Processing and physicochemical characterization of a water-soluble extract based on Pará nut ( Bertholletia excelsa ) and baru nut (

Despite the nutritional and functional qualities of Pará nut (Bertholletia excelsa) and of Baru nut (Dipteryx alata Vogel), the industrial exploitation of these two nuts is still scarce. With this, the elaboration of a water-soluble extract based on Pará and Baru nuts could be consider an alternative for the consumption of foods based on these vegetables, both little used industrially. Thus, by way of a simplex centroid experimental mixture design, this study aimed to elaborate water-soluble vegetable extracts based on Pará and Baru nuts. This study allowed evaluate the behavior of the physicochemical and physical characteristics of the processed vegetable extracts. The special cubic model obtained the best fit for the response variables, both for the physicochemical characterizations of moisture, lipid, protein, ash, carbohydrate, pH value and acidity, and for the physical characteristics of luminosity, chroma and hue angle. The results obtained showed that the model is highly predictive (all R > 97%). The water content directly influenced the nutritional aspects of the water-soluble Research, Society and Development, v. 9, n. 8, e163984848, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i8.4848 3 extracts, where a decrease in the proportion of water in the mixture provided greater content of proximate composition. The behavior of the color parameter predicted by the model showed that variations in the proportions of the nuts concentrations changes the behavior of the luminosity indexes. The pH had higher values when there was an increase in the concentration of Pará nuts, while an increase in the proportion of baru nuts decreased the acidity of the water-soluble extract. The mathematical models used allow to evaluate the effects of the interactions between variables and responses, highlighting their applicability in the food industry.

extracts, where a decrease in the proportion of water in the mixture provided greater content of proximate composition. The behavior of the color parameter predicted by the model showed that variations in the proportions of the nuts concentrations changes the behavior of the luminosity indexes. The pH had higher values when there was an increase in the concentration of Pará nuts, while an increase in the proportion of baru nuts decreased the acidity of the water-soluble extract. The mathematical models used allow to evaluate the effects of the interactions between variables and responses, highlighting their applicability in the food industry.

Introduction
Water-soluble extracts appear as a plant-based beverage that have commercial and nutritional appeal and that also have been used as alternative to lactose-free diets . Nondairy beverages are processed by extracting plant material, such as nut, soy, rice, and so on in water. The plant materials are homogenized and thermally treated to enhance suspension of particles and to increase shelf life (Mäkinen, Wanhalinna, Zannini, & Arendt, 2016).
Water-soluble vegetable extracts are protein products of vegetable origin defined by Brazilian National Health Surveillance Agency -ANVISA -RDC nº 268 (Brasil, 2005) as "foods obtained from the protein parts of vegetable species, which can be presented in the granular, powdered, liquid or other forms, with the exception of those not conventionally used as food, to which other ingredients can be added, so long as this does not change the character of the product".
The use of soy milk was first reported about 2000 years ago in China. Soy milk was the first plant-based milk which serves the purpose of providing nutrients to the population where the milk supply was inadequate (Sethi , Tyagi & Anurag, 2016). However, it has limitation in consumer acceptance due to its undesirable aftertaste (Sebastian, Barus, Mulyono & Yanti, 2018).
Following the world tendency in the segment of water-soluble vegetable extracts, water-soluble extracts based on Pará and Baru nuts were obtain as a pioneer work on the subject, with the aim of developing a beverage of vegetable origin with commercial and nutritional appeal with respect to certain health aspects such as the lack of animal fat and high mineral contents. With this in view, it is essential to know the behavior of the ingredients, both individually and in combinations, since this knowledge is essential for their exploitation, favoring their development during the various phases of product development.
Mixtures or blends are defined as a set of two or more components. When the properties of interest are basically a function of the mixture or of a combination of components, experimental mixture designs are used for product development in substitution of the development of the mixture formulation by trial and error. The Simplex-centroid mixture design makes it possible to obtain formulations with an exact number of experiments and one can choose the best condition for a given response through variations in the formulation of specific components ( Barros Neto, Scarminio & Bruns, 2010;Moraes Filho, Busanello, Prudencio & Garcia, 2018). Development, v. 9, n. 8, e163984848, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i8.4848 In the development of new products, food developers must have knowledge about the quality and storage requirements of raw materials and final products, product, formulation and the processes needed to safely manufacture the final food product (Azanedo, Garcia-Garcia, Stone, & Rahimifard., 2020).
Considering that a water-soluble vegetable extract is a beverage with a high unsaturated fatty acid content, and thus a highly perishable food due to oxidative processes, which can result in the appearance of rancid odors and flavors, the elaboration of ternary mixtures of Pará nut, Baru nut and water was proposed by way of an experimental mixture design to determine the best possible formulation for this mixed beverage. Hence physical, physicochemical and microbiological determinations were carried out so as to verify the quality of the product. Due to the aforementioned, this study aimed to evaluate the behavior of the physicochemical and physical composition of water-soluble extracts based on Pará nuts and baru using a simplex centroid type design.

