Introduction
Pigeonpea [Cajanus cajan (L.) Millspaugh], a major grain legume crop of tropics and subtropics is grown in about 82 countries lying between 30°N and 30°S latitudes. It has wide adaptability to diverse climates and is mainly grown for its multiple uses. Crop statistics are reported only for 21 countries and, during 2009, pigeonpea was cultivated in 4.86 mha with an annual production of 4.10 million tons and an average productivity of 844 kg/ha (FAO, 2009). India has the largest area under pigeonpea (3.50 mha), followed by Myanmar (0.57 mha), Kenya (0.19 mha), Malawi (0.12 mha), Uganda (0.09 mha), Tanzania (0.07 mha), Nepal (0.02 mha) and Dominican Republic (0.02 mha) (FAO, 2007).
Crop wild relatives (CWR) are important components of agro-ecosystems as potential gene contributors for crop improvement programmes. When the genetic variability is narrow and levels of resistance to various biotic and abiotic stresses in cultivated germplasm are low, the identification and incorporation of useful genes from wild species becomes key to sustain crop productivity. Wild relatives become more important when they posses traits of agronomic importance in addition to resistance to the biotic and abiotic stresses and can be hybridized easily with cultivated species. Despite their importance as the source for several beneficial genes, CWR have not received due attention. Though considerable progress in pigeonpea improvement has been made, the crop still suffers from several problems such as biotic and abiotic stresses, low pod setting, long duration, etc. Many wild relatives of pigeonpea have evolved to survive drought, floods, extreme heat and cold, and they have become adapted to cope with many natural hazards and often developed resistance to various biotic and abiotic stresses. Some wild species of pigeonpea such as Cajanus scarabaeoides (L.) Thou. possess genes for earliness and high pod setting (Upadhyaya, Reference Upadhyaya2006).
Genebank at ICRISAT conserves 13,632 accessions of pigeonpea germplasm from 74 countries, including 555 accessions of wild relatives belonging to 66 species of six genera. The collection of wild relatives includes 102 accessions of C. scarabaeoides. C. scarabaeoides is crossable with cultivated pigeonpea (Pundir and Singh, Reference Pundir and Singh1985) and possess several useful traits such as early flowering ( < 80 d to 50% flowering) and high pod setting (74%) (Upadhyaya, Reference Upadhyaya2006); tolerance to pod borer ( < 10% pod damage), pod fly and pod wasp (Sharma, Reference Sharma2006); water logging (Reddy et al., Reference Reddy, Bramel, Saxena and Ortiz2000); salinity (Srivastava et al., Reference Srivastava, Vadez, Upadhyaya and Saxena2006); resistance to Fusarium wilt, sterility mosaic disease (SMD), phytophthora blight and alternaria blight (Reddy et al., Reference Reddy, Bramel, Saxena and Ortiz2000) and cytoplasmic male sterility (CMS) (Tikka et al., Reference Tikka, Parmer and Chauhan1997; Saxena and Kumar, Reference Saxena and Kumar2003). World genebanks are conserving a limited proportion of wild relatives of total genetic variability and only a small proportion of conserved accessions have been characterized, resulting in limited use of wild species in crop improvement. Because of limitations such as low priority for characterization of wild species germplasm in the past, unavailability of sufficient seeds for characterization, etc., more works were not done using C. scarabaeoides germplasm. Increasing threats to natural habitats due to climate change and concomitant natural catastrophes (droughts, floods, fire hazards, etc.), human settlements, overgrazing, irrigation projects, etc. make it imperative to assess the species diversity and identify gaps in the collections before the valuable material is lost for ever (Upadhyaya and Gowda, Reference Upadhyaya and Gowda2009). Therefore, the present study was aimed at critical assessment of existing C. scarabaeoides collection at the ICRISAT genebank for diversity and geographical gaps in the collection for enhanced utilization and possible exploration in future.
