Evaluation of Global Composite Collection Reveals Agronomically Superior Germplasm Accessions for Chickpea Improvement

: The rich genetic diversity existing within exotic, indigenous, and diverse germplasm lays the foundation for the continuous improvement of crop cultivars. The composite collection has been suggested as a gateway to identifying superior germplasm for use in crop improvement programs. Here, a chickpea global composite collection was evaluated at ﬁve locations in India over two years for ﬁve agronomic traits to identify agronomically superior accessions. The desi, kabuli, and intermediate types of chickpea accessions differed signiﬁcantly for plant height (PLHT) and 100-seed weight (100 SW). In contrast, the intermediate type differed substantially from kabuli for days to maturity (DM). Several highly signiﬁcant trait correlations were detected across different locations. The most stable and promising accessions from each of the ﬁve locations were prioritised based on their superior performance over the best-performing check cultivar. Accordingly, the selected germplasm accessions of desi type showed up to 176% higher seed yield (SY), 29% lower ﬂowering time, 21% fewer maturity days, 64% increase in PLHT, and 183% larger seeds than the check cultivar JG11 or Annigeri. The prioritised kabuli accessions displayed up to 270% more yield, 13% less ﬂowering time, 8% fewer maturity days, 111% increase in PLHT, and 41% larger seeds over the check cultivar KAK2. While the intermediate type accessions had up to 169% better yield, 1% early ﬂowering, 3% early maturity, 54% taller plants, and 25% bigger seeds over the check cultivar JG 11 or KAK2. These accessions can be utilised in chickpea improvement programs to develop high-yielding, early ﬂowering, short duration, taller, and large-seeded varieties with a broad genetic base.


Introduction
Food grain legumes are a rich source of high-quality proteins, carbohydrates, dietary fibres, and minerals for millions of people in developing countries, especially in South Asia and sub-Saharan Africa [1]. Chickpea (Cicer arietinum L.) is one of the most important food grain legumes, widely grown on marginal lands in more than 56 countries. The world cropping area under chickpea cultivation is~13.72 million hectares, with a total annual production of~14.25 million tonnes, of which Asia contributes 83% of production [2]. The key chickpea-producing countries include India, Turkey, Russian Federation, Myanmar, and Pakistan. Cultivated chickpea can be classified into two distinct forms: desi types (characterised by purple flowers, small seed size, angular shape, and coloured seeds with a high level of fibre) and kabuli types (characterised by white flowers, large seed size, owl's head shape, and beige-coloured seeds possessing a low level of fibre).
Further, all desi types have anthocyanin pigmentation on other plant parts, whereas pigmentation is absent in kabuli chickpea. A third form, called intermediate type, is characterised by small-to medium-size round (pea)-shaped dark-or light-coloured seeds. The desi type chickpea is mainly cultivated in South Asia, while the kabuli type is primarily grown in the Mediterranean [2].
Although chickpea's productivity has continuously increased over the last six decades, its sensitivity to several biotic and abiotic stresses has also been enhanced simultaneously, leading to yield variations [3]. The major biotic stresses constraining chickpea production include the diseases Ascochyta blight (Ascochyta rabiei (Pass.) Labr.), fusarium wilt (Fusarium oxysporum f. sp. Ciceris), botrytis grey mould (Botrytis cinerea Pers. ex Fr.), collar rot (Sclerotium rolfsii Sacc.), and dry root rot (Rhizoctonia bataticola (Taubenhaus)); while pod borer or gram pod borer (Helicoverpa armigera (Hübner)), leaf miner (Liriomyza cicerina Rondani), aphid (Aphis craccivora Koch.), and bruchid (Callosobruchus chinensis L.) are among the important pests [4]. In addition, drought, heat, cold, and salinity stress, among abiotic factors, represent key constraints to chickpea productivity [5,6]. Worldwide, these stresses result in potential yield losses costing over USD 2559 million annually. About 46% of losses are projected to be recovered by developing improved chickpea cultivars with multiple resistance/tolerance to these stresses [7]. However, improved genotypes are hampered primarily because of the limited genetic diversity available in the modern crop gene pools [8]. Extensive use of limited and closely associated germplasm/improved breeding accessions in chickpea improvement programs has led to a plateau in genetic improvements of target traits. Therefore, the newly developed cultivars display increased susceptibility to pests, diseases, and unstable yields under changing environments. To this end, using exotic and diverse genetic resources in breeding is the key to boosting the productivity and diversity of chickpea cultivars [9,10].
One of the major reasons for the limited utilisation of germplasm accessions in breeding efforts is the lack of knowledge of economically important traits, which often display genotype × environment interactions and need multi-environment testing [11][12][13][14]. Apart from higher yields, early maturity, taller plants, and large seed size represent some key agronomic traits of economic importance. Developing high-yielding and early-maturing chickpea to escape terminal drought is important in breeding programs targeted at semiarid tropical environments [15]. Varieties with tall and erect growth habits are suitable for mechanical harvesting and help minimise labour costs [16]. Chickpea seed size is a crucial trait for international trade, with the large-seeded desi and kabuli types receiving a premium price in the markets. For instance, the large-seeded (>50 g 100-seed −1 ) kabuli cultivars receive a premium that is thrice the cost of desi and two times the cost of kabuli types, having a 100 SW of 25-40 g [17]. For developing improved chickpea cultivars for a target environment, identifying superior germplasm/parental accessions is a prerequisite. Different genetic backgrounds among parental accessions will offer the genetic diversity required to generate desired allelic combinations, which will increase the possibility of developing unique cultivars in proportion to the number of alleles by which the parents vary.
To increase the use of chickpea germplasm accessions in breeding programs, a composite collection of 3000 accessions represented variation in gene diversity ranging from 53% to 97% of the entire collection [18,19]. This global composite collection has been sequenced using whole-genome sequencing, and extensive analysis has been undertaken [10]. Evaluating such composite collections under multi-environmental trials is a promising approach to determining trait-specific promising germplasm accessions. The present study evaluated a composite collection under multi-environment field trials to identify agronomically superior germplasm accessions for increasing chickpea's genetic potential.

