properties, that are used for the aqueous gels formation in the food industry 10,11 . Some of the pharmaceutical applications of pectin are to diminish lipid digestion 12 , to improve lipid hepatic accumulation 13 , glucose tolerance for anti-diabetic effects 14 and its anti-inflammatory activity in high methoxyl pectins 15 .
The variables that affect the extraction yield are pH, temperature, time and liquid-solid ratio (LSR). The quality parameters of pectin include extraction yield and properties such as ash content 16 free acidity 17 , methoxyl content 18 , degree of esterification 19 , anhydro galacturonic acid content 20,21 and Fourier transform infrared spectroscopy (FTIR) analysis 22,23 . Pectin is classified through its degree of esterification (DE) due to define its gelling properties. High methoxyl pectins, characterized by forming gels in high soluble solids and acidic systems, which have galacturonic acid units that are more than 50% (DE > 50%) esterified with methyl groups. In low methoxyl pectins (DE < 50%), the gelation occurs on a widespread pH range than high methoxyl pectins, and require the presence of divalent cations 7,8,24 .
The primary sources of industrial pectin extraction are apple pomace and citrus peels 24,25 . However, pectin from agro-industrial processes wastes is generating a great interest because their use can reduce environmental impacts provoked by themselves and give added value to the agro-industrial production chain. Many biological residues such as banana peels 26 , mango peels 27,28 and melon peels 11 have been utilized to obtain pectin. Hence, watermelon rind can be used to produce pectin, while adding value to the watermelon's agribusiness.
Several works have reported the extraction of pectin from watermelon wastes by using different methods such as acid hydrolysis 29,30 and assisted microwave extraction 31,32 . However, the acid hydrolysis is the most employed method for the pectin extraction from food waste. Therefore, the objectives of this study are to optimize and characterize the pectin extraction from watermelon rind using citric acid as solvent for extraction through the response surface methodology (RSM) and determine optimal conditions for the highest pectin yield and anhydrouronic acid content (AUA) simultaneously. Table 1 presents the experimental values obtained for pectin yield, methoxyl content, degree of esterification, and anhydrouronic acid content at each Box Behnken design (BBD) point. The pectin yield range (4.19-27.86%) was within the typical values reported of 15-20% and 30-35% for dried apple pomace and citric peels, respectively 24 . The results indicate that the watermelon rind pectin can be classified as high methoxyl pectin 25 due to its DE and methoxyl content (MeO) were higher than the reference values of 50% and 6.7% 33 , respectively. Furthermore, according to the Food Agriculture Organization (FAO), and the World Health Organization (WHO), the AUA suggests that a high-quality pectin (AUA > 65% FAO/WHO) was obtained 34 Table 1. Experimental conditions of pectin extraction with citric acid, and responses for the Box Behnken experimental design. a X1 (pH with three levels: 2.0, 2.5 and 3.0), b X2 (extraction time with three levels: 45, 60 and 75 minutes) and c X3 (LSR with three levels: 10, 25 and 40 g/mL).

Statistical analysis
The analysis of variance (ANOVA) of the results was used to determine the effects of pH (X 1 ), extraction time (X 2 ) and liquid-solid ratio (X 3 ) on each of the responses of watermelon rind pectin ( Table 2). ANOVA revealed that the linear effect (X 1 ) was significant (p < 0.05) for pectin yield, MeO and AUA. Interactions (X 2 :X 3 ) had significant for DE and MeO. Furthermore, square effects X 1 2 and X 2 2 ; X 1 2 and X 3 2 and X 2 2 were also significant for MeO, AUA and DE, respectively. By using multiple regression analysis on the experimental responses, the second-order model was utilized to fit experimental data and predict the effects of dependent variables. The validity of the model could be confirmed because the lack-of-fit tests were not significant, and the following determination coefficients (R 2 ) 0.9162, 0.8841, 0.9338 and 0.9301 were obtained for pectin yield, DE, MeO and AUA, respectively. These results show that the calculated responses by the model were reliable and adequate. The second-order equations (coded factors) for pectin yield (Eq. 1), DE (Eq. 2), MeO (Eq. 3) and AUA (Eq. 4) were given as follows Effect of independent variables on pectin yield ANOVA showed that pH (linear effect) was the only variable that affected pectin yield significantly 35 .
The determination coefficient indicates that 91.62% of the total variation can be explained by the quadratic model (Eq. 1), which suggest that the model is suitable to predict the yield pectin from watermelon rind under the experimental conditions evaluated in this work. Fig. 1a shows that the pectin yield was favored at lower pH (negative regression term) and, the higher pectin yield was obtained at the lowest pH of 2.0. The carboxyl groups present in pectin are hydrated due to the acidified extraction solvent at lower pH; the loss of charges in carboxyl groups tends to reduce the repulsive forces, promoting the pectin precipitation 10 . Indeed, more pectin dissolution by hydrolysis of non-soluble pectin and increasing of pectin mass transfer from the plant source at lowering pH have shown by the literature 19,20,[35][36][37] . These conditions were shown to have led to increased the pectin release and recovery.
These results were similar to other reported pectins, which showed a significant rise of pectin yield at lower pH for extracted pectin from mango peels 27 and watermelon rind 31 . Fig. 1a illustrates that at high points of extraction time (X 2 ) and low values of LSR (X 3 ) the pectin yield decreased; this behavior is in agreement with results reported in the pectin extraction from seed watermelon peel 38 and pectin extraction from carrot pomace 19 . This is likely because of extended extraction time; the citric acid causes the breakdown of glycoside bonds and ester bonds of pectin 39 . On the other hand, excessive or insufficient amounts of LSR influenced the pectin mass transfer due to the solution saturation or low dissolution capacity. Therefore, the pectin yields were smaller at extreme LSR conditions 38 .

