The plant material comprised of 18 accessions of pigeonpea including 16 cultivars of C. cajan and 2 accessions of its wild progenitor (C. albicans and C. lineatus) (Table 1). All the above material was procured from core collection maintained at Indian Institute of Pulses Research, Kanpur, India.

Pigeonpea seeds were surface sterilized with 0.2% HgCl₂ and washed 3 - 4 times and 3grown in pots in sunlight for three weeks. DNA was isolated from young leaves using HiPurA™ Plant genomic DNA Miniprep Purification spin kit (Himedia Laroratories Pvt. Ltd). Quality and quantity of isolated DNA was checked by spectrophotometry as well as by 0.8% agarose gel electrophoresis. The DNA yield obtained was in the range of 1.0 μg to 3.0 μg.

A total of 30 decamer primers (GC content 60% - 90%) were used for RAPD analysis. Out of 30 primers 19 (OPA-19, OPB-14, OPB-17, OPB-19, OPC-05, OPH02, OPH-03, OPH-05, OPH-10, OPH-11, OPH-12, OPM-07, OPP-07, OPP-09, OPAQ-05 OPAQ-18, OPAQ-19, OPAZ-05 and OPAZ-18) belong to Operon series (Operon Technologies USA) and 11 (P-23;GTAGGCGTCG, P-24;GGCTCGTACC, P-25;GACCCCGGCA, P-26; CAGGGGACGA, P-27;CGCCACGTTC, P-28;GCCTCCTACC, P-29;GGCGTCGGGG, P-30;CAGGGCCGCT, P-31;CTCTCCGCCA, P-32;CTCGGCTGGA and P-33;AGGCCCGATG) were selected randomly. The PCR amplification protocol is described in our published paper [26] .

ISSR-PCR was performed by means of 15 primers out of which 11 were 3’-anchored [GAGAGAGAGAGAGA- GAT, GAGAGAGAGAGAGAGAC, GAGAGAGAGAGAGAGAA, ACACACACACACACACCT, AGAGA- GAGAGAGAGAGCT, GTGTGTGTGTGTGTGTA, GTGTGTGTGTGTGTGTCT, CACCACCACGC, CTCT- CTCTCTCTCTCTGC, GAGAGAGAGAGACC, AGAGAGAGAGAGAGAGTG] and four were non-anchored [GACAGACAGACAGACA, GACTGACTGACTGACT, GTGTGTGTGTGTGTGTGT, ACTGACTGACTGA- CTG] primers. Each amplification reaction contained 10 mM Tris-HCl, pH 9.0, 2 mM MgCl₂, 50 mM KCl, 0.01% gelatin, 200 µM of each dNTP, 4 µM of primer, 0.6 Units of Taq DNA polymerase (Bangalore Genei, Bangalore, India) and 30 ng of genomic DNA in total volume of 25 µl. The PCR amplification was carried out for 40 cycles in a thermal cycler (PTC-200, Bio-Rad USA). The reaction had initial denaturation step at 94˚C for 5 min, followed by 40 cycles of 94˚C for 1 min, annealing temperature for 1 min, 72˚C for 2 min. The final extension step was at 72˚C for 5 min. Amplified products were separated on 1.8% agarose gels having 0.5 µg・ml⁻¹ of the ethidium bromide at 50 V for 3 h. The gels were observed under a UV light source in a gel documentation system (BIOVIS Gel, Expert Vision Labs Pvt. Ltd, India).

Clearly resolved, unambiguous polymorphic band were scored visually for their presence or absence with each

RAPD and ISSR primer. The scores were obtained in the form of a matrix with “1” and “0”, which indicate the presence and absence of bands in each variety respectively. The individual RAPD and ISSR profiles as well as pooled data (RAPD+ISSR) were analysed for genetic diversity analysis. Jaccard’s similarity coefficient [27] was estimated from these binary data using FreeTree [28] software. The resulting similarity matrix was used for UPGMA based dendrogram construction. Support for clusters obtained was evaluated by bootstrap analysis with FreeTree and Tree View [29] software through generating two thousand samples by re-sampling. PCA was also done to check the results of UPGMA based clustering using EIGEN module of NTSYSpc [30] .