Experimental Design, Target Traits, and Experiment Locations
The entire composite collection and very promising checks include four desi (Annigeri, G130, ICCV10, and JG11) and two kabuli (KAK2 and L550) accessions, which were grown in 2 m single-row plots in an augmented block design. The experiment was conducted at Patancheru, Amlaha, Junagadh, and Sehore during the 2014-15 and 2015-16 years postrainy season (between October-November). The experiment was conducted at Kanpur for two years during the 2015-16 and 2016-17 years post-rainy season (during November). The meteorological conditions during the experiment were recorded for field trials conducted at Patancheru, India. During the 2014-15 crop season, the average day/night temperature recorded was 35.4/5.2 • C with a relative humidity of 18.0/98.0% and an average rainfall of 0-33.8 mm. While during the 2015-16 crop season, the average day/night temperature and relative humidity were 37.8/6.6 • C and 20.0/98.0%, respectively, and the average rainfall ranged between 0 and 40 mm. The agronomic data were recorded for days to 50% flowering (DF; days counted from sowing up to 50% of plants producing their first flower), DM (DM; days counted from sowing to the appearance of golden colour pods), plant height (PLHT; measured in centimetre from the soil surface to the tip of the plant foliage), 100-seed weight (100 SW; measured by taking a random sample of 100 seeds from a single row and by weighing them), and seed yield (SY).

Statistical Analysis
The statistical analyses were performed using GenStat 19th edition [20]. Analysis of variance was performed for the phenotyping data collected from all the locations over two years and for all five traits by considering block as random and entry as fixed effects using the restricted maximum likelihood (REML) estimation procedure. Further entry effect was partitioned to examine the differences among checks, differences among tests, and differences between checks and tests. The least-square means were calculated for all entries and used separately for three different sets (desi, kabuli, and intermediate type). Ranking biplots [21] were constructed to examine the discriminating ability and representativeness of the entries in individual test environments (location and year combination) and an ideal test environment. The total number of ranking biplots at individual test environments was 150 (3 sets × 5 locations × 2 years × 5 traits). It was difficult to present all 150 ranking biplots in this article; therefore, the relevant information for desi type is shown in Supplemental Table S1. For kabuli and intermediate types, the information is presented in Supplemental Table S2. A flow chart depicting the selection of germplasm accessions at each level is provided in Table 2.

Variation in Agronomic Traits among Desi, Kabuli, and Intermediate Type Accessions
Chickpeas are classified into three types based on seed shape-desi, kabuli, and intermediate. We observed significant differences among the means of desi, kabuli, and intermediate types of chickpea accessions for five traits phenotyped at five different loca-tions in India for two years (Supplementary Tables S3 and S4). The three types of chickpeas showed significant variation for PLHT, 100 SW, and DM. For instance, the desi and kabuli chickpeas varied substantially for PLHT and 100 SW. The kabuli accessions were taller and possessed a higher 100 SW than desi. The intermediate types differed from desi types for PLHT and 100 SW and had taller plants with greater 100 SW.
Furthermore, a significant difference was observed for DM, PLHT, and 100 SW between the intermediate and kabuli chickpeas. Compared to intermediate types, the kabuli were later in maturity, taller in height, and had higher 100 SW. No significant differences were detected among the means of three types of chickpeas for DF and SY.

