Crop Science Grow Your Career with CSSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 23 February 2005
Published in Crop Sci 45:722-727 (2005)
© 2005 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in Crop Science
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Collaku, A.
Right arrow Articles by Harrison, S. A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Collaku, A.
Right arrow Articles by Harrison, S. A.
Agricola
Right arrow Articles by Collaku, A.
Right arrow Articles by Harrison, S. A.
Related Collections
Right arrow Water Stress
Right arrow Wheat
Right arrow Crop Genetics
Right arrow Statistics

Heritability of Waterlogging Tolerance in Wheat

A. Collakua,* and S. A. Harrisonb

a J&J PRD, 1000 Route 202 S., Raritan, NJ 08869
b Dep. of Agronomy, Louisiana Agric. Exp. Stn. (LAES), Louisiana State University Agricultural Center, Baton Rouge, LA 70803

* Corresponding author (acollaku{at}prdus.jnj.com)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
Tolerance of wheat (Triticum aestivum L.) to waterlogging is related to many morphological and physiological traits that are under a strong environmental influence. Often their genetic control is confounded by environmental stress. The objective of this study was to estimate narrow-sense heritability for grain yield and yield components under waterlogging conditions and to provide selection criteria for waterlogging tolerance in early generations. We studied 80 families derived from four segregating soft red winter wheat populations in the F2 generation. The experiment was conducted under the effect of 5 wk of waterlogging stress. The Restricted Maximum Likelihood (REML) method was used to estimate genetic variance components. In contrast to traditional methods, REML has no limitations on the mating design and accounts for the relationships among families in a breeding population. Grain yield had the lowest heritability (h2 = 0.25). The highest heritability estimates were found for kernel weight (0.47), chlorophyll content (0.37), and tiller number (0.31). Strong genetic correlations were observed between grain yield and kernel weight (r = 0.56), and between grain yield and tiller number (r = 1). Selection for a relatively highly heritable trait, such as kernel weight, would be an effective way to improve waterlogging tolerance in early generations, as grain yield has a low heritability. Genetic and phenotypic information about traits was used to construct selection indices. A yield improvement of 17% is expected by selection on the basis of the index: grain yield-kernel weight-tiller number.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
WATERLOGGING STRESS is one of the limiting factors influencing wheat production, especially in the lower Mississippi valley. About 12% of agricultural soils in the USA are affected by excess soil water (Boyer, 1982). Average grain yield losses of 39 to 44% were found in wheat under waterlogging field conditions (Musgrave and Ding, 1998; Collaku and Harrison, 2002). In several studies, number of kernels was identified as the primary factor responsible for yield losses from waterlogging (Gardner and Flood, 1993; Musgrave and Ding, 1998), while Collaku and Harrison (2002) found a combined effect of reduced tiller and kernel number in decreasing grain yield under waterlogging conditions.

Genotypic variation for tolerance to waterlogging in wheat has been reported in several studies. Davies and Hillman (1988) demonstrated differences in vegetative growth and grain yield of various wheat species under continuous flooding conditions. Other varietal studies for waterlogging in wheat have been reported by Thompson et al. (1992), Huang et al. (1994), and Musgrave and Ding (1998). By screening different wheat genotypes under field conditions, Collaku and Harrison (2002) identified tolerant cultivars useful for waterlogged environments and revealed the potential for waterlogging tolerance in breeding material.

There is some evidence for genetic control of waterlogging tolerance. Alcohol fermentation genes, such as Adh found in wheat (Hart, 1980), barley (Hordeum vulgare L., Good and Crosby, 1989), and rice (Oryza sativa L., Umeda and Uchimiya, 1994), are frequently associated with waterlogging tolerance. Mujer et al. (1993) presented evidence supporting the presence of a few genes controlling hypoxia tolerance in grasses. Taeb et al. (1993) associated chromosome 2E and 4E of Thinopyrum elongatum (Host) D.R. Dewey with enhanced root growth under waterlogged conditions. Boru et al. (2001) studied the segregation ratios of all possible crosses between three waterlogging tolerant spring wheat lines and two sensitive cultivars and suggested that tolerance to waterlogging is controlled by a small number of genes. VanToai et al. (2001) identified a QTL associated with tolerance of soybean [Glycine max (L.) Merr.] to waterlogging.

