The establishment of the species-delimits and varietal-identities of the cultivated germplasm of Luffa acutangula and Luffa aegyptiaca in Sri Lanka using morphometric, organoleptic and phylogenetic approaches

Luffa acutangula and L. aegyptiaca are two vegetable species commonly found in South and South East Asia. L. acutangula is widely grown; however, L. aegyptiaca is considered as an underutilized crop. The species delimits, phylogenetic positions, and the varietal identities of L. acutangula and L. aegyptiaca in Sri Lanka are not known. Thus, in the present study, we aimed to establish the species delimits and varietal identities of L. acutangula and L. aegyptiaca varieties grown in Sri Lanka using morphometric, phylogenetic and organoleptic assessments. We assessed five varieties of L. acutangula and three varieties of L. aegyptiaca. The vegetative and reproductive data were collected for the morphometric analysis and DNA sequence polymorphism of the makers rbcL, trnH-psbA and ITS for the phylogenetic analysis. We also conducted an organoleptic assessment based on taste parameters; aroma, bitterness, color, texture, and overall preference using the dishes prepared according to the most common Sri Lankan recipe for Luffa. The variation of the vegetative and reproductive traits grouped L. acutangula varieties into two distinct clusters. The trnH-psbA polymorphism provided the basis for the species delimits of L. acutangula and L. aegyptiaca. The rbcL and ITS polymorphisms provided the basis for the identities of the varieties of L. aegyptiaca and L. acutangula respectively. In the phylogeny, the L. acutangula varieties of Sri Lanka formed a unique clade and the L. aegyptiaca varieties formed a reciprocal monophyletic group in comparison to worldwide L. aegyptiaca reported. The taste parameters aroma, texture, color, and overall preference were significantly different among the Luffa varieties. The L. aegyptiaca varieties received lower preference in the organoleptic assessment. The present study sets the species delimits, phylogenetic positions and the varietal identities of the cultivated germplasm of Luffa and revealed the distinct morphological and organoleptic properties of each variety.

Introduction internode length and petiole length as the parameters to evaluate the morphological variation of cultivated Luffa spp. [28]. Few studies have also been conducted to identify the interspecific relationships of Luffa spp. using their flavonoid patterns [30]. However, the interspecific relationships vary with the employed morphological and chemotaxonomic markers and their highly diverse nature [30,31]. Thus, morphological and chemotaxonomic markers do not provide an accurate description of Luffa varieties [31,6] demanding more consistent and precise techniques to assess the genetic variability among the Luffa varieties.
The molecular systematic approaches with morphological characterization are the most reliable tools to ascertain the accurate intra and interspecies delimitations. Thus, in the present study, we aimed to establish intra and interspecies delimits of L. acutangula and L. aegyptiaca in Sri Lanka and detect the varietal identities within the two species using morphological characterization based on vegetative and reproductive characteristics and molecular systematics. Moreover, an organoleptic assessment was carried out to understand the consumer preference on the varieties assessed in the present study. Overall it was aimed to establish an accurate varietal description of L. acutangula and L. aegyptiaca and to facilitate the breeding of improved Luffa varieties.

Plant establishment
The plants were established in the open fields at Regional Agriculture Research and Development Centre, Makandura (Agro-Ecological Region IL1a of Sri Lanka: average annual rainfall of 1960 mm; maximum and minimum average temperatures 31.7 and 23.0˚C respectively; Red

L. acuangula Asiri
Open-pollinated Sri Lanka [27] Gannoruwa Ari (GA) Open-pollinated Sri Lanka [32] Yellow Podsolic with Alluvial soil as a top layer) [35]. The seeds were soaked overnight in water and subjected to Cruiser Chemical Treatment recommended in the Integrated Pest Management (IPM) Package for controlling the pests of the family Cucurbitaceae. The plants/ varieties were arranged according to RCBD with two replicates each having eight beds. Each bed comprised of six planting pits arranged linearly. The spacing of 1.5 m × 1.5 m was maintained within and between adjacent lines. The pits were prepared each having a volume of 0.1 m 3 filled with cattle manure and topsoil in 1:1 ratio, followed by the addition of basal N-P-K mixture in recommended rates. The contents were mixed in the pit and allowed to settle for two days. After that, three treated seeds were planted in each pit, subsequently thinned to two well-established seedlings per pit. After two weeks of the establishment, the lateral branches were removed up to 0.6 m of the main stem from the soil level to train the plants to climb the trellis.

