The chemicals used were of analytical grade. Tris-HCl, 1-1-diphenyl-2-pycrylhydrazyl (DPPH), thiobarbituric acid (TBA), malonaldehyde bis-dimethyl acetal (MDA), trichloroacetic acid (TCA), quercetin, rutin and phenanthroline were obtained from Sigma (St. Louis, MO, USA). Methanol (HPLC grade), ethanol, iron(II)sulphate, gallic, caffeic and chlorogenic acids were purchased from Merck (Rio de Janeiro, Brazil). The water used was purified in Milli-Q plus system from Millipore (Milford, MA, USA). The membranefilter (PRFE 0:45 mm) was obtained from Waters Co. (Milford, MA, USA).

The determination of pH was performed using the potentiometric method under the rules of the Instituto Adolfo Lutz (1985[20]) and the total acidity was carried out according to titrimetric analysis using NaOH solution 0.1 M (Helrich, 1990[19]).

The concentration of total phenols was determined by the spectrophotometric method based on procedures described by Singleton et al. (1999[35]). Varying concentrations of methanolic fractions were mixed with 200 of Folin-Ciocalteu solution 10 % and 400 uL of 7.5 % sodium carbonate solution. The mixture was incubated at 45 °C and reacted protected from light for 15 minutes. The blank test was simultaneously prepared using distilled water instead of the samples. The absorbance was determined in spectrophotometer with wavelength adjusted to 765 nm. The same procedure was repeated for the gallic acid used as standard for comparison of phenolic compounds.

The total concentration of flavonoids was determined by the spectrophotometric method of aluminum chloride based on procedures described by Venkatachalam et al. (2012[37]). Concentration of the methanolic fractions were prepared in distilled water and mixed with 40 uL of sodium acetate, 0.1 M solution and 40 uL of a 10 % solution of aluminum chloride. The mixture was incubated at room temperature for 45 minutes and reacted protected from light. The blank test was prepared concurrently by using distilled water instead of the samples. The absorbance was measured at wavelength of 415 nm. The same procedure was repeated to obtain the standard curve with rutin.

The concentration of ascorbic acid was determined by the spectrophotometric method described in Leme Junior and Malavolta (1950[24]). 10 grams of lyophilized pulp in the three stages of maturation were immersed in the oxalic acid solution at 2 %. After 24 hours of extraction, the samples were centrifuged and the supernatant was collected for test continuity. Increasing amounts of the supernatant were added in the test tube (200 μL, 400 μL, 600 μL, 800 μL, and 1000 μL) and supplemented with oxalic acid solution to form a final volume of 1000 μL. The absorbance was measured at wavelength of 418 nm 20 seconds after adding 2000 μL dye solution of 2,6-diclorobenzenoindofenol 0.25 mg/mL. The blank test was simultaneously prepared using oxalic acid and the dye solution. The entire test was conducted at room temperature and protected from light. The same procedure was repeated to obtain the standard curve with ascorbic acid.

Reverse phase chromatographic analyses were carried out under gradient conditions using C₁₈ column (4.6 mm x 250 mm) packed with 5 μm diameter particles; the mobile phase was water containing 2 % acetic acid (A) and methanol (B), and the composition gradient was: 5 % (B) for 2 min; 25 % (B) until 10 min; 40, 50, 60, 70 and 80 % (B) every 10 min; following the method described by Sabir et al. (2012[33]) with slight modifications. The extract of Psidum guajava (HM, HDV and HV) and mobile phase were filtered through 0.45 μm membrane filter (Millipore) and then degassed by ultrasonic bath prior to use, the extracts of Psidum guajava were analyzed at a concentration of 20 mg/mL. The flow rate was 0.8 mL/min and the injection volume was 50 μL. The sample and mobile phase were filtered through 0.45 μm membrane filter (Millipore) and then degassed by ultrasonic bath prior to use. Stock solutions of standard references were prepared in the HPLC mobile phase at a concentration range of 0.030 - 0.250 mg/mL catechin, epicatechin, quercetin, quercitrin, isoquercitrin, kaempferol and rutin, and 0.045 - 0.500 mg/mL for gallic, caffeic and chlorogenic acids. Quantification was carried out by integration of the peaks using the external standard method, at 257 nm for gallic acid, 280 nm for catechin and epicatechin, 325 nm for chlorogenic and caffeic acids, and 365 for quercetin, quercitrin, isoquercitrin, kaempferol and rutin. The chromatography peaks were confirmed by comparing its retention time with those of reference standards and by DAD spectra (200 to 600 nm). Calibration curve for gallic acid: Y = 13591x + 1187.5 (r = 0.9996); caffeic acid: Y = 11748x + 1340.3 (r = 0.9997); chlorogenic acid: Y = 12685x + 1319.4 (r = 0.9998); catechin: Y = 12547x + 1337.4 (r = 0.9993); epicatechin: Y = 11958x + 1329.7 (r = 0.9990); rutin: Y = 12756x + 1163.8 (r = 0.9991); quercetin: Y = 11891x + 1241.6 (r = 0.9985); quercitrin: Y = 12694x + 1351.0 (r = 0.9989), isoquercitrin: Y = 13079x + 1451.6 (r = 0.9991) and kaempferol: Y = 13657x + 1341.8 (r = 0.9999). All chromatography operations were carried out at room temperature and in triplicate.

Carotenoid analysis was performed at reverse phase chromatographic analysis carried out under gradient conditions using C₁₈ column (4.6 mm x 150 mm) packed with 5 μm diameter particles. The mobile phase consisted of mixtures of ACN: H₂O (9:1, v/v) with 0.25 % triethylamine (A) and EtAc with 0.25 % triethylamine (B). The gradient started with 90 % A at 0 min to 50 % A at 10 min. The percentage of A decreased from 50 % at 10 min to 10 % A at 20 min. The flow-rate was 0.8 mL/min and the injection volume was 40 μL. Signals were detected at 450 nm, following the method described by Janovik et al. (2012[22]) with slight modifications. Solutions of standards references (tocopherol, licopene and β-carotene) were prepared in HPLC mobile phase at a concentration range of 0.035 - 0.350 mg/mL. P. guajava extracts (HM, HDV and HV) were analyzed at a concentration of 10 mg/mL, carotenoids were identified and quantified in the extracts of P. guajava by comparison of retention times and UV spectra with the standard solution. All chromatography operations were carried out at room temperature and in triplicate.

LOD and LOQ were calculated based on the standard deviation of the responses and the slope using three independent analytical curves, as defined by Boligon et al. (2012[5]). LOD and LOQ were calculated as 3.3 and 10 σ/S, respectively, where σ is the standard deviation of the response and S is the slope of the calibration curve.

The antioxidant capacity of the methanolic fractions was determined by spectrophotometric method of DPPH radical kidnapping according to Mensor et al. (2001[25]). Fractions of methanolic solutions were prepared in concentrations (µg/mL) of 250, 100, 50, 25, 10, 5 and 2.5 using methanol as solvent. The solutions were mixed with 1500 µL of 0.3 mM DPPH solution and incubated at room temperature for 30 minutes and reacted protected from light. The blank test was simultaneously prepared using water and methanol P.A instead of samples. The absorbance was measured at a wavelength of 517 nm. The same procedure was repeated to obtain the antioxidant capacity of ascorbic acid which was used as positive control.

The ability of the methanolic fractions extracted from P. guajava fruits to chelate Fe²⁺ ions was determined using the method described by Puntel et al. (2005[30]) with some modifications made. The test was performed with the addition of 150 µL of ferrous sulfate (FeSO₄) 2 mM in a reaction mixture containing 168 µL of 0.1 M Tris-HCl (pH 7.4), 218 µL saline solutions and the samples with concentrations (µg/mL) of 250, 100, 50, 25, 10, 5 and using distilled water as the solvent. The reaction mixture was incubated, protected from light, for 5 minutes before the addition of 13 µL of 1,1-O-phenanthroline 0.25 %.

Two blank tests were performed in which the first contained the samples solutions in the same concentrations used in the assay and the solution of 1,1-O-phenanthroline 0.25 %. Second, contained the samples solutions in the same concentrations of the samples used in the test and the FeSO₄ solution also used in the test. The absorbance was subsequently measured at 510 nm on a spectrophotometer and the chelating ability of the samples was expressed as percentage (%) (Batool et al., 2010[3]).

The lipid peroxidation was determined by the measurement of products from the oxidation of lipids that react with thiobarbituric acid (TBARS) between them, the malondialdehyde (MDA). The methodology used in this research uses as a parameter the reading of the solutions absorbance containing the extracts, phospholipids and thiobarbituric acid.

The phospholipids were obtained from egg yolk. A egg yolk solution was prepared (1g: 10 mL) of a solvent mixture containing hexane and isopropyl alcohol at a ratio of 3:2, respectively (Puntel et al., 2007[31]). TBARS production was determined using a modified method of Ohkawa et al., (1979[26]), The methanolic solutions of the fractions of P. guajava pulps at concentrations (mg/ mL) of 250, 100, 50, 25, 10 were mixed together with a solution of egg phospholipids and deionized water (total volume of 500 µL) and pre-incubated for 1 hour in a temperature of 37 °C. After this period, 500 µL of acetic acid buffer and 500 µL of 0.6 % TBA were added to the reaction mixture and taken for further incubation at 100 °C for 1 hour. After completion of the incubation, 1.5 mL of butanol was added and the mixture was subjected to centrifugation at 2000 rpm for 2 minutes. The standard curve followed the same procedure as the test being withdrawn the phospholipids solution and the solutions of the methanolic fractions, those replaced by a MDA solution 320 µL. The absorbance was measured at a wavelength of 532 nm.

The methanolic fractions of P. guajava fruits in immature stages, partially mature stages and mature stages were able to neutralize the free radical DPPH. The concentrations for LC₅₀ indicated similar values (859.33, 854.63 and 848.78 µg/mL), with no statistically significant difference between them, with better results for the mature stage.

The free radical DPPH neutralization reaction keeps a degree of proportionality to the amount of electron or proton donor compounds, the higher the donating capacity, the higher the neutralization. Phenolic compounds have in general a chemical structure composed of hydroxylic, carboxylic and carbonyl groups attached to the aromatic ring which are easily attacked by reactive species, the presence of these groups are directly linked to the power of neutralization of reactive species (van Acker et al., 1996[36]).

Cao et al. (1997[9]) correlate the amount of hydroxyl groups attached to the flavonoids basic structure with its antioxidant capacity and notes that the flavonoids with the largest number of hydroxyl has greater antioxidant potential. Gutierrez et al. (2008[17]) list the presence of flavonoids with hydroxylic multiligands (quercetin, leucocyanidin, kaempferol and guaijaverin), confirmed by the analysis in HPLC, in guava fruits. These flavonoids increase in concentration during ripening which can be related to the higher neutralizing activity of DPPH radical been observed in the mature stage.

The results obtained for the chelating capacity of Fe²⁺ ions by the methanolic fractions of P. guajava fruits in different maturation stages are shown in Table 2(Tab. 2).

The highest percentage of chelation was found in the immature stage in the concentration of 250 mg/mL with 21.93 %, as the concentration was reduced, the percentage of chelation also decreased proportionally. Unlike the immature stage, the partially mature and mature stage had lower percentages of Fe²⁺ ions chelation. At its highest concentration, the partially mature stage showed 9.20 % as well as the mature stage which showed 6.57 %. For both, not all concentrations were capable of chelating iron ions.

The chelation of these ions by various compounds present in the methanolic fractions of P. guajava fruits, is associated with a protective effect against ROS, harmful to biological systems. The immature stage showed greater activity through this antioxidant route in relation to the direct neutralization of EROS observed in the test of free radical DPPH kidnapping. Disler et al. (1975[11]) showed that in mice, tannins present in tea are capable of acting as iron chelators inhibiting the intestinal absorption of the metal. Jain et al. (2003[21]) apud Gutierrez et al. (2008[17]) cites the green fruit as rich in tannins and their correlation with in vivo activity observed in rats, we can assume that most chelating activity presented by the fruit in the immature stage is due to the presence of tannins.

In the inhibition assay of the lipid peroxidation process was possible to observe the reduction of the baseline levels for concentrations of 250 mg/mL in the three stages, especially the mature stage. A wide variation of results was observed, and the antioxidant activity was not considered significant for any of the samples and tested concentrations. The data obtained is shown in Figure 2(Fig. 2).

According to reports from Gull et al. (2012[16]) the antioxidant activity of P. guajava fruits at different stages of maturity, harvested in three different regions, was due to the oxidation of linoleic acid induced by thiocyanate. The values obtained showed significant activity being the immature stage the best result. Although his studies involved a different methodology, comparison with the results obtained in this study is valid, since the antioxidant route being studied (inhibition of lipid peroxidation) is common to work.