Watermelon Rind and Flesh Volatile Proﬁles and Cultivar Difference

: Watermelon rind is treated as agricultural waste and commonly discarded, causing environmental issues and biomass loss. This study aimed to identify volatile proﬁles of watermelon rind and ﬂesh and their cultivar difference. Volatiles were analyzed using solid-phase microextraction–gas chromatography–mass spectrometry (SPME-GC-MS). A total of 132 volatiles were identiﬁed, including aldehydes, alcohols, ketones, terpenes/terpenoids, esters, lactones, acids, and sulﬁdes. In both rind and ﬂesh, the most dominant compounds in numbers and abundance (peak area) were aldehydes and alcohols, which accounted 94–96% of the total volatile abundance in the rind and 85–87% in the ﬂesh. Total volatile in watermelon rind was only 28–58% of the corresponding ﬂesh samples. Both rind and ﬂesh shared nine-carbon aldehydes and alcohols, though the rind lacked additional diversity. Volatile difference between rind and ﬂesh was greater than the difference among cultivars, although volatiles in the rind could be two times difference between Fascination and other three watermelons (Captivation, Exclamation, and Excursion). This study provides the ﬁrst-hand knowledge regarding watermelon rind volatile proﬁles and cultivar difference and shows the potential use of rind in food or beverages due to its naturally contained nine-carbon compounds.


Introduction
Watermelon (Citrullus lanatus) is globally consumed, possessing unique organoleptic properties, hydration function, and well-studied health benefits owed to its red flesh, though verities with orange and yellow flesh also produced [1]. Roughly, watermelon consists of skin, rind, flesh, and seeds. Watermelon flesh accounts for approximately 60% of the total watermelon mass [2]. Watermelon flesh contains a large amount of water, which is approximately 93% of the total weight of the flesh [3]; it also contains micronutrients such as vitamins (A, B, C, E), mineral salts (K, Mg, Ca, Fe), amino acids (citrulline and arginine), lycopene, and phenolic compounds [2].
The edible rind makes up approximately 40% of the total watermelon mass yet is often discarded as waste [4]. Direct disposal of the rind waste is causing environmental issues, though several approaches of reusing watermelon rind have been investigated in a laboratory scale. The specialized function of the rind's polysaccharide composition (pectin and fiber) has been considered a potential reason for its reuse [2,5]. It would be favorable to take advantage of the nutritional potential of rind and create commercial value, rather than limiting it to agricultural waste. Approaches have been introduced to reduce the accumulation of solid watermelon waste by converting the rind's polysaccharides into other products such as biosorbent [6,7], bioremediation [8], biochar [9], and bioethanol [4]. Additionally, watermelon rind has been studied as a source of nutritional food ingredients such as antioxidants [10], amino acids [11][12][13][14], and pectin [15,16], especially citrulline. In processed foods, rind has been tested in pickled form and in jam [17,18]. Watermelon rind in powder form has been examined to apply in carbohydrate-based goods including  [12,19], cookies [20,21], noodles [22,23], beef patties [24], and pork patties [25]. Furthermore, a few studies have investigated watermelon rind as a possible growth medium for microbials [26,27].
Our previous study shows watermelon variety difference based on their sensory qualities [30]; however, volatiles in these varieties have not been investigated yet. It is unknown if volatile difference between flesh and rind exists and if volatiles in watermelon rind show a variety difference. Understanding volatile composition in watermelon rind will provide evidence for potential uses of watermelon rind in food and beverages. Therefore, this study aimed to (1) identify volatile compounds in watermelon rinds and the corresponding flesh, and (2) investigate part difference (rind vs. flesh) and volatile variety difference among four watermelon varieties.

Watermelon Samples
The watermelon cultivars (Captivation, Exclamation, Excursion, and Fascination) used previously in descriptive sensory analysis [30], were subjected to volatile analysis in this study. The four watermelon cultivars were picked from the Texas A&M Agrilife Research and Extension Center at Lubbock, Texas (33.6935 • N, 101.8249 • W) in September 2019, where at least six watermelon fruits per cultivar were randomly harvested. One melon (~6.0 kg) of each of the four cultivars was prepared by washing with tap water, drying with paper towels, and peeling with a knife. The entire flesh (red colored portion of the fruit,~3.5 kg/melon) of each melon was diced into 1 cm 3 pieces using a knife and mixed thoroughly using a spatula to ensure homogeneity. Separately, watermelon rinds (~1 cm thick green/white colored outer edge portion of the fruit,~0.8 kg/melon, with the steaked green external peel removed) of each cultivar were diced into 1 cm 3 pieces using a knife and mixed thoroughly with a spatula. The flesh and rind were blended separately for~20 s using a mini blender (GForce, Hong Kong, China) until a homogeneous consistency was reached. • Brix of the watermelon flesh ranged 10.5-11.5, while rinds had • Brix of 5.0-5.5. • Brix was measured using a pocket refractometer (Atago Co., Ltd., Tokyo, Japan). The samples were stored at −20 • C until volatile analysis.