Effect of independent variables on DE
As can be seen, in Table 2 and Fig. 1b, the extracted pectin is considered like a high methoxyl pectin with DE ranged from 66.67% to 75.44%. According to ANOVA, linear effects were not significant, and the interaction between time extraction and LSR (X 2 :X 3 ) and the quadratic effect of the time extraction (X 2 2 ) affected the DE (Table 2). High methoxyl pectin (DE > 50%) has been obtained from pomelo peel 36 , pomelo albedo 17 and watermelon rind 29 . Indicating that our findings are in agreement with these previous works. It was noticed that the DE increased for more extensive extraction time. These results are in accordance with the reported values given in pectin from cocoa husks 16 . Under the conditions tested in this study, the DE showed that the obtained watermelon rind pectin was characterized with a low degree of de-esterification of polygalacturonic chains.

Effect of independent variables on MeO
The methoxyl content listed in Table 2, varied from 6.75 to 10.66%. Based on the Regression analysis (Eq. 3), it can be suggested that pH (linear and quadratic coefficients) negatively influenced the MeO (Fig.1c). In the other word, decreasing pH led to get higher the methoxyl content. Other significant factors were interaction coefficients terms (X 2 :X 3 , Fig. 1c) and quadratic coefficient of the time extraction (X 2 2 ). In addition, the R squared pointed out that 6.62% of data variability could not be explained by the Eq. (3). The acidic extraction (citric acid) was favorable towards the methoxylation of side acid groups of polygalacturonic chains. This is probably due to the methoxylation was catalyzed by acidity medium (low pH), which increases the reactivity of the acid groups by shifting the equilibrium forward to methyl ester formation. The methoxyl content of citrus peel pectin and apple pectin 40 were comparable with obtained results.
Effect of independent variables on AUA Similar to the pectin yield, pH was the most significant variable for the anhydrouronic acid content. Unlike the yield, the quadratic coefficient term of LSR (X 3 2 ) was also significant. In Fig. 1d observes that AUA (purity pectin) increased at lower levels of pH and low-intermediate levels of LSR. However, the highest value of AUA (82.50%) was obtained at the lowest pH (2.0), greatest LSR (40 g/mL) and time extraction of 60 minutes. Several works from different plant sources like orange peels 37 , ponkan peels 20 and banana peels 26 have shown that low pH and LSR between 20 and 25 g/mL increased the AUA 16,20,37 . By contrast, it has been previously shown that longer extraction time improved pectin purity 16,21,26 , which did not agree with the results in the present study. Pectin optimization Based on the polynomial models fitted for each of the responses, a separate optimization process was carried out to find out the optimal conditions that maximized pectin yield, DE, MeO and AUA independently. Different optimal conditions were obtained (Table 3). These optimal predictions confirmed the strong influence of pH (low values) on the pectin yield, MeO content and AUA content as well as previously shown. It is noteworthy that for all responses except DE, the optimal conditions were at low pH, intermediate values of time extraction (nearly 60 min) and midpoints to high levels of LSR for the study regions tested in this work.
In order to unify to one optimal condition (theoretical optimum), an analysis of the multiple responses through the desirability discontinuous functions 41 using R-programming software 42 was performed. The conditions obtained for the highest pectin yield and highest AUA simultaneously were found at pH of 2.0, extraction time of 62.31 min and LSR of 35.07 g/mL (Table 3). Theoretical optimum was quite similar to the optimal conditions predicted for pectin yield model in Eq. (1) as is shown in Table 3. In fact, the theoretical optimum pectin yield (24.30%) can be considered equivalent to highest predicted by the Eq. (1), which is slightly lower than the experimental value, 27.16%. In addition, regarding the other responses, the theoretical predictions are very good, similar to the experimental values (Table 1). According to these findings, the models were well fitted to predict the effect of independent variables and their responses.  Table 3. Optimized responses and theoretical optimum. † Optimum calculated by each model separately ‡ Responses at pH = 2.0, extraction time = 62.31 min and LSR = 35.07 g/mL.
The optimal pectin yield (24.30%) was comparable with other studies that also used the desirability function to optimize simultaneous responses. The pectin extraction from melon peels 11 , mango peels 28 and ponkan peels 20 reported maximum pectin yields of 29.48%, 30.0% and 25.6%, respectively. Previous works from watermelon rind have reported pectin yields of approximately 19% 31 and 25% 30,32 . It has shown that citric acid was a better solvent extractant than hydrochloric acid in the pectin extraction from watermelon rind 29 . Nevertheless, Their maximum pectin yield (8.38%) obtained was very low compared to the reported values by this study and previous works, despite their optimal conditions (pH of 2.0, extraction time of 180 min, LSR of 25 and temperature of 80 °C) were pretty similar to ours. This might be due to the Colombian variety of watermelon and using methanol as a precipitating agent, which has shown to be a better precipitating agent than ethanol in the extraction of pectin from pomelo albedo 17 . For this reason, the highest properties of pectin were obtained from watermelon rind.