Correlation among Traits Evaluated at Five Locations
A Pearson correlation analysis was conducted to identify the correlations between agronomic traits within and between experimental trials. The correlation coefficients among the five traits evaluated for two years across five different locations revealed the presence of significant and meaningful correlations (0.50 ≤ r ≤ -0.50) (Supplementary Table S5). For instance, 100 SW measured across two years at Amlaha (Madhya Pradesh) displayed a significant positive correlation. At Patancheru (Telangana), significant positive correlations were detected between the following pairs of traits: DF and DM, DF and PLHT, and DM and PLHT. The 100 SW evaluated for two years at Kanpur (Uttar Pradesh) showed a substantial positive correlation. At Junagadh (Gujarat), DF was significantly and positively correlated with DM, while DF and 100 SW showed a significant positive correlation across years. Furthermore, at Sehore (Madhya Pradesh), DF, PLHT, and 100 SW showed a significant correlation across two years. Several other traits displayed a low to moderate correlation (r = ± 0.01 to r = ± 0.49) across years at different locations evaluated.  (Table 3). Based on DM, 20 elite accessions were identified with maturity days ranging from 108 d to 117 d and SY in the 16-42 g plant −1 . Of these 20 accessions, 9 (ICC14014, ICC16076, ICC13107, ICC5878, ICC11091, ICC14507, ICC8366, ICC10953, and ICC4074) produced an SY in the range of 50-163% over the check cultivar JG11 (Table 3). Furthermore, 20 germplasm accessions were selected based on PLHT, whose height ranged from 48 cm to 62 cm, and SY ranged from 27 to 36 g plant −1 . The PLHT of all the accessions was at par or more than JG11, and the accessions displayed a yield gain of up to 125% over the check cultivar JG11 (Table 3). Twenty accessions prioritised based on 100 SW had seed weight values ranging from 18 to 65 g 100-seeds −1 and SY ranging from 20 to 37 g plant −1 . Of these, ten accessions (ICC12452, ICCV88202, ICC8474, ICC14385, ICC11903, ICCV91902, ICC1000, ICC15540, ICC3164, and ICC16903) produced SY in the range of 50-131% more than JG11 (Table 3).

Kabuli Type
During the 2014-15 year, KAK2 was the best check cultivar, with an SY of 16 g plant −1 , DF of 79 d, DM of 119 d, PLHT of 50 cm, and 100 SW of 35 g. Eight accessions selected based on flowering time had DF in the range of 75-93 d and SY in the range of 16-32 g plant −1 . Three accessions (ICC11879, ICC15518, and IG10187) were prioritised, and SY was 56-100% more than the check cultivar KAK2 (Table 3). Furthermore, eight germplasm accessions were selected based upon DM (112-127 d), whose SY varied from 17-22 g plant −1 . None of these eight accessions produced a yield with a 50% increase over KAK2. Based on PLHT, eight accessions were selected whose height ranged between 48 and 64 cm, with a yield of 16-32 g plant −1 . Two of these eight accessions (ICC11879 and ICC2496) produced SY with a 75-100% increase over KAK2 (Table 3). Further, ten accessions were selected based upon 100 SW (26-39 g), whose SY ranged between 16-26 g. Two (IG70815 and IG6078) out of ten accessions had a 50-62% yield advantage compared to KAK2 (Table 3).
KAK2 was designated the best check during 2015-16, with an SY of 9 g plant −1 , a flowering time of 55 d, DM of 111 d, PLHT of 38 cm, and 100 SW of 37 g. Eight accessions were selected based on DF (50-57 d), which produced an SY of 6-11 g plant −1 . Ten accessions were prioritised based on maturity days (98-110 d), with a yield ranging between 6 and 11 g plant −1 . Further, seven accessions were selected based on PLHT (41-49 cm), yielding between 6 and 17 g plant −1 . Based on 100 SW, seven accessions were prioritised whose seed weight was 24-30 g and whose yield varied between 6 and 11 g plant −1 . Interestingly, during 2015-16, none of the accessions produced more than 50% over the check cultivar (Table 3).

Intermediate Type
The check cultivar JG11 yielded 15 g plant −1 , DF of 41 d, DM of 100 d, PLHT of 42 cm, and 100 SW of 25 g during 2014-15 (Table 4). One accession (ICC9778) was selected based on flowering time (51 d), which produced a 73% increase in yield over JG11. In addition, one accession (ICC9778) with a DM of 111 d had a yield gain of >73% over the check cultivar JG11. An accession (ICC9778) was prioritised based on PLHT (50 cm), producing more than 73% greater yield than JG11. During 2015-16, JG11 had an SY of 15 g plant −1 , days to flowering of 38 d, DM of 101 d, PLHT of 39 cm, and 100 SW of 24 g. One accession (ICC4878) was selected based upon flowering time (54 d) and produced about 73% more yield than JG11. Further, two (ICC4878 and ICC5727) accessions were selected based on maturity days (102-107 d) and produced more than 73% higher yield than JG11. One accession (ICC7524) was prioritised, with a PLHT of 60 cm yield gain of more than 100% over JG11. The two accessions (ICC4878 and ICC12431) selected based on 100 SW produced a 73-86% yield over the check cultivar JG11 (Table 4).  (Table 5). Fourteen accessions were selected based upon DF (54-74 d), which produced 15-24 g plant −1 . One accession (ICC16734) was found to possess a flowering time of 62 d and yielded 24 g plant −1 (>50% over JG11) ( Table 5). Based on DM, 13 accessions were selected with maturity days in the 112-120 d and produced a yield between 16 and 28 g plant −1 . One accession (ICC299) with days of maturity of 112 d had more than 75% yield over the check JG11 (Table 5). Further, 11 accessions were selected based on PLHT (44-67 cm), which yielded 15-43 g plant −1 . Three accessions (ICC2766, ICC9139, and ICC2878) were detected to produce a 94-169% higher yield than JG11 (Table 5). Thirteen accessions were selected with 100 SW in the 12-31 g and had a yield of 15-41 g plant −1 . Notably, two accessions (ICC9139 and ICC15610) had 94-156% more yield than the JG11 check (Table 5).