Morphometric measurements
The  Table lists all the abbreviations). All the morphometric measurements were collected according to International Plant Genetic Resources (IPGR) descriptors [36]. Furthermore, photographs were also taken to illustrate the stages of fruit development of each variety.

DNA sequencing
The DNA barcoding markers rbcL, trnH-psbA and ITS were chosen for sequencing considering the availability of comparison sequences for phylogenetic analysis [3] (S3 Table). The PCR products were purified using QIAquick PCR purification kit (Catalog No: 28104, Qiagen, Hilden, Germany) and cycle sequenced using the automated genetic analyzer 3500 (Catalog number: 622-0010, Applied Bio System).

Data analysis
The quantitative data collected for the reproductive and morphological parameters were subjected to normality testing, and LS-means/pdiff mean separation under General Linear Model (GLM) procedure in the statistical package SAS 9.4 (SAS Institute, NC, Cary, USA). A dendrogram was constructed combining both vegetative and reproductive data (normalized to 0-1 range) using Complete Linkage and Euclidean Distance method in the statistical package Minitab 17 (Minitab Inc., USA). The Principal Components One and Two (PC1 and PC2) calculated from all the quantitative data were used to draw the PC biplot in Minitab.

Phylogenetic analysis
The raw sequencing data yielded for the markers rbcL, trnH-psbA, and ITS were initially visualized in MEGA 7 [40] to detect the initial and end noises. The datasets were trimmed and subjected independently to a Basic Local Alignment Search Tool (BLAST) search to verify the identities of the sequences. The sequences were aligned with the data given in Filipowicz et al. (2014) [3]. For each marker, separate alignments were carried out in MEGA 7 using Clustal W algorithm [41]. The sequences were manually checked to avoid the incorporation of unwanted gaps in the alignment. The three final marker data sets were combined using Sequence Matrix software [42], and the data partition matrix was made. Since chloroplast and nuclear markers were employed in the downstream phylogenetic analyses, the phylogenetic concordance must be assessed for the combined datasets. Thus the combined alignment was subjected to partition homogeneity (ILD test) analysis [43] to check the phylogenetic congruence of three markers. The combined datasets of the Sri Lankan Luffa spp. were analyzed using Unweighted Pair Group Method with Arithmetic mean (UPGMA) algorithm with uncorrected pairwise distances of each sequence. The combined alignment with all the Luffa spp. in the world (adapted from Filipowicz et al. (2014) [3]) were uploaded to PAUP (version 4.0a) [44] to carry out the neighbor-joining (NJ) tree construction. Our approach considered all the substitutions for the tree construction, and the gaps were treated as partial deletions. Maximum Likelihood (ML) tree search was also carried out in RAxML [45] using the rapid bootstrap algorithm [46]. The analysis was run for 1000 iterations, and the DNA model was selected as GTRGAMMA [47]. The bootstrap replicates were included into one tree topology using bipartition option in RAxML. Moreover, a Bayesian analysis was carried out in MrBays [48]. Since the Bayesian tree construction associates with DNA substitution model, the model selection was carried out in Akaike Information Criteria (AIC) [49] and corrected Akaike Information Criteria (AICc) [50] in J model test [51]. Then the best nucleotide parameters were appointed in the Bayesian tree search and ran four Markov Chain Monte Carlo (MCMC) chains for fifty million cycles to probe heuristically in tree space. The trees were probed after every 5000 chain-runs, and the initial 10% of trees probed were discarded as burn-in to achieve the sampling from the independent and sufficient sample size. The analysis was set to draw a 50% majority rule consensus tree as the final output. The Maximum Likelihood, Bayesian tree searches, and model selection were carried out in CIPRES science gateway [52]. The final tree output was further modified in Figtree version 1.4.3 [53].

Evaluation of consumer preference
L. acutangula and L. aegyptiaca fruit samples from each variety were cooked according to the most common recipe for Luffa dishes in Sri Lanka. One kilogram of fresh fruit sample from each variety was separately cooked for five mins in slow heat by adding 50ml coconut milk, 20g of green chili, 20g of red onion and 10g of salt and served for a total of 30 panelists. They ranked the samples for five parameters: color, aroma, texture, bitterness and overall preference using a three-tier scoring system (the highest, medium and lowest levels of choice were indicated by a score of three, two and one, respectively for each parameter). The generated data were subjected to FREQ procedure in SAS to decipher the associations with varietal preferences. For each Luffa variety, a weighted score was calculated by multiplying raw percentages with the associated rank for each taste parameter. Thereby weighted scores were assigned separately for color, aroma, texture, bitterness and overall preference for each variety. The weighted scores were subjected to PCA using Minitab, and the PC biplot between two major PCs (PC1 and PC2), Eigenvalue plot and the Scree plot were obtained to depict the contribution of each organoleptic parameter to the total variance of preference among eight Luffa varieties.