Materials and methods
The pigeonpea wild genepool at ICRISAT genebank comprises 555 accessions of 66 species belonging to genus Rhynchosia (33 species, 303 accessions), Cajanus (19 species, 213 accessions), Flemingia (eight species, 18 accessions), Eriosema (three species, seven accessions), Dunbaria (two species, 12 accessions) and Paracalyx (one species, two accessions). The C. scarabaeoides collection comprises 102 accessions from eight countries. Passport and characterization data of 100 accessions in which seed quantity was adequate was considered for the present study. To assess the diversity in the collection, accessions were characterized during 2008 rainy season (June to November) under field conditions in alfisol-Patancheru soil series (Udic Rhodustolf) field at ICRISAT, Patancheru (17.53°N, 78.27°E, 545 m.a.s.l, and 600 km away from the sea), Andhra Pradesh, India. Experiment was laid out in unreplicated augmented design with systematic control after every 20 test accessions. Each plant in all accessions was given support with bamboo pegs to grow as climber to overcome the problem of pod damage due to soil contamination. ICP 15695, an early flowering (52 d) and pod borer [Helicoverpa armigera (Hun.)] tolerant (no egg laying, no larvae and < 10% pod damage) (Sharma, Reference Sharma2006) accession was used as control. Accessions were grown in a single row of 8-m length with a row-to-row distance of 100 cm and plant-to-plant distance of 50 cm within the row. Crop received a basal dose of Di-ammonium Phosphate (DAP) (150 kg/ha). Life saving irrigation was provided. The crop was protected from weeds, pests and diseases.
Observations were recorded on 13 quantitative traits (leaflet length, leaflet width, days to 50% flowering, days to 5% maturity, pods per raceme, pod length, pod width, seeds per pod, 100-seed weight, seed protein content, trichome density, trichome length and percentage of healthy pods per plant) and ten qualitative traits (growth habit, stem pigmentation, leaflet shape, flowering pattern, flower streak pattern, primary seed colour, seed coat colour pattern, seed shape, seed strophiole and biomass score). Flower streak pattern was recorded based on the density of streaks on vexillum. Seed coat colour was recorded as primary seed colour. Pattern of seed coat colour was recorded as mottled, plain and plain+mottled. C. scarabaeoides serves as forage for goats and other animals. Therefore, accessions were also scored for biomass production on 1–5 scale (1 = poorest and 5 = high biomass producer). Pods without egg layings, larvae, scares and holes on pod wall due to insects and diseases were considered as healthy pods. As per the standard maturity classification developed and being followed for pigeonpea at ICRISAT, accessions that flower in < 80 d were considered as extra early and those flower between 80 and 100 d as early maturing (ICRISAT, 1978). Data on trichomes were recorded on 7–10 d old pods. A longitudinal section on middle portion of the pod was made with the help of blade and placed in 45% acetic acid. The vials holding pod sections were placed in oven run at 40°C to soften the tissues. The section was transferred to water in a petri plate. From petri plate, three good sections were picked and placed on diluted glycerol drops on glass slide keeping the trichome side up. The sections were covered with glass cover slip and examined under simple microscope. The trichome density and length could be seen conveniently at 40 × magnification. For measuring density of trichomes, eye piece micrometer was placed in the microscope and the number of trichomes per grid (10 × 10 squares) was counted. For measuring the length of trichome, the number of squares covered by a trichome was counted and converted to microns. Days to 50% flowering, days to 5% maturity, all seed characters and biomass scores were recorded on plot basis and seed protein content was estimated at Crop Quality Unit of ICRISAT (IBPGR and ICRISAT, 1993).