Kabuli Type
The cultivar KAK2 was the best check during 2015-16 and had an SY of 15 g plant −1 , a flowering time of 59 d, DM of 116 d, PLHT of 44 cm, and 100 SW of 37 g (Table 5). Here, nine accessions were selected based upon days to flowering (56-73 d), which yielded 14-17 g plant −1 . Five accessions were prioritised based upon maturity days (111-121 d), which produced a yield range of 14-17 g plant −1 (Table 5). Based on PLHT, six accessions were selected with a height of 55-71 cm and a yield of 16-24 g plant −1 . One accession (ICC16637) with 71 cm PLHT produced >60% yield over the check KAK2 (Table 5). Seven accessions were selected based on 100 SW (25-38 g), whose yield varied in the range of 14-24 g plant −1 . Two accessions produced about 53-60% higher yield than the check cultivar (Table 5).

Intermediate Type
The best check JG11 displayed an SY of 16 g plant −1 , DF of 64 d, DM of 117 d, PLHT of 45 cm, and 100 SW of 24 g during 2015-16 (Table 5). One accession (ICC5923) with a DM of 122 d produced about 63% higher yield than check JG11. During 2016-17, JG11 displayed a yield of 16 g plant −1 , DF of 76 d, DM of 120 d, PLHT of 51 cm, and 100 SW of 24 g (Table 5). One accession (ICC7574) with 61 cm PLHT produced about 69% higher yield than JG11. Further, two accessions were selected based upon 100 SW, which displayed superior yield performance over JG11 (Table 5).  (Table 6). Eight accessions were selected based on DF (46-65 d), which produced a yield of 19-27 g plant −1 . A total of 16 accessions were prioritised based on DM (82-110 d), which yielded in the range of 19-31 g plant −1 . Based on PLHT, 12 accessions were selected whose height ranged from 37 cm to 59 cm, and whose yield ranged from 19 to 31 g plant −1 . Fifteen accessions were selected based on 100 SW (14-30 g), having an SY of 19-36 g plant −1 . One accession (ICCV95605), having a 100 SW of 16 g, produced about 57% higher yield than Annigeri (Table 6).
During the 2015-16 year, the best check Annigeri produced a yield of 17 g plant −1 , a flowering time of 50 d, DM of 102 d, PLHT of 47 cm, and 100 SW of 21 g (Table 6). Twenty accessions were selected based on flowering time (47-62 days), producing a yield between 20 and 47 g plant −1 . Of these 20 accessions, six (ICC5934, ICC14550, ICC2072, ICC45, ICC494, and ICC273) produced about 59-176% higher yield over Annigeri. Twenty accessions were prioritised based on crop duration, whose DM ranged between 86 and 108 d, and SY varied between 21 and 47 g plant −1 . Of these 20 accessions, 10 produced more than 50% compared with the check Annigeri (Table 6). Nineteen accessions were selected based upon PLHT (44-76 cm) with an SY of 22-37 g plant −1 . Of these, nine accessions showed about 50% higher yield over Annigeri. Based on 100 SW, 20 accessions were selected whose seed weight values ranged between 17 and 33 g, and whose SY ranged from 17 to 34 g plant −1 . Out of 20 accessions, 5 accessions produced an SY of 53-100% more than the check cultivar Annigeri (Table 6).

