Analysis of genetic variability and agronomic performance of Indian lettuce (Lactuca indica L.)

Lactuca indica L. is an undomesticated medicinal crop in the Asteraceae family. The study was carried out to identify elite genotypes for lettuce cultivation and breeding improvement. Data was recorded for 19 morphological and developmental traits across 38 accessions (Acc). The genotypic mean square variance was significant for all characters. The higher extent of genotypic and phenotypic coefficient of variation were obtained for basal branch, leaf blade width, and node number. The broad-sense heritability (H2B) ranged from 45.85% (seed length) to 98.59% (node number), whereas genetic advance as a percentage of the mean (GAM%) ranged from 9.33 to 191. Vegetative characters such as node number, plant height, basal branches were conjugated with high H2B and high GAM% indicating additive gene effect and selection of these traits based on phenotypic observation is effective for better gain. Reproductive traits, including bolting duration, flowering duration, and seed weight were linked with high H2B, and moderate GAM% revealing that these traits are amenable to genetic improvement, these traits also showed a significant and high positive correlation. Acc 55 and 8 showed the best performance for the majority of the attributes could be good material for further research and breeding. In the Wards’ phylogenetic tree of morphological traits, accessions were clustered based on their phenotypic characters rather than the geographic origin.


Introduction
Indian lettuce (Lactuca indica L.) is an important oriental medicinal species of the family Asteraceae. The plant is an annual or perennial, erect, lactiferous herb, with radical rosette when young, up to 2 m tall when flowering. Self-fertilization is common in L. indica, although insects visit the inflorescences, and the possibility of cross-pollination does occur (Thompson 1941). It is diploid with 2n = 29 = 18 chromosomes and was originated in Africa (South Africa, Mozambique, Madagascar, Mauritius, Seychelles) and widespread in Asian countries probably by Chinese immigrants and relatively common to Indonesia, Malaysia, China, Taiwan, Japan, and Korea (Jeffrey 1966;Oliya et al. 2018Oliya et al. , 2021. Moreover, it grows in East Himalaya (Darjeeling, Arunachal Pradesh), Tibetan plateau, Assam, Burma, North Asia (Siberia, Russia, Far East), and South East Asia (Ohwi 1984).
The common means of propagation is by seed, which germinates 3-4 days after sowing. Seed is usually sown in soil-bed at the end of April to early May and then transplanted in the field after 3-4 weeks. It can be cultivated from the lowlands up to 2000 m altitude. It is used traditionally in the form of soup and salad and homemade medicine to treat fever, cough, cold, diabetes and stomach disorder (Lin and Kan 1990). Many studies proved that this plant is an essential source in pharmaceutical and health functional foods because of the anti-oxidant, anti-inflammatory, cytotoxic, hepatoprotective, anti-bacterial, anti-diabetic, digestive, diuretic, necrotic, and sedative properties of the plant (Bown 1995;Wang et al. 2003;Kim et al. 2007;Choi et al. 2016;Oliya et al. 2021). The plant is rich in nutritional compositions, phenolic compound, flavonoids, and sesquiterpene lactones (Michalska et al. 2009;Kim et al. 2012;Ha et al. 2017;Oliya et al. 2021). Moreover, the properties such as enrichment of biotic and abiotic stressresponsive genes, wide molecular base and resistance to downey mildew are reported for this plant (Van Treuren et al. 2013;Ha et al. 2017;Oliya et al. 2018).
Among the various species in the genus Lactuca, L. sativa is the only cultivated. The success of domestication of lettuce depends on the amount of latex content, low or absence of leaf and stem spines, bigger seed size, non-shuttering seed, shortening of internodes, bunching of leaves and selection for late bolting Mou 2011). Breeders have been interested in searching the trait of high economic value in wild crops to improve cultivated ones (Dempewolf et al. 2017). This crop improvement approach has much importance because it improves the cultivated crops and studies the possibility of cultivation of wild crops for many aspects. Better knowledge of phenotypic variation, genetic variability, and diversity among the available germplasm and genotype is necessary for developing new cultivar with high yield potential and better adaptation to the biotic and abiotic stress. Therefore, investigating morphological characters on wild lettuce is important for lettuce breeding (Chowdhury et al. 2002;Bhattarai et al. 2016;Dempewolf et al. 2017;Zhang et al. 2018). To the best of our knowledge, very little information is reported for morphological characters, and no studies have been made for genetic variability and agronomic analysis using morphological traits for L. indica. Therefore, the objective of the present study is to investigate morphological characteristics for analyzing genetic variability and agronomic performance across L. indica accessions. In addition, this study also aims to elucidate genetic relationships among 38 accessions collected across Korea using morphological data.

