Introduction
Mulberry (Morus spp.) is a perennial crop cultivated mainly in the Asian continent for sericulture and is also exploited for its valued timber and fruits. Mulberry foliage forms the only source of food for the domesticated silkworm Bombyx mori L. for the production of natural silk – ‘the queen of textiles’. Silk production is a sustainable activity for labour-rich rural India with attractive financial returns at regular intervals. India earns a foreign exchange of about US$480 million (CSB, 2012) through the export of silk yarn and finished products.
The genus Morus belongs to the order Rosales, the family Moraceae and the tribe Moreae (Zhang et al., Reference Zhang, Soltis, Yang, Li and Yi2011) and is distributed mainly in the temperate and subtropical regions of the Northern Hemisphere. A considerable difference of opinion exists among the taxonomists with regard to the precise number of species under the genus Morus. However, the widely accepted classification by Koidzumi (Reference Koidzumi1917, Reference Koidzumi1923) recognizes 24 mulberry species and a variety under a species. In India, most of the cultivated mulberry belongs to Morus alba and Morus indica. The centre of origin of mulberry is in the region bordering China–Korea–Japan (Vavilov, Reference Vavilov1926). Cultivated mulberry is not native to India and is believed to be introduced by Buddhist monks around 140 BCE through the Silk Road (Hyde, Reference Hyde1984). It is presumed that mulberry was first introduced in the form of seeds to north-eastern India through the Gangtok pass and the Kashmir valley through Ladak. Sericulture was introduced to the south from eastern India during 1720–1740 by the ruler Tippu Sultan of the Kingdom of Mysore (Datta and Nanavaty, Reference Datta and Nanavaty2005). Significant diversity in wild mulberry has been recorded in the Indo-Himalayan region (Ravindran et al., Reference Ravindran, Rao, Naik, Tikader, Mukherjee and Thangavelu1997). The wild mulberry Morus serrata is restricted to the sub-Himalayan regions and Morus laevigata is distributed in the mainland as well as Andaman Islands (Ravindran et al., Reference Ravindran, Tikader, Naik, Rao and Mukherjee1998).
The characterization and evaluation of mulberry germplasm using phenotypic markers is mostly unreliable due to high genotype × environmental interactions and specific cultivation procedures practised in the crop (Naik and Dandin, Reference Naik and Dandin2005). In the last decade, the dominant marker systems such as random amplified polymorphic DNA (RAPD) and inter-simple sequence repeat (ISSR) (Bhattacharya and Ranade, Reference Bhattacharya and Ranade2001; Naik et al., Reference Naik, Sarkar and Sathyanarayana2002; Awasthi et al., Reference Awasthi, Nagaraja, Naik, Kanginakudru, Thangavelu and Nagaraju2004; Srivastava et al., Reference Srivastava, Vijayan, Awasthi and Saratchandra2004), amplified fragment length polymorphism (AFLP) (Sharma et al., Reference Sharma, Sharma and Machii2000; Botton et al., Reference Botton, Barcaccia, Cappellozza, Tos, Bonghi and Ramina2005) and sequence-related amplified polymorphism marker (Zhao et al., Reference Zhao, Fang, Pan, Chung, Chung and Park2009) have been utilized for mulberry genetic characterization. The development of mulberry-specific microsatellite (SSR, simple sequence repeat) markers was first reported by Aggarwal et al. (Reference Aggarwal, Udaykumar, Hendre, Sarkar and Singh2004) and then by Zhao et al. (Reference Zhao, Miao, Jia, Pan and Huang2005). Unlike other agriculturally important crops, genomic resources are limited in mulberry.
Over the last five decades, mulberry breeders in India invariably maintained a small set of germplasm with alleles of interest, precisely termed as ‘breeders’ collection’. The present study aims to utilize mulberry-specific SSR markers to estimate the genetic diversity and structure of the breeders' collection that has been extensively used in mulberry crop improvement programmes in India. This study also reconstructs the pedigree and kinship of some of the important mulberry cultivars that was hitherto mostly not available or uncertain.
