Introduction

Pigeonpea (Cajanus cajan (L.) Millsp.) is an important legume crop of tropics and subtropics, grown as a field as well as backyard crop in about 82 countries (Nene and Shiela 1990). As a field crop, it was grown in 19 countries, over 4.4 m ha during 2004 (FAO 2005). India, which is considered the primary center of diversity (van der Maesen 1980) has the largest area under pigeonpea (3.3 m ha) followed by Myanmar (0.58 m ha), Kenya (0.15 m ha), Malawi (0.12 m ha), Uganda (0.08 m ha), Tanzania (0.07 m ha), Nepal (0.03 m ha), and Dominican Republic (0.01 m ha) (FAO 2005). Pigeonpea has wide adaptability to diverse climates and soils. Because of its multiple uses as food, fodder, fuel wood, rearing lac insects, hedges, windbreaks, soil conservation, roof thatches and green manure, it plays an important role in subsistence agriculture. However, the average world productivity of pigeonpea is rather low (709 kg ha−1) (FAO 2004), indicating an urgent need for improving the genetic potential of the crop.

The contribution of landraces as source material for crop improvement has been substantial. In the past, the majority of released pigeonpea cultivars have been developed through selection from landraces (Pathak 1970; Laxman Singh et al. 1990). To meet the challenges in crop improvement, efforts were made to widen the genetic base by collecting and conserving germplasm across the world, which led to assembly of large collections at national and international genebanks. The genebank at ICRISAT holds 13,632 accessions of pigeonpea including 555 accessions of wild relatives from 74 countries. Of these 9,759 accessions were acquired mainly by donations from various institutes in different countries, and 3,873 accessions by launching 99 germplasm collecting missions in 33 countries (Upadhyaya et al. 2005). The cultivated accessions (13,077) are from 66 countries. However, utilization of germplasm in crop improvement is still very limited. The accessibility and utilization of a collection is inversely related to its size (Frankel and Soule 1981). Development of a core collection is one way to enhance the use of genetic resources in crop improvement (Frankel 1984). The core collection derived from an existing entire collection comprises a chosen set of accessions that represent a wide-cross section of genetic spectrum available in a given crop species with minimum amount of duplication (Brown 1989). Core collection has value in crop improvement programs and due to its reduced size, it can be evaluated extensively with limited resources. Pigeonpea core collection comprising 1,290 accessions of cultivated pigeonpea from 53 countries was developed at ICRISAT (Reddy et al. 2005). The main aim of the present study was to assess diversity for different morpho-agronomic characters in the pigeonpea core collection to enhance the use of genetic resources in pigeonpea improvement.

Materials and methods

The experimental material consists of the pigeonpea core collection comprising 1,290 accessions including advanced cultivars (9), breeding material (466), landraces (810) and others (5), representing 12,153 accessions that are conserved in the ICRISAT genebank (Reddy et al. 2005). The experiment was conducted at ICRISAT research farm, Patancheru (18° N, 78° E, 545 m a.s.l., and about 600 km away from the sea) in an augmented design. Four control cultivars; representing different maturity groups were used. Each control cultivar was sown after every nine-test accessions. Thus, there were four checks and 36 test accessions in a block of 40 entries. The checks used were ICP 11543 (extra early), ICP 6971 (early), ICP 8863 (medium maturing) and ICP 7221 (late maturing). Sowing was done on 30 June 2004 in rainy season on vertisols (Kasireddipally Series-Isohyperthermic-Type-Pellustert). Each accession was sown as single row of 4-meter length, spaced 75 cm apart. Plant to plant spacing within the row was 50 cm. Crop was fertilized with 20 kg N and 40 kg P2O5 ha−1 as a basal dose. The experiment was provided with life saving irrigation and protected from weeds, pests and diseases. The crop was harvested during November to February 2005 depending on the maturity of the accessions.

