Validation of Core

The core sets developed by three strategies [i.e. species-specific PowerCore (Core P), modified PowerCore (Core PM) and PowerCore involving stratified random sampling based on passport and clustering (Core PG)] were validated by different criteria based on summary of statistics. Means of the entire collection and core subset were compared using Newman-Keul's procedure (Newman 1939; Keuls 1952) for the 12 traits. The homogeneity of variances of the entire collection and core subset was tested with the Levene's test (Levene 1960). It is worth noting that the HCC method gave the same range, minimum and maximum values for the core set generated and the entire collection, indicating its capability to capture almost all of the existing variations. In order to compare the efficiency of “PowerCore” for developing core collection with modified stepwise method and PowerCore with grouping approach method, mean and statistical parameters for entire population, core developed using “PowerCore” and core developed using PowerCore with modified strategy of stepwise method and PowerCore with grouping were compared. The results showed that there was no significant difference (α = 0.05) for the means of all traits between core and entire collections. The variances of the entire collection and core subset were homogeneous only for five traits viz. days to maturity, plant height, grains per spike, biomass and harvest index. The reason might be due to the large number of germplasm in the entire collection in comparison to that of the core collection. The range of the characters was the same in the entire collection as well as in the core collection implying that the core captured extreme diversity of the total collection (Table 4.2). Four statistical parameters viz., MD (%), VD (%), CR (%) and VR (%), were analyzed using “PowerCore” to compare the mean and variance ratio between core and entire collections. The percentage of the significant difference between the core sets and the entire collection was calculated for the mean difference percentage (MD%) and the variance difference percentage (VD%) of traits. Coincidence rate (CR%) and variable range (VR%) were estimated to evaluate the properties of the core set against the entire collection (Hu et al. 2000).

Mean Difference Percentage (MD %) – which is estimated as:

1 m MeMc

MD (%) = ∑ ×100

m j =1 Mc

Where, Me = Mean of entire collection; Mc = Mean of core collection, and m= number of traits.

Variance Difference (VD %) – estimated as:

1 m VeVc

VD (%) = ∑ ×100

m j =1 Vc

Where, Ve =Variance of entire collection, Vc =Variance of core collection, and

m = number of traits.

Table 4.2 Descriptive statistics for quantitative traits and their validation in entire and core collection using PowerCore-M approach

Traits

Mean

Sig*

Variance

F value

Sig**

Std. Dev.

Minimum

Maximum

Entire

Core-PM

Entire

Core-PM

Entire

Core-PM

Entire

Core-PM

Entire

Core-PM

SE

109.8

111.3

ns

130.8

267.6

8.77

s

11.44

16.36

54

54

164

164. 00

DM

146.5

147.4

ns

81.66

141.4

5.87

ns

9.04

11.89

120

120

199

199

PH

114.9

115.8

ns

517.7

740.4

0

ns

22.75

27.21

33.5

33.5

197.8

197.8

ET

13.79

14.83

ns

14.42

37.26

70.45

s

3.8

6.1

3

3

48.4

48.4

SL

11.22

11.48

ns

4.21

8.2

11.03

s

2.05

2.86

2.72

2.72

24.6

24.6

SS

19.79

20.25

ns

5.65

14.76

51.82

s

2.38

3.84

5.4

5.4

60.4

60.4

GS

45.56

45.45

ns

131.6

276.7

0.78

ns

11.47

16.63

3.8

3.8

111.8

111.8

GW

1.79

1.88

ns

0.37

0.85

19.93

s

0.61

0.92

0.04

0.04

7.16

7.16

TW

39.81

41.93

ns

151.3

355.3

32.42

s

12.3

18.85

2.72

2.72

204.1

204.1

DM

353.4

347.4

ns

11,120

20,376

6.08

ns

105.5

142.7

20

20

1,042

1,042

SY

91.06

88.71

ns

1,463

2,578

9.8

s

38.26

50.77

1.33

1.33

283.2

283.2

HI

23.92

24.44

ns

93.37

156.8

1.76

ns

9.66

12.52

0.81

1.17

58.34

58.15

Core-PM core developed by Powercore with modified stepwise approach

*Significant at 5 % level, **Significant at 1 % level

Coincidence rate (CR %) – estimated as:

1 m Rc

CR (%) = ∑ ×100

j =1

Where, Re = Range of entire collection, Rc = Range of core collection, and m = number of traits.