Raw materials
The Pará nuts used in this study were in natura and partially skinned. One part was donated by the company Delta Castanhas do Brasil Ltda (Iporá, GO, Brazil) and the other part acquired from the company Naturalista Produtos Naturais Ltda (Goiânia, GO, Brazil) together with the roasted, non-skinned Baru nuts.

Processing of the water-soluble extract based on Pará and Baru nuts
The methodology used for the production of the water-soluble extract based on Pará and Baru nuts was developed by Cardoso et al. (2020).
To obtain the water-soluble extract, the nuts were first sanitized using a 300 ppm chlorine solution, and then ground for 10 minutes in an industrial stainless steel blender (Skymsen model LSR 25, Metalúrgica Siemsen Ltda, Brusque, SC, Brazil) together with previously pasteurized water heated to 45 ºC, according to the proportions established by the proposed experimental mixture design. The entire homogeneous contents of the blender were then centrifuged (Vicini model VCC-7000, China), obtaining two co-products, the water- Development, v. 9, n. 8, e163984848, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i8.4848 6 soluble extract (liquid phase) and the moist residue of Pará and Baru nuts (solid phase). The liquid phase was heated to 85 ºC. Carboxymethylcellulose (CMC) was added to the water-soluble extract in a proportion of 0.2% (w/v) in order to minimize separation of the oily and aqueous phases, together with the preservatives citric acid, sodium benzoate and potassium sorbate in the proportions of 0.05% (w/v), 0.15% (w/v) and 0.10% (w/v), respectively. The concentrations of each preservative used were established according to current legislation for coconut milk (Brasil 2013).
The water-soluble extracts were further homogenized for 5 minutes and then filled into glass bottles with crown-type metal caps. The bottled product was pasteurized in a jacketed stainless-steel tank using a binomial of 80 ºC for 25 minutes, then cooled and stored under refrigeration (4 ± 2ºC).

Physicochemical characterization and colorimetry
The percentages of moisture, ash and protein and the pH value were determined according to Association of Official Analytical Chemists -AOAC (2010), and the lipid content was determined using the methodology proposed by Bligh & Dyer (1959). The total titratable acidity was determined directly by titration and the result expressed in percent oleic acid, according to Analytical standards of the Adolfo Lutz Institute and the carbohydrate content was determined by difference by subtracting the sum of the moisture, protein, lipid and ash contents from 100. The instrumental color parameters (L*, a* and b*) were determined using a Hunterlab colorimeter model Colorquest II (Hunter Associates Laboratory Inc, Reston, Virginia, USA). The results were expressed as L* (Luminosity) which varies from 0 (black) to 100 (white); a* which varies from -a* (green) to +a* (red); and b* which varies from -b* (blue) to +b* (yellow). The values for chroma and the hue angle were calculated based on the coordinates determined (Arivalagan et al., 2018).