Materials and methods
Plant material
Breeders' collection consisting of 36 mulberry genotypes (Table S1, available online) was maintained in an experimental plot at the Central Sericultural Research and Training Institute (CSRTI), Mysore. These genotypes were maintained as medium bushes by regular pruning in a paired row system and replicated clonally with a spacing of 150 cm between a paired row and 90 cm within a paired row. The plants of the same genotype were replicated 60 cm apart in a row. The plants were grown under the standard package of practices recommended for mulberry cultivation in South India (Sarkar et al., Reference Sarkar, Kumar and Datta2000).
Morphological characterization
A total of 11 key morphological characters that are least affected by the environmental factors were recorded according to the procedure followed by Thangavelu et al. (Reference Thangavelu, Mukherjee, Tikader, Ravindran, Goel, Ananda Rao, Naik and Shekar1997). The characteristics such as branching nature, curvedness/straightness of the branch, phyllotaxy, and nature, lobation, colour, apex, base, margin, shape and surface of the leaf were recorded. The characteristic states were conveniently coded and used for the analysis.
DNA isolation
High-molecular-weight genomic DNA was extracted from fresh young leaves using the HiPurA Plant Genomic DNA Miniprep Purification Spin Kit (HiMedia, Mumbai, Maharashtra, India). Genomic DNA was assayed on 1% agarose gel stained with ethidium bromide for quality check and quantification of the stock solution. An aliquot of each of the stock solution was diluted to a uniform concentration of 10 ng/μl for the amplification of the polymerase chain reaction (PCR).
Optimization of SSR marker amplification
Microsatellite marker profiling on polyacrylamide gel electrophoresis was optimized using different approaches including touchdown PCR and standardization of PCR conditions (annealing temperature (T a), primer and template concentration and number of PCR cycles). PCR conditions for 140 mulberry-specific SSR markers (Table S2, available online) were optimized and marker polymorphism was assessed using a set of diverse mulberry germplasm which included distinct species and varieties (data not shown).
SSR marker profiling
PCR amplifications were carried out in 10 μl reaction volumes, containing 10 ng of template DNA, 0.5 pmol of each primer, 2 mM MgCl2, 100 μM of each deoxyribonucleotide triphosphates, 1 × PCR buffer and 0.5 U Taq DNA polymerase (HiMedia, India) on DNA Engine PTC – 200 (MJ Research, Waltham, Massachusetts, USA) programmed to the following cycling profile: initial denaturation at 94°C for 5 min followed by 25 cycles of 94°C for 45 s denaturation, primer-specific annealing temperature (Table S2, available online) for 30 s and 72°C for 45 s extension followed by the final extension step of 72°C for 5 min. The PCR products were resolved on 8% non-denaturing polyacrylamide gels using the Sequi-Gen GT electrophoresis system (Bio-Rad, Hercules, California, USA). The gels were silver-stained according to the method of Bassam and Gresshoff (Reference Bassam and Gresshoff2007) with modification and scanned. The allele sizes were estimated with reference to the pBR322 DNA-MspI digest marker (New England Biolabs, Ipswich, Massachusetts, USA) and recorded in base pairs.
Data analysis
Genetic diversity
Morphological traits were analysed by calculating simple matching dissimilarity coefficients using DARwin v. 5.0.158 (Perrier and Jacquemound-Collet, Reference Perrier and Jacquemound-Collet2006). Number of alleles, observed heterozygosity (H O), expected heterozygosity (H E) and polymorphic information content (PIC) were calculated for SSR marker data using PowerMarker v. 3.25 (Liu and Muse, Reference Liu and Muse2005). A simple matching dissimilarity coefficient was calculated for all pairs of individuals, a hierarchical tree based on the unweighted pair group method with arithmetic mean (UPGMA) method was constructed using DARwin v. 5.0.158 and the tree was visualized using FigTree (http://tree.bio.ed.ac.uk/software/figtree/). The principal coordinate analysis (PCoA) was carried out using GeneAlex v. 6.5b3 (Peakall and Smouse, Reference Peakall and Smouse2006), and the first two principal components were plotted using ggplot2 (Wickham, 2009) in the R program (R Development Core Team, 2008). Rare and private allelic richness were calculated for each individual using HP-Rare v. 1.1 (Kalinowski, Reference Kalinowski2005).