The observations on 18 qualitative (vigor, growth habit, plant pigmentation, stem thickness, flower base color, streak color, streak pattern, flowering pattern, pod color, pod shape, pod hairiness, seed color pattern, primary seed color, secondary seed color, seed eye color, seed eye color width, seed shape and seed hilum) and 16 quantitative characters were recorded following the ‘Descriptors for Pigeonpea’ (IBPGR and ICRISAT 1993). Observations on all qualitative and six quantitative characters (days to 50% flowering, days to 75% maturity, 100-seed weight, harvest index, shelling percent, and plot seed yield) was recorded on plot basis. Observations on remaining ten quantitative traits (leaf size, plant height, number of primary, secondary and tertiary branches, number of racemes, pod bearing length, pods per plant, pod length, seeds per pod) were recorded on three random plants from each plot.

Using the standard maturity classification that was developed for pigeonpea and being followed at ICRISAT (ICRISAT 1978), the accessions of core collection were grouped as extra early (<80 days to 50% flowering), early (81–100 days to 50% flowering), medium (101–130 days to 50% flowering) and late maturity (>130 days to 50% flowering) group. Days to 50% flowering refer to the number of days from the date of sowing to the date when 50% of the plants in a given plot have atleast one open flower. This classification of core collection resulted in 37 accessions belonging to the extra early group, 87 accessions to the early group, 669 accessions to the medium maturity group, and 497 accessions to the late maturity group. Phenotypic proportions for the 18 qualitative traits were calculated based on accessions in each maturity group.

The data were analyzed following REML (Residual Maximum Likelihood) approach using GENSTAT 6.1. Mean, range, and variances for all quantitative traits were calculated for each of the maturity groups. The means of different maturity groups for all traits were compared using the Newman–Keuls procedure (Newman 1939; Keuls 1952). The homogeneity of variances of maturity groups was tested using Levene’s test (Levene 1960). Phenotypic correlations were estimated among all traits in the entire core as well as in all maturity groups and identified number of significant and meaningful correlations. Principal coordinate analysis (PCoA) of 18 qualitative traits and principal component analysis (PCA) of 16 quantitative traits were performed to find out the relative importance of different traits in capturing the variation in core collection or its four subsets. A phenotypic distance matrix was created by calculating the differences between each pair of accessions for each character. The diversity index was calculated by averaging all the differences in the phenotypic values for each trait divided by respective range (Johns et al. 1997). Shannon and Weaver diversity index (H′) was used as a measure of phenotypic diversity for each trait (Shannon and Weaver 1949). The index was estimated for each trait over all accessions and for all traits within each maturity group.

Results and discussion

Because pigeonpea is often cross-pollinated, variation for different characters was considerably high. Out-crossing in pigeonpea ranges from 3% to 26% and varies according to location, genotype, and intensity of insect population and time of flowering (Bhatia et al. 1981; Saxena et al. 1990; Reddy et al. 2004). Therefore, heterogeneity for almost all traits exists in the germplasm samples collected from farmer’s fields. The intra-accession variation that was available in the collected accessions is maintained in the genebank. When the accessions were characterized, more than one class of traits within each accession was observed. Frequency distribution of different classes within each qualitative character, mean values, variance, character associations, principal components and diversity in four maturity groups and the entire core subset is discussed hereunder.

Qualitative characters

Vigor

Vigor of plants in accessions at 50% flowering was recorded visually as low (score 3), intermediate (score 5) and high (score 7). The medium and long duration genotypes predominant in the world collection, and show more plant vigor, due to their larger seeds and leaves. Highly vigorous accessions were predominant in the entire core (70.3%), medium maturity group (77.95%) and late maturity group (75.7%) (Remanandan et al. 1988). On the other hand, accessions with relatively smaller seeds in the extra early and early groups showed intermediate vigor (Appendix 1).

Growth habit

It was recorded in three classes: compact, spreading and semi-spreading. In the entire core, 91.1% accessions were of semi-spreading habit followed by compact type with 8.7%. A similar pattern was exhibited in all the maturity groups. The extra early maturity group contained a larger proportion of compact cultivars (24.3%), no accessions in this and the early groups were of spreading habit. Most of the accessions in extra early and early maturity groups were the products of breeding programs, which aimed at developing early maturing, short statured and determinate varieties (Remanandan et al. 1988).

Plant pigmentation

The stem color recorded as plant pigmentation was classified into five classes (green, purple, green + purple, green + red, purple + green). About 97% of the core accessions have plants with green stem. Remaining classes are negligible and are found only in the medium and late maturity groups. Conspicuously, all accessions of extra early and early maturity groups had green stem.

