Published in Crop Sci. 44:2127-2137 (2004).
© 2004 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
CROP ECOLOGY, MANAGEMENT & QUALITY
Use of a Water Stress Index to Identify Barley Genotypes Adapted to Rainfed and Irrigated Conditions
F. Rizzaa,*,
F. W. Badeckb,
L. Cattivellia,
O. Lidestric,
N. Di Fonzoc and
A. M. Stancaa
a Istituto Sperimentale per la Cerealicoltura, Via S. Protaso, 302-29017 Fiorenzuola d'Arda, Piacenza, Italy
b Potsdam Institute for Climate Impact Research (PIK) PF 60 12 03-14412 Potsdam, Germany
c Istituto Sperimentale per la Cerealicoltura, SS 16 Km 675-71100 Foggia, Italy
* Corresponding author (fulvia.rizza{at}libero.it)
 |
ABSTRACT
|
|---|
Future climate changes are expected to increase risks of drought, which already represent the most common stress factor for stable barley (Hordeum vulgare L.) production in Mediterranean areas. It is important, therefore, to evaluate if there are needs of specific adaptive measures in selecting genotypes for these more stressful environments. Our objective was to study diversity of yield performance under rainfed (R) and irrigated (I) conditions in 89 barley genotypes of different origin, growth habit, and year of release, representing a sample of cultivars grown in Europe. The experiment was conducted at Foggia (southern Italy) for 3 yr. For each trial, a water stress index (WSI) was calculated on the basis of the daily potential and actual evapotranspiration in the growing season, estimated by Thornthwaite's method. The WSI explained most of the variation in yield (R2 = 0.89**) among years and treatments. We examined, using the yield vs. WSI regression, the behavior of a given genotype across trials. The intercept and slope values were used as measures of yield potential and adaptability under drought, respectively. Several cultivars showing high yield potential and minimal genotype x environment (GE) interaction were identified. Notably, they were characterized by a high slope of the yield vs. WSI regression. Furthermore, within the range of water stress here examined, high yield potential played a preeminent role in the performance of these barley genotypes. This explains why, in this specific context, a selection based on minimum yield decrease under stress with respect to favorable conditions failed to identify the best genotypes.
Abbreviations: AWC, available water capacity • GE, genotype x environment • I, irrigated • R, rainfed • rel. YI, yield of the cultivar in I trials • rel. YR, yield of the cultivars in R trials • S, Fisher and Maurer's drought susceptibility index • WSI, water stress index
|
INTRODUCTION
|
|---|
THE MEDITERRANEAN REGION OF EUROPE is particularly sensitive to drought and potentially very vulnerable to future climate changes. Even if rainfall does not change, increased risks for drought will result from an increased atmospheric evaporative demand in a warmer future climate. As a consequence, soils will dry out faster and prolonged summer droughts might become more frequent (Bolle, 2003).
Breeding for drought resistance is therefore required for both mild and severe stress conditions. This implies a need for a better characterization of the biodiversity available for drought and a deeper comprehension of the physiological mechanisms, which are crucial to assure yield when drought occurs. However, high yield and drought adaptation are often based on different and, to some extent, conflicting mechanisms. Traits related to drought resistance, such as small plant size, reduced leaf area, and early maturity, lead to reduced total seasonal evapotranspiration. Prolonged stomatal closure allows plants to limit water loss but also reduces dry matter production (Karamanos and Papatheohari, 1999). These traits, however, are associated with a lower yield potential (Fischer and Wood, 1979). Furthermore, assimilate accumulation in the stems before anthesis is advantageous if drought occurs during the following phases, but it could reduce spike weight at anthesis (Slafer and Araus, 1998).