Although some progress has been made in molecular studies to identify QTL associated with waterlogging tolerance, little is known about the heritability of this trait, particularly in wheat. In the absence of a reliable marker, different morphological and physiological traits have been used in genetic studies for waterlogging tolerance. Most of these traits are quantitative and their expression is usually under strong environmental influences (Trought and Drew, 1980; Kozlowski, 1984; Waters et al., 1991). By estimating heritability of such traits, it is possible to account for genetic and nongenetic factors influencing waterlogging tolerance. The objectives of this study were: (i) to estimate heritability of waterlogging tolerance for grain yield and other quantitative traits of wheat in early breeding generations and (ii) to provide selection criteria for waterlogging tolerance in wheat breeding programs.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Genetic Material
The initial genetic material was derived from the wheat breeding program at the Louisiana Agricultural Experiment Station (LAES), Agricultural Center, Louisiana State University. It consisted of 80 families derived from four related segregating F2 populations: Population 1: Tchere/Savannah//GA 85240; Population 2: Tchere/Savannah//Pioneer 2643; Population 3: Tchere/DS2368//GA 85240; and Population 4: Tchere/DS2368//Pioneer 2691.

‘Tchere’ is known as a waterlogging tolerant variety but is very susceptible to disease, especially leaf rust (Puccinia triticina Eriks.). ‘DS 2368’ was included in crosses as a sensitive variety to waterlogging, while ‘Savannah’, ‘GA 85240’, ‘Pioneer 2643’, and ‘Pioneer 2691’ were used as parental components because of their high grain yield performance in wheat trials (Harrison et al., 1997).

In 1996, seed from all four crosses was planted in Styrofoam nursery trays in the greenhouse, and thinned to one plant per cell. Four weeks after germination, plants were transplanted in the field at the Ben Hur Research Farm, LAES Central Station, at Baton Rouge. Plots were 5.0 x 2.0 m. Late-maturing plants and those that had disease incidence were eliminated from the study. About 120 to 150 plants were grown for each population. From these, 20 plants per population were randomly chosen and hand harvested separately to produce F2:3 families. In total, 80 plants were chosen from the four populations.

The genetic structure of families was half-sibs between Populations 1 and 2 and between Populations 3 and 4. The 80 families in 1997-1998 were F2:3, i.e., F2–derived F3 families, as no selection was applied to the base population of 120 to 150 F2 plants per population. In the 1998-1999 season, the 80 families were F2:4, and in the 1999-2000 season, they were F2:5. Variation among F2–derived lines possesses the same amount of additive variance as subsequent generations, regardless of generation (i.e., F3, F4, or F5) when the measurements were taken, since no selection was applied during the 4 yr of the study (Cockerham, 1963, 1983; Wricke and Weber, 1986, p. 41–169, 355–360). Estimation of heritability was based on the following genetic assumptions: (i) parents involved in crosses were homozygous; (ii) no relationship existed among the six genotypes involved in crosses; (iii) populations from each cross were large enough, so that genetic drift could be considered zero; and (iv) no epistatic interaction of additive x additive, additive x dominance, and dominance x dominance existed.

Experimental Design
The experiment was conducted from the 1997–1998 through the 1999–2000 seasons, at the Ben Hur Research Farm, LAES Central Station, at Baton Rouge. The 80 families from the four populations were studied in a randomized complete block design with three replications (Hinkelmann and Kempthorne, 1994) under the effect of 5 wk of waterlogging stress. Plots consisted of three 22-cm-row spacing x 1-m length. Soil type was a Commerce silt loam (fine salty, mixed, nonacid, thermic Aeric Fluvaquent). A pre-plant fertilizer (8-24-24) at a rate of 240 kg ha–1, equating 19.2 kg ha–1 N, and 57.6 kg ha–1 P and K and a clorsulfuron herbicide (Glean) at 200 L ha–1, were applied. Levees of 30 to 40 cm high were constructed for each replication. Waterlogging started at the 4- to 6-leaf stage by pumping water and flooding the plots within the levees. The soil was kept saturated with water above field capacity by continuous flooding, usually every day, to create an oxygen deficiency environment. A top dressing of 90 kg ha–1 N was applied immediately after the waterlogging period ended.

Soil oxygen content was measured with gas samplers constructed of porous bronze cups attached to sampling taps. Gas samplers were buried in the soil between rows at a depth of 10 cm in every replication (Collaku and Harrison, 2002). Measurements on redox potential were taken weekly during the waterlogging period. Gas samples of 20 mL were withdrawn through the sampling taps with a 50-mL syringe. Oxygen concentration of the withdrawn samples was determined with an oxygen probe (DO-166, Lazar Research Laboratories, Los Angeles, CA), as described by Musgrave and Ding (1998).