Formulation and experimental design
The proportions of the mixtures were defined by employing a mixture design of the simplex-centroid type (Lima, Afonso, Costa, & Carvalho, 2019), with a view to evaluating the behavior of the interaction between the components.
The concentrations of water (x1), Pará nut (x2) and Baru nut (x3) were established as the independent variables, where the minimal amount of water was 65% (w/w) for each formulation, whereas the minimal amount of each of the nut components was 5% (w/w). The moisture, ash, lipid, carbohydrate and protein contents, plus the acidity, pH value, luminosity, chroma and hue angle were defined as the response variables. Table 1 shows the 7 different combinations used, both in terms of the original and pseudo-components Where x1, x2 and x3 represent the water, Pará nut and Baru nut concentrations, respectively, in terms of the real concentrations, and x'1, x'2 and x'3 correspond to the contents of each component (water, Pará nut and Baru nut, respectively) in terms of pseudocomponents.  The changes in moisture content presented a direct relationship with the proportion of water established by the mixture design, the smaller the proportion of water the smaller the moisture content found in the vegetable extract (Trials 2, 3 and 6). Thus, this variable presented results that were inversely proportional when compared with the other response variables, that is, for the trials with a larger water composition and hence a higher moisture content, the other response variables showed their smallest values (Trial 1). Considering these results, note that the amount of water had a direct influence on the nutritional aspects of the water-soluble extract and Labuza & Altunakar (2020), considers that the amount of water is related to the stability, quality and composition of the food. Note also that the moisture content obtained a significant difference in almost all the trials (p<0.05), with the exception of trials 2 and 3, which showed no significant difference between them, it being evident that the interaction between the nuts did not cause changes in the variable of moisture content, that is, the contents and types of nut were not important, only the water content being important.  Development, v. 9, n. 8, e163984848, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i8.4848 Observing the variations in the lipid contents presented in Table 2, all the formulations proposed by the design presented significant differences. Both the proportion of nut added to the formulations and the type of nut directly influenced the lipid composition of the extracts.