Kabuli Type
The best check cultivar, KAK2, had an SY of 23 g plant −1 , a flowering time of 48 d, DM of 86 d, PLHT of 42 cm, and 100 SW of 36 g (Table 6). Ten accessions were selected based on flowering time (51-67 d), producing yields in the range of 21-32 g plant −1 . Eight accessions were selected based upon DM (81-110 d), having a yield of 18-31 g plant −1 . Ten accessions were selected based on PLHT, whose height ranged from 51 to 57 cm and whose yield spanned from 22 to 46 g plant −1 . One accession (IG6401) was identified with a height of 57 cm and a yield of 46 g plant −1 , which was about 100% more than KAK2. Further, eight accessions selected based upon 100 SW (28-38 g) had an SY in the range of 19-31 g plant −1 ( Table 6). During the 2015-16 year, KAK2 yielded 10 g plant −1 , a flowering time of 50 d, DM of 96 d, PLHT of 50 cm, and 100 SW of 37 g ( Table 6). Ten accessions were selected based on flowering time (58-68 d), with SY in the range of 16-37 g plant −1 . All the accessions produced about 60-270% higher yield over the check KAK2 (Table 6). Ten accessions selected based upon maturity days (91-108 d) yielded in the range of 14-37 g plant −1 , and all the accessions, excluding one, yielded more than 50% of the check cultivar. Similarly, ten accessions selected based on PLHT (53-61 cm) had SY in the range of 18-29 g plant −1 ; all these accessions produced more than 50% high yield over the check KAK2. Ten accessions were prioritised based upon 100 SW (29-40 g) and yielded 13-19 g plant −1 . Seven of these ten accessions (ICC8483, ICC14220, ICC7654, ICC10747, IG72070, IG69692, and IG70436) exhibited more than 50% yield gain over KAK2 (Table 6). Table 6. Shortlisted germplasm accessions based on ranking biplot for five traits across two years for Junagadh, Gujarat. The check cultivars are highlighted in bold.
During the 2015-16 year, the best check KAK2 produced a yield of 7 g plant −1 , DF of 57 d, DM of 116 d, PLHT of 35 cm, and 100 SW of 34 g (Table 7). Ten accessions selected based upon flowering time (65-74 d) produced yield in the 4-8 g plant −1 . Furthermore, ten accessions were selected based on maturity days (102-115 d) and showed a yield variation of 4-10 g plant −1 ; 10 accessions were selected based on PLHT (40-53 cm) and produced a yield of 4-8 g plant −1 . Based on 100 SW, nine accessions were selected with seed weight values in 20-33 g 100-seeds −1 and yield variation in the 4-10 g plant −1 (Table 7).

Intermediate Type
During the 2014-15 year, KAK2 was the best check cultivar that produced a yield of 12 g plant −1 , DF of 64 d, DM of 122 d, PLHT of 38 cm, and 100 SW of 36 g ( Table 7). One accession (ICC4878) with DF of 79 d and producing SY of 18 g plant −1 (>50% over KAK2) was selected. In addition, one accession (ICC5616) with a PLHT of 40 cm and a yield of 19 g plant −1 displayed~58% yield gain over the check cultivar KAK2 (Table 7).  (ICC14014, ICC8950, ICC12527, ICC13107, ICC8332, ICC1560, ICC15851, ICC5878, ICC8556, and ICC14176) were selected based on early maturity, which showed up to a 10% decrease in crop duration and up to a 106% increase in yield over JG11 (Table 3). For Patancheru, two accessions (ICCV92503 and ICC8348) were prioritised based on early flowering, displaying up to a 15% decrease in flowering time and a 53% increase in yield over the check cultivar JG11. Only one accession (ICCV88202) was selected based on DM, which showed a 10% decrease in maturity days and a 7% higher yield over JG11 (Table 4). For Kanpur, six promising accessions (ICCV92503, ICC12159, ICC8348, ICCV88503, ICCV91902, and ICC10963) were selected for early flowering, which indicated up to 16% reduction in flowering time and up to 44% increase in SY over the check cultivar JG11. One accession (ICC12878) was prioritised based on DM, which revealed a 7% early maturity and a 69% yield increase over JG11 (Table 5). For Junagadh, one accession (ICC8348) was selected that showed a 6% decrease in flowering time and yield performance of 35% over check cultivar Annigeri. Based on DM, seven accessions (ICC1560, ICC6384, ICC4526, ICC95, ICC16841, ICC14495, and ICC5934) were selected, with up to 21% early maturity producing up to 88% higher yield over Annigeri (Table 6). For Sehore, five accessions were selected based on DF, which revealed a 29% decrease in flowering time and a 67% increase in yield over JG11 (Table 7).

