Introduction
Sorghum (Sorghum bicolor (L.) Moench) is one of the world's most important nutritious coarse cereals. Recently, in India and in other parts of the world, yield levels of sorghum have been stagnant due to the narrow genetic base of breeding material utilized in sorghum breeding programmes (Aruna and Audilakshmi, Reference Aruna and Audilakshmi2008). There is an urgent need for the diversification of the genetic base of the breeding material to break this yield plateau. Genetic improvement depends on the availability of genetic diversity in the selection material, and its efficient exploitation. In the recent past, sorghum breeders have started crossing an elite variety with another elite variety to develop rapidly and release new varieties for commercial cultivation. Such elite × elite crosses have an advantage of accumulating genes involved in grain yield expression, provided elite lines are from diverse sources. Crossing two elite lines with related parentage does not give significant advantage in terms of grain yield (Audilakshmi et al., Reference Audilakshmi, Aruna and Kiran2003). Hence, the narrow genetic base in the germplasm of a breeding programme affects the potential genetic gain through selection.
Traditionally, cultivars are differentiated based on their morphological characters. However, as only a few morphological characters are stable in their expression over seasons and locations, relying exclusively on morphological characters will result in ambiguity in the identification of cultivars. Therefore, other types of markers are necessary. Many molecular marker technologies have been developed and applied for studying the patterns of genetic diversity in sorghum germplasm collections and in breeding programmes, including restriction fragment length polymorphisms (Deu et al., Reference Deu, Rattunde and Chantereau2006), random amplification of polymorphic DNAs (RAPDs; Ayana et al., Reference Ayana, Bryngelsson and Bekele2000), inter-simple sequence repeats (Yang et al., Reference Yang, de Oliveira, Godwin, Schertz and Bennetzen1996), simple sequence repeats (SSRs; Menz et al., Reference Menz, Klein, Unruh, Rooney, Klein and Mullet2004; Klein et al., Reference Klein, Mullet, Jordan, Miller, Rooney, Menz, Franks and Klein2008; Smith et al., Reference Smith, Promino, Monk, Nelson, Jone and Porter2010) and amplified fragment length polymorphisms (AFLPs; Menz et al., Reference Menz, Klein, Unruh, Rooney, Klein and Mullet2004). Accurate assessment of levels and patterns of genetic diversity can be invaluable in crop breeding for various purposes, including (1) analysis of the genetic variability of cultivars, (2) identification of diverse parental combinations to develop segregating progenies with maximum genetic variability for further utilization, and (3) introgression of desirable genes or chromosome segments from diverse sources into elite germplasm.
In India, inbred lines released by public sector institutions are being used in breeding programmes by public and private institutions (multinational and national seed companies). Assessment of genetic relationships among diverse accessions allows breeders to select diverse genes from different sources and to accumulate those favourable alleles in one cultivar. In the present study, genetic diversity among elite sorghum genotypes (varieties, hybrids and hybrid parents) developed in different regions of India was initially assessed using a conventional method (using morphological characters). Subsequently, SSR analyses were carried out for the assessment of genetic diversity, and the results were compared with those obtained with morphological markers. Further, we discuss how such information could be used in the genetic improvement of sorghum, especially by choosing appropriate and diverse genotypes in breeding. The objectives of the study were (1) to compare morphological and molecular diversity analysis, and (2) to identify the heterotic gene pools from the genotypes evaluated.
Materials and methods
Materials
In total, eighty sorghum genotypes (65 rainy-season sorghums comprising eight maintainer female (B) lines, ten fertility restorers (R lines), 33 varieties, four hybrids and ten forage varieties; and 15 post-rainy-season varieties) were evaluated for 40 morphological traits. Ninety-three (79 from morphological studies and 14 new cultivars obtained during 2008) sorghum cultivars comprising ten rainy-season elite B lines, 13 rainy-season R lines, 36 rainy-season varieties, 7 rainy-season hybrids, ten forage varieties and 17 post-rainy-season genotypes were studied for molecular analysis during 2008. The list of the material studied is given in Supplementary Table S1 (available online only at http://journals.cambridge.org).
