The hibiscus was supplied for Association Lomba do Pinheiro, Brazil. Once received in the laboratory, the calyces were separated from the seeds. Then, calyces were cleaned, dried and packed in polyethylene bags, sealed and storage at -18 °C.

Enzyme complex of cellulose, hemicellulase and pectinase (Novozym 33095) was donated by Novozymes (Spain). HPLC-grade methanol was supplied by J.T. Baker (Trinidad and Tobago), formic acid and acetonitrile from Panreac (Darmstadt, Germany). Millipore (Massachusetts, USA) membranes (0.22 μm and 0.45 μm) were used before the HPLC analyses. The ABTS (2,2’azino-bis-3-ethylbenzo-thiazoline-6-sulfonic acid), trolox (6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid), standards caffeic acid, coumaric acid, ferulic acid, quercetin and cyanidin were purchased from Sigma-Aldrich (St. Louis, USA). All reagents were analytical grade.

The calyces were submitted to steam blanching for 4 min by autoclave at 100 ºC and right away, cooled in an ice bath for 3 min. The acidified water (2% of citric acid, w/v) [24Cid-Ortega S, Guerrero-Beltrán JA. Antioxidant and physicochemical properties of Hibiscus Sabdariffa extracts from two particle sizes. J Food Res 2016; 5: 98-109.
[http://dx.doi.org/10.5539/jfr.v5n2p98] ] was added to the calyces in a proportion of 1:5 (w/w), and the mixture was mixed and triturated in a blender (Britânia).

During the extraction procedure, the agitation was realized in a mechanical stirrer (RW20, IKA) with a rod stirrer (R1342). The temperature was controlled by a water bath (Laborota 4000, Heidoplph). After the extraction process, the extracts were centrifuged at 6.000 rpm (CR21GIII, Hitachi Koki) to separate the water-insoluble solids.

Table 1
Treatments defined by fractional factorial experimental design with range of coded and real values to aqueous extraction of hibiscus calyces.

Table 2
Color parameters of the hibiscus calyces extracts prepared following the treatments defined by fractional factorial experimental design showed in (Table 1).

The measured color of the extracts was made using a colorimeter (CR400/410, Minolta), according to the CIELAB (L*, a*, b*) system. Before measurement, the instrument was calibrated using a white ceramic plate. Chroma (C* = [a*² + b*²]^1/2) was calculated, which indicates color’s purity or saturation.

The total anthocyanin content was determined by the method of differential pH [25Lee J, Durst RW, Wrolstad RE. Determination of total monomeric anthocyanin pigment content of fruit juices, beverages, natural colorants, and wines by the pH differential method: collaborative study. J AOAC Int 2005; 88(5): 1269-78.
[PMID: 16385975] ]. The extracts were mixed with buffer solutions pH 1.0 and 4.5. The absorbance was measured in a spectrophotometer (Genesys S10, Thermo Scientific) at 520 and 700 nm. The concentrations were expressed in mg of delphinidin 3-sambubioside per g extract on a dry basis, using molecular weight 577 g/mol, and molar absorptivity 26000 L/mol/cm [26Cisse M, Vaillant F, Acosta O, Dhuique-Mayer C, Dornier M. Thermal degradation kinetics of anthocyanins from blood orange, blackberry, and roselle using the arrhenius, eyring, and ball models. J Agric Food Chem 2009; 57(14): 6285-91.
[http://dx.doi.org/10.1021/jf900836b] [PMID: 19545116] ].

The radical scavenging ability of the extracts was measured by the ABTS (2,2’azino-bis-3-ethylbenzo-thiazoline-6-sulfonic acid) antioxidant capacity assay following the methodology recommended by Re et al. [27Re R, Pellegrini N, Proteggente A, Pannala A, Yang M, Rice-Evans C. Antioxidant activity applying an improved ABTS radical cation decolorization assay. Free Radic Biol Med 1999; 26(9-10): 1231-7.
[http://dx.doi.org/10.1016/S0891-5849(98)00315-3] [PMID: 10381 194] ]. The results were calculated based on a calibration curve of Trolox (1.5 – 20 µM) (R² = 0.9995), and Trolox-equivalent antioxidant capacity (TEAC) values were expressed as µmol of Trolox per g extract on a dry basis.