Structure
The structure of the breeders' collection was analysed using Bayesian model-based clustering implemented through the program STRUCTURE v. 2.3.3 (Pritchard et al., Reference Pritchard, Stephens and Donnelly2000; Falush et al., Reference Falush, Stephens and Pritchard2003). The admixture model was chosen to run the program STRUCTURE for ten independent simulations for each subpopulation (k) value ranging from 1 to 10 having iterations of 10,000 burn-in length and 10,000 Markov chain Monte Carlo. The optimal subpopulation size was determined by considering the following: (1) maximum-likelihood value on the plot; (2) ad hoc statistic (Δk) (Evanno et al., Reference Evanno, Regnaut and Goudet2005); (3) pattern of clustering realized by the UPGMA method; (iv) grouping depicted by PCoA. Among the ten replicated runs, the run with the maximum-likelihood value was chosen to assign the posterior membership quotient (Q) to each genotype. Nei's genetic distance among the subpopulations was estimated and pairwise F ST was calculated using Alrequin v. 3.1 software (Excoffier et al., Reference Excoffier, Laval and Schneider2006).
The analysis of molecular variance (AMOVA) was used to assess the genetic variance within and among the STRUCTURE-derived subpopulations and was calculated using Alrequin v. 3.1 with 10,000 bootstrapping to obtain a reliable estimate.
Pedigree reconstruction
The Micro-Checker program (Oosterhout et al., Reference Oosterhout, Hutchinson, Wills and Shipley2004) was used to identify the null alleles and to exclude them from parentage analysis. Allelic marker data were computed along with the known pedigree information to reconstruct the pedigree of the breeders' collection using the program COLONY v. 2.0.2.3 (Wang and Santure, Reference Wang and Santure2008). COLONY assigns parentage and estimates full- and half-sibs based on the maximum-likelihood method. The pedigree chart was drawn using Pedigraph v. 2.4 (Garbe and Da, 2008) based on the assignments by COLONY.
Mantel test
The significance of the correlation of clusters generated by genotypic and phenotypic marker data was tested using GeneAlex v. 6.5b3 by the Mantel test (Mantel, Reference Mantel1967). The Mantel test was also performed to check the correlation between the results of pairwise Nei's genetic distance and F ST for the STRUCTURE-defined subgroups.
Results
Morphological analysis
The dissimilarity coefficient matrix based on simple matching using morphological characters was generated and presented in the upper diagonal of the heat map (Fig. S1, available online). The mean pairwise dissimilarity was 0.408. The dissimilarity coefficient was maximum (0.909) between Mysore Local and Kanva-2 × Kousen (Sahana) and was minimum (0) between three pairs of genotypes, namely (1) S-34 and S-30, (2) S-799 and C-776 and (3) S-523 and RFS-135. The frequency distribution of morphological characteristics in the breeders' collection is shown in the stacked bar plot (Fig. S2, available online).
Mulberry-specific SSR markers and polymorphism
Initially, 140 SSR primer pairs were screened using a diverse set of 12 genotypes. However, based on marker resolution and polymorphism, only 70 SSR primer combinations (Table S3, available online) were utilized to profile the breeders' collection. Among the selected SSR markers, 26 (37.1%) represented di-repeat motifs followed by 13 (18.6%) each of tri- and compound repeats. Hexa- and penta-repeat motifs were found in 12 (17.1%) and 4 (5.7%) markers, respectively. Tetra-repeat motifs were rare and found only in one (1.4%) among the markers employed in the study. The 70 SSR loci generated a total of 215 alleles with a mean of 3.1 alleles per primer pair. Among the SSR loci, 15 were monomorphic and the rest produced alleles ranging from 2 to 9. A total of 66 rare alleles (A R ≤ 10%) were recorded among the 36 SSR markers and the maximum (5) was observed in MulSTR 3. The H O varied from 0 to 0.84 and the H E ranged from 0 to 1. The PIC was maximum (0.82) in MESTSSR 31 and minimum (0) in the monomorphic markers.
The correlation of repeat motifs with the PIC was studied. The mean PIC value ranged from a maximum of 0.37 for di-repeats to a minimum of 0.06 for hexa-repeat motifs. The compound repeat motifs with a mean PIC value of 0.35 were excluded from the analysis. Pearson's correlation analysis was carried out to establish the correlation of repeat motif size with the PIC value. Pearson's correlation coefficient (r= –0.44) indicated a negative correlation between the two parameters. The analysis suggests that the di-repeat motifs were more polymorphic and the PIC value decreased with an increase in the repeat motif size.