Thus, traits related to drought resistance and to high yield potential should be alternatively favored in cereal breeding programs, based on the ideotype for a target area and a specific type of stress. According to some authors, yield in low and high yielding environments can be considered as separate traits which are not necessarily maximized by identical sets of alleles (Falconer, 1990), consequently plant breeding strategies should be different when targeting stress and nonstress environments (Ceccarelli et al., 1991; Ceccarelli et al., 1998). Other authors claim that selection under favorable conditions is required to select genotypes with good performance under both stress and nonstress conditions (Cattivelli et al., 1994; Braun et al., 1997; Sayre et al., 1995). There is some agreement that a high yield potential is advantageous under moderate stress, while cultivars with low yield potential and high drought tolerance may be useful when stress is severe (<300 mm) or at yield level below 3 Mg ha–1 (Voltas et al., 1999; Panthuwan et al., 2002).
Field selection is also complicated by the high variability associated with multiple interactions contributing to drought resistance of crops, as the occurrence of drought events at different phases during the growing season or the spatial variability, which is amplified when water is limiting. This contributes to a large GE interaction that may explain the slow progress in developing new cultivars of cereals for drought conditions (Fukai et al., 1999). Most often no clear cause of the GE interaction has been identified because of lack of information about the environment (such as weather, soil) or the genotypes themselves (Voltas et al., 2002). Several indices have been proposed to describe the behavior of a given genotype under stress and nonstress conditions (Finlay and Wilkinson, 1963; Fischer and Maurer, 1978; Soika et al., 1981; Bidinger et al., 1987; Lin and Binn, 1988; Yadav and Bhatnagar, 2001). The need to identify a quantitative variable to characterize a specific environment led some authors to express yield with regard to a given physiological trait related to the environment, such as canopy temperature (Nilsen and Anderson, 1989) or water potential (Karamanos and Papatheohari, 1999). Motzo et al. (2001), investigating the factors responsible for GE interaction in multienvironment trials, proposed a seasonal water stress index based on soil–plant–atmosphere interaction, after the growth model developed by Amir and Sinclair (1991a)( 1991b).
In comparison with other small-grain cereals, barley has higher adaptability to water stress because of a more extensive root system and its early development that permits drought escape (Fischer and Maurer, 1978; Lopez-Castaneda and Richards, 1994).
In the present work, we evaluated the diversity for yield performance under rainfed conditions and with supplementary irrigation in a set of barley cultivars in a Mediterranean environment subjected to mild drought. In this environment, more severe drought events may occasionally occur, especially during grain filling, when a gradual rise in temperature is associated with a severe depletion of soil water resources. We assumed, therefore, that the ideotype for these environments should have minimal GE interaction, so that genotypes with both high yield potential and stable yield would be selected.
|
MATERIALS AND METHODS
|
|---|
Site, Treatments, and Agronomy
Field experiments were conducted at Foggia (southern Italy) for three consecutive growing seasons (1998-1999, 1999-2000, 2000-2001) under rainfed nonirrigated (R) and irrigated (I) conditions. The soil was a Vertisol, Typic Calcixererts, fine, mixed, thermic, composed of 450 g kg–1 clay, 330 g kg–1 silt, and 220 g kg–1 sand. Eighty-nine genotypes of barley that included cultivars released during the last 40 yr in several European countries were used (Table 1).
View this table:
[in this window]
[in a new window]
|
Table 1. List of the 89 genotypes grown during 3 yr in rainfed (R) and irrigated (I) trials. Genotypes are listed according to their ranking for the average grain yield of 3 yr under both conditions (Mean R, I). The mean grain yield (Mg ha–1) of 3 yr under R and I conditions are reported with the corresponding ranking order (in brackets). Slope, intercept and R2 values were calculated for each genotype after the grain yield vs. water stress index (WSI) regression.
|
|
The experimental design was a randomized complete block with three replications. Each experimental unit consisted of a 4-m2 plot. Seedling density was 300 seeds m–2. The sowing dates were 3, 13, and 21 Dec. 1998, 1999, and 2000, respectively. The previous crop was durum (Triticum durum Desf.) wheat. Fertilizer applications were made at presowing (36 kg ha–1 N and 92 kg ha–1 P2O5) and top dressing (52 kg ha–1 N) at Zadoks growth stages 2.2 and 3.1.