Volatile Extraction Using SPME
Frozen watermelon rind and flesh were thawed to room temperature for 30 min prior to analysis. A 3 g portion of each sample was placed in 20 mL glass headspace vials which contained 1 g of NaCl (analytical grade, ACROS Organics, Fair Lawn, NJ, USA). The vials were sealed with 18 mm magnetic screw craps fitted with a PTFE blue silicone septum. A solid-phase microextraction (SPME) fiber coated with divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS, 50/30 µm film thickness, Supelco, Bellefonte, PA, USA) was used in combination with an automated AOC-6000 sampler (Shimadzu, Columbia, MD, USA). The SPME fiber was pre-conditioned at 200 • C for 3 min. The sample vial was incubated to reach equilibrium at 40 • C for 15 min within a SPME chamber. Volatile headspace extraction was performed by inserting and holding the SPME fiber in the sample vial headspace to a depth of 22 mm for 20 min, maintaining agitation at

GC-MS Analysis
Volatile separation and identification were performed with gas chromatography-mass spectrometry (GC-MS), which was a 2010 Plus GC coupled with a QP2020 MS (Shimadzu, Columbia, MD, USA). The GC-MS was coupled with an auto sampler (AOC-6000). A ZB-Wax column (30.0 m × 0.25 mm i.d. × 0.25 µm film thickness, Phenomenex, Torrance, CA, USA) was used to separate the compounds. A helium carrier gas flowed at 1.0 mL/min. The oven temperature was held for 1 min at 40 • C, then ramped up at 5 • C/min to 230 • C and held for 5 min. The mass spectrometer used an ionization source at 200 • C, an interface temperature of 230 • C, and a scan model over a fragment range of m/z 35-350 [31].
A 1:100 dilution of a purchased straight-chain C6-C26 alkane mixture (Supelco, Bellefonte, PA, USA) was ran through the same SPME-GC-MS using 1 min equilibration and 5 min extraction times for SPME. The linear retention indices (LRI) of the identified compounds were calculated based on the retention times of these alkanes. Compounds were tentatively identified by matching their ion spectra chromatogram with those from the National Institute of Standards and Technology (NIST) library and confirmed by comparing their LRI with values from literature, especially from NIST Chemistry Webbook, SRC 69. Relative compound quantities (percent of peak area) were calculated based on the ion peak areas integrated by the GC-MS program.

Statistical Analysis
Each sample was prepared and analyzed in triplicate with three watermelon fruits per cultivar. Volatile compounds in samples were compared by measurements of peak areas (abundance) and relative peak areas (percentages). Two-way analysis of variance (ANOVA) was conducted to understand if volatile abundance was dependent on cultivar, fruit part (flesh vs. rind), or both. One-way ANOVA and Tukey's HSD (honestly significant difference) test were conducted within cultivar samples of rind and flesh to determine significant differences in their volatile abundance. Principal component analysis (PCA) was conducted to visualize sample differences and volatiles associated with each sample, filtering out observations with squared cosines sum (cos 2 ) < 0.5. Heat map cluster analysis was conducted to visualize patterns of volatile abundance across the eight samples. Volatiles with an interquartile range below 0.01 were removed from heat map analysis due to their low variability. All ANOVA was conducted using SPSS version 25 (IMB SPSS, Armonk, NY, USA), while PCA and heat map cluster analysis were conducted using XLSTAT version 2019 (Addinsoft, New York, NY, USA). All tests assumed a significance level of α ≤ 0.05.