Genetic diversity and structure
The mean pairwise dissimilarity (Fig. S1, available online) computed based on simple matching was 0.26 among the breeders' collection. The least dissimilarity of 0.01 was recorded between RFS-135 and RFS-175, two cultivars evolved from the open-pollinated hybridization of Kanva-2. The next pair having minimum genetic dissimilarity was Punjab Local and Himachal Local with a dissimilarity coefficient value of 0.02. The maximum genetic divergence was recorded between Lamia Bay and Muki with a dissimilarity coefficient value of 0.61. Lamia Bay showed significantly higher dissimilarity coefficient values with all the collections except Doomar Nali. Lamia Bay is a wild mulberry collection belonging to M. laevigata from the Andaman Islands. Hierarchical clustering based on the UPGMA method (Fig. 1) showed four major groups (C1–C4). The first group (C1) consisted of 31 genotypes, which can be further divided into six subgroups. The majority of the genotypes under the first group (C1) are important cultivated varieties of India with region-specific characteristics. Only two collections, namely Morus multicaulis and Rosou, are of exotic origin and extensively used as donor parents in mulberry breeding programmes. Group II (C2) was represented by two accessions: English Black and S-763. Cluster III (C3) was represented only by S-1635, a national check for mulberry improvement programmes in India until recently. The last group (C4) consisted of two wild mulberry genotypes from the Andaman Islands known to harbour genes for resistance to biotic stress and tolerance to abiotic stress.
(colour online) UPGMA clustering of the breeders' collection based on the microsatellite allelic data.
The scatter plot (Fig. 2) based on the PCoA revealed four major groups (P1–P4) among the breeders' collection. The two collections of wild mulberry from the Andaman Islands represented a distinct group (P3) of the PCoA and corroborated with grouping obtained by the UPGMA method. The important mulberry varieties of India, which were mostly confined to group I of the UPGMA clustering, formed two groups in the PCoA scatter plot (P2 and P4). Two genotypes (English Black and S-763) from group II of the UPGMA clustering formed a separate group (P1) in the PCoA along with six other genotypes from groups I and III of UPGMA. The PCoA results were broadly in concurrence with UPGMA analysis.
(colour online) Scatter plot of the first two principal coordinates of mulberry genotypes. Numbers correspond to the serial number of the mulberry genotype given in Table S1 (available online).
The genetic structure of the breeders' collection was estimated using 70 locus-specific SSR markers by the STRUCTURE (Fig. 3). The LnP(D) was maximum at k= 5 and decreased gradually until k= 10. The ad hoc statistic (Δk) was found to be maximum at k= 4 (Fig. S3(a) and (b), available online). However, UPGMA clustering, PCoA, geographical origin and known pedigree of the genotypes supported the assumption of k= 4. In the STRUCTURE analysis also, the wild mulberry genotypes formed a distinct group with very little sharing of genomes with the other genotypes. However, the organization of the rest of the genotypes represented a hybrid structure (admixtures). The grouping obtained by the structure analysis with k= 4 was designated as G1, G2, G3 and G4, respectively.
(colour online) Genetic structure analysis of the breeders' collection based on SSR markers with subpopulation size (k) = 4. The Y-axis indicates the membership coefficients and the X-axis corresponds to the serial number of mulberry genotypes in Table S1 (available online).
Nei's genetic distance and F ST were calculated (Table 1) based on SSR allelic data corresponding to the groups (k= 4) of the STRUCTURE analysis. Among the groups, Nei's genetic distance (0.217) and pairwise F ST (0.502) were found to be maximum between G2 and G3. G2 consisted of the genotypes mainly belonging to M. alba from South India with exceptions of Kousen and Sujanpur-5. However, G3 consisted of two collections of M. laevigata and the divergence estimated by the two methods was well explained. However, minimum Nei's genetic distance (0.014) was recorded between G1 and G4 and the minimum pairwise F ST (0.049) was between G2 and G4. The pairwise F ST was marginally higher between G1 and G4 (0.543). The minimum values recorded for both the parameters can be appreciated from the fact that these genotypes belonging to different groups (G1 – North India, G2 – South India and G3 – exotic with exceptions) belonged to cultivated mulberry, namely M. alba and species with close affinity to this taxon. The AMOVA revealed that about 83% of the total genetic variance was accounted for within the population and the rest 17% was accounted for among the populations (Table 2).
