Physicochemical Properties and Oxidative Storage Stability of Milled Roselle (Hibiscus sabdariffa L.) Seeds

Milled Roselle (Hibiscus sabdariffa L.) seeds of the UMKL cultivar were analyzed for proximate composition, water and oil absorption capacity, and the influence of storage conditions on storage stability. The storage stability was determined under four types of conditions: light/oxygen (air) (LO), light/nitrogen (LN), darkness/oxygen (air) (DO), and darkness/nitrogen (DN) while monitoring for seven consecutive months. During the storage period, the formation of volatiles was determined using dynamic headspace sampling and Gas Chromatography-Mass Spectrometry (GC-MS) analysis. In total, 85 volatiles were identified, mainly aldehydes, alcohols, ketones, furans, and acids indicating lipid oxidation. It is recommended that milled Roselle seeds should be flushed with nitrogen and stored in darkness. Under these conditions, the seeds can be stored for at least three months without changes in volatile profile. This is important to ensure the good quality of milled Roselle seeds for further commercialization.


Introduction
Roselle (Hibiscus sabdariffa L.) is an annual erect, bushy, herbaceous shrub that can be found in almost all warm countries such as India, Saudi Arabia, Malaysia, Indonesia, Thailand, Philippines, Vietnam, Sudan, Egypt, and Mexico. Roselle is a very useful plant [1][2][3]. The most exploited part of the Roselle plant is its calyces but an underutilized by-product, the seeds, could also be valuable. In Malaysia, calyces are normally processed to produce juices and various food products; the seeds are removed and disposed of as a by-product. The Roselle seeds are, however, also edible [4]. Previous studies have shown that Roselle seeds contain high levels of protein .4%), dietary fiber (18.3-42.6%), lipid (17.4-29.6%), and minerals (23.7-128 mg Ca, 596-672 mg P, 2.08-4.0 mg Zn, 0.21-3.1 mg Cu, 26.34-396 mg Mg, 0.08-0.18 mg Cr, and 0.36-0.51 mg riboflavin) [2,[5][6][7][8]. In Nigeria, a decoction of the seeds is used to enhance lactation in cases of poor milk production and maternal mortality [9] and Nyam et al. [8] found that the seeds could be used as flour replacement for production of cookies enriched in protein and dietary fiber. In some countries, the seeds are roasted and consumed as a substitute for coffee [10].
Research done by Eltayeib and Elaziz [11] showed that Roselle seeds have a low content of free fatty acids, being indicative of low enzymatic hydrolysis, and are therefore well suited for preparation of food products.