The residual maximum likelihood method (REML; Patterson and Thompson, Reference Patterson and Thompson1971) was used to analyze the data of 13 quantitative traits. Variance components due to genotype (σ2g) and its standard errors (SE) were estimated. Principal component analysis (PCA) of 13 quantitative traits was performed using GENSTAT 13.1. Cluster analysis was performed according to Ward (Reference Ward1963) using scores of first six principal components (PCs). The mean, range and variances were calculated for 13 quantitative characters of each cluster and for the entire collection. The means for different traits were compared using the Newman–Keuls procedure (Newman, Reference Newman1939; Keuls, Reference Keuls1952). Homogeneity of variances was tested by Levene's test (Levene, Reference Levene and Olkin1960). The Shannon–Weaver diversity index (H′) (Shannon and Weaver, Reference Shannon and Weaver1949) was used to measure and compare phenotypic diversity for each trait. The diversity index was estimated for 13 quantitative traits over all the accessions and for each cluster. Phenotypic proportions were estimated for ten qualitative traits (Snedecor and Cochran, Reference Snedecor and Cochran1980).
Gaps in the collection were identified using the GIS tools such as MS Encarta Interactive Atlas, ArcGIS and FloraMap. The basic input in the GIS tools was the geographic coordinates (latitude and longitude) of the sampling site with a unique identifier (accession number). Passport data, particularly the information on precise location of collecting site and corresponding geographic coordinates were updated verifying all related records and collection reports. Using Microsoft Encarta®, an electronic atlas (MS Encarta® Interactive World Atlas, 2000), geographic coordinates were retrieved to fill the gaps for accessions having location information. Finally, the 76 accessions from India (44), Sri Lanka (15), Indonesia (14), Philippines (two) and Myanmar (one) were used to predict the probability of C. scarabaeoides occurrence and to identify gaps in the collection. Georeference data were validated by plotting the accessions on world map. Using the FloraMap, a GIS tool developed at Centro Internacional de Agricultura Tropical ( Jones and Gladkov, Reference Jones, Gladkov and Jones1999), probability of occurrence of C. scarabaeoides was mapped and overlaid the collecting sites of 76 accessions. Multiple accessions having same coordinates were considered as single collection site. Probability more than 75% was considered for interpretation of results and recommendation of gaps for launching collection missions. FloraMap is built on the assumption that climate is a robust indicator of environmental range of wild species. Likely alternative sites for finding a particular species will have climate profiles closely matching to those of the locations where the wild accessions were already collected. FloraMap uses a climate grid (18 × 18 km2) of monthly climate surfaces to produce probability map showing where else in the world the species might be found. While working on the passport dataset, equal weights were allocated to the climatic variables (monthly rainfall, minimum and maximum temperature and diurnal range in temperature) and an exponential transformation with a power of 0.3 was applied to the monthly rainfall data. More than 95% of total variation was explained by first five PCs. ArcGIS was used to prepare the layout and cartographic outputs. High probability area covering provinces/districts (shaded area in Figs. 2 and 3) in different countries is summarized in Table 5.
Results
Diversity in the collection
Qualitative traits
Growth habit
All accessions of C. scarabaeoides have showed creeper–climber growth habit.
Stem pigmentation
Accessions having green stems were found in maximum proportion (60%), followed by dark green (23%) and light green (10%). Remaining 7% accessions had a mixture of green and light green stem plants.
Leaflet shape
C. scarabaeoides produced pinnately trifoliate leaves (van der Maesen, Reference van der Maesen1986). Fifty-two percentage of accessions produced lanceolate leaflets, followed by ovate (25%), broad lanceolate (22%) and linear lanceolate (1%) shaped leaflets.
Flowering pattern
All accessions showed indeterminate flowering habit.
Flower streak pattern
Accessions having few streaks were maximum (32%), followed by dense streaks (22%), plain (12%), mixture of plants with dense streaks and few streaks (20%), plain and few steaks (10%) and plain and dense streaks (4%).
Primary seed colour
Grey colour was predominant (31%), followed by dark grey (26%), grey+grey yellow (23%), dark grey+grey (10%) and dark grey+grey yellow (10%).
Seed coat colour pattern
Most of the accessions produced a mixture of plain+mottled (35%) coat colour pattern, followed by plain (34%) and mottled (31%).