Water was distributed by drip irrigation as follow: in 1999 five water applications between April and May supplying a total of 175 mm; in 2000 four applications between March and May (135 mm); in 2001, water was supplied two times in March for a total of 80 mm. The decision to irrigate was based on a visual estimate of soil moisture. Plants were harvested on 15, 11, and 15 June in 1999, 2000, and 2001, respectively.
Daily maximum and minimum temperatures and rainfall were recorded at the experimental farm (Azienda Sperimentale "Podere 124") of the Experimental Agronomic Institute of Bari, placed within 300 m of the experimental fields.
Water availability for plants in the field throughout the barley life cycle was characterized by a water stress index, calculated on the basis of soil water balance. Soil water balance was calculated with a simple bucket model, the Thornthwaite (1948) approach for estimation of potential evapotranspiration (PET) and a function that reduces the actual evapotranspiration (AET) when the actual plant-available soil water content (WS) drops below 70% of the available water capacity (AWC). This cut-off point was chosen on the basis of frequent observations of onset of stomatal closure when soil water drops below a threshold value between 0.5 and 0.7 of the available water capacity (Jamieson et al., 1995, Turner et al., 1985, Gollan et al., 1985). For barley the onset of stomatal closure occurred at about 0.66 times the maximal recorded soil water contents (Borel et al., 1997).
 |
where WS is the actual plant available soil water content, i.e., soil water content minus water content at permanent wilting point. The actual WSI is defined as
 |
To establish water stress accumulated during the growing season, the integrated growing season WSI was calculated by summing daily WSI from 10 d after the sowing date (average date of emergence).
The soil water content at permanent wilting point and field capacity was 18 and 38 weight-percent of dry soil weight, respectively. Dry soil density was 1.2 g cm–3. From these measurements, plant-available volumetric soil moisture was estimated as 24% of field capacity. An independent estimate of the water content at field capacity was calculated from the soil composition and average values for water content (volume-percent) tabulated by Scheffer and Schachtschabel (1998), resulting in 26% volume. The available water capacity, AWC, was estimated from these estimates and rooting depth. An average estimated AWC of 192 and 208 mm resulted for a rooting depth of 80 cm with 24 and 26% of plant-available water, respectively. An uncertainty range of 144 to 260 mm was calculated with upper and lower bounds for field capacity and rooting depths, where lower and upper bounds for rooting depths were set to 60 and 100 cm (Lopez-Castaneda and Richards, 1994).
The following traits were determined for each plot: plant height, days to heading, grain yield, 1000 kernel weight, kernel number per square meter. Date of heading was recorded when about half of the culms showed emerging spikes. Plant height (excluding spike) was measured in all plots, at Zadoks growth stage 7.5. Two random samples of 100 kernels were weighed to estimate kernel weight. The kernel number per unit area (m2) was calculated by dividing yield by kernel weight.
Statistical Analyses
Combined analysis of variance (ANOVA) was calculated by the software MSTATC. Main effects were genotypes, treatments (R or I) and years. Genotype was considered a fixed factor; year and treatment (or environment) considered random. Where the F-test was significant, a least significance difference (LSD) was calculated to compare means.
Susceptibility Indices
The response of the genotypes was studied by considering the following aspects.
- The grain yield under R and I conditions, to characterize the genotypes for yield stability and yield potential. These traits were also expressed as relative yield (%) under R conditions (rel YR) and I conditions (rel YI) by standardizing, for each year and treatment, the grain yield of a genotype to the average yield of all genotypes.
- The yield decrease due to absence of supplementary irrigation, to identify the genotypes less affected by drought. For this purpose the Fischer and Maurer index (1978) was applied
where D (stress intensity) = 1 – (H
R/I), S = susceptibility index, YR = yield of the cultivar in R trials, YI = yield of the cultivar in I trials, and HR = mean yield of all cultivars in R trials. I = mean yield of all cultivars in I trials.