Nei's genetic distance (upper diagonal) and F ST (lower diagonal) between the STRUCTURE-inferred subgroups
AMOVA among and within the STRUCTURE-inferred subgroups (k=4)
a Degrees of freedom.
Pedigree reconstruction
Genotypes with well-recorded pedigree information were used for anchoring and parentage prediction based on SSR allelic inheritance. The Micro-Checker program detected null alleles in three SSR primers (MESTSSR 13, MESTSSR 31 and MESTSSR 42) and were excluded from the parentage analysis. COLONY was run independently three times and the program was able to consistently assign the relationships. The program assigned half-, full-sibs and parents based on microsatellite marker exclusion. A total of six full-sib and 37 half-sib relationships were established. The evolution and genetic relationships among the breeders' collection was determined (Fig. S4, available online).
Mantel test
The correlation of microsatellite markers with important morphological (qualitative) characteristics was tested and resulted in significantly low correlation (R 2= 0.004). Nei's genetic distance showed a high correlation (R 2= 0.993) with F ST.
Discussion
SSR marker polymorphism
The present investigation is the first to utilize a large number of SSR markers (genic and genomic) in mulberry. The study revealed a negative correlation (r= –0.44) of repeat motif size with PIC values. Di-repeat motifs were more informative in mulberry compared with tri- and tri-repeats which in turn were more informative than tetra-repeats and so on. The mean repeat motif number decreased from 13.6 (di) to 6 (tri), 4 (tetra), 3.5 (penta) and 2.6 (hexa). Studies on the distribution of microsatellites in plant genomes have suggested that di-repeats are more prevalent in non-coding regions and expressed sequence tag (EST) sequences are comparatively richer in tri-repeats (Victoria et al., Reference Victoria, Maia and Oliveira2011). We analysed 3311 mulberry-specific ESTs in the GenBank database and found that the majority of SSRs (42%) were tri-repeats (unpublished data). It has been found that the variation in repeat motif number is higher in di-repeat motifs than in other repeat motifs (Yu et al., Reference Yu, Won, Jun, Lim and Kwak2011), and allelic polymorphism is directly proportional to the number of repeat units (Sung et al., Reference Sung, Tucker, Bergeron, Lynch and Thomas2010). This offers a plausible explanation for the negative correlation of repeat motif size with the PIC value in the study.
Genetic diversity and structure
The breeders' collection mainly composed of an assembly of genotypes which are commercially important, having special attributes such as higher leaf yield, quality, combining ability, resistance to biotic stress, tolerance to alkalinity, moisture stress, etc. (Table S1, available online). The majority of these genotypes belonged to the cultivated species of mulberry (M. alba, M. indica, M. multicaulis and M. latifolia). Only two recent collections, namely Lamia Bay (♂) and Doomar Nali (♀), belonging to the wild species (M. laevigata) have been included for the enhancement of gene pool availability in breeding programmes. These two genotypes have been selected for possessing some unique characteristics such as reproductive compatibility with cultivated types, resistance to termites, large-sized leaves and high regeneration ability (Ravindran et al., Reference Ravindran, Tikader, Naik, Rao and Mukherjee1998). The two wild genotypes are diploids unlike many of the M. laevigata collections from the mainland which are mostly triploids or higher polyploids. The overall mean dissimilarity based on SSR alleles among the breeders' collection was estimated to be 0.26, which is low. The mean dissimilarity was further reduced to 0.22 if the two wild genotypes were removed from the analysis. Until recently, most of the efforts in mulberry improvement in India were towards increasing the leaf yield and many of the other commercial traits were often neglected (Tikader and Kamble, Reference Tikader and Kamble2008). In addition, mulberry breeders have preferred to use the same set of gene pool for achieving the objectives of crop improvement programmes resulting in a narrow genetic base of the evolved varieties. Therefore, there is a strong need for broadening the genetic base of mulberry by the assessment of total diversity available in the mulberry species and the construction of a core subset for effective utilization of the natural gene pool in crop improvement.