Results and Discussion
An observation of trials 2 and 3 shows that the greater the amount of nut added, the greater the lipid content and also that the type of nut directly interfered, such that the larger the amount of Pará nut added for the same amount of water, the greater the lipid content. This fact can be proven from an observation of Trials 2 and 3, which both had the same proportion of water (65%) and the same total proportion of nuts (35%), but the nuts in Trial 2 were composed of 30% Pará nut and 5% Baru nut, whilst the nuts in Trial 3 were composed of 30% Baru nut and 5% Pará nut. These trials presented lipid contents of 23.34% and 16.57%, respectively, since the lipid contents of the individual nuts were discrepant, being 66.20% for Pará nut and 35.72% for Baru nut, justifying this difference (Santos, Corrêa, & Lannes 2011;D'Oliveira, 2015). Carneiro et al. (2014) reported that the amount of fat in commercial coconut milk varied from 20 to 22%, and hence to standardize the fat content it would be necessary to establish the amount of added water at about 65%. In the present study this suggests a variation in Pará nut greater than that in Trial 6 (17.5%) and lower than that in Trial 2 (30%).
When the sum of the Pará and Baru nut contents was equal to the maximum amount allowed of 35% (Trials 2, 3 and 6), note that the contents of all the responses presented their highest values, except for the moisture and carbohydrate contents. The protein content varied between 3.17 and 9.32%, these values varying in proportion to the amounts of nuts added to the formulation, that is, trials with 35% of nuts (2, 3 and 6) showed the highest protein contents, followed by Trial (7) with 22.66% of nuts, then Trials (4 and 5) with 22.5% of nuts and finally Trial (1) with 10% of nuts in its formulation, which showed the lowest protein content in the extract.
When water-soluble extracts containing the same proportion of water are compared (2, 3 and 6) and (4 and 5), the ash and protein contents of these trials did not show significant differences, except for the protein content of trial 2. The latter difference occurred because the formulation of trial 2 contained 30% of Pará nut but only 5% of Baru nut, resulting in a much smaller protein content as compared with trial 3, which contained the same total nut content of 35%, but was composed of 30% Baru nut and 5% of Pará nut. Baru nut contains almost twice the amount of protein as Pará nut (26.97% and 13.75%, respectively) (Lima, Freitas, Czeder, Fernandes, & Naves, 2010;Santos, 2015), explaining this difference in protein content between trials 2 and 3. Research, Society and Development, v. 9, n. 8, e163984848, 2020 (CC BY 4. The water-soluble extracts formulated in the present study conformed to the parameters described in the Brazilian legislation, which fixes the identity and minimum quality characteristics that protein products of vegetable origin should obey, and recommends a minimum protein content of 3% (Brasil, 2005).
In general, the extracts showed elevated protein contents, being notorious when compared with other results found in the literature, such as that described by Carvalho et al. (2011) working with extracts from broken rice grains, whole rice grains and soybeans, who found protein contents to the order of 0.73%, 0.84% and 2.51%, respectively. Carneiro et al.
The differences found in the chemical compositions of the water-soluble vegetable extracts prepared in the present study and others discussed in the literature may be explained by many factors, such as the composition of the raw materials, the mode of processing of the water-soluble extract as well as the analytical methodology applied. Table 3 shows the results obtained for the determinations of pH value, acidity and the color variation of the samples produced.   similar to those found in the present study (6.14) whereas the acidity was 1.70. Thus, this occurred because the Baru nuts have a more acid pH value than the Pará nuts.
While evaluating the concentration of a water-soluble soybean extract, Barana, Oliveira, & Granato (2018), combined the nutritional advantages of whey and soybean by developing a type of chocolate beverage with water-soluble soybean and the pH values vary 6.5 to 6.6. Santos (2015), evaluating the stability of a water-soluble Pará nut extract, found an acidity of 0.30 and pH value of 6.34 for the treatment without preservatives and acidity of 0.53 and pH value of 5.87 with preservatives.
The elevated values for acidity found in the present study could be justified by differences in the raw materials used to elaborate the beverages and to differences in the processing procedure used, where some steps such as blanching and autoclaving were not applied in this study.
In general, the values found for Luminosity (L) varied between 59.19 and 80.71 distant from white. As the values become more distant from white, so the samples become more saturated, that is, they gained chroma and showed shades of grey. Increases in Luminosity were observed in experiments with higher concentrations of Pará nut and minimum proportions of Baru nut (Trials 1, 2 and 4). Carneiro et al. (2014) found a value of 73.35 for the attribute of Luminosity for a mixed beverage containing 60% of babassu and 40% of Pará nut, as chosen by the sensory evaluation. Silva, et al. (2020) studying water-soluble baru extract found values ranging from 67 to 70 for luminosity. These variations are mainly due to characteristics of the raw materials, since Pará nut shows higher values for luminosity than Baru nut (Santos, 2012).
The values for chroma define the color intensity, those closer to 0 being indicative of more neutral (greyer) colors and those closer to 60 indicating more vibrant and intense colors.
The chroma values increased as the Baru nut content increased and the Pará nut content decreased, indicating more vibrant colours.
Considering the results shown in Table 3, the mean angle obtained for the hue angle between all the trials was 75.78, corresponding to a color varying between red and yellow.
This property indicates the tonality and can vary from 0º to 360º, 0º corresponding to red, 90º to yellow, 180º to green and 270º to blue (Souza, 2010).Considering the values found for Luminosity, chroma and hue angle, the product obtained in the present study showed a yellowish white color with a reddish tendency.
With the aim of analyzing the fit of the experimental data obtained, the linear, quadratic and special cubic models were tested by applying the analysis of variance (ANOVA) at the 5% level of significance (p ≤ 0.05). The experimental (R²) and fitted (R²fitted) determination coefficients and the p value were analyzed in order to choose the best-fitting model. Table 4 shows the results obtained for the coefficients tested, for each response variable. A model is considered to fit the data satisfactorily if the fitted determination coefficients (R 2 ) show values above 0.85. All the models obtained fitted the data of the response variables except the quadratic model for the attributes of acidity and chroma, for which the p values were not significant (p > 0.05). The values for R 2 and R 2 fitted were close to unity for all the models, confirming their good fit and good predictive capacity.