Methods
Morphological analysis
Sixty-five rainy-season genotypes were evaluated in the 2006 and 2007 rainy seasons, and 15 post-rainy-season genotypes in the 2003 and 2004 post-rainy seasons at the Directorate of Sorghum Research (DSR), Hyderabad, India. The experiments were laid out in a randomized complete block design with three replications. Data were recorded on ten random plants in each replication. The forty morphological traits studied on eighty genotypes were those given in the Guidelines for the Conduct of Test for Distinctiveness, Uniformity and Stability on Sorghum developed by the DSR, Indian Council of Agricultural Research, Hyderabad, India. The guidelines were developed based on the UPOV (the International Union for the Protection of New Varieties of Plants, Geneva) guidelines (Audilakshmi et al., Reference Audilakshmi, Elangovan and Kanna Babu2004). The lists of the morphological traits studied are given in Supplementary Table S2 (quantitative traits) and Table S3 (qualitative traits) (available online only at http://journals.cambridge.org).
Molecular analysis
Sampling for DNA: Seeds of the 93 sorghum genotypes were germinated in small pots in a greenhouse. Fresh plant tissues were harvested by cutting five seedlings between 10 and 14 d after emergence, placed in plastic containers and frozen with liquid nitrogen. Genomic DNA was extracted from young leaf tissue (1–2 g) using the cetyltrimethylammonium bromide (CTAB) procedure (Saghai-Maroof et al., Reference Saghai-Maroof, Soliman, Jorgensen and Allard1984).
A set of 48 sorghum SSR markers (http://sat.cirad.fr/sat/sorghum_SSR_kit/) was used for genotyping the 93 sorghum genotypes using the genotyping service facility at the ICRISAT, Hyderabad, India. The map location of 48 primers represents at least three primers per chromosome. Forward primers were labelled with 6-carboxyfluorescein (FAM), 4,7,2',4',5',7'-hexachloro-6-carboxyfluorescein (HEX) or 7', 8”-benzo, 5'-fluoro-2', 4,7 trichloro-3-carboxyflourescein (NED) (PE-Applied Biosystems, Foster City, CA, USA). Polymerase chain reaction (PCR) products were separated by pooling post-PCR products based on dye and/or fragment size. PCRs were set up in 5-μl volumes in 384-well PCR plates (ABGene, Rochester, NY, USA) using a robotic liquid handling system (Tecan, Switzerland). DNA fragments were denatured and size-fractioned using capillary electrophoresis on an ABI 3700 automatic DNA sequencer (PE-Applied Bio-systems). GENESCAN 3.1 software (PE-Applied Bio-systems) was applied to size peak patterns, using the internal ROX 400 HD size standard and GENOTYPER 3.1 (PE-Applied Bio-systems) for allele calling. To verify the repeatability of each PCR and each capillary electrophoresis run, two control samples (accessions BT × 623 and R16) were included during the PCR of each SSR marker and during each capillary electrophoresis run.
Statistical analysis
Morphological data: Morphological data was subjected to multivariate analysis (Mahalanobis, Reference Mahalanobis1930) and the genotypes were further grouped into different clusters based on Ward's minimum variance method (Hair et al., Reference Hair, Anderson and Tatham1987).
Molecular data: For molecular diversity analysis, matrices of binary data were constructed with rows equal to accessions and columns equal to distinct SSR primers. For the 93 sorghum accessions, the binary matrix contained zeroes and ones, corresponding to the absence or presence of the marker band (alleles), respectively. Dissimilarity matrices were constructed from the binary data with Nei and Li (Reference Nei and Li1979) similarity coefficients. From these matrices of dissimilarity coefficients, the mean genetic distances (GDs), standard deviations and the distribution of dissimilarity values were calculated. Finally, to determine the efficiency of each marker type in detecting polymorphisms, the polymorphic information content (PIC) was calculated.