Determination of the phenolic compounds was realized according to Rodrigues et al. [28Rodrigues E, Mariutti LRB, Mercadante AZ. Carotenoids and phenolic compounds from Solanum sessiliflorum, an unexploited Amazonian fruit, and their scavenging capacities against reactive oxygen and nitrogen species. J Agric Food Chem 2013; 61(12): 3022-9.
[http://dx.doi.org/10.1021/jf3054214] [PMID: 23432472] ]. The liquid chromatographic analysis was carried out in a Shimadzu HPLC system with a diode array detector, using an Atlantis T3-RP C18 column (250 × 4.6 mm, 5 μm). In order to improve the peak of separation a Synergi Hydro-RP C18 column (250× 4.6 mm, 4 μm) was also used. Mobile phase consisting of two solvents (A) water acidified with formic acid (0.5%, v/v) and (B) acetonitrile acidified with formic acid (0.5%, v/v), and the following gradient was used: A/B from 99:1 to 50:50 (0 - 50 min), from 50:50 to 1:99 (50 - 55 min) and finally held for 5 min. Column conditions were of 29 °C, 0.7 mL/min and 20 μL for temperature, the flow rate of mobile phase and injection volume, respectively. The UV-vis spectra were monitored from 200 to 800 nm and the chromatograms were obtained at 260, 320, 360 and 520 nm [29Dal Magro L, Dalagnol LMG, Manfroi V, Hertz PF, Klein MP, Rodrigues RC. Synergistic effects of pectinex ultra Clear and Lallzyme Beta on yield and bioactive compounds extraction of Concord grape juice. Lebensm Wiss Technol 2016; 72: 157-65.
[http://dx.doi.org/10.1016/j.lwt.2016.04.046] ].

HPLC was connected in series to a diode array detector and a mass spectrometer with a Q-TOF analyzer and electrospray ionization source (micrOTOF-QIII, Bruker Daltonics). Electrospray ionization and MS analysis conditions were: capillary voltage 2.0kV (positive mode) and 3.0kV (negative mode), nitrogen was used as both the nebulizing gas (flow rate of 8 mL/min) and dry gas (pressure of 2 bar), probe temperature of 310 °C and the mass spectra scan from 100 to 700 m/z. MS2 was set in automatic mode applying fragmentation energy of 34 V.

The phenolic compounds were identified based on retention time and elution order in the reversed phase column, UV−Vis and MS spectrum characteristic compared to standards analyzed on similar conditions and data of the literature [30Piovesana A, Rodrigues E, Noreña CPZ. Composition analysis of carotenoids and phenolic compounds and antioxidant activity from hibiscus calyces (Hibiscus sabdariffa L.) by HPLC DAD MS/MS. Phytochemical Analysis 2018 https://onlinelibrary.wiley.com /doi/abs/10.1002/pca.2806].

The phenolic compounds were quantified by HPLC-DAD, using analytical curves with nine-point of caffeoylquinic acid, ferulic acid, quercetin and cyanidin. Data analysis was very well fixed to linear models (r² > 0.99).

For this study, a 2^4-1 fractional factorial design was utilized (Table 1), resulting in eight treatments. Four independent factors at two levels were studied: enzyme (x₁) (0; 50 µL/1000 g extract), extraction temperature (x₂) (35; 55 °C), stirring speed (x₃) (200; 400 rpm) and extraction time (x₄) (3; 5 h).

The variables x_i were coded as X_i, according to the equation:

(1)

Where is the mean value of each independent variable, and ∆x_i is the step change value. The levels of the independent variables in coded (-1, +1) and real values are shown in Table 1.

The following model was fitted to the data (Equation 1) through regression analysis:

(2)

Where Y is the response variable, β₀, β_i, and β_ij are coefficients of the term independent, the linear and interaction effects, respectively, and X_i and X_j the coded level of variables x_i and x_j.

The analysis of Chroma, TMA (total monomeric anthocyanins), ABTS and phenolic compounds were the responses variables. The resulting models were used to plot response surfaces. Data were subjected to ANOVA and Pearson correlation, and the treatments to Tukey’s multiple comparison tests, using the software SAS 9.3. The experiments were performed in triplicates and a 95% confidence level was used.

Table 2 shows the color parameters for different extraction conditions. L* values were significantly lower (darker color) when either higher levels of temperature or stirring speed were used. This might be due to the formation of dark compounds during the extraction. The higher values of a*, b* and Chroma were obtained at 35 ºC. The results also showed that they are located in the first quadrant of the hue circle, indicating a tendency to yellowness and redness. The best extracts with red color may be related to the extraction of anthocyanins [31Ramirez-Rodrigues MM, Plaza ML, Azeredo A, Balaban MO, Marshall MR. Physicochemical and phytochemical properties of cold and hot water extraction from Hibiscus sabdariffa. J Food Sci 2011; 76(3): C428-35.
[http://dx.doi.org/10.1111/j.1750-3841.2011.02091.x] [PMID: 21535 810] ].