- The yield response as a function of the WSI, based on soil water balance calculation, for years and treatments.
|
RESULTS
|
|---|
Precipitation and Temperature
The total mean rainfall during the 3 yr of this study in Foggia was 440 mm yearly (Fig. 1a)
, 193 mm of which were received during the barley growing season. This was lower than the mean rainfall over the last 30 yr at Foggia of 507 mm. Rainfall was not evenly distributed over the various phases of plant development. In the 1998-1999 growing season, low rainfall from the second half of January until mid June affected most developmental phases with more intense water stress during grain filling, when high temperatures occurred concomitantly with low precipitation. In 1999-2000, the climatic conditions were generally favorable with relatively high precipitation levels from February to May. In the 2000-2001 season the precipitation was concentrated in December and January, causing problems at emergence. Precipitation was low in February and March and associated with relatively high temperatures, while high rainfall occurred in April. In all three seasons, high temperatures between 30 and 35°C were frequent during the last part of the growth cycle.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 1. a) Trend in temperature (minimum and maximum) and rainfall at Foggia during the three growing seasons (1998-2001) where the trials were conducted. The symbols I, II, III in the x-axis indicate for each month the first, second and third 10-d period, respectively; b) Available soil water content calculated daily for the rainfed (R) and irrigated (I) treatments during the whole period of the trials.
|
|
Drought Effects on Plant Development and Grain Yield
In Table 2, the general means for the analyzed traits are summarized. The impact of climatic conditions on barley development and productivity was different in each of the 3 yr. The differences among years and between treatments, as well as their interaction, were highly significant for all traits (data not shown). Heading occurred about 1 wk later during the first year than the two subsequent years, although sowing was earlier than in the other two years. In general, water stress affected plant earliness: heading occurred 1 or 2 d earlier under R conditions with respect to the I treatments every year. In both R and I trials, the mean plant height was highest in 1999 and lowest in 2001. In R plots, plant height was reduced by 12, 13, and 45% in the first, second and third year, respectively. In both R and I trials the highest grain yield was recorded in 2000 while in 2001 a large yield decrease was observed. The effect of drought was more relevant during the first and the third year, as shown by the reduction of grain yield (27 and 33%, respectively) compared to 2000 (18%) (Table 2). The kernel number per square meter was highly variable across years. The highest values were observed in the second year, and the lowest in the third year. In the R treatments the kernel number per square meter was reduced by 6, 10, and 16% in 1999, 2000, and 2001 respectively, compared with the I treatment. The highest mean value of kernel weight was found in I during the first year; similarly to yield, the kernel weight was less affected by drought in the second year, being reduced in R trials by 22, 9, and 21% in the first, second, and third year respectively.
View this table:
[in this window]
[in a new window]
|
Table 2. Mean values of 89 genotypes for grain yield and the morphophysiological traits studied under irrigated (I) and non irrigated (R) conditions in 3 yr of experiments. The ratio R/I is also reported.
|
|
Yield and Water Stress Index
Significant genotypic variation existed for grain yield (Table 3). The magnitude of variation attributable to the year and treatment factors, estimated as a percentage of the total sum of squares was 38 and 21%, respectively. Highly significant interactions of GxY, GxT and GxYxT were observed; consistent with means of Table 2, the sum of squares for the GxY interaction was about twice compared to the GxT interaction (6.9 and 3%, respectively).
View this table:
[in this window]
[in a new window]
|
Table 3. Combined analysis of variance (ANOVA) for grain yield of 89 genotypes across years and treatments (R, I).
|
|
A more precise description of the water availability throughout the barley life cycle was obtained by calculating the daily potential and actual transpiration. The calculated soil water content under I and R conditions over the whole period of the trials is illustrated in Fig. 1b. During the first year, low water availability occurred in R trials since March, leading to drought stress during the reproductive and grain filling phases. In the second year, plants experienced a low level of stress, even under rainfed conditions. The 2001 season was characterized by two periods of drought, at reproductive and at grain filling phases; since the supplementary irrigation in the I plots was possible only during the first drought period, even I plants were subjected to late season drought stress. This may explain the low grain yield in the third year. The differences in yield and associated traits between R and I in 2001 can be attributed to the drought stress experienced by R plants in preanthesis. Drought stress also may explain the strong reduction of the average plant height in R with respect to the I plants (Table 2).