Watermelon Rind and Flesh Total and Grouped Volatiles
As shown in Figure 1 for the total volatiles identified in the rinds and flesh of four watermelon cultivars, the total volatile abundance (peak area) in the rinds was only 28-58% of the volatile abundance recorded in the corresponding flesh samples. Captivation had the highest volatile abundance in its rind and flesh, while Fascination had the lowest amount for both. rind and flesh, while ketones were the third most abundant in flesh. A greater numb volatiles were identified in the flesh than rind; the flesh also had greater total volatil tensity (cumulative peak area). The remaining volatile groups (terpenes/terpenoids ters, lactones, acids, ands sulfides) were minimal for all eight samples. Watermelon rind primarily consisted of aldehydes (31.4-65.1% of identificat and alcohols (31.1-64.7% of identifications), summing to 94-96% of total peak area shown in Figure 1. The remaining 3.8-6.1% of volatiles were represented by ketones penes/terpenoids, esters, lactones, acids, sulfurs, furans, and aromatics. Within four samples, it presented additional differences in total peak area and presence of volatil specific cultivars. Captivation had the greatest abundance (peak area) of volatiles lowed by Excursion, Exclamation, and Fascination. Fascination only had about hal amount compared to the other three cultivars. Interestingly, aldehydes were more a dant in the rinds of Captivation, Exclamation, and Fascination (65.1%, 59.8%, and 60 respectively) compared to the rind of Excursion (31.4%), in which alcohols were abundant (64.7%). Captivation, Exclamation, and Fascination, however, only conta 31.1%, 34.4%, and 33.0% alcohols, respectively.
Similarly, aldehyde and alcohols were dominant in watermelon flesh, and were sent in similar abundances (peak area) across all cultivars (52.2-57.1% aldehydes and 3 33.3% alcohols), as shown in Figure 1. The total amount of aldehydes and alcohols 85.4-7.4%, while summed aldehydes, alcohols, and ketones accounted for 95.1-95.8 the total volatiles. Terpenes/terpenoids, esters, lactones, acids, sulfur-containing c pounds, and others represented the remaining 4.2-4.9% of the total volatiles. Each cul had a unique volatile profile. Captivation had the greatest total amount of volatiles lowed by Excursion, Exclamation, and Fascination.

Watermelon Rind Major Volatiles and Cultivar Difference
A total of 132 volatiles were identified in eight samples (Table 1). There were volatiles present in four rind samples, while only 75 volatiles were found in common tween all four cultivars of rind. The most dominant (at least 2% peak area in all sam Volatiles were classified into nine groups based on their chemical structure, namely, aldehydes, alcohols, ketones, terpenes/terpenoids, esters, lactones, acids, sulfides, and others (mainly, furans and aromatics), as shown in Figure 1. The major volatile groups were aldehydes and alcohols based on numbers and quantity (peak abundance) in both rind and flesh, while ketones were the third most abundant in flesh. A greater number of volatiles were identified in the flesh than rind; the flesh also had greater total volatile intensity (cumulative peak area). The remaining volatile groups (terpenes/terpenoids, esters, lactones, acids, ands sulfides) were minimal for all eight samples.
Watermelon rind primarily consisted of aldehydes (31.4-65.1% of identifications) and alcohols (31.1-64.7% of identifications), summing to 94-96% of total peak areas as shown in Figure 1. The remaining 3.8-6.1% of volatiles were represented by ketones, terpenes/terpenoids, esters, lactones, acids, sulfurs, furans, and aromatics. Within four rind samples, it presented additional differences in total peak area and presence of volatiles in specific cultivars. Captivation had the greatest abundance (peak area) of volatiles, followed by Excursion, Exclamation, and Fascination. Fascination only had about half the amount compared to the other three cultivars. Interestingly, aldehydes were more abundant in the rinds of Captivation, Exclamation, and Fascination (65.1%, 59.8%, and 60.9%, respectively) compared to the rind of Excursion (31.4%), in which alcohols were most abundant (64.7%). Captivation, Exclamation, and Fascination, however, only contained 31.1%, 34.4%, and 33.0% alcohols, respectively.
Similarly, aldehyde and alcohols were dominant in watermelon flesh, and were present in similar abundances (peak area) across all cultivars (52.2-57.1% aldehydes and 30.2-33.3% alcohols), as shown in Figure 1. The total amount of aldehydes and alcohols was 85.4-7.4%, while summed aldehydes, alcohols, and ketones accounted for 95.1-95.8% of the total volatiles. Terpenes/terpenoids, esters, lactones, acids, sulfur-containing compounds, and others represented the remaining 4.2-4.9% of the total volatiles. Each cultivar had a unique volatile profile. Captivation had the greatest total amount of volatiles, followed by Excursion, Exclamation, and Fascination.