Characterization of pectin
The characteristics of extracted pectin from watermelon rind using citric acid under the highest yield experimental conditions (pH of 2.0, time of 60 min and LSR of 40 g/mL) were comparable with commercial pectin (Merck). Table 4 shows the remarkable properties of the pectin obtained. The ash content and alkalinity of ash were 1.24% and 0.39%, respectively, which were lower than the accepted levels for the standard pectin. The ash content is a measure of the degree of purity 40 and quality 43 of pectin. Based on these results, the extracted pectin presented low amounts of soluble solids and high purity. Free acid registered a high value compared to standard pectin; this might be due to the chemical nature of carboxyl groups, which in harsh pH conditions are hydrolyzed, increasing pectin acidity.
High methoxyl and rapid-set pectin was obtained from watermelon rind because of the MeO (10.66%), and DE (73.33%) values are ranged for this pectin type (Table 4). Rapid-Set pectin is used in high sugar jams, jellies, and marmalades 7,8 , indicating that pectin from watermelon rind could be employed in food applications. High methoxyl pectin from watermelon rind has been previously reported [29][30][31][32] , and these findings are in agreement with obtained results. From the results of AUA, the extracted pectin had a value of 82.5%, which was higher than those reported by similar works with watermelon rind [29][30][31] and as well as the percent requirements for commercial food or pharmaceutical purpose 33,34 . It should be pointed out that the AUA is a determining quality parameter and gelling properties of pectin 19,33 . Therefore, the results related to AUA also indicated that high-quality pectin was extracted from watermelon rind by using citric acid.

Properties
Standard  Table 4. Chemical properties pectin obtained from watermelon rind compared with standard pectin.

FTIR analysis
The information on the functional groups presents in the extracted pectin from watermelon rind was evaluated by FTIR analysis at highest yield experimental conditions (pH of 2.0, time extraction of 60 min and LSR of 40 g/mL). The FTIR spectra of watermelon rind pectin is illustrated in Figure 2. The peaks related to O-H and C-H stretching vibrations were between 3500-3250 cm -1 and 3000-2700 cm -1 , respectively. Stretching vibrations (=CO) of esterified carboxyl groups and free carboxyl groups (1800-1600 cm -1 ) were observed in the fingerprint regions of the pectin spectrum 10,44-46 . The peak of CH bending vibrations for pyranose ring (approximately 1338 cm -1 ) and the peak of COOstretching vibration for ester groups (1330-1210 cm -1 ) can be assigned in Fig. 2 44 . Furthermore, the characteristic overlapped peaks of glycoside bonds and pyranose cycles around 1000 cm -1 and weak peaks (830-500 cm -1 ) associated with α-and β-configurations were identified, indicating that the pectin is the principal component 44,47 . The bands of FTIR spectra of the obtained pectin were similar to reported with analogous work of pectin extraction from watermelon rind 29 .