Thus, the analysis of grain yield on the basis of the actual levels of water stress experienced by plants in all trials (Table 2) may be more appropriate than the comparison, for the same year, of the R treatment vs. the corresponding I treatment (Table 2). Grain yield was linearly and negatively related to the integrated growing season WSI calculated for each trial during the growing season (Fig. 2)
. The average grain yield varied between 2.9 and 6.4 Mg ha–1 for integrated WSI values varying between 6.1 and 64.6.This stress index explained a high proportion of the variation in yield (R2 = 0.89**; n = 6) among years and treatments. Sensitivity of results to uncertainty in the parameters used for description of soil properties was assessed by varying the AWC in the uncertainty range (as described in Materials and Methods) and then comparing the resulting R2 values for the regression of yield on WSI. R2 was always greater than 0.85 (P 
0.01) for any combination of initial soil water content and AWC.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 2. The relationship between average grain yield (Mg ha–1) and the integrated growing season Water Stress Index (WSI) calculated for each trial (3 yr, two treatments) on the basis of the available soil water during the growing season.
|
|
Genotypic Response
The ANOVA identified, by the LSD test, a group of five high yielding genotypes, on the basis of the average yield in both treatments and the three years: ‘Nure’, ‘Tea’, ‘Apex’, ‘Vertige’, and ‘Magda’ (Table 1). The analysis of the genotypic response (in R and I treatments) showed 31 genotypes in the R treatment and 11 genotypes in the I treatment with significantly highest yield (Table 1). Then, among these high yielding genotypes eight (Nure, Tea, Apex, Vertige, ‘Formula’, ‘Tidone’, ‘Federal’, ‘Barke’) ranked among the best in both R and I trials, showing high yield potential and stability; 23 genotypes showed high yield only in the R trials, therefore representing a group of cultivars resistant to mild drought but with limited yield potential; three others were superior in I fields but not in the absence of supplementary irrigation, showing high yield potential but lower yield stability (Table 1).
More generally, considering all the genotypes used in the experiment, high variability was observed in yield potential and yield stability (Table 1). The yield decrease in the absence of irrigation varied from 2.5% in ‘Gaiano’ to 38% for ‘Solen’. Only the low yielding ‘Dhalia’ performed better under R than I conditions; genotypes such as ‘Digersano’, ‘Onice’, and ‘Perga’, were slightly affected by drought, but their yield was very low in both R and I trials. The variability in the genotypic response is represented in Fig. 3a
where the relative yield under irrigation (rel. YI) is plotted against the relative yield in rainfed conditions (rel. YR). Four main types of genotypic response were observed: Class 1 (first quadrant) includes the genotypes with high and stable yield; cultivars of Class 2 showed good adaptability to water stress, but lower yield potential; Class 3 includes genotypes with constant low yield; and Class 4 groups cultivars with high yield potential and low adaptability to water stress. Class 1 was the largest with 41 genotypes including the eight identified by the ANOVA (Table 1) as superior for grain yield over 3 yr (Fig. 3a, open circles). The stress intensity, D, used to calculate, for each year, the susceptibility index (S) was 0.27, 0.18, and 0.33 in 1999, 2000, and 2001, respectively. The grain yield of each genotype expressed as a percentage of the mean under stressed conditions (rel. YR) was plotted against the susceptibility index (S) and 24 genotypes with rel. YR index higher than 100 and S index lower than 1 were identified as superior for stress resistance, after the approach of Gavuzzi et al. (1993) (Fig. 3b, Quadrant 1). The eight genotypes previously identified with the best performance for yield stability and yield potential were not in Quadrant 1 (Fig. 3b); however, because their S values were higher than 1 (equal to 1 for Vertige). For example the genotype Tea (S = 1.10) was not identified as one of the most resistant genotypes, although it had high yield under R and I conditions, unlike the lower yielding ‘Alpha’ (S = 0.77), Fjord (S = 0.45) or ‘Elan’ (S = 0.19)
View larger version (33K):
[in this window]
[in a new window]
|
Fig. 3. a) The relationship between ‘rel. YI’ and ‘rel. YR’ for the genotypes during 3 yr. Both indices were calculated for each year and the means of 3 yr are reported. In the graph, the four Quadrant, 1, 2, 3, 4 identifies four types of response in terms of yield potential and stability. b) The relationship between ‘rel. YR’ and the Fischer and Maurer's susceptibility index (S) calculated for the genotypes during 3 yr. Both indices were calculated for each year and the means of 3 yr are represented. The first Quadrant (1) indicates the genotypes identified as the most adapted to drought according to Gavuzzi et al. (1993). The 8 genotypes identified by the ANOVA as superior for yield performance under rainfed and irrigated conditions are indicated by open circles.