Plant materials
This study uses the same plant material used in the earlier study by Oliya et al. (2018). However, the analysis of morphological traits requires many replications from each accession to achieve a quality result. In an earlier study, we used 73 accessions from 41 collection sites; nevertheless, for this study, we only used 38 accessions covering 38 collections (Table S2-1). We excluded the remaining 29 accessions because of the unavailability of replications (because of germination failure) and three accessions because of failure of the reproductive phase. The studied accessions have different areas of origin characterized by different geographical and ecological characteristics (Table S1). The field experiment was conducted in Seoul National University experimental farm, Suwon, the Republic of Korea (N 37°16 0 12.094 00 , E 126°59 0 20.756 00 ) in 2016. A randomized complete block design with three replicates was used. In each replication (block), four competitive plants were randomly selected and tagged.

Data collection
Visual observation for leaf shape variation, (incision of lobe) was performed at the flowering time by following the standard evaluation system for wild lettuce (Dolezalova et al. 2002). Individuals were categorized, and other notable variations such as anthocyanin pigmentation, leaf margin, and inflorescence were observed. For the quantitative traits, a total of nineteen characteristics were recorded as follows: at fifty percentage flowering stage, leaf length (LL), leaf tip length (LTL), leaf blade width (LBW), leaf lobe number (LOB); at harvest maturity, basal branching (BB), total branching (TB), plant height (PH) basal stem diameter (BSD), total node number (TNN) basal internode distance (BIND), median internode distance (MIND); developmental stage: seeding to bolting duration (BOLT), seeding to first flowering time (FFD), seeding to first fruit set duration (FSD), seeding to fifty percent seed maturity duration (FMD); seed per capitulum (head) (SPC), seed length (SL), pappus length (PL) and thousand seed weight (TSW).

Data analysis
The data collected for nineteen quantitative characters were subjected to analysis of variance (ANOVA) using the statistical software R release R-3.5.1 and ANOVA models were fit using the aov() function (Chambers 1992). The phenotypic variance (PV) and genotypic variance (GV) were estimated based on the formula given by Syukur et al. (2012). The phenotypic (PCV) and genotypic coefficient of variation (GCV) were computed by formula suggested by Burton (1952). The coefficient of variation was determined as an indicator of variability. Heritability in a broad sense (H 2 B) was calculated using the formula given by Allard (1960). Similarly, genetic advance (GA) was determined as described by (Johnson et al. 1955) and the genetic advance as a percentage of the mean (GAM), also called genetic gain was calculated as described by Johnson et al. (1955). The formula used for calculation is given in Supplementary Table1. For visualization of phenotypic distribution, box and whisker plot was performed using SPSS version 23 (Armonk 2015) from which the outlier can be taken as the essential accession for a particular trait. For the attribute which did not show the outlier, the accession with maximum or minimum value (based on the economic importance of each attribute) plotted near the whisker in its approximate value scale. Again, to quantify accessions that fit the maximum traits, the quartile distribution was performed in MS-excel, from which few accessions throughout the range of characteristics can be screened. The simple correlation coefficients between the characters were calculated using the Pearson correlation coefficient using SPSS version 23 (Armonk 2015).
The principal component analysis (PCA) for morphological attributes was performed using an MSexcel add-in statistical package XLSTAT version 2018.2 and higher (Addinsoft, 2018). Scatter plots of the first two principal components (PCs) were estimated using the same software. The phylogenetic tree was computed using the Ward method based on the distance between the individuals constructed by Euclidean coefficients (Ward 1963).