The dendrogram constructed is based on Jaccard's similarity coefficient (Ward's minimum variance) cluster analysis. The program INDOSTAT was used to calculate the 93 pairwise similarity matrix with Jaccard's coefficient (Jaccard, Reference Jaccard1908). This similarity measure was chosen because it does not count 0; 0 matches between pairs of the genotypes (Devin et al., Reference Devin, Wang, Karl and Brian2007). The GD between each pair of the accessions was calculated as 1 minus Jaccard's similarity measure. The similarity matrix was inputted with Ward's minimum variance method (Ward, Reference Ward1963).
For a comparison between morphological and molecular analysis, data of the 79 genotypes were used. The simple correlation coefficient (r) between the morphological data and molecular data of the 79 genotypes was calculated between genetic similarities and morphological similarities to compare genetic diversity analysis using morphological and molecular markers (Hamza et al., Reference Hamza, Hamida, Rebai and Moncef2004).
Results
Genetic divergence using morphological traits
Rainy-season sorghum
The range and mean performance of the sorghum genotypes are given Supplementary Tables S2 and S3 (available online only at http://journals.cambridge.org). The 65 genotypes of the rainy-season sorghum were distributed into nine different clusters based on the D 2 values (Fig. 1). Cluster I was the largest with 19 genotypes, followed by cluster V with 12 genotypes and cluster IX with eight genotypes.
Fig. 1 Dendrogram based on phenotypic data on Mahalanobis Euclidean2 Ward's minimum variance for 65 rainy-season sorghum genotypes. Blue, rainy-season B lines; black, rainy-season R lines; red, rainy-season varieties; brown, rainy-season hybrids; green, forage varieties. A colour version of this figure can be found online at http://journals.cambridge.org/pgr.
Post-rainy-season sorghum
Fifteen genotypes were distributed in four different clusters (Fig. 2). Cluster III was the largest with 11 genotypes, while cluster I has two genotypes (104B and CSH19R) and clusters II and IV have one genotype each.
Fig. 2 Dendrogram based on phenotypic data on Mahalanobis Euclidean2 Ward's minimum variance for 15 post-rainy-season sorghum genotypes. Purple, post-rainy-season varieties. A colour version of this figure can be found online at http://journals.cambridge.org/pgr.
Average intra- and inter-cluster distances
Rainy-season sorghum
Intra-cluster values ranged from 0 to 0.77 for the rainy-season sorghum. The highest intra-cluster distance was observed in cluster V (0.77). The inter-cluster distance ranged from 1.17 to 10.95. The maximum inter-cluster D 2 distance was between cluster I and cluster VIII (10.95), while the minimum inter-cluster distance was observed between cluster VII and cluster VI (1.17).
Post-rainy-season sorghum
Intra-cluster values ranged from 0 to 20.66 for the post-rainy-season sorghum. The highest intra-cluster distance was recorded for cluster III (20.66). The inter-cluster distance ranged from 102.11 to 2177.64. The maximum inter-cluster D 2 distance was between cluster IV and cluster II, and the minimum between cluster III and cluster I.
Genetic divergence studies using molecular markers
Out of the 48 primer pairs, 47 were found to be polymorphic accounting for 97.9% polymorphism. The PIC varied from 0.0 (monomorphic) to 0.81 with an average of 0.51. Marker Isep0310 was monomorphic (one allele), while marker Xtxp265 was highly polymorphic with a PIC value of 0.81 yielding 15 different alleles among the tested genotypes. The PIC values were significantly correlated with the number of alleles (r = 0.74; P = 0.000). Of the 48 markers tested, eight were highly informative (PIC ≥ 0.75, mean PIC = 0.78). The SSR loci Xtxp320, mSbCIR238, gpsb151, Xtxp295, sb5-206, gpsb069, Xtxp057 and Xtxp265 produced more alleles per locus (range 10–15 with a mean of 12.50). For the markers tested in this study, 310 alleles were observed with an average of 6.5 alleles per locus. The number of alleles ranged between 1 (Isep0310) and 15 (Xtxp012 and Xtxp265). The SSR alleles detected were 137, 156, 198, 187 and 188 among the ten rainy-season B lines, 13 rainy-season R lines, 31 rainy-season varieties, ten forage varieties and 21 post-rainy-season genotypes, respectively. The maximum number of alleles per locus was six in the B lines with an average of 2.85; eight in the R lines with an average of 3.25; nine each in grain (with an average of 4.1) and forage varieties (with an average of 3.9); and 11 in the post-rainy-season varieties (with an average of 3.9). Of the 48 markers tested, 12.5% were monomorphic in the B lines, 16.7% in the R lines, 12.4% in the rainy-season varieties and 6.3% in the forage and post-rainy-season sorghums.
