Previous PageTable Of ContentsNext Page

Heritability of Drought–Resistance Traits in Peanut

Alan Cruickshank1, Arnaud Dowkiw12, Graeme Wright1, NageswaraRao C. Rachaputi1 and Shyam Nigam3

1 Queensland Department of Primary Industries and Fisheries, Agency for Food and Fibre Sciences, PO Box 23 Kingaroy, Qld 4610 Email Alan.Cruickshank@dpi.qld.gov.au, Graeme.Wright@dpi.qld.gov.au, Rao.Rachaputi@dpi.qld.gov.au
2
Present address Email A-Dowkiw@ifrance.com
3
ICRISAT, Patancheru, Andhra Pradesh, INDIA Email S.Nigam@cgiar.org

Abstract

The paper presents results from a case study on genetic enhancement of drought resistance in peanut using combined physiological and breeding methods. Peanut genotypes with high levels of components of a resource capture model, transpiration (T), transpiration efficiency (TE) and harvest index (HI), were used as parents in a hybridisation program. F2-derived families of peanuts from 4 crosses were subjected to trait-based among family index selection in the F3 and F4. In order to calculate the selection index, estimates of harvest index, transpiration efficiency and transpiration were obtained. Heritability estimates for these traits were estimated using 3 different bases i.e. broad-sense heritability on an F3 row basis, broad-sense heritability on an F4 family mean basis using variance components and standard unit heritability estimates using correlation among generations.

The heritability estimates varied significantly between crosses and traits depending on levels of genetic variation in parents. In some cases the variation seems to be caused by errors in sampling method.

Kernel yield (KY) generally had the best correlations between F3 and F4 (maximum 0.65), HI and TE were intermediate and correlations for T were very poor (one cross in one environment 0.56 the remainder not significant).

Consideration of these differences in heritability and the costs of sampling lead to the proposal that a simpler index based on KY and TE may be a useful compromise for selection of superior drought performance in peanut. Use of an index including TE would prevent selection for decreasing TE when selecting for KY, where there are casual negative associations between TE and the other two components.

Media summary

It is possible to breed peanut varieties that produce more yield from the same amount of rainfall by selecting for a combination of high transpiration efficiency and kernel yield.

Key Words

Drought tolerance, peanut, breeding, transpiration efficiency, kernel yield.

Introduction

The yield of peanut in India and Australia is usually severely limited by water deficits during crop growth, arising from unpredictable rainfall, high evaporative demands and production on low water holding capacity soils. The breeding of more drought resistant genotypes is a strategy to increase productivity in drought prone environments. New breeding approaches utilising physiological traits have been proposed to improve the efficiency of selection for superior drought tolerant genotypes. Most of these efforts to date have however been unsuccessful, as the specified traits have been considered in isolation and have not necessarily been related to superior performance under drought stress.

New opportunities to develop higher yielding drought tolerant peanut genotypes emerged in the ACIAR-DPI project – ‘Selection for WUE in Food Legumes’, which developed a detailed understanding of the physiological factors determining yield in water-limited environments (Wright et al 1996). A resource capture model proposed by Passioura (1977) was used to analyse pod yield variation under water limited conditions in terms of three functional components i.e transpiration (T), transpiration efficiency (TE) and harvest index (HI). If these traits have significantly greater heritability than yield, or if they can be measured at lower cost, they will provide an avenue for more effective selection for yield under water-limited conditions. A subsequent project applied a trait-based selection strategy in 5 breeding programs (4 in India, 1 in Australia) (Rachaputi et al 2000). This paper presents results of different estimates of some genetic parameters for the water-capture-model traits from the F3 and F4 generations in the Australian breeding program.

Methods

Genetic material

Genotypes with high levels of T, TE and HI were selected from a wide range of germplasm screened in the ACIAR-funded project conducted in India and Australia (Wright and Nageswara Rao 1994). A selection of these genotypes were intercrossed to provide populations for selection studies. The four Australian crosses are presented in Table 1.

Table 1. Crosses used in the selection studies.

Cross code

Female parent

Traits

Male parent

Traits

AX1

Streeton

High T, medium HI, low TE

CSMG 84-1

High T, medium TE, low HI

AX2

ICGV 86031

High TE, low T & HI

TAG 24

High HI, medium TE, low T

AX3

TAG 24

High HI, medium TE, low T

CSMG 84-1

High T, medium TE, low HI

AX4

Streeton

High T & HI, low TE

ICGV 86031

High TE, low T & HI

Experimental design and traits measured

Up to 500 F3 rows (from unselected F2 plants) per cross were grown under non-limiting water conditions. All the selection was conducted at 90 cm row spacing. All were assessed for kernel yield (KY), total shoot dry matter (TDM), TE (via SPAD – a measure of chlorophyll content), HI and T estimated using the methods of Wright et al. 1996. TDM was estimated from 0.9 m2 quadrats at maturity, KY was estimated from the remainder of the plot. HI was calculated from the independent estimates of TDM and KY to avoid manual separation of pods and vegetative material. SPAD readings were measured twice during the crop growth cycle, the single reading with the greatest precision was used to estimate TE (Nageswara Rao et al. 2001). A value was calculated for each progeny using a selection index (S) derived from the traits T, TE and HI (Chandra et al 2003). The top 50 F2-derived families of were carried forward to the F2:4 generation.