Stem thickness

It was recorded visually as thin, intermediate and thick at 50% flowering. Thick stems are predominant in the entire core (90.7%), medium (95.6%), and late (96.8%) maturity groups. High proportion of accessions in the extra early group (32.4%) produced thin stems. Increase in stem thickness with maturity duration may be attributed to the longer growth period.

Flower base color

Of the six classes (yellow, orange yellow, red, light yellow, ivory and mixture of light yellow and yellow), yellow flower base color was dominant (>95% accessions) in the core subset as well as in all maturity groups. Only the accessions that produced flowers with yellow and orange yellow flower base color were found in the extra early and early maturity groups.

Flower streak color

Color of streaks on vexillum was recorded as red, purple, yellow, none (no other color on streaks) and red (more plants without any other color to the streaks and few plants with red streaks), red and none (more plants with red streaks and few plants without any other color to the streaks), and none (all plants without any other color to the streaks). Accessions having flowers with red streaks were maximum (>88%) in the core as well as in all four maturity groups, and all accessions in the extra early group had red streaks.

Flower streak pattern

In total, 23 classes were observed for this trait in the entire core and maximum accessions were with few streaks (69.7%) followed by medium streaks (12.4%). A similar pattern was observed in all four maturity groups (Appendix 1).

Flowering pattern

Four classes, [Indeterminate (NDT), determinate (DT), and mixture of both patterns (DT+NDT, and NDT+DT)] were observed for this character with a maximum of 95.8% NDT type in the entire core. Proportion of NDT was 80.5% in the early, 97.2% in the medium and 100% in the late maturity group. On the other hand, maximum DT (48.7%) accessions were observed in the extra early group, which could be due to high proportion of targeted selection of breeding programs (Remanandan et al. 1988).

Pod color

It was recorded in eight classes (green, light green, purple and dark purple and four are heterogeneous classes in different combinations of these colors). Green pod color was predominant (>89%) in the entire core, and in all maturity groups. The other pod color that appeared (<3%) in all the four maturity groups and in entire core was purple. All other colors appeared only in the late maturity group (Appendix 1).

Pod shape

Cylindrical pod shape was most common in the entire core (82.6%), extra early (83.8%), early (90.8%), medium (93.1%) and the late (67.2%) maturity groups. Remaining accessions had flat pods (Appendix 1).

Pod hairiness

This is a monomorphic trait and all accessions in the entire core subset had pubescent pods.

Seed color pattern

In total, 11 classes (mottled, speckled and plain and different combination of these three patterns) were observed in the core. Maximum number of accessions had plain (>59.4%) seed color pattern followed by speckled (>5.4%) in the core as well as in all maturity groups.

Primary seed color

Large diversity for primary seed color was observed in the core. A total of 64 classes (14 solitary and 50 classes having heterogeneity) were observed. In the core, accessions having orange seed color was maximum (51.0%) followed by cream (21.6%) and light cream (3.1%) seed colors. All the four maturity groups showed similar pattern as that of entire core collection (Appendix 1).

Secondary seed color

This trait was recorded in 20 classes (13 solitary and 7 heterogeneous). Absence of secondary seed color (i.e. uniformly one color of the seed coat) was more common (68.8%) in the core as well as in all the maturity groups. However, white color as secondary seed color was observed in considerable proportions in the core (20.9%), extra early (21.6%), early (11.5%), medium (12.7%) and the late (33.2%) maturity groups.

Seed eye color

It was recorded in 22 classes (13 solitary and nine heterogeneous). Brown eye color was predominant in the core and also in all the maturity groups (>83%). Brown and reddish brown seed eye color accessions were found in all maturity groups. All seed eye colors, except cream and the mixture of purple and brown appeared in the late maturity group (Appendix 1).

Seed eye color width

Six classes (three solitary and three heterogeneous) were recorded for this trait in the entire core. Narrow eye color width was predominant (>76%) in the entire core as well as in all maturity groups.

Seed shape

In this trait, six classes (oval, square, elongate, globular and two heterogeneous classes) were recorded in the core. Oval shape was most common (>94% accessions) in the entire core and in all the maturity groups. Square shape was the other seed shape that appeared in the entire core as well as in all maturity groups in some proportion (<3.0%).