CR% indicates whether the distribution ranges of each variable in the core set are well represented.

Variable rate of CV (VR %) – estimated as:

1 m CVc

VR (%) = mCVe ×100

Where, CVe = Coefficient of variation of entire collection, CVc = Coefficient of variation of core collection, and m = number of traits.

VR% allows a comparison between the coefficient of variation values existing in the core collections and the entire collections, and determines how well it is being represented in the core sets.

Hu et al. (2000) reported that an MD% smaller than 20 %, in his case 10.07 %, effectively represented the entire collection. The high value obtained for coincidence rate (CR) percentage (95.57 %) suggests that the core attained using the HCC method could be adopted as a representative of the whole collection. In this case, the estimated value for MD% was −6.25, which indicated that there is no difference in the mean values of entire and core collections. VD% was estimated to be 49.04, indicating that the variance for the entire and the core populations are not the same. The CR% obtained was 96.06 which suggests that the core has captured all accessions from all the classes and, thus, is a representative of the entire collection. High VR% (53.87) indicated that the coefficient of variation in the core set is higher compared to entire collections for all the variables. The coefficient of variance in core developed using PowerCore was highest in the case of PowerCore with grouping followed by PowerCore with modified approach and entire collection for all the descriptors. The histogram comparing CV for the entire and core sets is shown in Fig. 4.1. High value obtained for CR% (96.06) suggests that the core obtained using the heuristic approach method could be adopted as a representative of the whole collection.

Shannon-Weaver Diversity Index

The descriptor and descriptor states are parallel to the locus and alleles, respectively, in morphological evaluation. Allelic evenness and allelic richness are the most commonly used parameters for measuring diversity. The allelic evenness in

Fig. 4.1 Coefficient of variation (%) in entire, modified core (Core-PM) and group based core collection (Core-PG) for different traits. DSE days to 75 % spike emergence, DM days to 90 % maturity, PH plant height, EFT effective tillers per plant, SL spike length, SLS spikelets per spike, GRS grains per spike, GRW grain weight per spike, TGW 1,000 grain weight, DMY dry matter yield per m row length, SY seed yield per m row length and HI harvest index

this study was measured using the Shannon–Weaver diversity index, whereas the allelic richness was measured by counting the descriptor states for each descriptor without considering their individual frequencies. The Shannon-Weaver diversity index (H') was computed using the phenotypic frequencies to assess the phenotypic diversity for each character.

n

H = −Σpi ⋅ ln pi

i = 1

where pi is the proportion of accessions in the ith class of an n-class character and n is the number of phenotypic classes for a character. A comparison of ShannonWeaver (Shannon and Weaver 1949) diversity index for the entire collection, core developed using PowerCore, core developed using modified power core with stepwise approach and PowerCore with clustering method also indicated a high diversity for all the quantitative traits in core developed using PowerCore-M compared to core developed using PowerCore-G approach, except for a few variables, where it was observed at par (Fig. 4.2).

Fig. 4.2 Validation of modified core (Core-PM) and group based core collection (Core-PG) in comparison to entire collection by Shannon diversity index for quantitative traits (traits same as given in Fig. 4.1)

Conclusions

PowerCore is a new and faster approach for developing core collection, which effectively simplifies the generation process of a core set with reduced number of core entries while maintaining high percent of diversity compared to other methods used. Using PowerCore as a tool, three sets of core collections viz. Core P, Core PM and Core PG have been developed. Due to its high Shannon-diversity index, Core PM proved to be the best. These core sets can be further grown with involvement of breeders to select the genotypes with desired background suiting to their requirement. The core sets can be used as a guide for developing trait specific reference/ core sets and subsequent allele mining. The best core set could be used as an initial starting material for large-scale genetic base broadening. Thus, it can be concluded that this modified heuristic algorithm can be applied for the selection of genotype data (allelic richness), the reduction of redundancy and the development of approaches for more extensive analysis in the management and utilization of large collection of plant genetic resources.

Acknowledgments We acknowledge with thanks the financial support from the National Initiative on Climate Resilient Agriculture (NICRA) project of the Indian Council of Agricultural Research (ICAR). The technical guidance received from Bioversity International, South Asia Office, New Delhi in the use of software is gratefully acknowledged.

 
< Prev   CONTENTS   Next >