Variation of the vegetative parameters
The vegetative parameters assessed in the present study displayed high variance (Fig 1, Table 2). The two hybrids, Naga-F1 and Nadee-F1, had the significantly highest LL (16.21 cm and 16.09 cm respectively) while LF3522, an L. aegyptiaca variety possessed the significantly lowest LL (9.63 cm) (P<0.001). The L. aegyptiaca varieties had significantly lowest LW. There was no significant difference between the varieties in PL. The NLS was significantly highest in NWYP and NWGP (58.25 and 52.50, respectively) (P<0.001). Significantly lower RDW values were recorded for Asiri, GA, LA33, Naga-F1 and LF3522 (P<0.001). There was no noted difference in the RTL among the eight varieties.

Variation of the reproductive parameters
The significant differences were observed for all the fruit parameters assessed (S1 Table) in the analysis except ST (P<0.001) ( Table 3). NWGP and the NWYP had the significantly highest value for FN (29.77 and 23.94, respectively). Furthermore, the TW of NWGP and NWYP also recorded significantly highest values (6792.27 g and 6081.11 g, respectively). However, the varieties belonging to L. acutangula, the three local varieties (Asiri, GA and LA33) and the two exotic hybrids Naga-F1 and Nadee-F1 had significantly similar FT. The FT of the two L. aegyptiaca varieties were significantly different from each other. The FL was significantly lowest in NWGP, NWYP, and Asiri (Fig 2, S1 Fig). The ST of all the varieties of Luffa spp. indicated no significant differences (Fig 2, S1 Fig). The RN was also significantly similar among the fruits of five L. acutangula varieties while there were no ribs observed in the fruits of L. aegyptiaca varieties (Fig 2, S1 Fig). The HSW was significantly different among all the varieties (P<0.001). There were no significant differences observed among the eight varieties for SL, SW and NP (P<0.001) ( Table 3, Fig 2).
The varieties NWYP, NWGP, and LF3522 of L. aegyptiaca and Asiri and LA33 of L. acutangula had taken the significantly highest number of days for the opening of male flowers (DMF) and highest number of days for the opening of female flowers (NFF) ( Table 4, P<0.001). The two hybrids got the least number of days for the DHV (10-11 days) while all the three varieties of L. aegyptiaca got the significantly highest number of days for DHV (18-19 days) (Table 4). Furthermore, there was no apparent variation between the two species on the NDH. There was no significant difference among the varieties in the parameters; FFN, VLF, and VGF (Table 4).

Morphological diversity structure
The PCA yielded 32 principal components (PCs) (S4 Table) collectively for the vegetative and reproductive parameters. In the PC biplot between PC1 and PC2, the eight varieties were grouped into three discrete clusters (Fig 3). All the varieties belonging to L. acutangula were clustered together, at 29.04% similarity coefficient as revealed in the dendrogram generated using the PCs (Fig 4). However, the three L. aegyptiaca varieties were split into two main clusters where the two land races of L. aegyptiaca were grouped together, and the LF3522 formed a separate cluster (Fig 4).
The dendrogram developed using Complete Linkage, and Euclidean Distance based on the principal components (S4 Table; S2A Fig) showed that all the five varieties of L. acutangula were fallen into a single cluster at 53% morphological similarity coefficient. However, the L. aegyptiaca variety LF3522 can be observed as an out-group although NWYP and NWGP clustered together at 68% morphological similarity coefficient. The two-hybrid varieties of L. acutangula (Nadee-F1 and Naga-F1) showed a close similarity at 71.04% of the morphological similarity coefficient (Fig 4).