The dendrogram generated nine clusters for all the 93 genotypes (Fig. 3). Cluster IV was the largest one which included 22 genotypes; cluster VI was the next largest in size comprising 19 genotypes followed by cluster IX (nine genotypes), clusters I, III and VIII (eight genotypes each), and clusters II and IX (seven genotypes each). Cluster V was the smallest (only five genotypes).
Fig. 3 Dendrogram based on genotypic data for Jaccard's similarity coefficient for 93 sorghum genotypes. IJ, distance between Ith and Jth genotype; Blue, rainy-season B lines; black, rainy-season R lines; red, rainy-season varieties; brown, rainy-season hybrids; green, forage varieties; purple, post-rainy-season varieties. A colour version of this figure can be found online at http://journals.cambridge.org/pgr.
Generally, the B and R lines were clustered in separate groups. Most of the rainy-season varieties were clustered with the R lines in clusters VI, VII and VIII. The post-rainy-season genotypes and the forage varieties formed yet other groups. The B lines were clustered in three clusters (clusters I, III and V) with their respective hybrids. Most of the maintainer lines of male sterile lines were found in the same cluster. The hybrids were placed in the same cluster with their female parent.
In cluster II, six genotypes were of the forage sorghum originated from Pantnagar and Hisar. One genotype, PSR34, was distinct from the other grain sorghum varieties and clustered with the forage sorghum. Genotypes 296B, IMS9B, and SB401B clustered with their hybrids in cluster III. Almost all of the post-rainy-season varieties (durra type) were found in cluster IV. Interestingly, forage sorghum HC136 from Hisar, and GFS4 from Surat were found in this cluster. The B lines AKMS14B and 7B, along with their hybrids (early flowering), were clustered in cluster V. Most of the R lines were in cluster VI. Interestingly, restorer lines did not find similarity with their respective hybrids. Exceptionally, only one B line, 463B, from DSR was found along with the R lines in this cluster. One R line, AKR150, parent for CSH14, found in cluster VII showed dissimilarity with the other R lines, and formed cluster with the grain sorghum varieties from Surat, Parbhani, Indore and Dharwad in cluster VII. Two R lines, RS673 and ICSR89085, showed similar results to those of variety CSV13 from the ICRISAT and GJ lines from Surat formed cluster VIII. Nine varieties, developed at the DSR and other centres were found in cluster IX.
Genetic similarity among the genotypes
The genetic similarity matrix revealed the lowest percentage of similarity (8%) in the pairs PSB9 and SSG59-3, JJ741 and GFS4, and GFS4 and NR486. The highest genetic similarity was found among the varieties both in the rainy-season and post-rainy-season sorghums.
The B lines were grouped in clusters I, III and V. Genotype 104B showed 15% similarity with AKMS14B, 17% with 7B and 463B, and the highest similarity with 296B (50%). When comparing the least genetic similarity (up to 15%) between the B and R lines/varieties, the 27B male sterile line was least similar to JJ741 (variety) and AKR73 (R line). One post-rainy-season restorer line, RS585, and another rainy-season R line, AKR73, were dissimilar to most of the R lines.
Diversity analysis by morphological markers versus molecular markers
A significant positive correlation (r = 0.463; P = 0.001) between the genetic similarities and morphological similarities was obtained.