Monitoring of Oxygen and Nitrogen Levels in Packages
The content of oxygen in the packages of milled Roselle seeds during storage is shown in Figure 1. For the packages that contained atmospheric air, a decrease in oxygen was seen during seven months of storage, most pronounced in samples stored in light. In the packages with nitrogen, the oxygen levels increased during storage, most in samples stored in the dark. The changes are a result of the balance between oxygen permeating through the packaging materials and oxygen removed due to oxidative processes within the package. The data in Figure 1 reveals that oxygen consumption is higher when packages are stored in light. In packages with air, the oxygen concentration decreases to a value close to zero, and it happens faster in packages stored in light. In packages with nitrogen, the higher oxygen increase in packages stored in the dark must be due to less oxygen consumption because the packages had identical geometry and therefore the same influx of oxygen. In fact, the oxygen level seems to stabilize in the samples stored in the light, indicating equal influx and consumption. In the samples stored in the dark, the oxidative processes are too slow to maintain a constant oxygen concentration.
In the PCA score plot (Figure 2A), a clear tendency can be seen that samples move to the right when storage time increases (higher values of PC1). It is also seen that samples stored in nitrogen move to the lower right (increasingly negative values of PC2) while samples stored in air have rather constant (small) positive values of PC2. The influence of light is less clear from these plots. From the score plot, it can be concluded that storage of up to three months only has a small effect on the volatile profile, except if samples are stored in air and light ('LO').
Up to three months, all linear aldehydes develop least in samples stored in the dark and most in the samples stored in air and light. After three months, the pattern becomes a little unclear, presumably because some aldehydes are broken further down, but, after seven months, all compounds except (E,E)-2,4-decadienal have the highest concentration in samples stored in air and light ('LO') (see Figure 3 showing hexanal and (E,E)-2,4-decadienal as examples). If samples are stored in nitrogen in the dark, no significant changes are seen during the first three months. Linear aldehydes are formed by oxidation of fatty acids. The oxidation can either be autoxidation, photooxidation or oxidation catalyzed by lipoxygenase, but since all create the same classes of secondary oxidation products, the present data cannot pinpoint which mechanism is the one dominating. It can be hypothesized that oxidation catalyzed by lipoxygenase is not very dominant, since this is the fastest.
In the PCA score plot (Figure 2A), a clear tendency can be seen that samples move to the right when storage time increases (higher values of PC1). It is also seen that samples stored in nitrogen move to the lower right (increasingly negative values of PC2) while samples stored in air have rather constant (small) positive values of PC2. The influence of light is less clear from these plots. From the score plot, it can be concluded that storage of up to three months only has a small effect on the volatile profile, except if samples are stored in air and light ('LO').
Up to three months, all linear aldehydes develop least in samples stored in the dark and most in the samples stored in air and light. After three months, the pattern becomes a little unclear, presumably because some aldehydes are broken further down, but, after seven months, all compounds except (E,E)-2,4-decadienal have the highest concentration in samples stored in air and light ('LO') (see Figure 3 showing hexanal and (E,E)-2,4-decadienal as examples). If samples are stored in nitrogen in the dark, no significant changes are seen during the first three months. Linear aldehydes are formed by oxidation of fatty acids. The oxidation can either be autoxidation, photooxidation or oxidation catalyzed by lipoxygenase, but since all create the same classes of secondary oxidation products, the present data cannot pinpoint which mechanism is the one dominating. It can be hypothesized that oxidation catalyzed by lipoxygenase is not very dominant, since this is the fastest. The fact that there is an effect of light indicates a combination of autoxidation and photooxidation to be the most prominent. Hexanal is often used as an indicator of oxidation, but many of the other aldehydes have also been reported to cause rancid off-flavors in a broad range of products [13,20,21].
The fact that there is an effect of light indicates a combination of autoxidation and photooxidation to be the most prominent. Hexanal is often used as an indicator of oxidation, but many of the other aldehydes have also been reported to cause rancid off-flavors in a broad range of products [13,20,21].  Table 2.
Among the Strecker aldehydes, 2-methylbutanal, 3-methylbutanal, and 2-methylpentanal levels increase significantly in samples stored in air, while the increase is moderate or absent during storage in nitrogen. 2-Methylpropanal increases somewhat over time, but there is no clear relation to the storage conditions. As an example of a Strecker aldehyde, the development of 3-methylbutanal is shown in Figure 4. It is seen that, if stored in nitrogen, no significant increase will occur at any point of time. This is an even more pronounced effect of the nitrogen atmosphere than seen for the linear aldehydes. There is no clear effect of light. A positive effect of oxygen on the formation of Strecker aldehydes was also demonstrated by Wietstock et al. [22] during beer production and storage.
The furans are generally increasing over time, and it is a common tendency that the levels are dropping in the last part of the storage when samples are stored with air (see Figure 5 showing 2pentylfuran as an example). Again, there is no change during the first three months if the samples are stored in nitrogen. The effect of light is not clear.  Table 2. Among the Strecker aldehydes, 2-methylbutanal, 3-methylbutanal, and 2-methylpentanal levels increase significantly in samples stored in air, while the increase is moderate or absent during storage in nitrogen. 2-Methylpropanal increases somewhat over time, but there is no clear relation to the storage conditions. As an example of a Strecker aldehyde, the development of 3-methylbutanal is shown in Figure 4. It is seen that, if stored in nitrogen, no significant increase will occur at any point of time. This is an even more pronounced effect of the nitrogen atmosphere than seen for the linear aldehydes. There is no clear effect of light. A positive effect of oxygen on the formation of Strecker aldehydes was also demonstrated by Wietstock et al. [22] during beer production and storage.
The furans are generally increasing over time, and it is a common tendency that the levels are dropping in the last part of the storage when samples are stored with air (see Figure 5 showing 2-pentylfuran as an example). Again, there is no change during the first three months if the samples are stored in nitrogen. The effect of light is not clear. Table 2. Volatile compounds identified by Gas Chromatography-Mass Spectrometry (GC-MS) in the packed milled Roselle (Hibiscus sabdariffa L.) seeds stored for seven months in the presence or absence of oxygen and under different light conditions (light or darkness). Volatiles in bold were detected in our previous work on aroma profile of Roselle seeds [19].