Even though the heterozygous nature of mulberry has often been reported in the literature (Naik et al., Reference Naik, Sarkar and Sathyanarayana2002; Naik and Dandin, Reference Naik and Dandin2005), no quantifiable assessment has been made so far. The present study is the first to estimate heterozygosity in mulberry based on microsatellite allelic richness (A R; Table 3). The mean A R was calculated as 1.36 and all the genotypes recorded a value >1, indicating that the mulberry varieties analysed were heterozygous. S-13, a variety recommended for rain-fed areas, was most heterozygous (A R= 1.79) of all the genotypes. The assessment of private allelic richness (A PR) indicated that the wild genotype Doomar Nali recorded the maximum number of unique alleles among the breeders' collection. This genotype has several unique characteristics and is a useful donor parent in breeding with the largest sized leaves in mulberry (Ravindran et al., Reference Ravindran, Rao, Naik, Tikader, Mukherjee and Thangavelu1997). Among the STRUCTURE-derived four subgroups (G1–G4), G3 having wild genotypes showed high private allelic richness than the other subgroups (Table S4, available online).
Allele and private allele statistics of the mulberry genotypes
a A R, allelic richness.
b A PR, private allelic richness.
Many studies have used dominant marker systems such as RAPD (Chatterjee et al., Reference Chatterjee, Nagaraja, Srivastava and Naik2004; Orhan et al., Reference Orhan, Ercisli, Yildirim and Aga2007), ISSR (Kar et al., Reference Kar, Srivastava, Awasthi and Urs2008) and AFLP (Sharma et al., Reference Sharma, Sharma and Machii2000; Botton et al., Reference Botton, Barcaccia, Cappellozza, Tos, Bonghi and Ramina2005; Kafkas et al., Reference Kafkas, Ozgen, Dogan, Ozcan, Ercisli and Serce2008) to characterize indigenous and exotic mulberry. Based on fluorescent-AFLP marker analysis of 45 mulberry germplasm accessions across different eco-geographic regions of Japan and other parts of the world, Sharma et al. (Reference Sharma, Sharma and Machii2000) inferred that clustering of these genotypes into distinct groups was not apparent. Furthermore, the study could not establish any correlation between estimated genetic relationships and the geographical origin of the genotypes. On the contrary, Vijayan (Reference Vijayan2004) analysed 18 mulberry genotypes from India and Japan, and clustering based on ISSR and RAPD markers clearly demarcated the accessions based on the countries of origin. However, studies on cultivated mulberry (Bhattacharya et al., Reference Bhattacharya and Ranade2001; Botton et al., Reference Botton, Barcaccia, Cappellozza, Tos, Bonghi and Ramina2005) utilizing the dominant markers could not establish the correlation between genetic divergence and geographical distribution. These inconsistencies in the performance of random markers have limited their use. To the best of our knowledge, to date, there is no report substantiating the known mulberry genetic relationships based on the pedigree using dominant markers. The lack of availability of sufficient co-dominant markers in mulberry has limited its use until now. Aggarwal et al. (Reference Aggarwal, Udaykumar, Hendre, Sarkar and Singh2004) first reported the development of six mulberry-specific SSR markers and the analysis was limited to the testing of cross-species affinity using a set of 43 mulberry genotypes. Similarly, Zhao et al. (Reference Zhao, Miao, Jia, Pan and Huang2005) validated the microsatellite polymorphism in mulberry using 27 genotypes of Chinese origin. Furthermore, Zhao et al. (Reference Zhao, Zhou, Xuexia, Zhang, Wang, Huang, Xiang, Pan and Huang2007) compared the performance of SSR and ISSR using the same set of Chinese mulberry genotypes and concluded that the former detected higher polymorphism, indicating higher discriminatory capacity than the latter. In the present study, the UPGMA clustering precisely supported the known genetic relationships of varieties: Sahana, G-2, G-4, RFS-135, RFS-175, etc. (Table S1, available online).