Rind and Flesh Volatile Comparison with PCA and Heat Map
Differences between watermelon flesh and rind were observed in the number of volatile compounds recovered, their total peak area, and the types that were most abundant ( Table 1). The differences in volatile abundance between samples is partially due to the samples being from different cultivars or being from different parts of the fruit (flesh or rind). There was a cultivar effect and fruit part effect for 66 and 69 of the volatiles, respectively (Table 1). Of the dominant volatiles in the flesh and rind samples, the abundance of hexanal, (E)-2-nonenal (except in Captivation), 6-methyl-5-hepten-2-one decreased in rind samples compared to flesh samples, while (Z)-3-nonen-1-ol, (E,Z)-2,6-nonadienal (except in Excursion), (Z)-6-nonenal (except in Excursion), nonanal (except in Excursion), (Z,Z)-3,6-nonadien-1-ol, and acetaldehyde increased in rind samples. Two-way ANOVA revealed that the interaction of cultivar and fruit part had a significant effect on the volatile abundance of 59 volatiles (Table 1).
PCA was conducted to visualize the relationship between volatiles in watermelon rind and flesh samples. PCA was conducted with all 132 volatiles as loading values and eight samples as score values. Then, all less important variables (cos 2 < 0.5) were removed. Less important variables meant those 26 volatiles common occurred in watermelon samples. A total of 106 volatiles were selected (cos 2 ≥ 0.5) and a new PCA was performed. As shown in Figure 2, the first two components accounted for 81.85% of the total variance. The first principal component (PC1) explained 62.09% of the total variance, which was the major component to differentiate samples.

B I P L O T ( A X E S F 1 A N D
Heat map cluster analysis was used to visualize the differences and commonalities of the relative percentage abundances of volatiles recovered from the flesh and rind samples ( Figure 3). The samples were grouped into two main clusters, one included all four flesh samples and the other included all four rind samples. Within the flesh cluster, Fascination and Exclamation were grouped together and separated from Excursion and Captivation, which were also separated. Within the rind cluster, Excursion and Exclamation were grouped together and separated from Captivation and Fascination, which were also separated. A total of 55 volatiles were selected to differentiate these samples. The main flesh cluster was positively correlated with 31 volatiles, which included an assortment of aldehydes, alcohols, ketones, terpenes/terpenoids, and lactones, while the main rind cluster showed a negative correlation with those volatiles. Captivation flesh showed positive correlations with benzaldehyde, 1-penten-3-one, (E)-2-pentenal, (E,E)-2,6-nonadienal, and hexanoic acid compared to the flesh of other cultivars, which were negatively correlated to those. A total of 24 volatiles were used to distinguish rind compared to flesh. The rind cluster had more positive correlations with the nine-carbon aldehydes and alcohols that greatly represent watermelon flavor compared to the flesh cluster overall, despite a lesser variation in volatiles in the main rind cluster. Fascination rind was positively correlated with less nine-carbon alcohols compared to Excursion rind. Captivation and Exclamation rinds shared the most negative correlations as well as the most positive correlations with nine-carbon aldehydes. The results were consistent with PCA ( Figure 2). Volatile compound names for V1-V132 could be found in Table 1. V = volatiles.

Discussion
The most abundant grouped volatiles in rind and flesh were aldehydes and alcohols, predominantly volatile nine-carbon compounds. Though volatiles in rind have not yet been reported, the findings of flesh volatiles were consistent with literature in that the most abundant volatile compounds in watermelon flesh are known to be nine-carbon aldehydes and alcohols [29]. Other grouped compounds such as ketones, terpenes, esters, lactones, acid, sulfur compounds, furans, aromatic, and hydrocarbons have also been reported [1,29]. Though the identification of these compounds depending on genotypes, growing conditions, and analytical method.
Though diverse groups of compounds occur in watermelon flesh, it has been established that the saturated and unsaturated nine-carbon aldehydes and alcohols are derived   Table 1. V = volatiles.