Morphological characterization
Out of total 38 accessions, ten accessions (acc 16, 19, 32, 46, 58, 59, 64, 69, and 70 were heterozygous within the accessions for the leaf shape morphology, and other accessions were homozygous (Table S1, Fig. S1). Individuals were categorized following the leaf type category proposed by Dolezalova et al. (2002) for wild lettuce. In our samples, lobed leaf (96.93%) was dominating over entire (3.07%). Four leaf groups pinnatifid (73.45%), pinnatisect (17.41%), pinnate lobed (3.07%) and pinnate part (2.63) were observed (Fig. 1). Individuals showed entire, dentate, and undulating with slight dentate margin. Among them, undulating with slight dentate type was the most common. Significant variability was recorded in anthocyanin pigmentation in rosette leaves and stems, but such variation was less during the flowering stage. Except individuals belonging to three accessions (acc 32, 51, 70) other showed dense anthocyanin distribution in the involucre bract at the full flowering and seed setting stage, where two patterns, namely spot and stripes, existed. Two types of inflorescences, pyramidal panicles (in most individuals) and compound corymbs panicles were observed only in three individuals belonging to acc 8.

Mean performance of Agro-morphological traits
The individual showed a wide range of variation in all the studied traits (Table 1). At the flowering stage, LL ranged from 9.53 to 31.00 cm, LTL ranged from 0.13 to 21.07 cm, LBW ranged from 0.52 to 5.67 cm, and LOB from 0.00 to 6.33 cm, respectively. Similarly, at the harvest maturity, the mean value for TB, BB and TNN were 28.68 cm, 1.61 cm and 72.62 cm, respectively. The PH ranged from 94.33 cm (acc 32) to 190.28 cm (acc 9) with mean value of 143.50 cm. The mean value for BIND and MIND were 1.53 cm and 3.28 cm, respectively. The minimum duration for BOLT was 87 days (acc 5) and the maximum duration was 122.14 days (acc 35) with a mean value of 108.87 days. The duration for FF ranged from 105.33 (acc 27) to 141.42 (acc 15) days with mean value of 125.33 days. Similarly, the SPC ranged from 14.40 (acc 36) to 28.53 (acc 36) with mean value of 22.1 and the TSW ranged from 0.63 (acc 2) to 2.38 g (acc 18) with mean value of 1.12 g.