Although all the models showed a good fit, the special cubic model, in addition to being significant (p < 0.05) in all cases, presented the highest values for R 2 and R 2 fitted, and hence the special cubic was considered to be the most adequate for this study D'Oliveira (2015), when developing a Baru nut-based beverage aromatized with chocolate, used response surface methodology to test the fit of the models to the sensory analysis attributes (color, aroma, flavor, texture, sweetness and global acceptance). The author found much lower values for the determination coefficients, varying between 0.71220 and 0.79257, than those found in the present study.
The high coefficients and ANOVA showed that the equations could be used for predictive ends.
The objective of analyzing the results was to numerically model the behavior of the mixtures of Pará nut and Baru nut in the elaboration of a mixed water-soluble extract. The region of combination between the three variables x1, x2 and x3 can be seen in the contour curves obtained using the mathematical models and presented in Figure 1, corresponding to moisture, protein, lipid, ash, carbohydrate, pH and acidity, respectively, and in Figure 2, corresponding to luminosity, chroma and the hue angle, in this order. Source: Authors (2020).
All the contour curves showed that concentrations close to 100% water, green-colored regions, were less efficient with respect to the composition of the extracts studied, providing smaller indexes for the proximate composition except for moisture. To the contrary, on observing the contour curve for the variable of moisture content, the highest proportion was found on the vertexes (1, 0, 0), represented by the red-colored region, signifying that the greater the amount of water added in the trials, the greater the moisture content found. This was proven by the model in the predictive equation for moisture (Table 5) by way of the elevated coefficient for the proportion of water in the mixture (x1) of 89.16.
Observing the contour curve for the variable lipid, it can be seen that the greatest levels for this variable were close to the variable Pará nut, as proven by the values obtained in Table 1 and by the equation (Table 5), where Trials 2 and 6 presented the largest values for this variable, presenting levels of 100% and 50% of Pará nut associated with the variable of water at its lowest level. This infers that the water acts as an ingredient that decreases the lipid content in the extract. The lowest levels for this variable can be found at the extremity for the totality of water and about 50% of the nut, as shown by the results obtained for trials 1 and 4.
The results showing greater efficiency for the contents of carbohydrate can be seen on the vertexes (0, 0, 1) with the greatest proportion of Baru nut (red region), since this parameter is calculated by difference (carbohydrates = 100moistureproteinlipidash) and the extracts with a greater proportion of Baru nut (Trials 2;3 and 4;5) presented lower lipid contents and consequently higher values for carbohydrate.
The results with the greatest efficiency for the acidity contents can be observed on the vertexes (0, 0, 1) with the greatest proportion of Baru nut (red region) as proven in trial 3, which had the highest acidity content.
In a search for the highest indexes of protein and ash, it was shown from the response surfaces plotted, that the most efficient effects for these parameters were due to the combination between elevated Pará nut and Baru nut contents (red region) showing synergistic action between these components.
From the equations generated for the color variation (Luminosity, Chroma and hue angle) one can see that the combination of the coefficients for Pará nut and Baru nut (x2x3) showed a synergistic effect, that is, it contributed to increases in the chroma and hue angle (x2 and x3 > zero) but not for luminosity, as shown by the contour curves presented in Figure 2.
An analysis of the contour curves shown in Figure 2 indicates that the luminosity (L*) decreased with increases in the quantity of Baru nut, as can be proven by trials 3 and 5.  The values for chroma presented an action to the contrary of that found for luminosity, since for this parameter, the greater the proportion of Baru nut the greater the value for chroma. Consequently, greater proportions of water and Pará nut lead to a beverage with a less intense color, as proven by trials 2 and 4.
With respect to the hue angle, the largest values for this parameter can be found in the dark red region, close to the centroid (⅓, ⅓, ⅓). Thus, greater proportions of Baru nut in the extract formulations contribute to a decrease in tonality, and hence the extracts tend to a red color.
With the objective of proving the efficiency of the model, the models generated were validated, showing a great proximity between the observed and predicted values, confirming the good predictive capacity of the models tested from their experimental and fitted determination coefficients, for which the error obtained was lower than 0.1697% for all the response variables studied.

Conclusions
The elaboration of a mixed water-soluble extract based on Pará and Baru nuts can be considered an alternative for the consumption of foods based on these vegetables.
The cubic models fitted the experimental data best and were shown to be highly predictive, considering the high coefficients and analysis of variance. In general, the mathematical models allowed for the evaluation of the effects of the interactions between the variables and the responses, emphasizing their applicability for the food industry.
It was shown that the water content had a direct influence on the nutritional aspects of the water-soluble extracts based on Pará and Baru nuts, where a decrease in the proportion of water in the mixture provided larger contents for the proximate composition.
An increase in the concentration of Pará nut raised the pH values and an increase in the proportion of Baru nut contributed to a decrease in the acidity of the water-soluble extract.
The behavior of the color parameters predicted by the model, showed that an increase in concentration of Pará nut increased the indexes of luminosity, resulting in the water-soluble extracts tending to a white color. On the other hand, an increase in the proportion of Baru nut resulted in more intense colors, contributing to a decrease in tonality of the water-soluble extracts.
Carbohydrates were calculated by using an equation. Thus, analytical measure could be necessary in order to best determine the concentration. Desirability test can be executed to estimate the best formulation based on the physicochemical composition

Disclosure of conflict of interest
The authors have no conflicts of interest to declare.