Morphometric characterization of Jatropha curcas germplasm of North-East India

The morphological variation among Jatropha curcas L. populations from 29 different locations of NorthEast India was determined. Four populations from other parts of India were also incorporated in the study as an out-group. The morphological trait based analysis of J. curcas revealed large variation of quantitative traits among the populations. Of the six morphological traits used in this study, the highest variation (% coefficient of variation=23.19) was observed in floral sex ratio (M: F) while canopy spread (CS) appeared as the least variable trait (% coefficient of variation=1.67). Based on morphometric trait values, the three populations from Assam (IITJC15, IITJC24 and IITJC28) and one population from Arunachal Pradesh (IITJC7) emerged as superior when compared with the other populations. Both cluster and principal component analyses depicted that the populations IITJC19, IITJC21, IITJC22 and IITJC24 from Assam maintained maximum inter-cluster distance from the rest of the populations and are, thus, substantially distinct. The analyses also depicted that no clear demarcation can be made between populations from North-Eastern India and other areas on the basis of morphometric variability alone. Morphometric characterization of J. curcas populations leads to the identification of seven promising populations (IITJC7, IITJC15, IITJC19, IITJC21, IITJC22, IITJC24 and IITJC28) from North-East India which can substantially contribute to Jatropha breeding in the future. From the study of variance components and broad sense heritability, it was suggested that the selection of elite plants on the basis of M: F ratio, 100 seed weight (100SW) and total seed yield (TSY) in tree improvement programs is likely to be more successful.

. The fact that J. curcas seed oil upon transesterification can provide better quality biodiesel with high cetane number when compared with other oilseed plants has created a surge of interest in this plant.J. curcas plant is a native of Mesoamerica, however, it has been distributed throughout the arid, semi-arid, tropical and subtropical regions of the world (Augustus et al., 2002).J. curcas was introduced to India by Portuguese seafarers during the sixteenth century (Sunil et al., 2008).In India, J. curcas has acclimatized itself in diverse eco-geographical zones with different edaphoclimatic conditions and, consequently, has over time amassed variability within the germplasm (Bhatt et al., 2014).
Evaluation of morphological traits of J. curcas provide ample information of the genetic diversity of J. curcas population (Divakara et al., 2010;Kaushik et al., 2007;Rao et al., 2008).Studies focussing on agronomic traits, qualitative traits, reproductive traits and genetic variability are of utmost significance for a thorough investigation of extent of phenetic diversity in germplasm and for utilizing the genetic information for crossing and breeding programmes (Franco et al., 2001;Montes and Melchinger, 2016).For successful crop enhancement and commercial exploitation of the biofuel plant, J. curcas improvement programs should aim at agronomic traits like low male to female flower ratio, high seed yield and oil content, abiotic and biotic stress resistance, and high natural ramification of branches with greater canopy spread.J. curcas crop improvement programs are largely dependent on the assessment of variability in wild sources and selection of superior genotypes (Divakara et al., 2010).Presently, J. curcas varieties with elite traits which can be grown in varied conditions in different parts of the world are not available for the growers, which makes this crop a risky business (Jongschaap et al., 2007;Moniruzzaman et al., 2016).Thus, assessment of trait based variability in J. curcas is a prerequisite for the screening and selection of agronomically elite genotypes which can later be exploited in breeding programs.To date, phenotypic diversity studies in J. curcas have mainly focused on the assessment of variability in seed traits and seed-oil content (Ginwal et al., 2005;Kaushik et al., 2007;Mazumdar et al., 2012).However, systematic studies involving other morphological and agronomically important traits in J. curcas germplasm such as plant height, canopy spread, male to female flower ratio, collar length, total seed yield and 100 seed weight for selection of elite planting material have been rare (Rao et al., 2008;Srivastava et al., 2011).
The North-East region of India comprises eight states viz., Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, Sikkim and Tripura.Diverse climatic conditions (tropical to temperate), high variation in altitude (50 m to 7000 m) along with high rainfall (800 mm to 4000 mm) has made this agroclimatic zone one of the richest reservoirs of plant diversity in the country.In Basu et al. 649 North-East India, J. curcas grows mainly in the wild and such plant populations, with a longer history of isolation and evolution, harbor a significant amount of diversity (Ranade et al., 2008;Goswami and Choudhury, 2015).Such diverse populations can contribute towards broadening of J. curcas genetic resources and so, morphometric trait-based characterization of J. curcas from distinct eco-geographical regions of North-East India has the potential of identifying agronomically elite J. curcas plants.Unfortunately, scant research has been conducted on the morphological characterization of J. curcas germplasm from North-East India (Saikia et al., 2009).The main objective of this study was to assess the morphometric diversity of J. curcas germplasm from North-East India along with the identification of elite populations.

Collection of plant material
Field trips were undertaken during the months of June-August, 2012 to gather J. curcas seeds.The latter were collected from six states of North-East India (viz., Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram and Tripura) covering 29 distinct eco-geographical regions.J. curcas seeds from four other states of India viz., Delhi, Punjab, Gujarat and Orissa, were also included in the study as an outgroup (Table 1).The experiments were carried out at Centre for Energy, IIT Guwahati with the experimental site co-ordinates (26°11'5'' N; 91°40'9'' E) which experiences a warm temperate climate with an average annual rainfall of 1698 mm.The region comes under the influence of southwest monsoon from first week of June to early September.The collected seeds were planted in polythene bags containing mixture of sand, soil and vermiculite in the ratio of 1:1:1 (by volume) for germination.The polythene bags were labelled according to the region of seed collection.After two months, the hardened and rooted plantlets were transferred into garden soil (pit size 50 x 50 x 50 cm) in the field of Centre for Energy, IIT Guwahati, in a randomized complete block design.The spacing between the plants was 2.5 m x 2.5 m.Ten different plants from each region were treated as a single population and assigned an accession number (Table 1).The type of soil in the field plots were a mixture of sandy laterite clay soil and very deep well drained forest clay soil.Irrigation was performed initially for two weeks manually followed by natural rain water irrigation from the monsoon rains.

Measurement of plant morphometric characters
The study on the phenotypic assessment of morphological descriptors plant height (PH), collar diameter (CD), canopy spread (CS) and floral sex ratios (M:F) were conducted during the flowering phase of growth year three for 33 J. curcas populations with ten plants per population.PH, CD and CS were measured using standard measuring tape.CD was calculated from the girth measurement of the main stem 5 cm above ground.The M:F ratio was calculated counting the male and female flowers in the inflorescences.The mature seeds were harvested from the same plants and sun-dried to constant weight.The weight of 100 seeds (100SW) and total seed yield (TSY) from ten plants per population were estimated.The seeds were separated from the fruit mechanically and cleaned manually to remove all foreign material.The cleaned seeds were dried under similar temperature (35°C) and humidity conditions to reach constant weight.All the quantitative data was statistically analyzed and the mean value of each morphological trait for individual populations, standard error and coe cient of variation was calculated (Table 2).

Statistical analysis
Phenotypic inter-relations using the quantitative data were assessed using Manhattan dissimilarity coefficients.The latter were calculated as, 1/ ∑ , where and are the observed values of two populations and , with respect to the th trait, and is the number of morphometric traits considered.The pairwise dissimilarity matrix based on the Manhattan coefficient was subjected to cluster analysis using unweighted pair group method with arithmetic mean analysis (UPGMA) (Sneath and Sokal, 1973).Principal Components Analysis (PCA) was performed to further elucidate phenotypic variability of J. curcas.All calculations were performed using NTSYS-pc version 2.02 (USA) (Rohlf and Version, 1997).Pairwise Pearson correlation coefficients were computed using SigmaPlot 11.0 for the determination of the linear relationship among the morphometric traits (Wass, 2009).The analysis of variance (ANOVA), broad-sense heritability, phenotypic and genetic variance was calculated for the six selected quantitative traits using the online software PBSTAT 1.2 (Syukur et al., 2015).

Determination of mean, minimum and maximum values and coefficient of variations
The examination of all six quantitative traits exhibited considerable morphological variability in 33 J. curcas populations under investigation (p< 0.01) (Table 2).The mean data on morphometric parameters showed broad variation in plant height (PH) (143 cm -385.7 cm).Significant differences were observed in collar diameter (CD) (8.61 cm -30.46 cm), seed weight per 100 seeds (100SW) (54.4 g-123.8g) and total seed yield (TSY) (91.1 g -283.8 g).Canopy spread (CS) appeared to be the least variable trait (104.5 cm -239.9cm) with % coefficient of variation (%CV) of 1.67.The highest variation (%CV=23.19) was observed in male to female flower (M:F) ratio (10.5 -25.4).The population, IITJC24, had the highest PH (385.7 cm) and CS (239.9 cm) with respect to all other populations.In contrast, IITJC15 had a maximum TSY (283.2 g) and a minimum M: F ratio (10.5:1).IITJC28 had a maximum CD (30.5 cm).The population IITJC7 from Arunachal Pradesh recorded the highest 100SW (123.8 g).Consequently, three populations from Assam (IITJC15, IITJC24 and IITJC28) and one population from Arunachal Pradesh (IITJC7) emerged superior on the basis of morphometric trait values.

Correlation coefficients
The Pearson correlation between growth attributes and seed characteristics of J. curcas was estimated.The correlation coefficients revealed a positive relationship between growth traits PH, CS and CD with TSY (Table 3).However, the M: F ratio was negatively correlated with all other morphological traits.Interestingly, the existence of a highly significant negative association was observed between M:F ratio and TSY.This allows us to hypothesize that a direct positive correlation exists between the number of female flowers and TSY.This is in accordance with Rao et al. (2008) who found a positive relationship between plant height and female to male flower ratio with seed yield.The fact that IITJC3 and IITJC17, the low seed-yielding populations, scored below average value for most of the other morphological traits was in accordance with the correlation analysis data which showed a positive association between all traits (with the exception of M:F ratio) with TSY.Thus, correlation analysis allowed direct assessment of positive and negative contribution of other quantitative traits on TSY.J. curcas is having a gestation lag of 3-4 years (Biswas et al., 2010).Thus, during tree improvement programs, seed-related traits cannot be used for preliminary screening of J. curcas planting materials from large-scale plantations at early stage of growth.Since growth related traits of J. curcas is having a positive correlation with TSY, it was suggested that morphometric traits like PH, CS and CD can be used as initial screening indices for the selection of J. curcas plants at an early stage of growth, that is, before completion of the gestation period.It was also suggested that during J. curcas improvement programs, increasing the total number of female flowers or producing a more extensive canopy will provide better opportunities for increasing total seed yield of the plant.

Cluster and principal component analyses
The UPGMA dendrogram, based on the Manhattan dissimilarity matrix, separated the 33 populations into three major clusters I, II and III with 16, 13 and 14 populations respectively.Cluster I and Cluster II again formed four (IA, IB, IC and ID) and three (IIA, IIB and IIC) distinct sub-clusters respectively (Figure 1).The grouping of 33 J. curcas populations in eight sub-clusters is shown in Table 4.The remaining populations dispersed themselves into sub-clusters IA and IB along with populations from Assam and Arunachal Pradesh.The sub-clusters IC, IIA and IIB consisted of mixed populations from the North-East.Conversely, subclusters ID, IIC and cluster III were specific to populations from Assam.
The sub-cluster wise mean values of the quantitative morphological traits were also estimated (Table 5).The highest mean plant value (356.86 cm) and mean canopy spread (224.7 cm) were observed in cluster III.The latter also showed high values for 100SW and low M: F ratio.Sub-cluster IC comprising of populations from Meghalaya, Tripura and Assam recorded the highest CD (18.3 cm).Sub-cluster IIA, consisting of two populations each from Arunachal Pradesh, Manipur and Assam, recorded the maximum 100SW (97.6 g).The highest value of TSY and the lowest M: F ratio was observed in sub-cluster ID containing a single population, IITJC15.Thus, cluster III and sub-cluster ID recorded high mean values for the majority of agronomic traits.The populations from cluster III (IITJC19, IITJC21, IITJC22 and IITJC24) and sub-cluster ID (IITJC15) were found to be promising for future tree improvement programs.
In order to gain a better understanding of the relationship between J. curcas populations, Principal Component Analysis (PCA) was undertaken by concurrently assessing all six morphometric traits (Figure 2).PCA showed separation of the populations into three discrete groups (I, II and III).Groups I and II were again sub-divided into four and three sub-groups respectively.The overall grouping pattern of the populations in PCA was in accordance with the major clades of the UPGMA dendrogram.It has been previously reported that the crossing of populations from the clusters, which exhibit maximum inter-cluster distance and high mean value of agronomic traits, would result in production of more divergent trees (Kaushik et al., 2007;Shabanimofrad et al., 2013;Srivastava et al., 2011).Both the analyses delineated that the four populations in cluster III, IITJC19, IITJC21, IITJC22 and IITJC24, have maintained maximum inter-cluster distance from other J. curcas populations.Thus, it can be theorized that the selection of parents from these four populations during breeding programs would lead to the development of J. curcas plants with greater genetic heterogeneity.
When the grouping pattern of outside North-East and North-East populations were compared, it was observed that in both cluster and principal component analyses, J. curcas populations from outside North-East have nested together with populations from North-East India.Thus, from morphological-character-derived cluster analysis and principal component analysis it was determined that for all morphometric traits, association among populations was independent of their geographic origin.The inability of the dendrogram and PCA plot to reveal a clear relationship between diversity pattern and  geographical origin led to the deduction that morphometric traits of J. curcas are relatively uncorrelated with geographic distribution.

Analysis of variance and estimation of variance components
An ANOVA for morphometric traits reflected highly significant differences between the J. curcas accessions under investigation at p≤0.01 (Table 6).ANOVA among J. curcas accessions for different morphometric characters have previously been reported (Shabanimofrad et al., 2013;Sunil et al., 2012).
In selection and breeding experiments, knowledge of heritability and phenotypic trait under selection is essential for predicting the selection response and improving the agronomic trait (Robinson et al., 1949;   V P as compared to V G were higher for all traits.Broadsense heritability is defined as the ratio of total genetic variation to total phenotypic variation (Brown et al., 2012).Thus, the high magnitude of h 2 bs (65%) for all morphological traits depicted the dominance of heritable variation in J. curcas.Comparatively high % of h 2 bs (>75%) was observed for seed yield related traits like M: F ratio, 100SW and TSY respectively.However, comparatively low h 2 bs (<72%) was detected for plant growth related traits like PH, CD and CS.High estimates of heritability for M: F ratio (88.07%), 100SW (82.19%) and TSY (75%.01) also revealed that environment is less influential on seed yield related traits.Therefore, from the present study it can be inferred that, after gestation period, the selection of elite J. curcas plants on the basis of M: F ratio, 100SW and TSY (h 2 bs >75%) for tree improvement and breeding programs will be more successful.

Conclusion
This study reflected a high level of morphometric variation among 33 J. curcas populations.Of these, based on the morphometric traits, IITJC7 (highest 100SW), IITJC15 (highest TSY and minimum M:F ratio), IITJC24 (highest PH and maximum CS) and IITJC28 (maximum CD) were identified as exceptional.Cluster and principal component analyses demarcated IITJC15, IITJC19, IITJC21, IITJC22 and IITJC24 as promising and diverse populations.The seven groups from North-East India (IITJC7, IITJC15, IITJC19, IITJC21, IITJC22, IITJC24 and IITJC28) identified in this investigation on the basis of morphometric trait values and cluster analysis results can be recommended as potential starting materials in tree breeding programs for the development of genetically diverse J. curcas genotypes with desirable agronomic traits.It is also suggested that during future tree evaluation programs, growth attributes like PH, CS and CD can be used for preliminary screening of young J. curcas plants from large scale plantations.However, after gestation period, once the plants start giving economic yields, further screening and selection of agronomically promising plants on the basis of male to female flower ratio, 100 seed weight and total seed yield is likely to be more effective.

Table 1 .
Geographical locations of Jatropha curcas used in diversity analysis.

Table 3 .
Pearson correlation coefficients for morphometric traits of J. curcas populations.

Table 4 .
Grouping of Jatropha curcas populations as depicted by cluster analysis.

Table 6 .
Analysis of variance for morphometric traits in 33 J. curcas populations.

Table 7 .
Estimation of variance components and broad sense heritability.