Analysis of variance and variability component
The individuals showed a considerable degree of variation in all traits studied (Table 1). In ANOVA analysis, the genotypic mean square variance for all traits was significant. Among the nineteen attributes studied, sixteen were significant at P = 0.001, two characteristics, PL, and TSW were significant at P = 0.01, and only one trait, SL, was significant at P = 0.05 (Table 1). This high morphological variation with a considerable difference between them revealed substantial genetic variability in our accessions. To compare the variation among morphological traits, the variability components, PV, GV, PCV, GCV, H 2 B, GA and GAM% estimated for 19 quantitative characters were presented in Table 2. The highest and lowest GV and PV was obtained for TNN and TSW, respectively. GCV ranged from 6.11 to 99.12%, and PCV ranged from 7.36 to 105.77%. In general, PCV was found higher than GCV for all traits. BB showed the highest value for GCV and PCV; however, PL and BOLT resembled the lowest value. According to Deshmukh et al. (1986), the estimate of GCV and PCV values greater than 20% is high and values between 10 and 20% to be medium, whereas values less than 10% are considered as low. In the present study, lower GCV and PCV values were recorded for BOLT, FFD, FSD, FMD, and PL, indicating the influence of the environment on trait expression. Other traits were moderate to high for GCV and PCV, indicating less environmental influence for expressing these traits. The highest and lower genotypic and phenotypic variation was obtained for TNN and TSW, respectively. GCV ranged from 6.11 to 99.12%, and PCV ranged from 7.36 to 105.77%. According to Dabholkar (1999), heritability estimates can be low, ranging from 5 to 30%; medium ranging from 30 to 60%; and high ranging from 60% to above. From this point of view, the broad-sense heritability estimates were medium for SL, PL, TSW, and high for other traits, with TNN (98.59%) showing the highest and SL (45.85%) delivering lowest value. The genetic advance ranged from 0.08 (PL) to 80.84 (TNN). Genetic advance as a percentage of the mean (GAM %) ranged from 9.33% (PL) to 191.35% (BB). The GAM% can be categorized according to (Johnson et al. 1955) as low (\ 10%), moderate (10-20%) and high ([ 20%). In this context, GAM% was low for SL and PL, moderate for BOLT, FFD, FSD, FMD, SL, and high for other traits. The top three genetic gains were obtained for BB (191.35%) followed by LBW (117.04%) and TNN (191.35%) ( Table 2). For obtaining the idea about expected genetic gain in nextgeneration, heritability has to be considered in conjugation with genetic gain/genetic advance as a percentage of the mean (Johnson et al. 1955). Traits such as LL, LTL, LBW, LOB, TB, PH, BSD, TNN, BB, BIND, and MIND showed high heritability associated with increased genetic gain indicating the additive gene effects. These traits based on phenotypic observation are effective for better gain in breeding. For other traits (BOLT, FFD, FSD, FMD, and SPC) showed high heritability was in conjugation with moderate genetic gain indicating the effect of nonadditive gene action; therefore, anticipated for improved productivity via hybridization, heterosis breeding, family selection and progeny testing methods. Similarly, SL and TSW with moderate heritability was associated with average genetic gain and PL with moderate heritability was linked with low genetic gain ( Table 2).

Quantification of potential accessions
Box and whisker plots were used to visualize distributions of values for each trait across the accessions. The line within the box represents the median, the box range includes the second and third quartile, and the whiskers represent the minimum-maximum values. The outliers are described in the hallowed circle and the star. The Box-whisker plot of agronomic performance showed the distribution of accessions within the interquartile range for most traits; however, some accessions distributed in the form of outliers (Fig. 2). Moreover, to quantify the best performing accessions throughout the range of traits, we plotted them into the quartile distribution, and quartile member sets were identified based on the economic importance of agronomic traits. Higher values were expected for all traits except LTL, LOB, BIND, and MIND, where a lower value is considered best for more significant LL leaf length, LTL leaf tip length, LBW leaf blade width, LOB leaf lobe number, BB basal branch, TB total branch, PH plant height, BSD basal stem diameter, TNN total node number, BIND basal internode distance, MIND median internode distance, BOLT duration to bolting, FFD duration to first flowering, FSD duration to first seed set, FMD duration to fifty percent seed maturity, SPC seed per capatilium, PL pappus length, SL seed length, TWS thousand seed weight, GV genotypic variance, PV phenotypic variance, GCV genotypic coefficient of variance, PCV phenotypic coefficient of variance, H 2 B: broad sense heritability, GA genetic advance, GAM genetic advance as percentage of mean economic gain. Thus, the fourth quartile for high measuring traits and first quartile for low measuring traits were taken. With this assumption, two accessions acc. 8 and 55 were found superior whose overall trait value in the fourth quartile were 12 and 11, respectively (Table S2). These accessions could be good breeding material (Fig. 2, 3).

Correlation analysis
The Pearson correlation coefficients between traits are presented in Table 3. The correlation r values are categorized as weak (0.0-0.4), moderate (0.4-0.6) and strong (0.6-1.0) (Belsley et al. 2005). In the present study, LL showed moderate and positive correlation however, these traits showed no correlation with other traits. PL showed low positive correlation with TB and SL (correlation of PL with TB and SL = 0.389 and 0.383 at P = 0.05) and TSW showed no correlation with any of the traits studied (Table 3).