The clustering derived from the UPGMA method (Fig. 1) could be broadly categorized into four groups and in concordance with genetic relationships based on the known pedigree. Cluster I (C1) was the largest with the maximum number (31) of genotypes. The genotypes in this cluster belonged to different species of cultivated mulberry (M. alba, M. indica, M. multicaulis and M. latifolia) and used extensively in breeding without an inter-specific sexual barrier. However, high reproductive compatibility does not recommend the species distinction. This has also been supported by Hirano's (Reference Hirano1980) observation based on isozymes and several sap proteins. The clustering of the large number of cultivated genotypes should be inferred in this background. Cluster I could be further resolved to smaller subgroups to establish accurate genetic relationships among the mulberry varieties. RFS-135 and RFS-175 (both Open Pollinated Hybrid (OPH) of Kanva-2) were the closest followed by Himachal Local and Punjab Local. The high-yielding ‘S’ series of mulberry varieties suitable for both irrigated and rain-fed conditions of South India formed a distinct subgroup, suggesting the utilization of a common gene pool for the evolution of these varieties by the breeders. In early mulberry crop improvement efforts, OPH seeds were utilized for selecting better recombinants and development of varieties (Kesavacharyalu et al., Reference Kesavacharyalu, Balakrishna, Dandin and Sarkar2006). Due to this reason, many of the popular cultivars developed in the 1980s lack clear pedigree records. The exotic collection (English Black) and an elite variety (S-763) in cluster II (C2) were significantly divergent with maximum intra-cluster distance. S-1635 in cluster III (C3) is a unique and widely adaptable variety to different agro-climatic conditions in the sericulture zones of India. The grouping of two wild genotypes from the Andaman Islands was distinct (C4), which also supported the relationship based on geographical distribution and taxonomic status (Naik, 2007). PCoA (Fig. 2) based on SSR allelic data further supported genetic relationships among the genotypes and was very similar to the one depicted by the UPGMA method. Broadly, the PCoA also plotted four groups (P1–P4) across the two axes on the Cartesian plane. The wild genotypes were clearly separated in the PCoA and formed a distinct group (P3). The rest of the mulberry genotypes were distributed in three other groups according to the known or recorded genetic relationships of the mulberry genotypes. The species distinction of cultivated mulberry could not be achieved both in the UPGMA and PCoA, suggesting the use of gene pools of different origin in Indian breeding programmes and inferring that most of them are hybrids.
The STRUCTURE analysis with assumed subpopulation size (k= 4) is shown in Fig. 3. Wild mulberry genotypes from the Andaman Islands formed a distinct group (G3), structured with little genome sharing with other groups. The grouping of these wild genotypes can be appreciated in the background of taxonomic distinction as well as reproductive isolation. The varieties evolved in the eastern and northern parts of the country were grouped (G1) and separated from southern genotypes (G2). The exotic genotypes with a few exceptions formed the last group (G4). The STRUCTURE analysis throws insight into the breeding and evolutionary history of mulberry varieties of South India. The varieties developed in the 1970s and early 1980s were from a narrow gene pool of Mysore Local (introduced to Mysore region during 1720–40 by Tippu Sultan), Kousen (Japanese variety) and Kanva-2 (traditional cultivar) along with Sujanpur-5 (Guruprasad, 2011). In the late 1980s and early 1990s, there was a shift in the approach of South Indian breeders to utilize more exotic genotypes and varieties originating from eastern India in breeding programmes for leaf quality improvement to produce the international grade bivoltine silk (2A–4A). As a consequence, a high-yielding variety, Victory-1 (V-1), with superior leaf quality was developed. In addition, the breeding of two mulberry varieties, namely G-2 and G-4, using M. multicaulis (an exotic donor parent of the eastern Indian breeding programme) resulted in the broadening of the genetic base of South Indian cultivars and the genotypes being grouped with eastern and northern Indian varieties in the STRUCTURE analysis with k= 4.
Nei's genetic distance (k= 4) indicated that wild genotypes of the Andaman Islands are divergent from the rest of the groups. The mulberry varieties developed in the eastern and northern regions of the country showed least genetic distance. F ST was highest (0.502) among the Andaman and South Indian genotypes, and they share least genetic diversity. The AMOVA revealed a high percentage of variation within the subpopulation (82.9) compared with a low percentage of variation among the subpopulations.