Phylogenetic analysis
The PCR products obtained for the markers; trnH-psbA, rbcL, trnS GCU -trnG UUC , atpB-rbcL, matK-trnT, ITS, atpB gene, trnL-trnF, and trnL, are depicted in S3 Fig. Out of these markers,  we sequenced ITS, rbcL and trnH-psbA. The NJ and Bayesian trees had an almost similar topology. Because the NJ tree was more resolved, we presented it as an unrooted tree diagram ( The nucleotide variations in nuclear and plastid regions derived from the sequence alignment of Luffa varieties are depicted in Fig 6. Accordingly, eight distinct haplotypes were identified. ITS was the most informative among the regions we sequenced due to the highest number of variable regions present. The sequence data was sufficient to capture the variance between the two landraces; NWYP and NWGP.

Organoleptic assessment on Luffa fruits as a vegetable
The organoleptic properties evaluated through association analysis indicated a significant association between each variety of Luffa spp. and the taste parameters assessed except bitterness. A significant association was observed between the Luffa variety and color (S4A Fig, χ    The highest level of preference regarding color was recorded for Naga-F1 while the least was recorded for NWYP (S4A Fig). Naga-F1 further recorded a significantly higher preference for aroma. However, LA33 indicated the highest preference for aroma. Nadee-F1, NWGP, and LF3522 received the lowest preference for their aroma (S4B Fig). Furthermore, Naga-F1 had the highest preference level for the texture compared to the other varieties. NWGP recorded the least preference regarding the texture (S4C Fig). Asiri had the highest level of overall preference whereas the NWGP had the lowest overall preference (S4E Fig). The PC biplot constructed based on the PCA for organoleptic parameters indicated the highest preference values for Naga-F1, Asiri, LA33, and GA. The least preferences were indicated for Nadee-F1 and the three L. aegyptiaca varieties; NWYP, NWGP, and LF3522. The PC biplot also divided the two species into clusters where all the L. acutangula varieties were clustered separately from the L. aegyptiaca varieties that were clustered together (Fig 7A). The loading plot constructed for the organoleptic parameters depicts that the overall preference for Luffa varieties was most closely associated with the aroma. The overall preference was not dependent on the bitterness of each of the variety (Fig 7B; S4D Fig).

Variation of the vegetative parameters
The statistical analysis of the leaf traits indicated that LL and LW are the decisive traits that can be used to distinguish the two species (Fig 1). The mean LL and LW of the three L. aegyptiaca varieties; NWYP, NWGP, and LF3522 were markedly different from the varieties of the L. acutangula and showed comparatively lower values for the two leaf traits (P<0.001). Thus, we could identify that the significantly lower LL and LW of L. aegyptiaca varieties can be employed as morphological parameters in distinguishing the Niyan Watakolu varieties from the cultivated varieties of L. acutangula. Further, analysis on the leaf morphology and leaf texture indicated that the L. acutangula has smoothly textured adaxial and abaxial surfaces which are shallowly lobed. The L. aegyptiaca had contrasting features where the leaves were rough in  Diversity analysis of Luffa acutangula and Luffa aegyptiaca in Sri Lanka texture with deep and prominent lobes (Fig 1). These features are consistent with the prior records on the Indian Luffa varieties [4]. The NLS of the two Niyan Watakolu varieties were significantly different from all the other varieties further suggesting their uniqueness from the other varieties of Luffa spp. However, INL and the two root traits (RTL and RDW) cannot be employed as vegetative parameters for species delimitation. L. aegyptiaca varieties, NWYP and NWGP were significantly similar to each other in all the vegetative parameters reported in the present study. However, the two varieties did not indicate a clear difference from the other varieties in terms of the vegetative parameters. The varieties of L. aegyptiaca shared significant similarities with the varieties of L. acutangula for the majority of the vegetative traits such as PL, INL, RTL, and RDW. Thereby, statistical analysis of the vegetative parameters revealed that only the vegetative parameters LL and LW could be employed in the identification of the NWGP and NWYP from the other varieties of Luffa spp. (Table 2) inferring that the vegetative parameters do not provide sufficient evidence to reveal the species delimits and varietal identification of Luffa spp.