The mean values of TMA varied from 3.07 to 3.82 mg/g extract on a dry basis for the hibiscus calyces extracts Fig. (1A), where the increase in the stirring speed and the temperature increased the TMA concentrations. These values are in agreement with Christian and Jackson [32Christian KR, Jackson JC. Changes in total phenolic and monomeric anthocyanin composition and antioxidant activity of three varieties of sorrel (Hibiscus sabdariffa) during maturity. J Food Compos Anal 2009; 22: 663-7.
[http://dx.doi.org/10.1016/j.jfca.2009.05.007] ] that reached mean values from 1.8 to 3.5 mg/g extract on a dry basis for TMA using organic solvents in the extraction. Thus, when used to ideal conditions in the hibiscus aqueous extraction process, it is possible to achieve similar levels of total monomeric anthocyanins compared with extraction with organic solvents. The highest anthocyanins extractions were obtained for T8 and T7 and they did not show a significant difference Fig. (1A).

Fourteen phenolic compounds, namely, 3-caffeoylquinic acid, delphinidin 3-sambubioside, 3-p-coumaroylquinic acid, cyanidin 3-sambubioside, 5-caffeoylquinic acid, 4-caffeoylquinic acid, ferulic acid derived, myricetin 3-sambubioside, 5-p-coumaroylquinic acid, 5-O-caffeoylshikimic acid, quercetin 3-sambubioside, quercetin 3-rutinoside, quercetin 3-glucoside and kaempferol 3-O-rutinoside were identified [30Piovesana A, Rodrigues E, Noreña CPZ. Composition analysis of carotenoids and phenolic compounds and antioxidant activity from hibiscus calyces (Hibiscus sabdariffa L.) by HPLC DAD MS/MS. Phytochemical Analysis 2018 https://onlinelibrary.wiley.com /doi/abs/10.1002/pca.2806] and quantified. The concentrations on a dry basis of the phenolic compounds obtained by different extraction assays are shown in Table 3.

The total phenolic acids varied from 4165 to 6929 μg/g extract on dry basis, being the 3-caffeoylquinic acid the main phenolic acid of the hibiscus extract (2089 to 3568 μg/g extract on dry basis), followed, in descending order, by 5-caffeoylquinic acid (1726 to 2751 μg/g extract on dry basis), 3-p-Coumaroylquinic acid (163 to 310 μg/g extract on dry basis), 5-p-coumaroylquinic acid (72 to 127 μg/g extract on dry basis), 4-caffeoylquinic acid (63 to 107 μg/g extract on dry basis), 5-O-caffeoylshikimic acid (40 to 59 μg/g extract on dry basis) and ferulic acid derived (10 to 25 μg/g extract on dry basis). The T8 and T7 extractions provided the highest values of total phenolic acid (6929 and 6751 μg/g extract on dry basis, respectively), in these treatments can be observed that the use of both the higher levels of temperature and stirring speed, showing the importance of these variables for the extraction of phenolic acids. Several studies reported the importance of phenolic acid in antioxidant activity [35Borrás-Linares I, Fernández-Arroyo S, Arráez-Roman D, et al. Characterization of phenolic compounds, anthocyanidin, antioxidant and antimicrobial activity of 25 varieties of Mexican Roselle (Hibiscus sabdariffa). Ind Crops Prod 2015; 69: 385-94.
[http://dx.doi.org/10.1016/j.indcrop.2015.02.053] , 5Fernández-Arroyo S, Herranz-López M, Beltrán-Debón R, et al. Bioavailability study of a polyphenol-enriched extract from Hibiscus sabdariffa in rats and associated antioxidant status. Mol Nutr Food Res 2012; 56(10): 1590-5.
[http://dx.doi.org/10.1002/mnfr.201200091] [PMID: 22893520] ].