Kabuli Type
The kabuli type chickpea, having both early flowering and maturity and producing higher/similar yields over the best check cultivar of the respective location, was selected based on two years' evaluation. For the Amlaha location, two accessions (ICC11879 and IG70436) were selected based on DF, which displayed a 9% decrease in flowering time and up to 100% yield gain over the check cultivar KAK2. Four accessions (ICC16644, IG71878, ICC6831, and ICC8527) were selected based on maturity days, which matured up to 6% early and produced up to 25% greater yield than KAK2 (Table 3). For Patancheru, only one accession (ICCV2NN) was selected based on maturity days, which showed 6% early maturity and a 40% yield increase over KAK2 (Table 4). At Kanpur, eight accessions showed up to 13% early flowering and 79% higher yield over KAK2. Six accessions were prioritised based on early maturity, which showed an 8% decrease in crop duration and a 43% increase in yield over KAK2 (Table 5). Two accessions for the Junagadh location were selected based on early maturity, indicating a 6% decrease in maturity days and up to 60% yield advantage over KAK2 (Table 6). At Sehore, one accession (ICC16644) with 9% early flowering and 50% yield gain over KAK2 was found. Six accessions (ICC16644, IG5868, IG72070, IG7458, ICC14220, and ICC15823) were selected based on early maturity that showed a 3% decrease in DM and up to a 50% increase in yield over the check cultivar KAK2 (Table 7).

Intermediate Type
Like desi and kabuli types, intermediate type cultivars that displayed early flowering and maturity and higher/similar yield over the check cultivar were selected for each of the evaluated locations. For Amlaha, one accession (ICC5955) with 1% early flowering and 3% early maturity with 38% higher yield than KAK2 was selected (Table 3). For Patancheru, none of the accessions was superior to JG11 (Table 4). For the Kanpur location, two accessions (ICC15234 and ICC7574) with early maturity showed a 3% decrease in maturity days and up to 69% yield gain over JG11 (Table 5). For Junagadh, one accession (ICC15234) with 1% early maturity and SY similar to KAK2 was selected ( Table 6). For Sehore, two accessions (ICC8407 and ICC12431) were selected based on early maturity, which indicated a 3% decrease in crop duration and up to 33% yield advantage over KAK2 (Table 7).
3.9. Selection of Mechanically Harvestable Desi, Kabuli, and Intermediate Type Accessions 3.9.1. Desi Type Desi accessions displaying at least a 10 cm increase in the PLHT over the check cultivar and having a higher/similar yield than the best check were selected. For Amlaha, three accessions (ICC13285, ICC5472, and ICC8474) were found to have a 29% increase in PLHT and up to 100% yield gain over the check cultivar JG11 (Table 3). For Patancheru, eight accessions (ICC9139, ICC9307, ICC12993, ICC5472, ICC16234, ICC1817, ICC5722, ICC2356, and ICC10197) were prioritised that had up to 64% taller plants and up to 60% higher yield over JG11 (Table 4). The selected ten accessions displayed up to 49% tall plants for the Kanpur location and 94% yield gain over the check JG11 (Table 5). For Junagadh, five accessions (ICC9307, ICC5491, ICC3558, ICC9261, and ICC2878) showed a 62% increase in height and up to 65% yield advantage when compared to Annigeri (Table 6). For the Sehore location, three accessions (ICC4233, ICC5541, and ICC2500) were selected based on the above-defined criteria, demonstrating up to 50% gain in PLHT and up to 33% increase in yield over JG11 (Table 7).

Kabuli Type
Kabuli accessions showing a minimum of 10 cm increase in PLHT over the best check and those having a higher/similar yield than the check cultivar KAK2 were selected. For Amlaha, two accessions (ICC16659 and IG72021) display up to a 28% increase in PLHT and up to a 31% yield advantage over KAK2 (Table 3). For Patancheru, 19 accessions revealed up to 111% taller plants and a 200% higher yield over the check cultivar KAK2 (Table 4). Furthermore, five accessions (ICC16637, ICC14190, IG70321, ICC16206, and IG6448) had more than 10 cm taller plants and produced up to 60% higher yield for the Kanpur location over the check cultivar KAK2 (Table 5). In the Junagadh location, six accessions (IG6401, ICC16659, ICC9330, IG69746, IG73294, and IG5868) were found to have up to a 36% increase in PLHT and up to 180% yield increase over the check cultivar KAK2 (Table 6). Sehore's seven accessions (ICC8483, ICC10778, IG73286, ICC4899, ICC11879, IG72070, and IG73294) showed a 51% increase in PLHT and up to 92% increase in SY over KAK2 (Table 7).