The F2:4 families were divided equally between irrigated and rainfed experiments, and the same measurements made as in the F3 generation. For selection in the F3 there was one experiment per cross, with F3 rows unreplicated with replicated check plots of the commercial variety Streeton. In the F4, 50 families per cross were divided into 25 each for an irrigated and a rainfed experiment. Each of the eight F4 experiments was a 3 replicate, 30 entry lattice with 5 check varieties.

Selection procedure

The three traits (T, HI and TE) were combined into the following selection index:-

S=∑j (xj-medj)/QRj, QRj={Q3(j)-Q1(j)}

Where, x is…. Q1 and Q3 are the first and third quartiles, and in our case there are j = 3 traits (T, HI and TE) included in the index, a modification of index S7 of Chandra et al (2003). The index, S, ensured that selection was not influenced by extreme values and gave equal weighting to each trait. Median and quartiles were used so that no assumptions were made about distribution(s) of the data.

Estimates of heritability

Following Nyquist (1991) broad-sense heritability was estimated in the F3 by using the variation of Streeton plots within each experiment as an estimate of environmental variance, and in the F4 restricted maximum likelihood (REML) variance components were used to calculate broad-sense heritability on a progeny mean basis. Standard unit heritabilities (Frey and Horner 1957) or intergeneration correlations were calculated by the Pearson's correlation between F3 and F4. These three estimates of heritability are not directly comparable as they each have a different basis, but comparisons among crosses are valid using the same basis.

Results and Discussion

Broad-sense heritability (H) estimates for KY varied from 17 to 78 % in the F3 and 37 to 91 % in the F4 (Table 2). Cross AX4 had higher H values in the F4 but the correlation between generations was not significant. Crosses AX2 and AX3 had higher H in the F4 with significantly positive correlation with the F3. While there was some consistency between the F4 H values and the correlations, neither aligned well with the estimates from the F3 The F3’s are unselected populations so the potential genetic variability in each population is not truncated but the lack of replication reduces the precision with which the variation is measured. Conversely in the F4, the replicated measurement ensured nearly all cross-environment combinations had a significant H of 70% or greater. Hence, for KY the F3 estimates may indicate the presence of genetic variability however they are not good predictors of the effectiveness of selection as measured by the correlations.

Table 2. Estimates and indicators of kernel yield heritability.

 

Broad-sense heritabilities (%)

F3-F4 Intergeneration Correlations B

Cross

F3

F4 Irrigated A

F4 Rainfed A

Irrigated

Rainfed

AX1

78

ns

49

ns

0.60

AX2

17

87

79

0.65

0.42

AX3

63

91

88

0.57

0.38

AX4

47

71

69

ns

ns

A ns = no significant progeny effect (P≤0.05). B ns = not significantly different from 0 (P≤0.05).

HI showed some similarities to KY with a wide range of estimates of H from the F3. Cross AX4 had the lowest values of H and crosses AX2 and AX3 were more consistent in the F4 over the two watering environments (Table 3). Despite similarities, generally the heritability of HI was poorer than that of KY particularly for the correlation between F3 and F4. One of the reasons for using physiological traits in selection is that they may have better heritability than direct yield measurement. The intergeneration correlation suggests this was not the case for HI with this genetic material.

Table 3. Estimates and indicators of Harvest Index heritability.

 

Broad-sense heritabilities (%)

F3-F4 Intergeneration Correlations B

Cross

F3

F4 Irrigated A

F4 Rainfed A

Irrigated

Rainfed

AX1

55

ns

61

ns

0.42

AX2

62

75

59

0.44

ns

AX3

62

80

85

0.56

0.42

AX4

16

ns

58

ns

ns

A ns = no significant progeny effect (P≤0.05). B ns = not significantly different from 0 (P≤0.05).

All four populations showed moderate to high H for TE in the F3 (Table 4). Only the AX4 cross had significant heritability for TE on all bases, possibly because it had the greatest contrast between the parental genotypes for this trait (Table 1). Eventual multi-site evaluation of these families (Cruickshank et al 2003) has shown that TE is the most stable model factor over sites and that the trait-based selection index was successful at improving TE but not the other physiological traits.

Table 4. Estimates and indicators of Transpiration Efficiency heritability.

 

Broad-sense heritabilities (%)

F3-F4 Intergeneration Correlations B

Cross

F3

F4 Irrigated A

F4 Rainfed A

Irrigated

Rainfed

AX1

50

86

49

ns

ns

AX2

53

ns

ns

0.55

ns

AX3

71

ns

ns

0.46

ns

AX4

61

77

67

0.53

0.38

A ns = no significant progeny effect (P≤0.05). B ns = not significantly different from 0 (P≤0.05).