The content of total anthocyanins quantified by HPLC ranged from 5456 to 10084 μg/g extract on dry basis. The highest content for total anthocyanins was also obtained by T8 and T7 treatments, 10084 and 9395 μg/g extract on dry basis, respectively. These results were much higher than those found by Salazar-González et al. [36Salazar-González C, Vergara-Balderas FT, Ortega-Regules AE, Guerrero-Beltrán JÁ. Antioxidant properties and color of Hibiscus sabdariffa extracts. Cienc Investig Agrar 2012; 39: 79-90.
[http://dx.doi.org/10.4067/S0718-16202012000100006] ] who reported 2147 μg/g of total anthocyanins in hibiscus aqueous extract. Extraction efficiency depends on several variables, such as the technique of separation, the raw material composition, and the type of solvent [37Soquetta MB, Terra LM, Bastos CP. Green technologies for the extraction of bioactive compounds in fruits and vegetables. CYTA J Food 2018; 1: 400-12.
[http://dx.doi.org/10.1080/19476337.2017.1411978] ]. Thus, phenolic compounds profile contained in raw materials, for having different polarities, may affect the extraction process [38Andrade RAMS, Maciel MIS, Santos AMP, Melo EA. Optimization of the extraction process of polyphenols from cashew apple agro-industrial residues. Food Sci Technol 2015; 35: 354-60.
[http://dx.doi.org/10.1590/1678-457X.6585] ].

Regarding the anthocyanins of hibiscus extract, two anthocyanins, highlighting the delphinidin 3-sambubioside (up to 7516 μg/g extract on dry basis) as the main anthocyanin of the hibiscus were identified and quantified, followed by cyanidin-3-sambubioside (up to 2568 μg/g extract on dry basis), which is found to be in agreement with the results obtained by Ramirez-Rodrigues et al. [31Ramirez-Rodrigues MM, Plaza ML, Azeredo A, Balaban MO, Marshall MR. Physicochemical and phytochemical properties of cold and hot water extraction from Hibiscus sabdariffa. J Food Sci 2011; 76(3): C428-35.
[http://dx.doi.org/10.1111/j.1750-3841.2011.02091.x] [PMID: 21535 810] ] and Christian et al. [39Christian KR, Nair MG, Jackson JC. Antioxidant and cyclooxygenase inhibitory activity of sorrel (Hibiscus sabdariffa). J Food Compos Anal 2006; 19: 778-83.
[http://dx.doi.org/10.1016/j.jfca.2006.04.004] ], who reported delphinidin and cyanidin sambubiosides as the anthocyanins in hibiscus aqueous extracts.

Anthocyanins are responsible for the blue, red, purple and their complementary colors of most fruits, vegetables, and flowers [40Flores FP, Singh RK, Kong F. Anthocyanin extraction, microencapsulation, and release properties during in vitrol digestion. Food Rev Int 2016; 32: 46-67.
[http://dx.doi.org/10.1080/87559129.2015.1041185] ]. They constitute a large percentage of phenolic compounds from hibiscus, representing the most important compounds of the phenolic profile in this flower. These molecules have polar characteristic and are very reactive and water soluble [41Ongkowijoyo P, Luna-Vital DA, Gonzalez de Mejia E. Extraction techniques and analysis of anthocyanins from food sources by mass spectrometry: An update. Food Chem 2018; 250: 113-26.
[http://dx.doi.org/10.1016/j.foodchem.2018.01.055] [PMID: 29412 900] ], and for this reason, the water might be used as a solvent for extraction. Moreover, add a small amount of acid would allow rising the stability of them [41Ongkowijoyo P, Luna-Vital DA, Gonzalez de Mejia E. Extraction techniques and analysis of anthocyanins from food sources by mass spectrometry: An update. Food Chem 2018; 250: 113-26.
[http://dx.doi.org/10.1016/j.foodchem.2018.01.055] [PMID: 29412 900] ]. Besides, they are unstable at different factors, such as temperature, light, oxygen, pH, water activity and attendance of enzymes and copigments [42Chung C, Rojanasasithara T, Mutilangi W, McClements DJ. Stability improvement of natural food colors: Impact of amino acid and peptide addition on anthocyanin stability in model beverages. Food Chem 2017; 218: 277-84.
[http://dx.doi.org/10.1016/j.foodchem.2016.09.087] [PMID: 27719 910] , 43Weber F, Boch K, Schieber A. Influence of copigmentation on the stability of spray dried anthocyanins from blackberry. Lebensm Wiss Technol 2017; 75: 72-7.
[http://dx.doi.org/10.1016/j.lwt.2016.08.042] ].