Discussion
Genetic divergence studies using morphological traits
Rainy season
The rainy-season sorghum genotypes were clustered into nine groups (Fig. 1), with 47.7% of the genotypes falling into two clusters (I and V). The genotypes developed at the DSR (Hyderabad), Rahuri, Surat, Aruppukottai, ICRISAT (Hyderabad) and Parbhani grouped into cluster I, which is in agreement with the Engels (Reference Engels1994) findings. This could also be due to many common parents used in developing these genotypes at various centres. All the GJ lines except GJ39 (cluster III) originating from Surat were formed in the same cluster (cluster I), showing a lack of variation among the cultivars. In this study, no correlation was found between genetic diversity and geographical origin.
Majority of the varieties were grouped in clusters III and IV, and the cluster distance between them was less, showing low diversity among them. The inter-cluster distances were higher than the intra-cluster distances, which indicated wide genetic diversity among the genotypes of different groups than among those of the same cluster. Genotypes from the divergent clusters, when crossed, may produce higher heterosis (Anantharaju and Meenakshiganeshan, Reference Anantharaju and Meenakshiganeshan2008). The cultivars from the distinct clusters, if chosen for hybridization, may result in a broad spectrum of variability in segregating progenies for DUS testing traits. The greater the distance between the clusters, the wider the genetic diversity in the genotypes.
GFS4 can be crossed with any forage line to get maximum variation. Similar results were obtained by Chandra Sekara Reddy et al. (Reference Chandra Sekara Reddy, Audilakshmi and Seetharama2009). Maximum variability can be obtained by crossing the MS lines from clusters I and V; and the R lines from clusters VI and VIII.
Post-rainy season
The D 2 analysis showed narrow variation for the post-rainy-season genotypes for DUS testing traits. Examining the pedigree (Supplementary Table S1, available online only at http://journals.cambridge.org), it showed that all post-rainy-season varieties have M35-1 or similar durra landrace as one of the parents. The post-rainy-season grain is used for human consumption and the genotypes are based mostly on local durra, M35-1, as the local landrace has superior grain quality (consumer prefers bold lustrous grain). Though grain size and shape can be incorporated from other germplasm, the trait lustre is present in M35-1/similar durra landraces. Because of its grain lustre trait and adaptability to adverse conditions, M35-1 (which is agronomically poor) is extensively used in developing rabi varieties, and such rabi varieties are inter-crossed (adding a double dose of M35-1) to develop new varieties that make post-rainy-season varieties non-distinct due to the presence of multiple M35-1 in the post-rainy-season genotype.
GFS4 in the rainy-season sorghums and BP53 in the post-rainy-season sorghums showed highest mean values for many characters. Anantharaju and Meenakshiganeshan (Reference Anantharaju and Meenakshiganeshan2008) suggested the use of parents for hybridization based on their mean and genetic divergence for different characters. Genetic variance in populations for any trait is the result of simultaneous segregation of many genes that affect those traits. Therefore, it is expected that as the GD between parents increases, the resulting populations will segregate for more genes and have greater genetic variance (Kisha et al., Reference Kisha, Sneller and Diers1997).
Genetic divergence studies using molecular traits
Gene diversity observed in this study (mean PIC = 0.51) is similar to the values (0.40, 0.46, 0.62 and 0.58) reported by Ali et al. (Reference Ali, Rajewski, Baenziger, Gill, Eskridge and Dweikat2008), Schloss et al. (Reference Schloss, Mitchell, White, Kukatla, Bowers, Paterson and Kresovich2002), Agrama and Tuinstra (Reference Agrama and Tuinstra2003) and Smith et al. (Reference Smith, Kresovich, Hopkins, Mitchell, Dean, Woodman, Lee and Porter2000), respectively. The markers used in this study represented all the ten sorghum chromosomes and were of great utility in understanding the genetic relationships among the tested genotypes. Allelic diversity observed in this study is similar to that reported by Smith et al. (Reference Smith, Kresovich, Hopkins, Mitchell, Dean, Woodman, Lee and Porter2000). However, Menz et al. (Reference Menz, Klein, Unruh, Rooney, Klein and Mullet2004) reported a slightly higher average of 7.8 alleles per locus, probably because they studied the diversity in elite lines and exotic germplasm. Menz et al. (Reference Menz, Klein, Unruh, Rooney, Klein and Mullet2004) reported similar results with a slightly higher diversity in the R lines compared with the B lines. The forage and post-rainy-season varieties present a higher diversity of 60%. This could be because the post-rainy-season varieties are developed from local landraces especially durras, and forage genotypes are developed from crosses between the cultivated sorghum and weedy–wild sorghum. Landraces are close to the wild sorghums, and the evaluation of SSR polymorphisms indicated that landraces retained 86% of the diversity observed in the wild sorghums (Casa et al., Reference Casa, Mitchell, Hamblin, Sun, Bowers, Paterson, Aquadro and Kresovich2005).