Seed strophiole

This is a primitive trait differentiating wild and the cultivated pigeonpea. It is prominent in almost all wild relatives. Observation on seed strophiole was recorded as absent or present. Maximum frequency of accessions (95.5%) in the core did not produce the strophiole. Only 4.4% in the core, 8.1% in the extra early group and 8.3% accessions in the late maturity group produced strophiole.

Quantitative characters

Range, mean and variance

REML analysis indicated that the variance components due to genotypes were significant for all quantitative characters in the entire core as well as in all the maturity groups suggesting significant variability for all traits in the core and in the four maturity groups (data not given). The mean, range and variances for 16 quantitative traits of the four maturity groups are given in Table 1. The four maturity groups differed significantly for different traits. The differences between extra early, early, medium and late maturity groups were significant for days to 50% flowering, plant height, days to 75% maturity, number of secondary branches and seed yield (Table 1). Mean values in the early group were significantly greater than in the extra early group for number of racemes, pods per plant and harvest index. Mean values in the late maturity group were significantly greater than medium maturity group for seven traits (leaf size, days to 50% flowering, plant height, days to 75% maturity, pod length, seeds per pod and 100 seed weight). The extra early group differed significantly with the late maturity group for all characters except 100-seed weight. High seed weight in the extra early group could be the result of selection by breeders in the recent years to develop short duration and determinate varieties (which can produce large seeds), adaptable to different cropping systems (Laxman Singh et al. 1990). Majority of determinate type accessions were extra early and early and are products of breeding programs (Remanandan et al. 1988). Medium maturity group having high proportion of semi-spreading and indeterminate accessions recorded significantly higher mean values for number of secondary branches, pod bearing length, shelling percent, and seed yield than all other maturity groups. In the present study, the extra early and the early maturity groups were not found promising for any of the other characters under study except days to 50% flowering, plant height and days to 75% maturity, which is in agreement with findings of Remanandan et al. (1988). Among four maturity groups, the medium maturity group was found more promising for important agronomic characters (for e.g. number of primary, secondary and tertiary branches, number of racemes, pod bearing length, pods per plant, harvest index, shelling percent and seed yield (kg ha−1). However, the late maturity group was good source for long pods and large seeds (for vegetable pigeonpea).

Table 1 Mean, range and variances for quantitative characters in different maturity groups of the pigeonpea core collection, at ICRISAT center Patancheru, India
Full size table

The variances were heterogeneous (P < 0.0001) for all quantitative traits, except for days to maturity and number of tertiary branches. The late maturing group recorded maximum variance for all traits, except plant height, days to 75% maturity, pod bearing length and shelling percent. Higher variance was observed in the extra early group for days to 75% maturity and shelling percent and the early group for plant height and pods per plant (Table 1).

Range of variation was highest in the early maturity group for plant height, days to 75% maturity and pod bearing length; in the medium maturity group for number of primary branches, pod length, seeds per pod, 100-seed weight, harvest index, shelling percent and seed yield; and in the late maturity group for leaf size, days to 50% flowering, number of secondary and tertiary branches number, number of racemes and pods per plant (Table 1). On the other hand, range was lowest in the extra early group for all the traits, except seeds per pod, harvest index, and shelling percent. Range for leaf size, days to 50% flowering, number of secondary and tertiary branches, racemes number and pods per plant increased with the increase in maturity duration. This increase in variation from the extra early group to the late maturity group for above traits may be attributed to the high proportion of highly photoperiod sensitive indeterminate accessions in medium and late maturity groups (Remanandan et al. 1988).

Correlations

Phenotypic correlations were estimated among all quantitative characters in all the maturity groups and in the entire core. Within the group, correlations were tested for significance (Snedecor and Cochran 1980). In the present study, 37 correlations in the extra early group, 99 in the early group, 105 in the medium group, 111 in the late maturing group and 118 in the entire core were found to be significant at 5% probability level. Similarly, 28 correlations in the extra early group, 85 in the early group, 98 in the medium maturity group, 109 in the late maturity group and 117 in the entire core were significant at 1% probability level. As more than 50% variation in one character is predicted by another character, the correlation coefficients greater than 0.707 and lower than  − 0.707 were considered as biologically meaningful (Skinner et al. 1999). Two such meaningful correlations were found in the entire core and in all the four maturity groups. Meaningful associations were also found between number of secondary branches and pods per plant in the extra early group (r = 0.756), late maturity group (r = 0.776) and entire core (r = 0.728); and between number of racemes and pods per plant in the extra early (r = 0.841), early (r = 0.890), medium (r = 0.866) and late (r = 0.923) maturity groups and entire core (r = 0.901) (Table 2). Magnitudes of correlations indicate maturity dependent positive associations among these characters. These results suggest that number of secondary branches and racemes per plant should be recorded ignoring the number of pods per plant, which takes relatively more time to record, to improve the seed yield (r value between number of pods and yield = 0.724).