|
|
A detailed analysis of the annual means across R and I treatments, for the eight higher yielding genotypes, showed that Nure, Tea, and Vertige ranked among the best each year. Moreover, Nure showed higher yield in each trial, under both R and I conditions (Table 4). The results confirm the high yield potential and stability across trials for these genotypes with respect to the mean of all genotypes (Table 4).
View this table:
[in this window]
[in a new window]
|
Table 4. Grain yield (Mg ha–1) of the genotypes with high yield potential and yield stability, identified on the basis of 3 yr data under irrigated (I) and rainfed (R) conditions. The means of this group of genotypes are compared with the means of the 89 genotypes employed.
|
|
Genotypic Response and Water Stress Index
Instead of comparing the R vs. I trials, the behavior of a given genotype across trials may be summarized in terms of the relationship between integrated growing season WSI and grain yield, shown in Fig. 2. The analysis of the average grain yield of each genotype as a function of the corresponding WSI results in a linear regression where the slope represents a measure of adaptability to changes in water availability and the intercept a measure of the yield potential. Then the response of each genotype within a range of water availability can be simply described by these two parameters.
Among the 89 genotypes tested, the slope values ranked from –0.023 (Onice) to –0.074 (‘Aliseo’, Table 1). The intercept values varied from 4.1 (Onice) to 7.7 (Nure) and the R2 values of the linear regression Yield vs. WSI for each genotype varied from 0.44 (Onice) to 0.97 (Digersano). The regression was significant at P = 0.05 or 0.01 for 46 and 23 genotypes, respectively. For the other 20 genotypes low R2 values were generally associated with a lower dependence of yield on water availability.
In Fig. 4
, the regression yield–WSI is displayed for four genotypes representative of different responses: (i) Nure showed the highest intercept value and a slope value of –0.069, although this value was more negative with respect to the mean slope (–0.051) and Nure ranked among the best also at increasing WSI; (ii) Aliseo showed good yield potential but low stress adaptation (slope –0.074); (iii) Elan surpassed Aliseo at high WSI but had low yield potential and better adaptability to stress (slope –0.036); and (iv) Onice was a low yielding genotype but was only slightly affected by increasing stress level (slope –0.023). The eight genotypes identified as highly productive in terms of yield potential and stability showed intercept values from 6.8 to 7.7, slopes more negative than –0.050 and significant R2 values, indicating a significant dependence of yield on soil water availability. The intercept values calculated for the 89 genotypes according to the relation yield–WSI were negatively correlated with the corresponding slope shown in Table 1 (r = –0.81***, n = 89). Notably, under the environmental conditions considered here, the ability to produce well during water limitation as well as under favorable conditions was associated with a higher slope value of the relation yield–WSI.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 4. The relationship between the integrated growing season Water Stress Index (WSI) and grain yield for 4 genotypes representative of different responses to changes in water availability.
|
|
A distinctive aspect is the behavior of the local checks ‘Micuccio’ and ‘Diomede’, which, although selected in the experimental environment subjected to drought, were surpassed by genotypes with high yield potential selected elsewhere. Significant breeding progress however is evident when comparing the old genotype ‘Micuccio’ with the modern cultivar Diomede characterized by good adaptability and an improved yield potential.