Seed shape
Three clear seed shapes (square = 35%, round = 4% and elongate = 1%) were found in the collection. There were accessions with a mixture of plants producing square and round shaped seeds (58%) and square and elongate (2%) shaped seeds in the collection.
Seed strophiole
This was the important diagnostic trait to identify wilds. Seed strophiole was observed in seeds of all accessions.
Biomass score
Only three accessions (ICP 15685, ICP 14690 and ICP 15706) scored high (5), while nine accessions scored 4, 59 accessions scored 3, 22 accessions scored 2 and 7 accessions scored 1.
Quantitative traits
REML analysis
REML analysis indicated highly significant genotypic variance for leaflet length, days to 5% maturity, seeds per pod, 100-seed weight, seed protein content, trichome density and length indicating the considerable variation in the collection for these traits. Variance for days to 50% flowering was significant at 5% only (Table 1).
Table 1 Range, mean, variance and Shannon–Weaver diversity index (H′) and variance due to genotypes (σ2g) for different characters of C. scarabaeoides germplasm assembled and evaluated at ICRISAT genebank, Patancheru, India
* P at 0.05.
** P at 0.01.
Cluster analysis
PCA carried out using standardized data of 13 quantitative traits captured 74% of total variation from first six PCs. The PC1 alone accounted for 19.9% variation, followed by PC2 with 15.6%, PC3 with 11.6%, PC4 with 10.1%, PC5 with 8.6% and PC6 with 8.1%. A hierarchical cluster analysis conducted on the scores of the first six PCs resulted in four clusters (Fig. 1 and Table 2) (Ward, Reference Ward1963). Clustering of accessions revealed no effect of geographic origin on agronomic performance. Eight accessions from India and one accession from Australia formed the first cluster; 19 accessions from India, 13 from Sri Lanka, four from Indonesia, two from Australia, one from Fiji and one accession of unknown origin formed the second cluster; six accessions from India, three from Indonesia, two from Philippines, one accession each from Australia and Myanmar formed the third cluster and 22 accessions from India, nine from Indonesia, three from Sri Lanka, two from Australia and one accession each from Fiji and United Kingdom formed the fourth cluster (Table 2).
Fig. 1 Dendrogram of 100 accessions of C. scarabaeoides conserved at ICRISAT genebank, Patancheru, India, based on scores of first six PCs.
Table 2 Cluster-wise pigeonpea germplasm accessions
Variances
The homogeneity of variances of the four clusters was tested for all the 13 quantitative traits by Levene's test (Levene, Reference Levene and Olkin1960). The variances were heterogeneous for leaf width, pods per raceme and percentage of healthy pods per plant (Table 3).
Table 3 Variances and Shannon–Weaver diversity index (H′) for different characters of C. scarabaeoides germplasm clusters, evaluated at ICRISAT genebank, Patancheru, India
aVariance homogeneity tested by Levene's test.
bCl1 = cluster 1, Cl2 = cluster 2, Cl3 = cluster 3 and Cl4 = cluster 4.
Range and means
In the entire collection, the important traits such as days to 50% flowering ranged from 52 (ICP 15695) to 100 d (ICP 15723) and days to 5% maturity varied from 103 (ICP 15747) to 167 d (ICP 15723). Seeds per pod varied from 3.8 (ICP 15743) to 5.6 (ICP 15908), seed protein content from 18.3 (ICP 15719) to 23.8% (ICP 15695) and percentage of healthy pods per plant from 73 (ICP 15706) to 93 (ICP 15692) (Table 1). Among the clusters, accessions of cluster 2 varied more for leaf width, days to 50% flowering, days to 5% maturity, 100-seed weight, seed protein content and trichome density. Accessions of cluster 4 varied widely for leaf length, pod length and width, seeds per pod, trichome length and percent healthy pods per plant (Table 4).
Table 4 Range and mean values for different characters of C. scarabaeoides germplasm clusters evaluated at ICRISAT genebank, Patancheru, India
Cl1 = cluster 1, Cl2 = cluster 2, Cl3 = cluster 3 and Cl4 = cluster 4.
a,b,cMeans were tested by Newman–Keul test and means followed by different letters are significantly different at P= 0.05.