Table 2 Meaningful correlations (r > 0.707) in different maturity groups of pigeonpea core collection and in the entire core collection, at ICRISAT center, Patancheru, India
Full size table

Principal coordinates and principal components

The percentage variation explained by the first five principal coordinates (PCo) was 69.9% in the entire core, 80.5% in the extra early group, 71.9% in the early maturity group, 82.6% in the medium maturity group and 71.0% in the late maturity group. The PCo 1, the most important coordinate accounted for 34.69% in the entire core, 32.39% in the extra early group, 31.64% in the early group, 39.46% in the medium maturity group and 38.73% in the late maturity group. Only seven traits (vigor, growth habit, stem thickness, flowering pattern, pod shape, primary seed color and secondary seed color) showed high correlation with PCo scores and occurred at least three times out of five, indicating their importance in explaining the variation.

The percentage of variation explained by the first five principal components (PC) and the vector loadings for each agronomic character and PC are given in Table 3. The first five PCs explained 78.81% variation in the entire core, 76.9% in the extra early group, 74.9% in the early group, 66.2% in the medium maturity group and 77.8% in the late maturity group and reduced the original 16 traits to 12 in the entire core and in all maturity groups, except medium maturity group. The PC1, which is most important component accounted for 38.62% variation in the entire core, 29.54% in the extra early group, 43.57% in the early group, 30.32% in the medium maturity group and 43.67% in the late maturity group. The eigen values for the entire core, extra early, early, medium and late maturity groups were 6.18, 4.73, 6.97, 4.85 and 6.83, respectively. The PC1 separates accessions on six traits (number of secondary and tertiary branches, number of racemes, pods per plant, pod length and 100-seed weight) in the entire core; on five traits (number of secondary and tertiary branches, number of racemes, pod bearing length and pods per plant) in the extra early group; on seven traits (days to 50% flowering, plant height, number of secondary branches, number of racemes, pods per plant, shelling percent and plot seed yield) in the early group; on five traits (leaf size, number of secondary branches, number of racemes, pods per plant and pod length) in the medium maturity group; and on seven traits (leaf size, number of secondary branches, number of racemes, pods per plant, pod length, 100 seed weight and yield) in the late maturity group. Considering the analyses of all maturity groups and the entire core, nine traits (days to 50% flowering, plant height, number of secondary and tertiary branches, number of racemes, pods per plant, pod length, 100-seed weight and yield) occurred at least four times in the first five PCs, indicating their importance for characterization of pigeonpea germplasm accessions. Remaining characters (leaf size, days to 75% maturity, number of primary branches, pod bearing length, seeds per pod, harvest index and shelling percent) had no contribution in explaining variation in the first five PCs of the entire core as well as all the four maturity groups, indicating their low importance as pigeonpea descriptors for characterization.

Table 3 Vector loadings and percentage of variation explained by the first five principal components in different maturity groups of pigeonpea core collection, at ICRISAT center, Patancheru, India
Full size table