%The relationships among the released varieties, hybrids and their parental lines were in accordance with their known pedigree information. The pedigrees of the 93 genotypes show that the rainy-season sorghums are based on inter-race/multiple race crosses (Supplementary Table S1, available online only at http://journals.cambridge.org). In India, several sorghum breeders have used numerous germplasm lines belonging to different races during the last 30 years. The majority of them belonged to durra-caudatum, durra and caudatum races (Audilakshmi et al., Reference Audilakshmi, Aruna and Kiran2003). Studies have shown that the guinea race contributes significantly (after caudatum) to higher mean and heterosis for grain yield (Reddy and Prasada Rao, Reference Reddy and Prasada Rao1993). The bicolor race and its intermediates are quite heterogenous as the molecular diversity of bicolor and associated intermediate races is not reflective of their common morphological classification (Perumal et al., Reference Perumal, Krishnaramanujam, Menz, Katilé, Dahlberg, Magill and Rooney2007). Aruna and Audilakshmi (Reference Aruna and Audilakshmi2008) reported that the germplasm lines belonging to guinea, kafir and inter-races with them are good sources for different yield components besides caudatum.
Cluster analysis showed a clear-cut demarcation of the B, R, rainy-season varieties, forage and post-rainy-season varieties. Unlike the findings of Menz et al. (Reference Menz, Klein, Unruh, Rooney, Klein and Mullet2004), the B and R lines could be demarcated in different clusters in this study. However, Menz et al. (Reference Menz, Klein, Unruh, Rooney, Klein and Mullet2004) reported that the B and R lines (US inbred lines) studied were based on sorghum races and working groups. In this study, the B and R lines formed distinct groups and two different heterotic groups could be identified which may be due to the usage of different sets of germplasm in the development of the B and R lines. Also, most of the B and R lines in the Indian breeding programme are inter-race crosses. The three different clusters, in this study, where different MS lines were grouped, are based on the pedigree of the B lines. Cluster I includes the B lines that were derived from temperate × converted tropical germplasm (African). Cluster III represents the B lines such as 296B and IMS9B, which are developed by using temperate × tropical germplasm (Indian locals). The B lines AKMS14B and 7B are developed from the MS line from the USA and found in separate cluster V. Two major heterotic pools for the R lines are identified. In cluster VI, the R lines clustered were mostly based on CS3541, the R line of many commercial hybrids. The most diversified R line AKR150 aligned with varieties.
SSR marker analysis of hybrids showed that the grouping of hybrids was reflective of known pedigrees (Smith et al., Reference Smith, Promino, Monk, Nelson, Jone and Porter2010). In the results from the present study, the highest percentage of similarity was observed between many varieties. All these varieties have a common parentage like the agronomically superior elite lines SPV475 and/or SPV462. Recently, breeders have started crossing one advanced variety with another to release rapidly new varieties. Such elite × elite crosses have an advantage of accumulating yield genes, provided each of the elite lines is from a diverse source. Crossing two elite lines with a related or shared parentage does not give any significant advantage (Audilakshmi et al., Reference Audilakshmi, Aruna and Kiran2003).