Heritabilities of T were generally poor (Table 5). This is not surprising as it is a second-order, derived estimate of T (i.e. T = TDM/TE). Also the 0.9 m2 quadrat for estimation of TDM could be a probable source of experimental error. Washing and drying of the TDM sample incurs a substantial cost. Doubling the size of the sample may have helped the accuracy of the T estimate; it would drastically increase the total cost associated with this selection protocol.

Table 5. Estimates and indicators of Estimated Transpiration heritability.

 

Broad-sense heritabilities (%)

F3-F4 Intergeneration Correlations B

Cross

F3

F4 Irrigated A

F4 Rainfed A

Irrigated

Rainfed

AX1

12

ns

ns

ns

ns

AX2

<0.001

53

46

ns

ns

AX3

54

ns

44

ns

0.56

AX4

70

49

ns

ns

ns

A ns = no significant progeny effect (P≤0.05). B ns = not significantly different from 0 (P≤0.05).

The smaller plant stature of TAG 24 and ICGV 86031 is much better suited to narrower row spacing and higher plant density. The lower T achieved under the wide row planting arrangement may have imposed a “maximum yield ceiling” on all progeny in cross AX2, and many progeny in AX3 and AX4, and reduced genetic variability (hence heritability) of T and KY in those 3 crosses. Cruickshank et al (2003) concluded that the choice of parents for those three crosses was not the most suitable for the Australian cropping system. Cross AX1 expressed only small variation in T but as the mean value for T was high (data not shown), low heritability for T in this cross points to potential fixation of desirable genes for high T.

This study demonstrated that it was possible to select and breed for higher TE using novel and cost effective selection tools such as SPAD chlorophyll meter readings (Wright et al., 1996). The SPAD meter provides a practical tool for breeding programs to now make improvements in TE and potentially T. Results from our studies clearly demonstrated that trait-based selection for high TE (via SPAD) was more efficient than empirical yield selection for improvement in TE (Cruickshank et al 2003). The challenge remains to be able to concurrently select for high levels of the three traits (T, TE, HI). TE and direct measurement of kernel yield had the most consistent intergenerational correlations while HI and T were less consistent. Heritabilities of T and HI are both dependent on the precision of TDM measurement, which cannot be improved without substantially increasing costs associated with sampling.

Conclusion

Kernel yield (KY) generally had significant correlations between F3 and F4, HI and TE were intermediate and correlations for T were very poor. This may be primarily due to the measurement of KY, which incorporates most of the plot, whereas HI and T estimation was dependent on a relatively small quadrat sample for shoot dry matter (TDM). Consideration of these differences in heritability and the costs associated with TDM sampling lead to the proposal that a simpler index based on KY and TE may be a useful compromise. KY integrates the effects of HI and T. Use of an index including TE will prevent selection for KY decreasing TE where there are casual negative associations between TE and the other two components.

References

Cruickshank AW, Rachaputi NC, Wright GC and Fresser DB (2003). In ‘Breeding of Drought Resistant Peanuts’. (Ed. AW Cruickshank, NC Rachaputi, GC Wright and SN Nigam) pp. 52-57. (ACIAR, Canberra).

Chandra S, Nigam SN, Cruickshank AW, Bandyapadhyaya A and Harikrishna S (2003). Selection Index for identifying high yielding groundnut genotypes in irrigated and rainfed environments. Annals of Applied Biology 143, 303-310.

Frey KJ and Horner T (1957) Heritability in standard units. Agronomy Journal 49, 59-62.

Nageswara Rao RC, Talwar HS and Wright GC (2001). Rapid assessment of specific leaf area and leaf nitrogen in peanut (Arachis hypogaea L.) using a chlorophyll meter. Journal of Agronomy and Crop Science 186, 175-182.

Nageswara Rao Rachaputi, Wright GC, Cruickshank AW, Basu MS, and Nigam SN (2000). Genetic Enhancement of Drought resistance in Peanuts. (Poster) Proc. International Crop Science Congress. Aug 17-22, 2000. Hamburg, Germany.

Nyquist WE(1991) Estimation of heritability and prediction of selection response in plant populations. Critical Reviews in Plant Sciences 10, 235-322.

Passioura JB (1977) Grain yield, harvest index and water use of wheat. Journal of the Australian Institute of Agricultural Science 43, 117-120.

Wright GC and Nageswara Rao RC (1994). "Selection for water-use efficiency in grain legumes". ACIAR Technical Report No. 27: The Australian Centre for International Agricultural Research. 70pp.

Wright GC, Nageswara Rao RC, and Basu MS (1996). In ‘Plant Adaptation and Crop improvement’. (Ed. M Cooper and GL Hammer) pp.365-381 (CAB International, Wallingford, UK).

Previous PageTop Of PageNext Page