Mean overall accessions (70 ± 6.65 d) indicated C. scarabaeoides collection as an excellent source for earliness (Table 1). As per the standard maturity classification for pigeonpea followed at ICRISAT, mean for days to 50% flowering suggested that 89 accessions were extra early ( < 80 d to 50% flowering) and 11 accessions were early (80–100 d to 50% flowering) maturing (ICRISAT, 1978). None of the accessions flowered earlier than the control ICP 15695, a promising accession against pod borer flowers in 52 d. All accessions from Australia, Fiji, Myanmar, Philippines and United Kingdom were found to be extra early. About 87% of accessions from India, Indonesia and Sri Lanka were found as extra early. Overall mean for seeds per pod (4.9 ± 0.16) indicated the relatively higher seeds per pod than majority of cultivated accessions. As many as 31 accessions produced more than five seeds per pod. Eight accessions (ICP 15692, ICP 15696, ICP 15698, ICP 15699, ICP 15712, ICP 15719, ICP 15732 and ICP 15758) produced more than 92% healthy pods per plant, which is comparable to that of pod borer tolerant control ICP 15695 (92.08 ± 4.5).
Newman–Keuls test of significance for mean values indicated significant differences among the clusters for one or more traits under study (Tables 2 and 4) (Newman, Reference Newman1939; Keuls, Reference Keuls1952). Cluster 1 for six traits (leaflet width, pods per raceme, pod length and width, 100-seed weight, seed protein content), cluster 2 for three traits (pods per raceme, seeds per pod and healthy pod per plant), cluster 3 for seven traits (leaflet length and width, pods per raceme, seeds per pod, seed protein content and trichome density and length) and cluster 4 for four traits (days to 50% flowering, days to 5% maturity, pods per raceme and seed protein content) were found as promising sources. Differences were not significant among the clusters for pods per raceme.
Phenotypic diversity
The H′ was calculated over all accessions and for each cluster to compare phenotypic diversity among the clusters for 13 quantitative traits (Tables 1 and 3) (Shannon and Weaver, Reference Shannon and Weaver1949). A low H′ indicates extremely unbalanced frequency classes for an individual trait and lack of genetic diversity in the collection. The diversity index values (H′) were variable among traits. In the entire collection, diversity index (H′) ranged from 0.551 ± 0.00 for seeds per pod to 0.635 ± 0.03 for leaflet width (Table 1). Mean diversity over all traits was maximum in cluster 2 (H′ = 0.57 ± 0.00) and it was lowest in cluster 1 (H′ = 0.48 ± 0.00). Mean diversity over all clusters was highest (H′ = 0.57 ± 0.00) for seeds per pod and lowest for days to 50% flowering (H′ = 0.48 ± 0.00) (Table 3).
Character associations
Phenotypic correlations were estimated among all quantitative characters and their significance was tested (Snedecor and Cochran, Reference Snedecor and Cochran1980). Highly significant positive correlation of leaflet length with leaflet width (0.638), pods per raceme (0.255) and pod width (0.297) was observed. Correlation of leaflet length with pod length (0.230) was significant at 5% level only. Leaflet width had highly significant positive correlation with pod length (0.293) and width (0.319). As expected, days to 50% flowering had highly significant positive association with days to 5% maturity (0.869). Pods per raceme had significant negative correlation with healthy pods per plant ( − 0.213), indicating the high pod damage in racemes having more pods. Pod length had highly significant positive association with pod width (0.618) and 100-seed weight (0.430) and significant positive correlation with seeds per pod (0.231). Pod width had highly significant positive association with 100-seed weight (0.314). Trichome density had highly significant positive association with trichome length (0.294), indicating pods having more trichomes will produce long trichomes. Trichome length had highly significant negative association with healthy pods per plant ( − 0.293), suggesting that the selection for trichome length and density may not result in selection for pod borer resistance.