Among the B lines, those grouped in clusters I, III, V and VI showed high to low similarity with each other. The most diverse B lines were 463B (rainy-season sorghums) and 104B (post-rainy season sorghums), which showed low similarity (17%) between them. Arya et al. (Reference Arya, Verma, Sandhia, Singh and Lakhanpaul2008) reported highest diversity in 104B (MS line) and C43 (R line) using AFLP markers. Usage of these two B lines (104B and 463B) in the crossing programme may yield highly diversified B lines.
In general, hybrids were placed in the same cluster with their female parents. Similar results were reported by Dhillon et al. (Reference Dhillon, Sharma, Folkertsma and Chandra2006). They reported that the F1 hybrids were closer to CMS than to restorers. On the contrary, Arya et al. (Reference Arya, Verma, Sandhia, Singh and Lakhanpaul2008) studied diversity with AFLP markers among 24 sorghum cultivars and found hybrids grouped with male parents. AFLP markers used in the study of Arya et al. (Reference Arya, Verma, Sandhia, Singh and Lakhanpaul2008) reflected the dominant traits present in male parents and their respective hybrids. Use of a large number of different types of markers, preferably whose map positions are known, will resolve these differences, if at all the differences are real. Similarity results (10% similarity) show that the cross-combination between rainy-season (mostly caudatum and its crosses with other races) and post-rainy-season genotypes (mostly durras) will give highly diversified genotypes.
Diversity analysis by morphological markers versus molecular markers
Both the morphological marker and molecular marker methods differentiated the groups such as forage varieties and grain varieties. Grouping of the B lines was also similar by both methods. The MS lines 27B, 2077B and 2219B were placed in one cluster, 296B and IMS9B in another cluster, and AKM14B in yet another cluster. Correlations between genetic similarities and morphological similarities showed a positive significant correlation (r = 0.463; P = 0.001). Similar results were obtained by Hamza et al. (Reference Hamza, Hamida, Rebai and Moncef2004) when they compared morphological diversity with SSR diversity in barley. Fufa et al. (Reference Fufa, Baenziger, Beecher, Dweikat, Graybosch and Eskridge2005) studied a correlation among genetic estimates obtained from the pedigree method, morphological diversity, Sodium dodecyl sulphate- polyacrylmide gel electrophoresis (SDS-PAGE), molecular markers, SSR and SRAP in winter wheat cultivars. They found that correlations were low, and SSR markers had better correlations with seed proteins. Agrama and Tuinstra (Reference Agrama and Tuinstra2003) reported that genetic diversity with SSR data was highly correlated with the distances based on geographical origin and race classification in sorghum. They concluded that SSR markers were useful for the estimation of genetic similarity among diverse genotypes of sorghum. Differences among sorghum accessions determined by four clustering procedures based on agronomic characteristics were compared with clusters developed by means of RAPD markers, and the results indicate that clusters by means of agronomic descriptors closely approximate the grouping produced by RAPD markers (Dahlberg et al., Reference Dahlberg, Zhang, Hart and Mullet2002). In this study also, many rainy-season varieties showed highest similarity among themselves at the molecular level (70–94%), and generally also expressed highest similarity (found in the same cluster) based on the morphological level. However, in this study, the differentiation of the genotypes based on SSR markers was higher since the SSRs were able to distinguish clearly all the genotypes even belonging to a single cluster based on morphological data. Since morphological diversity represents functional diversity, a combination of both (morphological and molecular) is advocated to select the parents that are functionally as well as genetically diverse for developing heterotic pools and to develop genotypes with increased yield potential to break the existing yield barrier.
In conclusion, the B and R lines for the rainy season were placed in different gene pools, and further, heterotic gene pools were identified among these B and R lines. The most divergent male sterile lines were AKMS14B, 463B and 104B. The Indian B lines were more divergent when compared with the B lines from the USA, as the Indian B lines are developed from inter-race/multiple race crosses. The B and R lines form distinct gene pools, as they are inter-race/crosses from different geographical regions such as India, Africa and the USA.
The correlation between the two diversity measures was highly significant and the correspondence between the clustering based on the SSR and morphological data was relatively good. However, differentiation among the sorghum genotypes was higher by molecular markers than by morphological markers.