Intermediate Type
Intermediate type accessions with at least a 10 cm increase in PLHT and having higher/similar yield over the check cultivar were selected. Accordingly, for Amlaha, none of the accessions was superior in height and yield performance over the best check KAK2 (Table 3). For Patancheru, three accessions (ICC7574, ICC8930, and ICC5616) were found to have a minimum of 10 cm increase in PLHT and a 100% increase in yield over the check JG11 (Table 4). One accession (ICC7574) with 61 cm PLHT and 69% higher yield than JG11 was detected for Kanpur (Table 5). For Junagadh, one accession (ICC5616) with a PLHT of 61 cm and a yield of 19 g plant −1 , 90% higher than the KAK2 check, was selected (Table 6). Sehore prioritised one accession (ICC7574) with a 32% increase in PLHT and an 8% higher yield than KAK2 (Table 7). Desi accessions displayed an increase of at least 3 g 100-seeds −1 over the best check and a higher/similar yield than the selected check cultivar. For Amlaha, eight accessions (ICC12452, ICC8348, ICC8474, ICC15762, ICCV88202, ICC11903, ICCV93958, and ICCV91902) were selected that showed up to 183% increase in 100 SW and up to 131% increase in yield over JG11 (Table 3). For Patancheru, five accessions (ICC5003LN, ICC5003HN, ICCV93958, ICC12159, and ICC8474) increased the 100 SW by up to 28% and produced up to 87% higher yield than JG11 (Table 4). For the Kanpur location, three accessions (ICC15065, ICC9139, and ICCV91902) were found to have up to 29% bigger seeds and up to 94% greater yield than the check cultivar JG11 (Table 5). For the Junagadh location, eight accessions (ICCV93958, ICC10301, ICC15762, ICC5003LN, ICC15376, ICC9139, ICC5003HN, and ICC9261) were detected to possess up to a 57% increase in 100 SW and up to 100% yield advantage over Annigeri (Table 6). Sehore identified one accession (ICCV95605) with a 100 SW of 41 g and a 50% higher yield than the JG11 check (Table 7).

Kabuli Type
Kabuli accessions showing higher/similar 100 SW than best check and greater/similar yield over the check cultivar were selected. For Amlaha, one accession (ICC8483) with an 11% increase in 100 SW and yield similar to KAK2 was detected (Table 3). For Patancheru, three accessions (ICC14220, IG70466, and ICC6239) were found to have up to a 29% increase in 100 SW and up to 160% higher yield over KAK2 in the year 2015-16 (Table 4). Furthermore, only one accession (IG70872) was selected from the Kanpur location, with up to 41% bigger seeds and up to 27% greater yield over the check cultivar KAK2 (Table 5). For the Junagadh location, three elite accessions (ICC8483, ICC14220, and ICC7654) with up to 8% bigger seeds and 80% higher yield than the check KAK2 were detected (Table 6). At the Sehore location, none of the accessions was superior in seed weight and yield than the check KAK2 (Table 7).

Intermediate Type
Intermediate type accessions having higher/similar 100 SW than the best check and greater/similar yield over the check cultivar were selected. Agronomically superior accession was detected only for the Kanpur location compared to the JG11 check. The accession (ICC7574) had a 25% better 100 SW and a 69% higher yield than the JG11 check ( Table 5). None of the accessions was suitable for the remaining four locations compared with the JG11 or KAK2 check.