Plant measurements were taken on the mid-row of each plot. Measurements for chlorophyll, height, and kernel number were taken on three randomly selected plants in each plot. Chlorophyll content was measured immediately after waterlogging. Three to four readings were taken for each plant, and the mean was used for the data analysis. A SPAD meter (Model 502, Minolta Corp.) was used to measure chlorophyll content (Collaku and Harrison, 2002). Measurements for height and kernel number were taken at harvest time. Tiller number was measured by counting the number of heads in a 20-cm-row length. Kernel weight was determined from a 100 seed sample for each plot. Grain yield was measured by hand harvesting and threshing the mid-row in each plot.

Statistical Analysis
Cases such as this in our study, where families from different populations have a degree of relationship, are very common in breeding programs. Traditional mating designs used to estimate genetic variance components are applicable only when parental components are unrelated. By using REML to estimate genetic variance components in a mixed model approach (Henderson, 1975; Harville, 1977; Bernardo, 1994), it is possible to account for the relationship among derived families. The mixed model, applied to estimate genetic variance components from 80 families studied in three environments (years), was:

Formula
where y is the vector of n observations for each family; X and β are the design matrix and the vector of trial effects (including environments and replications within environments), respectively; β is the vector of fixed effects b, where b = l x r, and 1 ≤ l ≤ 3 is the number of environments, and 1 ≤ r ≤ 3 is the number of replications within environments; Z1 and {alpha} are design matrix and vector of additive effects a (1 ≤ a ≤ 4), where a is number of populations; Z2 and {gamma} are design matrix and vector of dominance effects, d (1 ≤ d ≤ 4), where d is number of populations; Z3 and {delta} are design matrix and vector of genotype x environment (GE) interaction effects g as a result of cross combination of entries with environments (1 ≤ g ≤ 240); and {epsilon} is the vector of experimental error effects.

The following assumptions have been made for the random effects: the additive effects are normally distributed (N), with mean 0 and variance {sigma}2A, dominance effects are NFormula, GE interaction effects are independently identically distributed (iid) NFormula, and errors are iid NFormula. Random effects have the following variance-covariance matrix:

Formula

The additive matrix A has dimensions 80 x 80 with diagonal elements equal to 1 and off-diagonal elements equal to two times coancestry coefficient (2rxy) between 80 families in the study (Falconer and Mackay 1996; Kang, 1996). They have a value of 0.5 between full-sib families, i.e., between families 1 to 20, 21 to 40, 41–60 and 61–80 and smaller values as the relationship between families becomes weaker. Elements of the covariance matrix D represent double coancestry coefficients (u). Diagonal elements are 0.25 as double coancestry coefficients among full-sib families, while off-diagonal elements are 0 or greater than 0, depending on the relationship among different groups of families. Elements of matrix A and D were obtained by using PROC INBREED of SAS (SAS Institute, v.8, 2000, Appendix 1, p. 491–504; chapter 4, p. 135–169; chapter 18, p. 533–610), and then appended in the PROC MIXED (SAS Institute, v.8, 2000) to estimate additive and dominance variance based on the mixed model approach (Collaku, 2003). I is the identity matrix.

The vector of observations y is assumed to be multivariate normal with mean E(y) = Xβ, and variance-covariance VarFormula = Z1G1Z'1 + Z2G2Z'2 + Z3G3Z'3+ R (Searle et al., 1992). The estimates of {alpha}, β, {gamma}, and {delta} can be obtained by solving the following system of mixed model equations:

Formula

Formula

The REML method was used to estimate additive and dominance genetic variance components and non-genetic variance components.

Narrow-sense heritability (Hanson, 1963; Nyquist, 1991) was estimated on a plot basis, as:

Formula
where: Formula2P = Formula, r = 3 is the number of replications, and l = 3 is the number of years. Standard error of heritability was calculated according to Graybill and Wang (1979) and Knapp et al. (1987). Selection response was calculated as: R = ih2 Formula2P, for 5% and 10% selection intensity. Assuming a normal distribution and a truncated selection for the phenotypic values of traits studied, values of selection intensity were i(5%) = 2.063 and i(10%) = 1.755 (Falconer and Mackay, 1996).

Genetic correlation coefficients were calculated as:

Formula

Where CovXY is the additive covariance between two traits X and Y.

Correlated response were calculated as:

Formula

Where CRY/X is the expected grain yield selection response if the selection would be applied for trait X, hx and hy are the square root of heritability estimates of trait X and grain yield.