Effect of the Year of Release, Ear Type, and Growth Habit
The comparison of genotypes released over a 40-yr period (Table 1) indicated a slow continuous improvement in average yields, in both R and I trials, with significant regression coefficients, (R2 = 0.16*** and 0.22***, respectively) (Fig. 5)
. Genetic improvement was 4.2 and 2.1 g m–2 yr–1 under I and R conditions, respectively (0.8 and 0.6% yr–1, respectively). Data of genetic gain calculated for each of the six treatments varied from 1.5 to 5.1 g m–2 yr–1 (for the R treatment of the 2000-2001 and the I treatment of the 1999-2000, respectively).
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 5. The relationship between year of release and grain yield of the genotypes employed in this study under rainfed (R) and irrigated (I) conditions.
|
|
Similar trends and identical R2 (0.21***) were found in a regression analysis of the intercept values calculated for each genotype on the basis of the yield-WSI relation as a function of the year of release. This is indicative of a genetic improvement of yield potential.
The genotypes studied were equally distributed between two and six row-types. Among the most productive genotypes, however, the two row types were largely predominant (Table 1 and Fig. 6) . Also, the genotypes with higher intercept values (Table 1) were mainly the two row-types. The intercept values ranged from 4.1 to 7.1 (mean 5.8) for the six row-type and from 4.4 to 7.7 (mean 6.4) for the two row-type. The effect of growth habit was less pronounced even though the best genotypes were mostly the spring cultivars (Fig. 6b).
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 6. Distribution frequency for yield performance of 89 genotypes in rainfed (R) and irrigated (I) trials, with regard to the a) row-type; b) growth habit.
|
|
|
DISCUSSION
|
|---|
The effect of drought on a single genotype was studied on the basis of the hypothesis that the ideotype for mild drought conditions should minimize the GE interaction (Cattivelli et al., 1994). For the environment analyzed here, with rainfall normally higher than 300 mm but occasionally subjected to severe drought, the most suitable genotypes should maintain high yield under both favorable and stress conditions. When the data of 89 barley genotypes were evaluated considering the average grain yield (mean value between irrigated and rainfed) of 3 yr, five cultivars were selected. Moreover, the analysis of yield in rainfed and irrigated trials, identified eight genotypes with high yield potential and minimal GE interaction, ranking among the best in both treatments. Although reduced in the absence of supplementary irrigation, their yield was superior under all conditions tested. Indeed, within the range of water availability that occurred in our study, having a high yield potential played a substantial role in the overall performance of the genotypes. The correlation between R yield and I yield calculated for the means of 89 genotypes was highly significant (r = 0.73***). This explains why in this context a selection based on the minimum yield decrease under stress with respect to favorable conditions, rather than to the absolute performance of the genotypes across environments, has failed to identify the best genotypes (Fig. 3a and 3b). The "drought susceptibility index" (S) proposed by Fischer and Maurer (1978) calculates for a given genotype the yield under drought as a function of the yield without drought, but the existence of a negative correlation between this index and yield under high moisture conditions was emphasized by the same authors (Fischer and Wood, 1979). The S index is still widely used by breeders and often adapted for specific needs (Yadav and Bhatnagar, 2001; Gorny, 2001). To reduce the influence of yield reduction from stress to nonstress conditions, Yadav and Bhatnagar (2001) suggested to use S in combination with yield under stress. These two parameters were employed previously by Gavuzzi et al. (1993) to identify genotypes with superior drought adaptation in trials conducted in several locations of southern Italy. This approach penalized, as in our case, genotypes with high yield potential.