Discussion
Diverse germplasm accessions stored in gene banks (or germplasm repositories) hold the key to most future success in developing improved crop cultivars [22,23]. That said, a major challenge for the gene bank curators or plant breeders is to characterise and determine beneficial genetic variations from the vast germplasm collections, particularly for agronomically important traits that need multi-location evaluation. Developing a composite collection was suggested as a gateway for harnessing the available genetic diversity in applied plant breeding to expand the breeding populations' genetic base [18]. In the present study, the multi-location evaluation of a chickpea composite collection allowed a selection of elite germplasm accessions with superior agronomic performance, different from the check cultivars. This, in turn, provided novel sources of variation for economically important traits, which can be utilised to increase the genetic potential of desi, kabuli, and intermediate type chickpea. When such diverse accessions are deployed in chickpea breeding programs, it is predicted that there will be a high possibility of the appearance of transgressive segregants with useful traits because of the reshuffling of alleles due to recombination. Such beneficial traits can then be selected to identify high-yielding accessions with desired allelic combinations.
A breeder is mostly interested in selecting high-yielding and stable accessions compared to the best check of the location to reap economic benefits. Therefore, the desi, kabuli, and intermediate type accessions that yielded 50% above the check cultivar of a particular location were prioritised in the present study. The desi accessions were compared with either JG11 or Annigeri check cultivars. In contrast, the kabuli accessions were compared with the KAK2 check, representing some of the most widely used checks in India's northern and southern parts. According to a recent survey on the adoption of chickpea varieties, the JG11 cultivar was found to cover almost 82% of the total chickpea growing area in Andhra Pradesh [24]. The desi, kabuli, and intermediate type germplasm accessions producing up to 176%, 270%, and 169% higher yield over the check cultivars were identified in the current study. However, further field evaluation of the selected high-yielding accessions in large trials and optimal sites for the target population of the environment will be required to confirm their yield potential and their deployment in chickpea breeding programs.
Chickpea is usually grown on residual soil moisture in the post-rainy season and is mostly subjected to terminal drought and intermittent heat or is sown after harvest of long/medium duration rainy season crops that face heat stress towards the end of the cropping cycle [6,25]. Developing early maturing chickpea cultivars that can escape terminal drought or heat stresses is highly desirable. Efforts are being made to develop varieties with varying maturity duration (75-120 d) for cultivation under rainfed and irrigated conditions in different parts of the country. For example, early (90-100 d), extra early (85-90 d), and super early (75-80 d) maturing cultivars of desi and kabuli types will have an advantage in southern and central India. A key success story of the adoption of early-maturing chickpea is represented by the chickpea revolution being brought about in the Andhra Pradesh state of India, where over 80% of the chickpea growing areas are now cultivated with short-duration improved cultivars such as JG11 (desi) and KAK2 (kabuli) [26]. In the present study, desi accessions that showed at least three days early flowering and seven days early maturity when compared to JG11 or Annigeri, with higher/similar SY than the respective check, were selected.
On the other hand, kabuli and intermediate type accessions having early/similar flowering and maturity and higher/similar yield over the respective check cultivar were selected. Such short-duration chickpea could be used as a cash crop (after rice harvest and before wheat sowing). It may provide extra income to the farmers and utilisation of available resources. These accessions will also provide an opportunity to expand the chickpea area under late-sown conditions, which is immediately required to diversify the rice-wheat cropping system [27].
Although mechanical harvesting of pulse crops is extensively practised in countries such as the United States of America, Canada, and Australia [28], chickpea is still harvested manually in the Middle East, North Africa, and South and West Asian countries [29]. In most countries, labour cost is continuously rising, and manual harvesting has become an expensive and time-consuming field operation. Mechanical harvesting of chickpea offers a promising alternative to manual harvest. It is expected to reduce production costs and the chances of damage to the crop due to fluctuating environments, which usually occurs when there is a delay in harvesting due to the non-availability of labour. The harvest loss during machine harvest is higher for the semi-erect genotype (~20%) and low in tall and erect genotypes (2.6-5.0%) [16]. Therefore, germplasm accessions showing at least a 10 cm increase in PLHT over the check cultivars were selected in the present study. These selected accessions must be further evaluated under large-scale field trials for yield potential and machine harvesting to identify the percentage of economic gain associated with the harvest.
In chickpea, erectness and height of the first pod from the ground (ground clearance) are the two most crucial traits that primarily decide a cultivars' suitability for mechanical harvesting [29]. Therefore, there is also an emerging need to study the selected accessions' growth habits and the lower pods' height from the base before their recommendation to be included in breeding programs. The resulting taller and erect genotypes with higher yields are predicted to enhance chickpea area and production mostly in the dryland tropics, which is required to achieve self-sufficiency in pulse production.
The large-seeded desi and kabuli types have high consumer preferences because of the premium price to the farmers. The bold seeded chickpea is used as a green vegetable, parched, fried, roasted, and boiled as a snack food and chhole, while sprouted seeds are eaten as vegetables or added to salads. A farmer prefers to have large-seeded cultivars with high yield potential from the market point of view for gaining extra income. This is because the trading of extra-large seeded kabuli types dominates the international market [17,30]. The premium price of >USD 1000 per tonne is paid for kabuli types having extra-large seeds; whereas for kabuli types with medium-size seeds, the price is about USD 600 per tonne, and for desi types with medium-size seeds, the prices range between USD 400 and 500 per tonne. Keeping these things in mind, desi germplasm accessions showing at least a 3 g increase in 100 SW and higher/similar yield over the JG11 or Annigeri were selected. In addition, kabuli and intermediate type accessions with larger/similar 100 SW and SY over the check cultivars were selected. This resulted in identifying desi, kabuli, and intermediate type accessions showing up to 183%, 41%, and 25% increase in 100 SW, respectively, with higher/similar yields over the check cultivar. These germplasm accessions can be deployed in a breeding program to develop large-seeded high-yielding cultivars that combine earliness, high pod clearance, and 100 seed weight or can be used directly for cultivation after evaluating their performance in large-scale field trials.

Conclusions
Over two years, the chickpea composite collection was evaluated at five locations in India for five agronomic traits. The desi, kabuli, and intermediate type chickpea showed significant variation for PLHT and 100 SW. The selected germplasm accessions displayed 169-270% higher yield over check cultivars. About 25-183% higher 100 SW was observed in selected germplasm accessions compared to check cultivars. The identified superior accessions hold potential for enhancing the genetic potential of chickpea cultivars.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/agronomy12092013/s1, Table S1: Entry number and name of desi type chickpea lines used for ranking biplots; Table S2: Entry number and name of kabuli and intermediate type chickpea lines used for ranking biplots; Table S3: Mean for five agronomic traits among accessions belonging to desi, kabuli, and intermediate types in chickpea composite collection evaluated for two years at five locations in India; Table S4: Differences between means of desi, kabuli, and intermediate type chickpea for five agronomic traits evaluated for two years at five locations in India; Table S5: Pearson correlation analysis for five agronomic traits evaluated for two years at five locations in India.  Data Availability Statement: All the phenotyping data generated and used for analysis have been submitted to the ICRISAT Breeding Management system.