Gaps in the collection
The C. scarabaeoides collection at ICRISAT genebank comprises 102 accessions from India (57), Indonesia (16), Sri Lanka (16), Australia (6), Fiji (2), Philippines (2), Myanmar (1), United Kingdom (1) and one accession of unknown origin. The results indicated Asia as the major source with 90 accessions. Eight accessions are from Oceania and one accession from Europe. The collection under study included 72 accessions collected in 26 germplasm collection missions launched by ICRISAT and its partners in five countries for ICRISAT mandate crops germplasm during 1975–93. Maximum collections are from India (37), followed by Indonesia (16), Sri Lanka (16), Philippines (2) and Myanmar (1). A total of 30 accessions were introduced from four countries. All accessions from Australia (6), Fiji (2) and United Kingdom (1) are the introductions in the collection. Overlaying of collection sites (76) on predicted probability map revealed that most of the existing collections are from the primary centre of diversity for pigeonpea (Fig. 2). It also revealed high probability (>75%) areas in Asia and Oceania where no collections were made in the past (Figs. 2 and 3). The high probability areas comprise of 118 provinces covering 790 districts in ten countries, which include 25 provinces in Indonesia, 19 provinces in Cambodia, 18 provinces in Vietnam, 15 provinces in India, 12 provinces each in Laos and Philippines, seven provinces in Myanmar, four provinces each in Malaysia and Thailand and two provinces in Nepal. In Bangladesh, seven districts (district is the sub-nation unit in Bangladesh) were identified as gaps. Papua New Guinea was considered as one unit as there was no information on province and districts (Table 5).
Fig. 2 Geographical gaps (shaded area) identified in Indian subcontinent for C. scarabaeoides germplasm assembled at ICRISAT genebank, Patancheru, India.
Fig. 3 Geographical gaps (shaded area) identified in East Asian countries for C. scarabaeoides germplasm assembled at ICRISAT genebank, Patancheru, India. Is, island.
Table 5 Country-wise provinces showing high probability (>75%) for occurrence of C. scarabaeoides
Discussion
During the year 2009, the average productivity of pigeonpea was 844 kg/ha (FAO, 2009), which is considered as low. Insect pests like pod borer, pod fly, pod wasp, diseases such as wilt, SMD, phytophthora blight, alternaria blight and abiotic stresses such as salinity and water logging conditions are the major causes for low seed yields in pigeonpea. In addition, long duration of crop, low pod setting percent and less number of seeds per pod are the other reasons for low seed yield in pigeonpea. Pod borer alone causes an estimated yield loss of US$1000 million per year globally (Sharma, Reference Sharma2006). Farmers rely upon synthetic insecticides and fungicides to manage the pests and diseases, creating lot of environmental pollution and causing health hazards.
At ICRISAT, efforts of cultivated germplasm screening for resistance sources resulted in identification of 27 accessions resistant to pod borer, 21 accessions resistant to pod fly, 87 accessions resistant to wilt, 362 accessions resistant to SMD, 148 accessions resistant to phytophthora blight, 19 accessions resistant to alternaria blight and 29 accessions tolerant to salinity. C. scarabaeoides was found to have higher levels of resistance to pod borer, pod fly, pod wasp, wilt, SMD, phytophthora blight and root-knot nematode, in addition to having high seed protein content (28%) (Kameswara Rao et al., Reference Kameswara Rao, Reddy and Bramel2003). Upadhyaya (Reference Upadhyaya2006) reported early maturity (34 d), high pod setting (74%) and more seeds per pod (4–6 seeds) in C. scarabaeoides. Saxena et al. (Reference Saxena, Singh, Reddy, Singh, Lateef, Sharma and Remanandan1990) reported near immunity of C. scarabaeoides accessions to pod fly. Sharma et al. (Reference Sharma, Green, Stevenson and Simmonds2001) reported higher levels of resistance to pod borer in C. scarabaeoides, Cajanus sericeus (Benth Ex Bak.) van der Maesen, Cajanus acutifolius (F. von Muell.) van der Maesen and Cajanus Albicans (W & A) van der Maesen. Sharma et al. (Reference Sharma, Remanandan and McDonald1993) reported C. scarabaeoides accession ICPW 92 as highly resistant to Melaidogyne javanica and ICPW 38 and ICPW 92 as resistant to Rotylenchulus reniformis nematode. C. scarabaeoides crosses easily with cultivated pigeonpea and many interspecific crosses involving C. scarabaeoides and C. sericeus have resulted in the development of stable CMS lines (Pundir and Singh, Reference Pundir and Singh1985; Tikka et al., Reference Tikka, Parmer and Chauhan1997). Saxena and Kumar (Reference Saxena and Kumar2003) developed CMS system in pigeonpea using ICPW 89, a germplasm accession of C. scarabaeoides. Development of C. scarabaeoides based CMS system has shown promise in breaking the yield barriers in pigeonpea. However, issues related to fertility restoration needs to be resolved for making C. scarabaeoides based hybrids acceptable to the farmers.