Diversity index

The grouping of similar genotypes depends on the dissimilarity among them, which can be determined by a phenotypic diversity index (Table 4). The average diversity was highest in the extra early group (0.2619) and lowest in the medium maturity group (0.1543). Range of phenotypic diversity was highest in the late maturity group (0.0363–0.6337) and lowest in the extra early group (0.1042–0.5769) and it was almost same in the medium maturity group (0.0007–0.5236) and the entire core (0.0005–0.5229) (Table 4). High range of phenotypic diversity in the late maturity group may attribute to indeterminate flowering pattern and high percent of out-crossing during prolonged flowering period. When compared within each maturity group between accessions, lowest phenotypic diversity index was observed between ICP 10923 and ICP 14829 in the extra early group (0.104); between ICP 12989 and ICP 6965 in the early group (0.066); between ICP 11790 and ICP 15016 in the medium maturity group (0.001); between ICP 11012 and ICP 12235 in the late maturity group (0.036); and between ICP 11790 and ICP 15016 in the entire core (0.001). Highest phenotypic diversity index was observed between ICP 14444 and ICP 14471 in the extra early group (0.577); between ICP 16198 and ICP 7782 in the early group (0.561); between ICP 6523 and ICP 8265 in the medium maturity group (0.524); between ICP 12079 and ICP 7941 in the late maturity group (0.634); and between ICP 14444 and ICP 11246 in the entire core (0.523). Mean phenotypic diversity index was highest (0.262) between accessions of extra early group. The accessions with high phenotypic diversity index in the extra early maturity group would be useful in pigeonpea breeding programs.

Table 4 Phenotypic diversity index in four maturity groups of core and the core collection of pigeonpea germplasm, at ICRISAT center, Patancheru, India
Full size table

The Shannon–Weaver diversity index (H′) was calculated to assess the phenotypic diversity for all characters in different maturity groups and the entire core collection (Table 5). The diversity index is used as a measure of allelic richness and evenness. Low H′ indicates extremely unbalanced frequency classes and lack of genetic diversity in a trait. The estimates of H′ were made for each trait and four maturity groups. The diversity values were variable among traits and among maturity groups. So, the diversity within a maturity group depends upon the traits. The H′ values averaged over the four maturity groups ranged from 0.04 ± 0.03 (plant pigmentation) to 0.73 ± 0.07 (primary seed color) among qualitative characters; and from 0.45 ± 0.03 (number of tertiary branches) to 0.61 ± 0.01 (number of primary branches among quantitative characters. The H′ values averaged over both qualitative and quantitative traits ranged from 0.36 ± 0.04 (extra early and early groups) to 0.42 ± 0.04 (late maturity group). In the late maturity group, possibly due to the increased out-crossing during prolonged flowering in indeterminate accessions, diversity was highest for all qualitative characters except, growth habit, stem thickness and flowering pattern. Diversity for plant pigmentation, flower base color, flower streak color, pod color and seed eye color increases with maturity duration among accessions.

Table 5 Shannon–Weaver diversity index (H′) for qualitative and quantitative characters in different maturity groups of pigeonpea core collection, at ICRISAT center, Patancheru, India
Full size table

Accessions in the medium maturity group were highly diverse (H′ = 0.59 ± 0.01) for quantitative traits followed by the late (H′ = 0.58 ± 0.01), early (H′ = 0.57 ± 0.01) and extra early (H′ = 0.53 ± 0.02) maturity groups. Diversity level increases from extra early group to late maturity group for leaf size, pod bearing length and pod length indicating the association of these traits with late maturity. High diversity for most of the characters in the medium and late maturing groups having maximum indeterminate accessions could be due to evolution of new variation because of prolonged out-crossing. Higher sensitivity of the crop for photoperiod and temperature during long crop growth period might have also contributed for high diversity in late maturity group. On the other hand, extra early and early maturing groups having maximum determinate type accessions, which are mostly the products of breeding programs and less sensitive to photoperiod and temperature has probably resulted in lower diversity (Saxena and Sharma 1990).

Conclusions

As evidenced in the present study, medium and late maturity groups are more promising than extra early and early groups for diversity, particularly, for agronomic traits. However, accessions in the extra early and early maturity groups are less sensitive to temperature and photoperiod and adaptable in different cropping systems (Saxena and Sharma 1990). Days to 50% flowering, plant height, number of secondary and tertiary branches, number of racemes, pods per plant, pod length, 100 seed weight and seed yield were found as more important for characterizing pigeonpea germplasm accessions. Future germplasm collections for pigeonpea should be preferably focus on the traits mentioned above to update the present core collection. The information generated in this study could be used to reduce the size of the core collection further to develop a mini core subset (Upadhyaya and Ortiz 2001). This will provide easy access to the pigeonpea collection through replicated multilocational evaluation of mini core to identify useful parents for use in crop improvement programs.