Discussion
The most abundant grouped volatiles in rind and flesh were aldehydes and alcohols, predominantly volatile nine-carbon compounds. Though volatiles in rind have not yet been reported, the findings of flesh volatiles were consistent with literature in that the most abundant volatile compounds in watermelon flesh are known to be nine-carbon aldehydes and alcohols [29]. Other grouped compounds such as ketones, terpenes, esters, lactones, acid, sulfur compounds, furans, aromatic, and hydrocarbons have also been reported [1,29]. Though the identification of these compounds depending on genotypes, growing conditions, and analytical method.
Though diverse groups of compounds occur in watermelon flesh, it has been established that the saturated and unsaturated nine-carbon aldehydes and alcohols are derived from fatty acids, especially polyunsaturated linoleic (C18:2) and linolenic (C18:3). C18:2 and C18:3 are the main precursors of most fruit aroma volatiles via αand β-oxidation in intact fruits, or the LOX pathway in fresh-cut fruits and juices [29]. Nevertheless, studies on the impact of genotypes and metabolism pathway of volatile aroma compounds in watermelon are scarce, though a few studies have reported the volatile compounds derived from the degradation of carotenoid pigments. Consequently, genotypic variation in carotenoid composition underscores differences in the terpenoic and norisoprenoid volatile profiles [1]. Hundreds of volatiles have been identified in watermelon.
In addition to nine-carbon aldehydes and alcohols, several other compounds from minor groups might be important to the aroma of watermelon flesh, though their contribution is under-explored. For example, 6-methyl-5-hepten-2-one, the most abundant ketone in watermelon flesh (77.3-83.5% of the total ketones in this study), can contribute green, musty, fruit odors [3,28]. Increased concentration of some volatiles such as acetophenone might contribute to off odors [34]. Terpenes/terpenoids such as β-ionone can contribute floral notes (violet-like) to watermelon [3], along with other terpenes/terpenoids such as linalool, citral, and geraniol; however, their contribution is not well defined. Esters are well known to contribute fruity notes to many fruits. Esters, however, are not the major grouped compounds in watermelon and only several esters have been identified in this study as well in literature [28,37]. Our current study identified two lactones. γ-Nonalactone had low peak abundance, but potentially, it could contribute sweet, creamy notes to watermelon. Volatile organic acids are barely found in watermelon literature. One of five acids identified in this study, hexanoic acid, might contribute fatty, cheesy notes to watermelon, though these notes are minimal for watermelon organoleptic characteristics. Three sulfur compounds were identified in this study, while many more have been identified by other researchers [28,38,39]. Sulfur compounds are considered off-notes for watermelon, especially processed watermelon juice [28,39].
In the present study, rind and flesh samples of different varieties differed significantly in their total volatile abundance. Results also showed that different watermelon varieties were differentiated by their volatile profile percentages of aldehydes, alcohols, and ketones (especially for flesh), indicating that volatiles are linked to the genetics of watermelon. The findings are in line with other researchers stating significant variation in the amounts and profiles of volatile among genotypes and fruit fractions [3,40,41]. Literature specifically pertaining to watermelon volatile variety difference is very limited. The total volatile content of five watermelon varieties was found to vary by a factor of up to one time [3]. Additionally, different parts of the watermelon contain different bioactive compound concentrations and antioxidant activities [41]; however, volatiles in different parts of watermelon have not yet been analyzed. Volatiles in the rind and flesh of four watermelon cultivars (Captivation, Exclamation, Excursion, and Fascination) were identified in this study for the first time, though their sensory profiles have been reported previously [30]. Horticultural traits of Captivation, Exclamation, and Fascination have also been reported [42][43][44].
Watermelon rind has advantages of high citrulline content compared to flesh [14], and contained major volatile compounds of nine-carbon aldehydes and alcohols as identified in this study. The results of the major nine-carbon volatiles in rind from the current study bring in a new potential use of watermelon rind; for example, watermelon rind might be used in highly nutritive plant waters, which could be marketed as consumer hydration options, with fresh, green flavor notes imparted by watermelon rind [45]. It should be pointed out that our previous consumer test shows that high amounts of watermelon rind imparts excessive green notes and off-flavor, negatively correlating with consumer hedonic rating [45]. The previous study also shows refreshing perception of watermelon juice blended with rind [45]. Combined our previous sensory study with this rind volatile study, it could be deduced that watermelon rind has the potential to impart refreshing perception in food products due to its high nine-carbon volatile content. Fruit volatiles are mainly association with aroma contribution, instead of other functions such as refreshing perception. The relationship between watermelon rind volatiles and refreshing perception is an interesting area, which should be further verified in future studies.

Conclusions
Volatile profiles of four watermelon rinds were investigated for the first time in this study. Both watermelon rind and flesh volatile profiles were predominantly nine-carbon aldehydes and alcohols. Fewer volatile compounds were recovered from watermelon rind compared to flesh, and volatiles were much lower in abundance in the rind. Variety difference for both rind and flesh was observed, with volatiles contents notably ranging two times within four rinds. These results imply that rind has lower volatile intensity, consequently having bland aromas, compared to flesh. The properties of bland aroma are ideal for using as a food ingredient in different food products. On the other hand, the "little" but "dominated" nine-carbon volatiles in the rind associated with fresh, green, and melon-like aroma have potential to make positive contributions to the food products, especially beverages; although, further studies related to aroma-active compounds are desirable to confirm the rind volatile aroma contribution. The finding provides knowledge that watermelon rind could be a promising supplemental ingredient for food and beverages for either using its bland aroma or its volatiles, potentially contributing to flavor, which depends on how it is used and types of its final products.
Author Contributions: J.R. carried out the study, performed data analysis and interpretation, and wrote the manuscript. X.D. designed and supervised the study, aided with data analysis and interpretation, and wrote and finalized the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This research project was funded by the National Watermelon Promotion Board and the grant ID# was 00704263.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Our data are available for review.