In RAPD analysis, our earlier results [26] based on 30 primers tested were included in the analysis. Out of these, 22 primers yielded a total of 212 fragments (average 9.6 bands/primer), of which 151 amplicons (71.2%) were polymorphic, the number of polymorphic bands per primer ranged from 2 to 17, the average being 6.8 (Table 2). Primers OPB17, OPB-19 OPM-07, OPAZ-05, OPAZ-18, P-24, P-26, P-27, P-29, and P-30 were the most informative primers as 75% or more of the amplicons were polymorphic. These ten highly polymorphic primers produced a total of 83.1% polymorphism. Amplification pattern obtained by primer P-27 is given in Figure 1. A total of nine intense, distinct, species-specific markers were amplified by primers OPP-09, P-23, P-24, P-25, P-27, P-28, P-29, P-30 in all the 16 cultivars of pigeonpea but absent in both the wild species.

ISSR primers produced different numbers of DNA fragments, depending on their simple sequence repeat motifs. Out of the 15 primers tested, 10 produced reproducible and polymorphic patterns. The 10 primers yielded a total of 100 fragments (average 10 bands/primer), of which 93 amplicons (93%) were polymorphic, the number of polymorphic bands per primer ranged from 5 to 15, the average being 10 (Table 2). All the 10 Primers [GAGAGAGAGAGAGAGAC, GAGAGAGAGAGAGAGAA, GACAGACAGACAGACA_, CACCACCACGC, GTGTGTGTGTGTGTGTGT, AGAGAGAGAGAGAGAGCT, ACTGACTGACTGACTG, AGAGAGAGA- GAGAGAGTG, GAGAGAGAGAGACC, ACACACACACACACACCT] were the most informative primers as 71% or more of the amplicons were polymorphic. Amplification pattern generated by primer GAGAGA- GAGAGAGAGAC is given in Figure 2.

The genetic diversity among pigeonpea cultivars and wild relatives was analyzed by UPGMA method. In RAPD analysis, 151 polymorphic markers amplified were analyzed for genetic diversity. Jaccard’s similarity coefficients among C. cajan cultivars ranged from 0.434 - 0.714 [26] Lowest similarity was obtained between GT-101 and Azad (0.434) followed by that between Birsa arhar and Vamban-1 (0.451). Highest similarity was obtained between Pusa-2002 and Manak (0.714) followed by that between ICPL-87 and Azad (0.703). In the dendrogram 16 genotypes were grouped into two distinct clusters; cluster I and cluster II with seven and nine genotypes, respectively, whereas the two wild species did not group into clusters (Figure 3).

Genetic diversity analysis based on 93 ISSR markers yielded Jaccard’s similarity coefficients range from 0.328 - 0.827 among C. cajan cultivars (Table not shown). Lowest similarity was obtained between Pusa-2002 and ICPL-85063 (0.328) followed by that between Pusa-2002 and ICPL-87 (0.379). Highest similarity was obtained between GT-101 and Vamban-1 (0.827) followed by that between GT-101 and BDN-2 (0.813). A dendrogram based on UPGMA analysis with ISSR data has revealed three main clusters viz., cluster I, II and III with two, three and ten genotypes, respectively (Figure 4). Cluster III was further divided into three sub-clusters (IIIA, IIIB and IIIC). The two wild species and one cultivar Pusa-2002 did not group into clusters.