1-Octen-3-ol is shown in
Acetic acid is shown in Figure 7 as an example of an acid. When oxygen is present, there is almost no production of acids, but, in nitrogen, significant production is observed when stored more than three months.
Most of the mentioned alcohols and acids can be related to lipid oxidation and/or microbial growth. It could be hypothesized that microbial counts do not influence levels until late in the storage period, and the reason that it dominates in the packages with nitrogen could be that oxygen levels actually are highest late in the storage period (Figure 1). Microbiology was, however, not included in the present study, but could the subject of future studies.
Molecules 2018, 23, x FOR PEER REVIEW 10 of 14 benzyl alcohol, and phenylethyl alcohol) and acids (acetic acid, 2-methylpropanoic acid, and butanoic acid). 1-Octen-3-ol is shown in Figure 6 as an example of an alcohol. It is seen that, although it reaches the same high level after dark and light storage in nitrogen, the increase starts earlier in samples stored in light. Samples stored in dark exhibit no increase during the first three months.
Acetic acid is shown in Figure 7 as an example of an acid. When oxygen is present, there is almost no production of acids, but, in nitrogen, significant production is observed when stored more than three months.
Most of the mentioned alcohols and acids can be related to lipid oxidation and/or microbial growth. It could be hypothesized that microbial counts do not influence levels until late in the storage period, and the reason that it dominates in the packages with nitrogen could be that oxygen levels actually are highest late in the storage period (Figure 1). Microbiology was, however, not included in the present study, but could the subject of future studies.  Overall, the observations from the PCA are confirmed: There are practically no significant changes within the first three months of storage if samples are stored in nitrogen and in the dark. Furthermore, an effect of light was seen primarily on linear aldehydes and alcohols. benzyl alcohol, and phenylethyl alcohol) and acids (acetic acid, 2-methylpropanoic acid, and butanoic acid). 1-Octen-3-ol is shown in Figure 6 as an example of an alcohol. It is seen that, although it reaches the same high level after dark and light storage in nitrogen, the increase starts earlier in samples stored in light. Samples stored in dark exhibit no increase during the first three months.
Acetic acid is shown in Figure 7 as an example of an acid. When oxygen is present, there is almost no production of acids, but, in nitrogen, significant production is observed when stored more than three months.
Most of the mentioned alcohols and acids can be related to lipid oxidation and/or microbial growth. It could be hypothesized that microbial counts do not influence levels until late in the storage period, and the reason that it dominates in the packages with nitrogen could be that oxygen levels actually are highest late in the storage period (Figure 1). Microbiology was, however, not included in the present study, but could the subject of future studies.  Overall, the observations from the PCA are confirmed: There are practically no significant changes within the first three months of storage if samples are stored in nitrogen and in the dark. Furthermore, an effect of light was seen primarily on linear aldehydes and alcohols. Overall, the observations from the PCA are confirmed: There are practically no significant changes within the first three months of storage if samples are stored in nitrogen and in the dark. Furthermore, an effect of light was seen primarily on linear aldehydes and alcohols.