For total flavonoids found in the hibiscus extracts, the values ranged from 349 to 598 μg/g extract on dry basis. The highest concentrations were achieved by T3 and T8 (518 and 581 μg/g extract on dry basis, respectively) while the lowest concentrations were obtained by T1 (349 μg/g extract on dry basis). Among the main flavonoids identified and quantified on the hibiscus calyces extracts, it can be highlighted the derivatives of quercetin, kaempferol, and myricetin, in accordance with the results reported by Borrás-Linares et al. [44Borrás-Linares I, Herranz-López M, Barrajón-Catalán E, et al. Permeability study of polyphenols derived from a phenolic-enriched Hibiscus sabdariffa extract by UHPLC-ESI-UHR-Qq-TOF-MS. Int J Mol Sci 2015; 16(8): 18396-411. b
[http://dx.doi.org/10.3390/ijms160818396] [PMID: 26262611] ].

In general, T8 provided the highest extraction of phenolic compounds (17595 μg/g extract on dry basis), followed by T7 (16656 μg/g extract on dry basis), showing a positive effect mainly of the temperature and stirring speed in the extraction of phenolic compounds. According to Pinelo et al. [45Pinelo M, Arnous A, Meyer AS. Upgrading of grape skins: significance of plant cell wall structural components and extraction techniques for phenol release. Trends Food Sci Technol 2006; 17: 579-90.
[http://dx.doi.org/10.1016/j.tifs.2006.05.003] ], the phenolic compounds can be linked or entangled in the polysaccharides of the cell walls, the cell vacuoles, or associated with cell nuclei.

Thus, it can be concluded that treatments in which high temperatures combined with high stirring speed provided greater degradation of hibiscus cells, increasing the release of intracellular components and of phenolic compounds bound to the cell wall polysaccharides. These results are in accordance with other studies that also emphasized the importance among temperature, time and stirring on the extraction of phenolic compounds [18Luthria DL, Mukhopadhyay S, Kwansa AL. A systematic approach for extraction of phenolic compounds using parsley (Petroselinum crispum) flakes as a model substrate. J Sci Food Agric 2006; 86: 1350-8.
[http://dx.doi.org/10.1002/jsfa.2521] , 45Pinelo M, Arnous A, Meyer AS. Upgrading of grape skins: significance of plant cell wall structural components and extraction techniques for phenol release. Trends Food Sci Technol 2006; 17: 579-90.
[http://dx.doi.org/10.1016/j.tifs.2006.05.003] ].

The extraction yields are shown in Table 3. The yields ranged from 28.36 to 52.42 and from 29.38 to 51.84 for anthocyanins and total phenolic compounds, respectively. The highest yield for these two compounds was obtained by T8 treatment.

The highest antioxidant capacities measured by ABTS were obtained by T7 and T8 Fig. (1B). These results are also in agreement with Wong et al. [46Wong PK, Yusof S, Ghazali HM, Man YBC. Optimization of hot water extraction of roselle juice using response surface methodology: a comparative study with other extraction methods. J Sci Food Agric 2003; 83: 1273-8.
[http://dx.doi.org/10.1002/jsfa.1416] ], who observed that the antioxidant capacity values increased with the increase of temperature, due to the higher mass transfer and the higher separation of antioxidant phenolic compounds.

Thus, the highest antioxidant capacities were obtained by treatments that provided the highest extraction of phenolic compounds. It is important to highlight that several studies have correlated the antioxidant activity with the phenolic content [16Prenesti E, Berto S, Daniele PG, Toso S. Antioxidant power quantification of decoction and cold infusions of Hibiscus sabdariffa flowers. Food Chem 2007; 100: 433-8.
[http://dx.doi.org/10.1016/j.foodchem.2005.09.063] , 47Sánchez-Mendoza J, Domínguez-López A, Navarro-Galindo S, López-Sandoval JA. Some physical properties of Roselle (Hibiscus sabdariffa L.) seeds as a function of moisture content. J Food Eng 2008; 87: 391-7.
[http://dx.doi.org/10.1016/j.jfoodeng.2007.12.023] ].

Table 4 shows the Pearson correlation coefficients between the color parameters, phenolic acids, total phenolics, anthocyanins, total monomeric, flavonoids and ABTS in several extractions. The highest correlation coefficients were observed between ABTS (0.90) with phenolic acids, total phenolics, and anthocyanins. On the other hand, both parameters a* or b* were also highly correlated, but showed the negative relationship (ranged between -0.98 and -0.85 and -0.95 and 0.84, respectively) with phenolic acids, total phenolics, anthocyanins total monomeric, flavonoids and ABTS.

The equations that describe the behavior in the extraction of the Chroma (Y₁), TMA (Y₂), ABTS (Y₃) and phenolic compounds (Y₄) in relation to the responses variables are presented in the equations below.