Variation of the reproductive parameters
We identified that NWGP and NWYP recorded a significant difference in the fruit traits FN, FW and TW in comparison to all other Luffa varieties. The reproductive parameters can be successfully utilized in distinguishing of two L. aegyptiaca varieties; NWGP and NWYP from the other Luffa varieties. Significantly similar values obtained for ST suggest that it cannot be used as a fruit trait in the identification of NWGP and NWYP. The relatively similar ST values for each variety is indicated in (S1A-S1E Fig) by the cross sections of each variety of Luffa spp. However, none of the seed traits such as SL, SW, SET, NP, and HSW can be employed for the identification of NWGP and NWYP from the other varieties. The reproductive parameters at the flowering and harvesting stages also indicate certain traits that are of relative significance regarding the species identification ( Table 4). The parameters; DMF, NFF, DHV, and NDH, are relatively higher for the L. aegyptiaca varieties than the L. acutangula varieties which are relatively disadvantageous in terms of improving the crop potential of NWGP and NWYP. Nevertheless, the other reproductive stage parameters; MFR, NFF, VLF, and VGF cannot be exploited in setting up species delimits as there is no significant difference between the two species. The significant difference in PDL for the two varieties, NWGP and NWYP, allows the successful differentiation between them. The overall analysis of the reproductive parameters indicated that out of the fruit, seed, flowering and harvesting stage traits, only FN, FW, and TW can be used in the identification of NWGP and NWYP from the other varieties, highlighting the limited applicability of morphological parameters.

Combined analysis of vegetative and reproductive parameters
The vegetative and reproductive parameters did not provide the resolution for the species delimitation. Thus, we further extended our study focusing on both the vegetative and reproductive data collectively for better identification of inter and intra-species variation. In PCA with all the morphometric data followed by clustering, a clear separation was evident between the two species and the grouping of NWGP and NWYP separately from LF3522. Our results conform with the previous findings [4,5].

Phylogenetic analysis
The unrooted NJ tree revealed the nesting of Luffa sequences generated in the present study into well-separated clades of L. acutangula and L. aegyptiaca thus confirming the genetic identity of the two species. Although, L. aegyptiaca is previously named as L. cylindrica in Dassanayake and Fosberg (1988) [10], here we synonymize this species as L. aegyptiaca [3]. Similarly, in a study conducted by An et al. (2017) [9], the cladding of L. aegyptiaca and L. acutangula into two well defined clusters was evident. However, the three clades appeared within the L. aegyptiaca cluster revealed the genetic divergence of our L. aegyptiaca varieties with respect to a set of different accessions of L. aegyptiaca in different geographical regions in the world, as previously reported by Filipowicz et al. (2014) [3]. Thus, it is possible that Sri Lankan L. aegyptiaca could be a subspecies of L. aegyptiaca, which could be a unique genetic form naturalized in Sri Lanka. It is interesting to see the existence of a reciprocal monophyletic group within L. aegyptiaca clade (with the node support values of, bs = 92 and PP = 99). The L. aegyptiaca in Sri Lanka is known to be the landrace and an underutilized crop for which hybridization attempts have been implemented recently by Regional Agriculture Research and Development Center, Sri Lanka. It is also visible that the difference exists between landraces and the hybrid of L. aegyptiaca in vegetative and reproductive parameters. The similar pattern is also observed in our phylogenetic analysis (Fig 5 and Fig 6) indicating that the landrace could be a unique genetic form. The LF3522 had a slight divergence from L. aegyptiaca landraces mainly because of the hybridization of exotic varieties with NWGP and NWYP. LF3522 was clustered separately from NWGP and NWYP in the combined morphological and phylogenetic diversity structures. In Fig 3, LF3522 got separately clustered from NWGP/NWYP indicating the exotic origin. However, the exact origin of LF3522 is not available. Due to the reciprocal monophyly observed within the NJ tree (Fig 5), it is possible that LF3522 could have been originated due to a hybridization event between an exotic parent and a Sri Lankan variety of L. aegyptiaca.
We selected three barcoding markers, ITS, rbcL, and trnH-psbA to assess the two species of Luffa, due to the ease of amplification, clear banding pattern (S3 Fig) and the polymorphism reported in literature [3,54]. According to the SNP/INDEL profile, we identified trnH-psbA as the most reliable and desirable DNA barcode due to its high inter-specific sequence variation and non-ambiguous nature, further verifying high efficacy of trnH-psbA in the identification of species delimits which was also reported in Kress and Erickson (2007) [55].