The vector of section indices was obtained by solving the equation:

Formula
where B is the unknown vector of phenotypic trait weights, or selection indices, bi, and P is the phenotypic variance-covariance matrix. The diagonal elements of P matrix represent phenotypic variances of traits included in the index, starting with grain yield, while the off-diagonal elements are the phenotypic covariances among all traits. G is the genetic variance-covariance matrix with additive variance of each trait as diagonal elements and additive covariances as off-diagonal elements. In the vector of economic weights e, grain yield has a value of 1, while all other traits have a value of 0.

The vector of expected selection response from selection index was calculated as: Ri = iGb/(b'Pb)1/2, where i is the selection intensity.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
Redox potential data ranged from 290 to 335 mV (Table 1). These data were very similar to redox potential data taken in flooded treatments of the waterlogging losses experiment conducted at the same time and in the same location (Collaku and Harrison, 2002). Grain yield losses from waterlogging in that experiment were 44%. Waterlogging stress also influenced other traits, such as kernel number and tiller number.


View this table:
[in this window]
[in a new window]

  		Table 1. Redox potential data from 5 wk of waterlogging of 80 families derived from four soft winter wheat populations at LAES Central Station, Louisiana State University at Baton Rouge. Data presented are ranges for each set of observations taken weekly on three replications.

 
Heritability Estimates
Presence of both additive and dominance variance for all the traits studied was anticipated, since populations studied were in the F2 generation. The absolute value of the additive variance component was greater than the dominance variance component for kernel weight, chlorophyll content, and height (Table 2). For some other traits, in particular for grain yield and kernel number, the additive variance component was considerably smaller than the dominance component, showing that these traits are under a stronger control of dominance effects. Selecting in early generations for grain yield or kernel number would not be as effective as selecting for traits such as kernel weight or chlorophyll content. The GE interaction was an important component of variance for all traits studied. This component had the highest relative magnitude for grain yield, where it contributed to 53% of the total variability (Table 2). As shown in earlier studies for wheat under stress conditions (Tillman and Harrison, 1996), the magnitude of GE interaction is important in case of severe conditions, and it can be a source of bias in heritability estimates, especially for grain yield.


View this table:
[in this window]
[in a new window]

  		Table 2. REML estimates of additive and dominance variance from 80 families derived without selection from four soft winter wheat populations under waterlogging conditions at LAES Central Station, Louisiana State University, at Baton Rouge.

 
In the presence of additive and dominance variance, narrow-sense heritability estimates are required to estimate the expected selection response. Among the traits studied, the highest heritability estimate of 0.54 was found for plant height (Table 3). However, heritability for this trait had a high standard error. The standard error for grain yield heritability was also high.


View this table:
[in this window]
[in a new window]

  		Table 3. Heritability estimates and expected selection response for grain yield, chlorophyll content, plant height, tiller number, kernels per spike, and kernel weight of 80 soft winter wheat families under waterlogging stress.

 
Heritability of kernel number (0.22) was the lowest among the traits studied. Previously, kernel number has been found to be highly heritable (Knot and Talukdar, 1971; Cantrell and Haro-Arias, 1986). The low heritability estimates reported in this study were probably attributed to waterlogging stress. In studies in wheat (Collaku, 1994), and in soybean (Kristin et al., 1997), it has been found that traits depressed the most by environmental stress had the lowest values of heritability. Grain yield and kernel number have been found as the most sensitive traits to waterlogging stress (Collaku and Harrison, 2002). Under environmental stress, the phenotypic variance of these traits generally increases more rapidly than the genotypic variance (Johnson and Frey, 1967), resulting in a reduced heritability.

Values of selection response were expressed as a percentage of the respective trait mean. A response 20% higher than the mean is expected from 5% selection intensity for chlorophyll content (Table 3). High selection response was observed for kernel weight (17%) and tiller number (18%), although data for this latter trait were available for the last year only. Grain yield components with high heritability can be used for a rapid screening of a large number of families in early generations of selection. Nass (1987) found early selection for large seed size in wheat to be more effective than three other selection methods, while Islam et al. (1985) and Gebre-Mariami et al. (1988) found that selection for kernel weight was more effective than direct selection for grain yield. Screening for waterlogging tolerance on the basis of traits such as kernel weight or chlorophyll content would be an effective method of selection in early generations.