Breeding history and pedigree reconstruction
Systematic mulberry breeding was initiated in India during the early 1960s, utilizing few exotic varieties from Japan (Kousen and Rosou), France (M. multicaulis and M. rotundiloba), Italy (Cattaneo) and Myanmar (Mandalay) (Das and Krishnaswami, Reference Das and Krishnaswami1965). However, the introduction and breeding history of mulberry in India have been poorly recorded and have few valid reasons: (1) introduction of cultivated mulberry to India during early historical times lacked proper records; (2) high heterozygous nature of the plant coupled with dioecy and outbreeding; (3) clonal propagation of mulberry in tropical climate and high phenotypic plasticity. In the absence of credible breeding records, different approaches for deciphering the kinship and pedigree using molecular techniques have been suggested (Jones et al., Reference Jones, Small, Paczolt and Ratterman2010). Lacombe et al. (Reference Lacombe, Boursiquot, Laucou, Vecchi-Staraz, Peros and This2012) successfully reconstructed the pedigree of grapevine cultivars with 20 nuclear SSR markers and a large number of markers increased the confidence on the assignments (Jones et al., Reference Jones, Small, Paczolt and Ratterman2010). Parentage analysis without parents (Waples and Waples, Reference Waples and Waples2011) and polymorphic nuclear microsatellite markers enabled us to reconstruct parentage for most of the mulberry genotypes without having all the potential parents in the analysis.
The parentage analysis confirmed the available pedigree records of eastern Indian varieties. Ber. S-1 (OPH of Mandalay) was evolved from an unknown parent 2 (UKP2), predicted to be Mandalay and shares a parent (UKP3) with Rosou. The analysis also revealed that Thaipech (variety from Thailand) was developed by the hybridization of Mandalay (UKP2) and UKP1. The allelic inheritance pattern reconfirmed the origin of an important eastern variety, C-776, by the hybridization of M. multicaulis and Black Cherry. The traditional cultivar ‘Mysore Local’ shares a common parent (UKP17) with Rosou. UKP17 is also the progenitor for mulberry varieties: M. indica and ‘Berhampore Local’. This supports the introduction of ‘Mysore Local’ to the south particularly to the Kingdom of Mysore.
The development of a mulberry variety, Kanva-2 (K-2) or M-5, had significantly improved sericulture in the Mysore State (present-day Karnataka) during the late 1960s. According to the pedigree record, Kanva-2 originated from the OPH seeds of Mysore Local in a Kanva farm near Mysore. However, the male parent that contributed to the development of this variety had remained elusive. The present investigation suggests that the varieties Kanva-2 and Kousen have a common paternal progenitor (UKP13). The finding is very important in terms of understanding the gene flow in mulberry and planning future breeding programmes. The present observation has been further substantiated by the close genetic relationship among Kanva-2 and Kousen as also inferred by the random marker analysis (Naik and Dandin, Reference Naik and Dandin2005). It is also interesting to note that the hybrid of Kanva-2 × Kousen (Sahana) was identified for cultivation under shaded conditions, especially in coconut plantations of Kerala (Das et al., Reference Das, Balakrishna, Mathur, Vedavyasa and Dandin2010).
The three popular varieties of South India for cultivation under rain-fed conditions, namely S-13, RFS-135 and RFS-175, are all OPHs of Kanva-2. However, the male parent of these varieties is so far unknown. The SSR allelic inheritance reveals a common male parent (UKP10), a mulberry genotype which was not included in the present investigation. The study has further validated the pedigree of many mulberry cultivars of India and specifically of South India. Furthermore, the present investigation has also provided insight into the genetic relationships of many other mulberry genotypes that were hitherto unknown. This is the first report on cultivated mulberry varieties of India wherein the pedigree records have been tested and validated based on microsatellite allele sharing and full parentages have been uncovered for most of the varieties. The course of mulberry breeding has been documented for possible revision and improvisation of breeding strategies in India.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S1479262113000415
Acknowledgements
The authors acknowledge the financial support received from the Department of Biotechnology, Ministry of Science and Technology, Government of India, New Delhi (sanction order no. BT/PR13197/PBD/19/205/2009, dated 20 September 2010) for undertaking the research work. R. R. K. received a Junior Research Fellowship under the project grant. The authors thank Dr Rajeev K. Varshney and Dr Pawan Khera of the International Crop Research Institute for Semi-Arid Tropics, Hyderabad for their useful comments for improving the draft of the manuscript. V. G. N. is grateful to Dr M. Udayakumar and Dr M. S. Sheshashayee, University of Agriculture Sciences, Bangalore for providing SSR primers of MESTSSR and Mul3SSR series. The authors also thank Mr Marian Vincent Pinto, CSRTI, Mysore for critically checking the manuscript.