All C. scarabaeoides accessions used in the present study are the best sources for extra early ( < 80 d to 50% flowering) and early maturity (80–100 d to 50% flowering). Eight accessions (ICP 15692, ICP 15696, ICP 15698, ICP 15699, ICP 15712, ICP 15719, ICP 15732 and ICP 15758) and the control, ICP 15695, which produced more than 92% healthy pods per plant, are also found as the good sources for higher number of seeds per pod (4–6 seeds). Accessions in cluster 2, 3 and 4 with low mean values for days to 50% flowering were found as the best sources for early flowering and maturity (Table 2). Accessions in cluster 2 and 3 for seeds per pod and cluster 2 for healthy pods per plant were found as promising sources for use in crop improvement. The genetic potential of CWR in crop improvement is now well demonstrated and is likely to contain the genetic diversity necessary to combat the climate change because of the diversity of habitats in which they grow and wide range of conditions they are adapted to (FAO, 2008). Wide latitudes of collecting sites for accessions under study, varying from 10°75′S in Indonesia to 32°53′N in India, suggested that the C. scarabaeoides accessions are from diverse climate and may adapt well to varying climate. For enhanced utilization of C. scarabaeoides germplasm in pigeonpea improvement, assembled germplasm needs systematic characterization for morpho-agronomic traits and evaluation for biotic and abiotic stress resistance.
The collection strategy for wild relatives involves difficulties in identifying the geographic distribution of species, precise location, time of maturity, etc. In addition, the wild relatives are rather uncommon in natural vegetation and the available location data are often very old and may not be precise. Physical survey for C. scarabaeoides occurrence will be more laborious and require lot of resources. Hence, the use of GIS tools to analyze passport data and corresponding climate data to map the potential distribution of species is a powerful method to assist germplasm collectors and genebank managers. As revealed in the present study, GIS tools have earlier been found to provide more precise cartographic representation of the eco-geographic regions and the germplasm collecting sites (Burle et al., Reference Burle, Cordeiro, Fonseca, de Melo, de Belem das Neves Alves and Abadie2003). The number of collection sites (76) included in the present study is quite considerable to predict the high probability of C. scarabaeoides occurrence. Hernandez et al. (Reference Hernandez, Graham, Master and Albert2006) reported a 30% prediction success using Bioclim, a GIS tool, with ten samples and this increased to over 80% when 75 samples are used. India is the primary centre of origin for pigeonpea and the probability image generated in the present study for occurrence of C. scarabaeoides matched quite close to the origin and dispersal of pigeonpea (Figs. 2 and 3) (van der Maesen, Reference van der Maesen, Nene, Hall and Shiela1990). van der Maesen et al. (Reference van der Maesen, Remanandan, Kameshwara Rao and Pundir1984) reported the occurrence of C. scarabaeoides in South and Southeast Asia, parts of Oceania, coastal Africa, Madagascar and Jamaica from 0 to 1000 m.a.s.l. It is the most widely spread species of genus Cajanus all over India. C. scarabaeoides is mostly confined to areas close to the shores of oceans and rivers (van der Maesen, Reference van der Maesen1980a). New countries identified as potential for C. scarabaeoides occurrence include Bangladesh, Cambodia, Laos, Malaysia, Nepal, Papua New Guinea, Thailand and Vietnam. Singh et al. (Reference Singh, Singh, Singh, Singh, Smartt and Dhillon2005) reported that Uttar Pradesh and Uttarakhand (districts of Mirzapur, Bundelkhand region and Tarai), Madhya Pradesh (Rewa, Sidhi, Gwalior, Morena and Bhind region), Maharashtra (Nasik and Dhule) and adjoining North Karnataka, southeastern districts of Bihar, parts of Orissa, Rajasthan, Gujarat, Nilgiri and Ragan hills and the northeast region of India are yet to be surveyed for pigeonpea and its wild relatives. In Uganda, emphasis should be in potential areas such as eastern and northern parts of the country (Singh et al., Reference Singh, Singh, Singh, Singh, Smartt and Dhillon2005).