The RAPD and ISSR data were combined for UPGMA cluster analysis. The 244 markers (151 RAPD + 93 ISSR) revealed varying degrees of genetic relatedness among wild and cultivated genotypes. Jaccard’s similarity coefficients among C. cajan cultivars revealed very narrow range from 0.477 - 0.720 and high range between C. cajan cultivars and its wild relatives C. albicans (0.240 - 0.331) and C. lineatus (0.163 - 0.193) (Table 3). Lowest similarity was obtained between Pusa-2002 and ICPL-85063 (0.477) followed by that between Pusa-2002 and Vamban-1 (0.487) and between Pusa-84 and ICPL-85063 (0.487). Highest similarity was obtained between Vamban-1 and AK-101 (0.720) followed by that between Vamban-1 and GT-101 (0.705). The UPGMA dendrogram obtained from the cluster analysis of RAPD+ISSR data gave three distinct clusters viz., cluster I, II and III with two (Pusa-2002 and Manak), six (Vamban-1, AK-101, GT-101, Paras, T-15-15 and ICPL-85063) and eight (Pusa-84, Birsa-Arhar, JKM-7, Azad, MA-3, BDN-2, C-11 and ICPL-87) genotypes, respectively, whereas the two wild species did not group into clusters (Figure 5). Principal Components Analysis (PCA) of the combined RAPD and ISSR revealed that the PC1, PC2, and PC3 accounted for 56.26%, 5.71%, and 4.97% of the variation, respectively. Together, the first three PCs accounted for 66.99% of the total variation. Two dimensional (Figure 6) and three dimensional (Figure 7) plots were prepared by using the first two and first three PCs, respectively. In both 2-D and 3-D plots the wild species C. lineatus and C. albicans occupied distant positions from the C. cajan. The wild relatives occupied lower left and C. cajan genotypes clustered to the upper right corner of the biplot. The pattern of genetic divergence obtained by PCA is in close agreement with the results of UPGMA based cluster analysis. In both the methods wild species formed out groups whereas all the 16 cultivated genotypes showed narrower range of diversity.

In the present investigation, two multilocus marker systems viz., RAPD and ISSR were evaluated for their efficiency in revealing the genetic diversity among the pigeonpea genotypes and its wild relatives studied. RAPD markers produced 71.2% polymorphic band with an average 9.63 band per primer. ISSR markers amplified 93% polymorphic band with an average 10 band per primer. Clearly the ISSR markers are more efficient than the RAPD assay, as they revealed more polymorphism than that of RAPD markers. Similar results were obtained in several other studies involving pigeonpea [31] [32] , groundnut [33] and wheat [22] . Many authors reported the higher reproducibility of ISSR-PCR is because of use of longer primers as compared to RAPD [34] - [36] .

Jaccard’s similarity coefficients among 16 C. cajan cultivars used ranged between 0.434 - 0.714 (RAPD), 0.328 - 0.827 (ISSR) and 0.477 - 0.720 (RAPD + ISSR). ISSR marker showed relatively higher range of similarity coefficients as compared to RAPD and RAPD + ISSR data in C. cajan cultivars. Similar results were also found in chickpea [37] and Vigna [24] . A possible explanation for the difference in resolution of RAPDs and ISSRs is that the two-marker techniques target different portions of the genome. The ability to resolve genetic variation among different genotype may be more directly related to the number of polymorphisms detected with each marker technique rather than a function of which technique is employed.

Cluster analyses revealed three main clusters viz., cluster I, II and III in RAPD + ISSR dendrogram, which has similar pattern to that of ISSR dendrogram whereas, the dendrogram based on RAPD showed some variation in the clustering of genotypes. Sixteen genotypes were grouped in three clusters in ISSR and ISSR + RAPD data; whereas the dendrogram based on RAPD data showed two main clusters. Clustering of genotypes within groups was not similar when RAPD and ISSR derived dendrogram were compared, whereas the pattern of clustering of the genotypes remained more or less the same in ISSR and combined data of RAPD + ISSR. In all the three dendrograms, both the wild species C. lineatus and C. albicans formed out groups as they did not cluster with any of the pigeonpea cultivars. Pusa-2002 showed more diversity from other cultivars in ISSR dendrogram; this pattern was not found in RAPD and RAPD + ISSR dendrograms. On comparing these results it can be concluded that ISSR marker has more differentiating power than RAPD. In ISSR and RAPD + ISSR dendrograms, we obtained the almost same sample distributions while dendrogram from RAPD is little different. The differences in clustering pattern of genotypes using RAPD and ISSR markers may be the different number of PCR products analysed, this confirms the importance of the number of loci and their coverage of the overall genome for producing reliable estimates of genetic relationships among cultivars [38] .

The dendrogram based on RAPD + ISSR markers clearly separated various biotic and abiotic stress tolerant genotypes together (Figure 5). All the members, except Pusa-84, of cluster III are tolerant to certain biotic and abiotic stresses viz., Birsa arhar, JKM-7, BDN-2, C-11, ICPL-87 are all tolerant to wilt while JKM-7 Azad and MA-3 are Phytophthora blight sterility mosaic and pod fly resistant, respectively. However, two wilt resistant genotype viz., Paras and AK-101 could not group in cluster III instead it occupied place in cluster II. More or