Organoleptic assessment
The association analysis of the organoleptic properties reveals that the NWGP and NWYP are the least preferred choices with respect to color, aroma, texture and overall preference (S4A, S4B, S4D and S4E Fig; respectively). Higher preference for the variety Asiri and the two hybrids Naga-F1 and Nadee-F1 suggests that the breeding protocols have played a key role in enhancing the organoleptic properties. We suggest that these varieties and the exotic hybrids have higher consumer preference due to the varietal improvements through breeding. The studies show that the consumers in India prefer both types of Luffa spp. in their immature forms regardless of the fruit size [56]. The organoleptic properties are more varied in terms of the culinary patterns of the different regions of the world. The Sri Lankan perspective with reference to the L. aegyptiaca varieties is contrasting compared to the organoleptic preferences of the other regions of the world. For instance, in Vietnam, L. aegyptiaca is highly preferred by the consumers for its inherent aroma, and thus cultivars with enhanced aroma and taste are being bred [57]. Similar approaches can be utilized in the improvement of the NWGP and NWYP varieties by cross-pollination with the varieties with the range of aroma and taste preferred by the Sri Lankan consumers. The PC biplot created for the variation of the organoleptic parameters based on PCA analysis led to the close clustering of L. aegyptiaca varieties together which received lower preference scores compared to the varieties of L. acutangula (Fig 7A, S2B  Fig, S5 table). The L. acutangula varieties got dispersedly positioned indicating that Asiri, GA, LA33, Naga-F1, and Nadee-F1 are preferred highly compared to NWYP, NWGP, and LF3522 ( Fig 7A). The bitterness was not significantly associated with Luffa whereas all other four parameters were significantly associated. The Scree plot also depicts that the bitterness has a meager influence on the overall taste. The aroma is almost overlapped with overall taste with its position indicating that the aroma mostly determines the overall preference. The diversity structure created for organoleptic parameters demonstrates the need of improving L. acutangula varieties Nadee-F1 and GA, and all the varieties of L. aegyptiaca for better consumer preference. Within L. acutangula, it is possible to make segregating populations between highly preferred and low preferred varieties as the parents to create segregating populations to detect the QTL for marker assisted breeding.

Avenues for improving the crop potential
Despite the favorable traits that support the crop potential of NWGP and NWYP, we identified several reproductive traits that are undesirable. The comparatively higher DMF, DFF, DHV, and DHF values are relatively disadvantageous in their utilization as a crop species. Thus, the enhancement of the crop potential of NWGP and NWYP requires the shortening of the DFF, DMF, DHV, and DHF. Therefore, we suggest the implementation of breeding protocols for the NWGP and NWYP varieties for shorter harvesting periods. Furthermore, the outcomes of the organoleptic evaluation indicate a barrier for the utilization of the L. aegyptiaca as a crop. The higher preference values for the varieties and hybrids of L. acutangula suggest that the selection and breeding for varietal improvement has played a major role in enhancing the organoleptic properties. The current Sri Lankan varieties including GA are bred for higher yields [27]. However, the L. aegyptiaca varieties are not currently subjected to any varietal improvement procedures. Therefore, the organoleptic assessment results stress the importance of introgressing the preferred organoleptic traits to NWGP and NWYP. Similar studies have been conducted in other countries such as Vietnam to improve aroma. The breeding programs have been implemented in Vietnam to breed for L. aegyptiaca hybrids with subsequent selection for preferred levels of taste and aroma [55]. Similar approaches can be utilized in the improvement of the NWGP and NWYP varieties by cross pollinating with the varieties with the aroma and taste levels preferred by the Sri Lankan consumers.

Conclusions
The combined data of vegetative and reproductive parameters classifies L. acutangula varieties into two distinct clusters at 29.04% of morphological similarity coefficient. The sequence polymorphism of trnH-psbA establishes the species delimits of L. acutangula and L. aegyptiaca where L. acutangula varieties have the GATTTT haplotype whereas L. aegyptiaca has the TGCA haplotype. The sequence polymorphism in rbcL establishes the varietal identification of L. aegyptiaca whereas ITS polymorphism establishes the varietal identities of L. acutangula. The phylogenetic analysis infers that the cultivated germplasm of L. acutangula forms a separate clade within the worldwide germplasm. The L. aegyptiaca varieties form a reciprocal monophyletic group with respect to other L. aegyptiaca germplasms found elsewhere in the world. L. aegyptiaca varieties studied in the present study could be identified as distinct genetic forms. The organoleptic assessment reveals that the aroma, texture, color, and overall preference are significantly different among the Luffa varieties assessed. Moreover, L. aegyptiaca varieties receive lower preference scores. The organoleptic parameters also differentiate two species where L. aegyptiaca varieties get tightly clustered. Therefore, the present study sets the species delimits, and the varietal identities based on the phylogenetic analysis and also shows the distinct morphological and organoleptic properties.