Sample Materials
Sun dried Roselle (Hibiscus sabdariffa L.) seeds of the UMKL cultivar (obtained from HERBagus Sdn. Bhd., Penang, Malaysia) were chosen for the study. Identification of plant species was based on taxonomic descriptions and photographic illustrations by botanist Dr Shamsul Khamis from Institute of Bioscience, University Putra Malaysia. Roselle seeds were received from HERBagus as one batch of 6 kg from which representative samples were drawn. The samples were manually sorted to remove dirt and other extraneous matter. The samples were stored at −24 • C in glass jars flushed with nitrogen. Prior to the physicochemical analysis and storage experiment, the seeds were taken out from the freezer and milled for 90 s using a laboratory blender Model 38BL41 (Waring, CT, USA). The particle size of the milled Roselle seeds was as follows: 2.2% of the milled seeds had a particle size less than 75 µm; 13.3% between 75 µm and 160 µm; 13.6% between 160 µm and 250 µm; 27.5% between 250 µm and 500 µm; 39.5% between 500 µm and 1000 µm, and only 3.9% had a particle size bigger than 1000 µm.

Proximate Analysis
Proximate analysis of the milled Roselle seeds was performed according to the standard Association of Official Analytical Chemist (AOAC) method [24]. Moisture content (hot-air oven method), ash (dry ashing method), lipid (Soxhlet extraction), protein (Micro-Kjeldahl method), total dietary fiber [25] and carbohydrates (by difference) were analyzed and calculated. All measurements were conducted in triplicate. The results were expressed as a percentage (wet weight).

Water and Oil Absorption Capacities
Bhat and Yahya's method [16] was employed for the determination of water and oil absorption capacity. One gram of the milled Roselle seeds was mixed with distilled water or oil (10 mL) in a centrifuge tube and allowed to stand at room temperature (25 • C) for 1 h. After this time period, samples were centrifuged (200× g rpm for 30 min). WAC or OAC were expressed as percentage of water or oil absorbed by 1 g of the milled seeds. All measurements were carried out in triplicates.

Packaging and Storage Conditions
The milled Roselle seeds were packed in transparent plastic laminate bags (PA/PE 20/70) with an oxygen transmission rate (OTR) of 32 mL/m 2 /24 h/atm (23 • C, 75% RH) and a water vapor transmission rate (WVTR) of 1.0g/m 2 /24 h (23 • C, 85% RH). Each bag was filled with 85 g of milled Roselle seeds and then either sealed under atmospheric conditions (21% oxygen) or flushed with nitrogen (<1% oxygen). The bags were packed using a Komet Digi-Gas Packaging Machine (Mirovac X200, Plochingen, Germany). Two replicate bags were prepared for each storage condition and for each month of analysis. The samples were represented as follows: The samples stored in light were placed in a room without direct sunlight and with only electric light (Philips Master TL-D Super 80, 36 W/380, Copenhagen, Denmark) turned on constantly. The samples that were stored in darkness were kept in the same room inside two black plastic bags. The samples were stored at room temperature during the entire storage period. During the storage experiment, the temperature readings were recorded by a portable thermometer (EL-EnviroPad-TC, Corintech, UK) via a thermocouple probe. A diagram of packaging and storage conditions is shown in Figure 8. The minimum, average, and maximum of the storage temperatures were 20.3 • C, 22.2 • C, and 24.2 • C, respectively. The samples were analyzed after approximately 0, 1, 3, 5, and 7 months of storage (or exactly 0, 32, 101, 168, and 225 days, respectively). In the following, the time points will be given as the approximate number of months. The two replicates of each storage condition were analyzed on each day of analysis and, from each of these, two sub-replicates were prepared for all the analyses. Overall, this resulted in four replicates for each storage treatment on each day of analysis. In the sample preparation and the analysis steps, the handling of the samples was randomized. Four samples resulted in very atypical values and were excluded from the dataset. The samples stored in light were placed in a room without direct sunlight and with only electric light (Philips Master TL-D Super 80, 36 W/380, Copenhagen, Denmark) turned on constantly. The samples that were stored in darkness were kept in the same room inside two black plastic bags. The samples were stored at room temperature during the entire storage period. During the storage experiment, the temperature readings were recorded by a portable thermometer (EL-EnviroPad-TC, Corintech, UK) via a thermocouple probe. A diagram of packaging and storage conditions is shown in Figure 8. The minimum, average, and maximum of the storage temperatures were 20.3 °C, 22.2 °C, and 24.2 °C, respectively. The samples were analyzed after approximately 0, 1, 3, 5, and 7 months of storage (or exactly 0, 32, 101, 168, and 225 days, respectively). In the following, the time points will be given as the approximate number of months. The two replicates of each storage condition were analyzed on each day of analysis and, from each of these, two sub-replicates were prepared for all the analyses. Overall, this resulted in four replicates for each storage treatment on each day of analysis. In the sample preparation and the analysis steps, the handling of the samples was randomized. Four samples resulted in very atypical values and were excluded from the dataset.