Principal component analysis (PCA)
In the PCA, 64% of total phenotypic variation was explained by first five components (  thousand seed weight *Correlation is significant at 0.05 and **Correlation is significant at 0.01 level covered by PC1 (4.20) in decreasing order followed by PC2, PC3, PC4, and PC5, respectively (Table 4).

Cluster analysis
Clustering analysis based on morphological traits revealed four distinct clusters. Euclidean distances of 160 were threshold values for grouping those accessions in different clusters. Cluster I contain 14 accessions among these accessions five of the acc 8, 12, 36, 37 and 70 also clustered in quartile (Q4) based on the traits TNN, FF, and FSD values. The other four accessions (4, 7, 11, 51) had LOB in Q1, PH in Q2, and TSW in Q3. Some of the accessions 15, 16, 32, 39 and 58 had TNN in Q2 except acc 39 and BOLT in Q3 except acc 58 was sub grouped. Cluster II comprises 13 accessions with FMD in Q1 except acc 65, LBW in Q3 except acc 6, BOLT, and FF in Q1 and Q2 were closed. Cluster III contains 10 accessions having values Q1 for PH, NN, LTL, TB, FF, FSD were grouped together. Cluster IV contains only one accession (acc 55) isolated in the form of an outlier had shown better performance for 11 traits with higher economic value out of 19 traits. Within cluster accessions were represented by unique color while accessions with two different colors indicating admixed forms from different provinces. The clustering pattern revealed geographic location was not basis for clustering (Fig. 4).

Discussions
L. indica is one of the potential wild lettuce with high nutraceutical, therapeutic and breeding potential. The assessment of genetic diversity and agronomic performance is essential which help to investigate its valuable traits, genetic structure, and better breeding material for L. indica production and commercialization. This study investigated ample variability among the studied accessions using morphological (both qualitative and quantitative) characters. Among all attributes, leaf shape variation was most distinct. Leaf shape variation plays a crucial role in shaping the plants functioning and taxonomical identification (Lindqvist 1958;Wright et al. 2004Wright et al. , 2005Niinemets 2015). Previously, high level of genetic diversity using genic-SSR marker data have been reported (Oliya  (Ha et al., 2017;Oliya et al. 2021). Morphological variability has been reported in other Lactuca species, including L. sativa, Lactuca serriola L., Lactuca saligna L., Lactucacanadensis L., Lactuca aculeata Boiss. et Ky (Dolezalova et al. 2002;Beharav et al. 2008Beharav et al. , 2010Lebeda et al. 2009Lebeda et al. , 2010Ogbodo et al. 2010).
In lettuce breeding, the high variation in the duration to flowering has economic importance because it produces a high concentration of sesquiterpene lactone (a major class of natural bitter compounds) in leaf at flowering time (Ha et al. 2017;Oliya Fig. 4 Phylogenetic relationship between 38 accessions of L. indica. A: Wards' phylogenetic tree using Euclidian distance matrix based on 19 morphological traits et al. 2021). So early flowering accessions can help speed up generation time in genetic, phytomedicinal, and growth studies (Choi et al. 2016). In contrast, lateflowering accessions are useful for leaf harvesting for vegetable purposes Thus, knowing the appropriate breeding material and the effect of genetic and environmental factor in expression of such trait is crucial. Finding important traits using genomic tools is costly for whole accessions. In this regard, the phenotypic study helps to identify the peculiar accessions for further studies (van Nocker and Gardiner 2014). The potential donor accessions screened in this study could be helpful in parental selection for breeding, introgression of the gene to cultivated lettuce, gene diversity, and marker-trait association studies (D'Andrea et al. 2008;Uwimana et al. 2012;Lebeda et al. 2012;Zhang et al. 2017). Significant marker-trait associations and genes responsible for anthocyanin distribution and flowering control in lettuce have been studied (Ryder and Milligan 2005;Kwon et al. 2013;Zhang et al. 2017;Han et al. 2021).
In a population, observed variation is caused by genetic and environmental factors. However, only genetic variation is inherited. High heritability indicates that a large proportion of phenotypic variance is attributed to genotypic variance, and reliable selection could be made for these traits on the basis of phenotypic variation. Nevertheless, estimated heritability itself alone is not very much useful because it includes the effect of both additive and non-additive genes. Therefore, estimation of heritability and genetic advance together could provide the best image of the amount of advancement to be expected through phenotypic selection (Johnson et al. 1955). Further, genetic advance in percentage of mean give more precise result in comparison to only genetic advance. In the present study, leaf blade width, total and basal branch, node number, plant height, basal and median internode distance, seed per capatilium and thousand seed weight were associated with high heritability and high genetic advance as percentage mean (GAM%). This indicates that observed traits have additive gene effects, and these traits based on phenotypic observation are effective for better gain in lettuce breeding improvement studies. High heritability coupled with high genetic advance as per cent of mean was recorded for leaves per plant, leaf yield per plant, vitamin C and carotenoids in cultivated lettuce (Lactuca sativa L.) by Gupta et al. (2008); plant height, number of branches per plant, and flower yield per plant in F2 segregating population of cross Arka Archana 9 AAC-1 in China Aster (Callistephus chinensis (L.) Nees (Harishkumar et al. 2017), 1000 seed weight in Cabbage (Kumar 2004) and plant height in Soybean (Glycine max (L.) Merrill) (Jain et al. 2018). In the present study, other characters viz. the duration for bolting, first flowering, seed set, and seed maturity having low genotypic and phenotypic coefficient of variation, high heritability linked with moderate GAM% and dispersed in the outlier in the box plot, indicating the effect of nonadditive gene action and therefore anticipated for improved productivity via hybridization, heterosis breeding, family selection and progeny testing methods (Acquaah 2009). High heritability coupled with moderate genetic advance as percent of mean was observed for days to 50% flowering, and days to maturity in rice (Bandi et al. 2018). Moreover, these reproductive traits showed strong, positive and highly significant correlation to each other. Therefore, plant selection based on any of these component is effective for the improvement of early or late flowering problem in lettuce. Significant positive relation of flowering time and seed maturity is reported in Soybean (Abugalieva et al. 2016).
The method of clustering using morphological traits showed the clustering of accessions with particular morphological character rather than the geographic distribution which could be attributed to the effect of gene flow in the population, gene reshuffling, environmental heterogeneity, habitat destruction, rapid climate change and variation in secondary metabolites (Fitzpatrick et al. 2019;Hoffmann et al 2021;Oliya et al 2021). In the previous finding with the same genotype by Oliya et al. (2018), clustering based on SSR polymorphism had differed with clustering based on morphological data in this finding; however, genetic data was more robust to separate accessions based on their geographic origin. This could be attributed to the influence of plant growth and development by environmental and climatic conditions and the morphological characteristics are associated with a relatively small number of specific gene loci as reported for other crops (Seyedimoradi et al. 2012;Kumar et al. 2013;Jain et al. 2017). The mode of inheritance in the numerous difference between shape and size as well as architectural arrangement and orientation (also called discontinuous variation) is governed by one or two gene loci. In contrast, the classical component of agricultural yield such as plant height, dimension, weight, and number generally exhibit the continuous variation in their expression and usually governed by multiple gene system (Gottlieb 1984).
In conclusion, the high level of variability and diversity on agro-morphological traits and influence of genetic factors on trait expression appeal to researcher for the efficient utilization and improvement of L. indica in the lettuce breeding program. Moreover, the best accessions screened in this study could help to certified L. indica cultivar production and commercialization. Thus, L. indica an undomesticated medicinal crop could be the most viable option to address the issues raised on lettuce breeding, community health, and fulfillment of natural drug demand.