CONCLUSIONS
In this study, the pectin extracted from watermelon rind with citric acid is considered as a high methoxyl pectin (DE > 50% and MeO > 6.7%) and has high quality (AUA > 65%, FAO/WHO, except for runs at pH of 3.0). Based on the results of BBD, pH was the most significant variable on the pectin yield and its properties. The simultaneous optimization to obtain the highest pectin yield and highest AUA showed that the optimal conditions were found at pH of 2.0 (lowest level), extraction time of 62.31 min (intermediate points) and 35.07 g/mL (high-midpoints), which were close to optimal experimental conditions (pH of 2.0, time of 60 min and LSR of 40 g/mL) for the highest yield pectin. Under this optimum the yield pectin, DE, MeO and AUA were 24.32%, 73.33% (rapid-set pectin), 10.66 and 82.50% (AUA > 74% USP, high quality), respectively. In addition, ash content and alkalinity of ash revealed high purity of pectin at the optimal condition.

MATERIALS AND METHODS Samples preparation
Watermelon rinds were obtained from local market of watermelon juices located in the Barranquilla center and around the Universidad del Atlántico, Colombia. They were washed, manually cut, and heated in distilled water until boiling to denature enzymes and inactive microorganisms. 48 The collected material was milled for homogenization and was dried in a convection oven at a temperature no higher than 80 °C for 1440 minutes. Finally, the treated watermelon rind was packed and stored in a desiccator for later use.

Chemicals and solvents
All chemicals and solvents such as citric acid, ethanol and methanol used were of analytical grade.

Pectin extraction
The pectin extraction was carried out through acid hydrolysis with citric acid, according to the methodology with some modifications. 49 The dried watermelon rind was stirred in citric acid solutions with the following defined conditions for all runs of the Box-Behnken design: pH (2.0, 2.5 and 3.0); extraction time (45, 60, 90 min) and liquid-solid ratio (1:10, 1:25 and 1:40 g/mL) at a constant temperature of 80 °C. The resultant slurry was vacuum filtered with a micro cloth using vacuum pressure. The residual liquid was precipitated with methanol in a 60% volume solution. The obtained precipitate was washed three times with 70% ethanol and subsequently was rewashed with 96% ethanol. The collected precipitate was dried in a convection oven until constant weight at 50 °C for 12 hours.

Pectin yield
The pectin yield of watermelon rind on dry weight basis was determined as follow (Eq. 5) 10,43 Pectin yield (%)= weight of dried pectin(g) weight of dried watermelon rind ( g) (5) Where the dried pectin was obtained after the treatment of filtration, precipitation and drying; the dried watermelon rind was collected as result of sample preparation.

Pectin characterization Determination of degree of esterification and anhydrouronic acid content
The degree of esterification and the anhydrouronic acid content were determined by titration relating the methoxyl content with the equivalent weight. 0.50g of watermelon rind pectin was dissolved into ethanol/water solution (1:20 v/v); 5 drops of phenol red indicator were added, and the sample was titrated with 0.1N sodium hydroxide (V1, mL) until the indicator changed. Then, 25 ml of 0.25N NaOH was added, and the sample was heated and stirred vigorously. Five drops of phenol red and 25 mL of 0.25N HCl were added again and, it was titrated with 0.1N NaOH (V2, mL) until the color change from yellow to faint pink endpoint. 50 The MeO and the DE were then calculated according to Eq. (6) 18,51 and Eq. (7) 11,19,52 , respectively.

Fourier transform infrared spectroscopy
Fourier transform infrared (FTIR) spectra of the extracted pectin was used to evaluate its structural chemical properties. FTIR analysis was performed in the SHIMADZU (IRAffinity-1) spectrometer with a resolution of 4cm -1 and 130 scans of wavelengths ranging from 7000 to 400 cm -1 .

Experimental design
The Box-Behnken design response surface design was used to evaluate the effect of three independent variables such as pH (X1: 2.0-3.0), the extraction time (X2: 45-75 min) and the liquid-solid ratio, LSR (X3: 10-40 g/ml) on simultaneous responses (pectin yield, the MeO, the DE and AUA). Table 5 summarizes the levels and code of three independent variables.  Table 5. Levels and code of variables chosen for the Box-Behnken design.

Dependent variables
The experimental design and variance analysis (ANOVA) were performed using R-programming software 4.0.2. The responses variables were fitted to the second-order polynomial model as given for the following Eq. (9).
Where: Y is the response variable; B 0 , B i , B ii and B ij are the coefficients terms for the regression model; X i and X j are the levels of independent variables. The responses were initially optimized separately to predict the 3D surface and the maximum values. Latterly, an optimization of simultaneous responses using the desirability function 41 was performed.