There may be some low probability ( < 75%) areas for the occurrence of C. scarabaeoides in countries such as Australia and Fiji. Though Australia was considered as one of the potential countries for C. scarabaeoides, it is not adequately represented in the global collection at ICRISAT genbank. Hence, attempts to collect and assemble Australian pigeonpea wild relatives, including C. scarabaeoides, should receive high priority before they are lost irretrievably (Singh et al., Reference Singh, Singh, Singh, Singh, Smartt and Dhillon2005). As per the dispersal of pigeonpea, southeast Africa and Central American countries, particularly the Caribbean region, are also expected to have pigeonpea wild relatives and countries of these regions are not represented in the world collection (van der Maesen, Reference van der Maesen, Arends, Boelema, de Groot and Leeuwenberg1980b; van der Maesen, Reference van der Maesen1986). Hence, there is a clear need for further collection of C. scarabaeoides to cover the full range of geographic diversity.
Many CWR, including C. scarabaeoides, are currently threatened with loss of diversity and/or extinction (Stolton et al., Reference Stolton, Maxted, Ford-Lyod, Kell and Dudley2006). Major threats to the CWR include deforestation, logging, plantation agriculture and forestry, industrialization, dryland destruction and desertification, fire hazards, urbanization, civil unrest, mining and quarrying, invasive species and climate change. Therefore, there is an urgent need to assemble as much diversity as possible in C. scarabaeoides through introduction of already collected germplasm if any, from different organizations and by launching germplasm collection missions in the potential areas (provinces) identified in the present study as well as in other studies (Table 5). Irrespective of probability of occurrence, collecting for resistant sources should concentrate on areas most favourable for pests and diseases. Disease evaluation data for previously collected C. scarabaeoides germplasm should be used to target collecting sites for disease resistance sources (Lenne and David, Reference Lenne and David1991). Although the information available on C. scarabaeoides occurrence through GIS tools is precise, it is important to go through different flora and fauna, catalogues and literature for the distribution of species before embarking on actual germplasm collection. Generally, different species mature at different times. Therefore, the overall knowledge of wild relatives of pigeonpea is essential to collect different species in one collection mission and make the collection mission a success. It is suggested to collect the complete passport data including georeference information while collecting the germplasm. Availability of georeference data is crucial in predicting the probability of occurrence of a particular species in nature, its diversity distribution and the areas of adaptation and identifying geographical gaps in the collection.
Acknowledgements
Authors acknowledge the contribution of all former and present staff of Genetic Resources Unit (GRU), ICRISAT in collection, assembly and conservation of pigeonpea genetic resources. The help of Jacob Mathew, D. Bapa Rao, G. Dasaratha Rao and G. Ram Reddy, Research Technicians, GRU, ICRISAT, Patancheru, India, in recording observations and documentation of the data for this study is highly appreciated.