Correlated Response
Strong genetic correlations were observed between grain yield and tiller number (r = 1.00), grain yield and seed weight (r = 0.56), and grain yield and chlorophyll (r = 0.40) (Table 4). Kernel weight had a significant genetic correlation with tiller number (r = 0.44) and chlorophyll (r = 0.31) as well as grain yield, showing that selecting for this trait would notably improve other important traits associated with grain yield. Genetic correlations among plant height and other traits were generally low. Although plant height has a high heritability, it seems to have little effect on grain yield or other traits.


View this table:
[in this window]
[in a new window]

  		Table 4. Genetic correlations among grain yield and other traits from 80 soft red winter wheat families under waterlogging stress.

 
A highly correlated response from grain yield is expected when selecting for kernel weight because of the high heritability of kernel weight and its strong genetic correlation with grain yield (Table 5). Selecting individually for traits such as kernel weight would improve grain yield. This is especially important in early generations when heritability of grain yield is low and the number of families to be screened is large.


View this table:
[in this window]
[in a new window]

  		Table 5. Efficiency of correlated selection response for grain yield estimated from 80 soft winter wheat families under waterlogging stress.

 
Selection Indices for Waterlogging Tolerance
Selection indices and their respective response for all possible combinations of grain yield with other traits were calculated. Some of the most important selection indices are presented in Table 6. A maximum improvement in grain yield of about 17% is expected for selection based on the Y-CHL-KWT-KN-T, Y-CHL-KWT-T, and Y-KWT-T indices. Considering grain yield losses from waterlogging at 44%, and assuming that the predicted gains will remain fairly constant for several cycles of selection, then at least three cycles of selection should produce a promising source population for waterlogging tolerance. Use of recurrent selection with the base population of 80 families would accelerate the selection process and maintain a better balance of genetic variation for grain yield. Since this process in wheat requires much manual work, it would be more convenient to use the Y-KWT-T index, selecting grain yield, kernel weight, and tiller number. Kernel weight and tiller number have significantly positive genetic correlations with chlorophyll content. The use of this selection index will contribute to a higher grain yield than direct selection for grain yield, while improving other important characteristics of the plant. In other studies, selection indices have been used to estimate the efficiency of improving grain yield, plant height, and protein content (Wells and Kofoid, 1986). In our study, plant height was not included in construction of selection indices, since it was not significantly correlated with grain yield or yield components.


View this table:
[in this window]
[in a new window]

  		Table 6. Predicted response in grain yield from applying different selection indices in a population of 80 soft winter wheat families grown under waterlogging stress.

 
The REML procedure, as used in this study, provides a robust method for estimating genetic variance components. In contrast with traditional methods, it has no limitations on the mating design and accounts for the relationships among families in a breeding population. Bromley et al. (2000) noted that ignoring these relationships has usually resulted in a reduction of estimated values of genetic variances. This method can be useful in many breeding programs where genetic material is made up of different families with a degree of relationship among them.

Heritability of grain yield under waterlogging conditions was low. Other traits with high heritability and strong correlation with grain yield, such as kernel weight, represent useful selection alternatives for introducing waterlogging tolerance in early generations. Use of selection indices will improve grain yield under waterlogging conditions more rapidly than selecting individually for each of waterlogging criteria. The estimated gain of selection indices needs to be confirmed. The relative efficiency of selection for various traits associated with grain yield depends on environmental conditions of specific breeding programs and on the genotypic variation.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 


Related articles in Crop Science:

THIS ISSUE IN CROP SCIENCE

Crop Science 2005 45: vi. [Full Text]  



This article has been cited by other articles:


Home page
 ANN BOT (LOND)Home page
T. L. Setter, I. Waters, S. K. Sharma, K. N. Singh, N. Kulshreshtha, N. P. S. Yaduvanshi, P. C. Ram, B. N. Singh, J. Rane, G. McDonald, et al.
Review of Wheat Improvement for Waterlogging Tolerance in Australia and India: The Importance of Anaerobiosis and Element Toxicities Associated with Different Soils
Ann. Bot., August 15, 2008; (2008) mcn137v1.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
M. X. Zhou, H. B. Li, and N. J. Mendham
Combining Ability of Waterlogging Tolerance in Barley
Crop Sci., February 6, 2007; 47(1): 278 - 284.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in Crop Science
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Collaku, A.
Right arrow Articles by Harrison, S. A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Collaku, A.
Right arrow Articles by Harrison, S. A.
Agricola
Right arrow Articles by Collaku, A.
Right arrow Articles by Harrison, S. A.
Related Collections
Right arrow Water Stress
Right arrow Wheat
Right arrow Crop Genetics
Right arrow Statistics

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome