Treatments and experimental design.

Six treatments were evaluated: two types of harvest and three varieties of roselle (factorial 2 x 3). The types of harvest were continuous harvest with 6 cuts, with intervals of seven days, with the first at 25 days of the beginning of flowering and a unique harvest at the end of the cycle. Used varieties of roselle were: China; UAN6 and UAN8. A randomized complete block experimental design was used with five repetitions.

The experiment was realized under seasonal conditions, the sowing of seeds was performed to steady flow and later thinning was realized at a distance of 1 meter between each plant. Separation between furrows was 1 meter, for an approximated density of 10 thousand plants per hectare. The experimental unit was of three furrows of 5 meters each and the central furrow was taken as useful plot.

Flowers labelling.

Once flowering started, a label was placed for each flower appearing to establish the date of cutting for the continuous harvest (CH) as well as for the unique harvest (UH).

Yield variables.

Number of calyxes and calyx length. Harvested calyxes were counted in each treatment, per plant and per cut as a control; later capsule or ovary was removed from the calyxes to measure their lengths with a digital Vernier brand Mitutoyo with a 130 mm scale and sensibility of 0.01 mm.

Calyx fresh and dry weights. Capsule was removed from the recently harvested calyxes from each treatment and were weighted in a digital weighting scale brand SCIENTECH model ZSA210, later they were stored in paper bags to be dehydrated in a drying oven brand Lumistel, model HTP-42; at 60 °C for 24 horas.

Chemical analysis.

Titratable acidity. Three grams of fresh calyxes were used, placed in 50 mL of distilled water put at boiling point for 3 minutes; acidity with con NaOH at 0.08 N Valued was measured for the obtained extraction. Three drops of phenolphthalein were added in each sample as an indicator and results were reported in meq/100 g of fresh simple, with a specific pH read of 8.3; Titratable acidity was estimated with base in the equation:

Where: TA = Titratable Acidity (meq/100 g fresh sample). V_NaOH = Normality of NaOH (0.08 meq/mL). W = Weight of the sample (g).

pH. pH of the roselle extracts, obtained for each sample was measured with a potentiometer brand Corning model 350, which was calibrated with buffer solutions of pH 7 and 4.

Total Soluble Solids (TSS) . A digital refractometre Brand Abbe monocular TPM-2WAJ was used. Six calyxes per treatment were chosen for the extraction, were halved and ground to extract the sample from which the percentage of soluble solids was determined.

Microbiological analysis.

Used microbiological indicators were total coliforms, aerobic mesophilic bacteria, fungi and yeasts, which were reported as CFU/g of dehydrated calyx. Before realizing analysis, calyxes were stored in cellophane bags during 4 months, with the purpose of determining microbiological conditions after this time. These analysis were performed, based on official standards: NOM-SSA1-1994, goods and services. Preparation and dilution of food samples for microbiological analysis; and Mexican standard: ^{NMX-FF-115-SCFI-2010}, agricultural products for human consumption - roselle (Hibiscus sabdariffa L.) flower (calyx)-Specifications and test methods.

Total coliforms. The method of the most probable number was used, based on standards: NOM-109-SSA1-1994; NOM-110-SSA1-1994; NOM-113-SSA1-1994. These official standards establish procedures for sampling, handling and transport of samples, preparation and dilution of samples, as well as the method for plate count of total coliforms.

Aerobic mesophilic bacteria. They were identified according to the official Mexican standard NOM-092-SSA1-1994, goods and services. Method for plate count of aerobic bacteria.

Fungi and yeasts. Fungi and yeasts determination were realized by means of reads at 3, 4 and 5 days, as specified by the official Mexican standard NOM-111-SSA1-1994.

Yield variables.

The analysis of variance of yield characteristics, number of calyxes, calyx fresh weight, calyx length and calyx dry weight detected significant differences among varieties. In the interaction type of harvest X variety, only the calyx length variable was significant. There were no significant differences, in none of yield performance variables, when comparing the type of harvest (Table 1).

Number of calyxes and calyx length. When comparing the method of harvest no significant differences were observed for calyx length, while when comparing varieties, UAN8 was the one with highest calyx length (Table 2). According to ^{Elsadig et al. (2013)}, the number of fruits per branch/plant, as well as the number of capsules in the main stem and fruits weight are important characteristics influencing on calyx yield per plant.

Calyx fresh weight and calyx dry weight. There were no significant differences between the methods of harvest. When comparing the three varieties of roselle, UAN8 variety was different from UAN6 and China varieties, for calyx fresh weight and calyx dry weight variables. The yield in dry weight, for the three varieties in average, for the total harvest, was of 1130 kg ha^-1, while a yield of 1160 kg ha^-1 was obtained for the continuous harvest (Table 2). ^{Castro (2004)} evaluated productivity and quality of roselle calyxes, on four planting dates (October 18th, November 15th, December 18th and January 15th), with two methods of harvest (unique final harvest an in ranks of 15 days, with the first cut at 138 days after planting date). Obtained yields were different depending on sowing dates, but similar between methods of harvest, coinciding with what was obtained in the present study.

UAN8 variety obtained the highest yield with 1,410 kg ha^-1, comparing with UAN6 and China varieties, with 900 kg ha^-1 and 1,120 kg ha^-1, respectively (Table 2). These yields are high when comparing them with the average national yield which is of 291 kg ha^-1 (^{FAO, 2004}). However, yield potential is even higher since yield higher than 1,500 kg ha^-1 have been reported (^{Castro et al., 2004}). The yield of cultivation is known to depend on various factors (climate, soil, variety, handling), as demonstrated by the results reported in roselle on sowing densities (^{Terán & Soto, 2004}); fertilization (^{Ruíz-González & Victorino-Ramírez, 2015}); varieties (^{Ariza-Flores et al., 2014}); factors which should be evaluated in producing regions to optimize yield and quality of roselle calyxes.

Chemical variables.

Evaluated chemical variables were significantly different depending on the type of harvest, varieties, as well as the interaction harvest X variety, excepted for titratable acidity values in the type of harvest. For microbiological variables, in the type of harvest, there were significant differences for total coliform and fungi. There were no differences among varieties, nonetheless, differences for aerobic mesophilic bacteria were detected in the interaction type of harvest X variety (Table 1).

For the type of harvest, pH and total soluble solids variables were significantly different, the result of titratable acidity was not significant. There were significant differences of the evaluated variables in the three varieties (Table 1).

The variety resulting with the highest titratable acidity was UAN8 with 40.70 meq/100 g of fresh simple, while the other two varieties resulted with a lower acidity without significant differences among them (Table 2).

In the case of soluble solids, the variety with the higher percentage resulted to be China variety with 11.52, while the other varieties obtained lower values without significant differences among them (Table 3).

pH. There were significant differences in pH results among the methods of harvest, the continuous harvest presented a lower pH with 2.55 than the value of unique harvest with 2.76. Of the three varieties, UAN6 had the lowest pH (2.55) and different from the China variety (2.70) and UAN8 variety (2.72) which were similar among them. Results reported for pH in roselle extracts coming from distinct varieties varied in the range 2.29 to 2.81 (^{Galicia-Flores et al., 2008}; ^{Abdallah et al., 2011}; ^{Ramírez-Cortez et al., 2011}; ^{Salinas-Moreno et al., 2012}). Although higher pH values have been reported (3.3 to 3.4) in roselle aqueous extracts (González-Palomares et al., 2009), values under 3.0 are desirable (^{Salinas-Moreno et al., 2012}).

Titratable acidity. There were no differences among the type of harvest for this variable. However, this characteristic was different among varieties. UAN8 variety had the highest titratable acidity with 40.7 meq/100 g. Values in different unities were reported by ^{Abdallah et al. (2011)} in varieties of red and white roselle, with 9 and 11 mg/100 g, respectively. Other studies have reported values of titratable acidity, as the percentage of citric acid, in the range of 13.9 to 23.7 (^{Salinas-Moreno et al., 2012}). To be able to compare these values, evaluations should have been performed under similar conditions of varieties and analytic technics, in addition of being reported in equal units.

The acidity of the extracts is related with the quantity of present acids, and in roselle it has been found: hibiscus acid, citric acid, ascorbic acid, stearic acid, benzoic acid, chlorogenic acid, 4-hidroxybenzoic acid, salicylic acid, vanillic acid and protocatechuic acid, among others (^{Ariza-Flores et al., 2014}; ^{Ramirez-Rodriguez et al., 2011}).

Total soluble solids. The analysis of variance detected significant differences in total soluble solids content (°Brix), among the type of harvest and among varieties. The continuous harvest had the highest value. Regarding varieties, China had the highest content. These values were in the range 7.38 to 11.52 %, which were higher than those found by ^{Abdallah et al. (2011)} in a white roselle variety (5.5) and a red roselle variety (5.0).

Microbiological analysis.

Regarding microbiological variables (total coliforms, aerobic mesophilic bacteria, fungi and yeasts), when comparing varieties, there were no significant differences among them; Meanwhile, when comparing the types of harvest, significant differences were observed in total coliforms and fungi, with higher contamination in the unique harvest (Table 4). Factors that favor the growth of fungi are humidity, type of substrate and temperature. These conditions are also these that foster toxin production in most of agricultural products (^{Ciegler, 1978}; ^{Tola & Kebede, 2016}).

Dehydrated roselle calyxes as well as other food are susceptible to damage by microorganisms, mainly fungi which can reduce its quality in terms of color, flavor or nutritive properties, in addition, they can sporulate and produce mycotoxins affecting human health (^{Adebayo-Tayo & Samuel, 2009}).

^{Bobadilla-Carrillo (2016)} pointed out that in roselle cultivations, the highest microbial load was presented when the calyx and the fruit were not separated yet. In addition, they found that after washing the calyxes, the number of microorganisms increased. They mentioned that the different maturation of calyxes can influence their contamination.

The main fungi, producers of mycotoxins, were from Aspergillus, Fusarium and Penicillium genera, although the potential for mycotoxin production considerably varied among species from the same genus (^{Frisvad et al., 2006}). One of the problems of the productive chain of roselle is related with a deficient handling of calyxes in post-harvest, which can involve a high content of humidity, in addition to the low phytosanitary care. Results of a study in fungi associated with roselle calyxes identified 16 genera of fungi, of which the most common were Aspergillus, Alternaria, Nodulusporium, Chaetomium and Thielaria (^{Ruíz-Ramírez et al., 2015}).

Total coliforms. There were significant differences in the number of total coliforms among the methods of harvest. In the continuous harvest, there was lower contamination. This was due to the fact that calyxes of the continuous harvest were maintained less time under contaminants exposure in the environment. Mexican official standard ^{NOM-112-SSA1-1994} establishes as a limit the absence of total coliforms and although none of the evaluated treatments achieved this disposition, the highest contamination was highlighted in the traditional method of harvest (Table 4). Similar results were found by Cid-Ortega et al. (2009), who reported the presence of total coliforms in roselle flower extracts.

Aerobic mesophilic bacteria. The low pH reduced at the minimum the bacterial attack (^{Fasoyiro et al., 2005}), and control the growth of aerobic mesophilic microorganisms (^{D’Heureux-Calix & Badrie, 2004}). In the results, aerobic mesophilic bacteria did not present significant differences neither among varieties nor in types of harvest. CFU/g found for these microorganisms (Table 4) did not exceed the critical limit established in the Mexican Standard, which is of 100 CFU/g. In the study of Cid-Ortega et al. (2009) 130 CFU/g of aerobic mesophilic bacteria were reported, exceeding the highlighted limits of the Mexican standard.

Fungi and yeasts. A higher number of fungi was found in the unique harvest compared to the continuous harvest. Fungi grow better in warm and humid conditions, and in the unique harvest, calyxes were exposed during more time to these conditions. The quantity of fungi among varieties was not different. Regarding yeasts, there were no differences among treatments (Table 4). Not with standing the afore mentioned, CFU/g values for fungi and yeasts were above the critical limit established in the Mexican standard NOM-111-SSA1-1994, which is of 10 CFU/g maximum. These results coincide with those obtained by Cid-Ortega et al. (2009) who reported high values of these microorganisms (5,800 CFU/g).

Contamination of roselle flower calyxes was attributed to the deficient post-harvest handling of the product, among those that can be mentioned, there were the open-air drying of calyxes, with the resulting damage by insects, dust and animal excrements