Gas Composition
Before the bags were opened for analysis, the gas composition was measured using a CheckMate 9900 (PBI Dansensor A/S, Ringsted, Denmark). After measurement, the sample was vacuum-packed and kept frozen until further analysis. The overall gas composition during the entire storage period is shown in Figure 1.

Volatile Compound Analysis
Dynamic headspace sampling (DHS) was applied to extract volatile compounds. The method was carried out as described by Juhari [19]. Twenty-five gram of milled seeds was put in a gas washing flask (500 mL) with a magnetic stirrer. The samples were equilibrated to 40 °C ± 1 °C in a circulating water bath and then purged with nitrogen (100 mL min −1 ) for 40 min. Volatile compounds were collected on a Tenax-TA traps (Markes International, Llantrisant, UK).

Gas Composition
Before the bags were opened for analysis, the gas composition was measured using a CheckMate 9900 (PBI Dansensor A/S, Ringsted, Denmark). After measurement, the sample was vacuum-packed and kept frozen until further analysis. The overall gas composition during the entire storage period is shown in Figure 1.

Volatile Compound Analysis
Dynamic headspace sampling (DHS) was applied to extract volatile compounds. The method was carried out as described by Juhari [19]. Twenty-five gram of milled seeds was put in a gas washing flask (500 mL) with a magnetic stirrer. The samples were equilibrated to 40 • C ± 1 • C in a circulating water bath and then purged with nitrogen (100 mL min −1 ) for 40 min. Volatile compounds were collected on a Tenax-TA traps (Markes International, Llantrisant, UK).
The trapped volatiles were thermally desorbed (TurboMatrix 350, Perkin Elmer, Shelton, CT, USA) and transferred to a gas chromatography-mass spectrometry (7890A GC-system interfaced with a 5975C VL MSD from Agilent Technologies, Palo Alto, CA, USA) including a DB-Wax column (30 m × 0.25 mm × 0.50 µm). Peak areas and mass spectra were extracted from the chromatograms using the PARAFAC2 based software PARADISe (University of Copenhagen, Copenhagen, Denmark) [26] and mass spectra were identified using the NIST05 database. Peak areas were used as relative measures of concentration. Volatile compound identification was confirmed by comparison with retention indices (RI) of authentic reference compounds or retention indices reported in the literature.

Data Analysis
One-way analysis of variance (ANOVA) was performed using the software JMP (version 13.0, SAS Institute Inc., Cary, NC, USA). Principal component analysis (PCA) with autoscaling was run using the Latentix software (LatentiX 2.0. Latent5, Copenhagen, Denmark).

Conclusions
To our knowledge, this is the first time that changes of volatile compounds during storage of milled Roselle seeds are reported.
The present study demonstrates that milled Roselle seeds are a potential ingredient for food purposes because they contain high levels of protein, lipid and total dietary fiber. Besides that, we have shown that milled Roselle seeds exhibit good water and oil absorption capacity, which can be beneficial if included in food products. Furthermore, this study points out the importance of controlling the storage conditions to conserve the quality of milled Roselle seeds. For optimal storage, milled Roselle seeds should be stored in darkness rather than light, and flushed with nitrogen because light and oxygen accelerate the oxidation processes in the packaged food. Storage under these conditions may provide a shelf life of three months without attaining off-flavor.