In Mediterranean environments, water regime and temperature explain a major proportion of the yield variation of cereals (Blum and Pnuel, 1990; Voltas et al., 2002; Araus et al., 2003). Taking into account that drought is also a function of the soil water holding capacity (Voltas et al., 2002) and of the specific water consumption in different phases of the crop development, we propose a simple, more integrated, approach based on atmospheric and soil data, estimating for each trial the actual soil water availability at each day during the crop cycle. The calculated WSI and the mean grain yield across trials were highly correlated (R2 = 0.89**). Araus et al. (2003) found that, among the environmental factors studied in several locations in Spain, the water input, expressed as the sum of rainfall and irrigation during the growing period, explained a large part of the yield variability. In our experiment, the regression of yield vs. water input was not significant. During the third growing season, high rainfall at emergence (Fig. 1a) had a smaller effect on yield than if it had occurred at other phenological phases. Thus, it contributed to an increase in water input but not to a significant regression. The effective stress conditions during the crop cycle were assessed in wheat by Karamanos and Papatheohari (1999). Instead of climatic and soil data, they analyzed the physiological response of the plant by integrated measurements of water potential (WPI, Water Potential Index). Through a combined study of both intercept and slope in the yield vs. WPI regression, they identified four hypothetical different response types in terms of yield potential and adaptability to drought. We applied a similar approach, based on the yield vs. WSI regression, to summarize the behavior of each cultivar analyzed in our experiments, expressing its yield potential and adaptability to water stress by means of the intercept and the slope of the regression, respectively.
The evaluation of a genotype using an environmental index that allows for the comparison of different trials subjected to a range of water stresses appears more advantageous than the simple comparison of yield in rainfed vs. irrigated trials. This latter approach may lead to a wrong interpretation of the yield data because, as occurred in our trials, the yields in rainfed and irrigated treatments are not always contrasting, when several years are included and variation between years may affect yield even more than the irrigation treatment (Table 3). Here, we propose to use the integrated growing season WSI as a parameter to quantify the stress level experienced by a crop in different trials and the yield vs. WSI regression as a tool for simultaneous comparison of the performance of the genotypes across water-stressed and nonstressed environments. The question arises if the best performing genotypes identified here by the ANOVA and the yield vs. WSI regression would maintain their superiority under more severe drought conditions. Work is in progress to test the genotypes in several environments and establish if the simple relation described here is conserved.
Ideally, a high performing genotype should have both a high intercept and a low slope. Notably, the intercept values calculated according to the yield vs. WSI regression, were negatively correlated with the slope values (r = –0.81***). The difficulty in finding genotypes that combine high yield potential and low sensitivity to drought may be due to the existence of distinct mechanisms responsible for these traits. As described by Fischer and Wood (1979) the strategies of resource allocation to assure high yield potential (i.e., high harvest index or number of grain per square meter) minimize investment in organ, tissues, or tissue reserve, which would buffer yield and yield components against the effects of water stress. Under the conditions of our trials, however, the higher sensitivity to decreasing water availability in high-yielding genotypes was associated with the absence of a significant GE interaction. This suggests that in high-yielding genotypes, physiological mechanisms operate in a way that confers the ability to perform well under water limitation. In this study, this ability was associated with a higher slope of the yield vs. WSI regression. Genotypes with high intercept and more negative slope may therefore be regarded as highly responsive to water availability, possibly because of their higher WUE or a greater ability to take up water (Araus et al., 2003).
The mean genetic gain measured under I and R (4.2 and 2.1 g m–2 yr–1) observed in this study was consistent with the reports of other authors. The influence of genetic improvement on the barley yield in different countries was reviewed by Abelado et al. (2002), who reported genetic gain values varying from 1.6 to 7.4 g m–2 yr–1. It is expected that in poorer environments there is lower chance for the contribution of breeding, as well as of management, to increase yield. (Slafer et al., 1994)
Most of the genotypes with superior yield capacity were spring two row-type varieties released in the two last decades. The genotypes were released in different European countries located at varying latitudes (Sweden, France, Netherlands, Italy). Notably, genotypes selected in environments very different with respect to the site analyzed here, have shown wide adaptability, performing better than the local checks included in the experiment. This indicates that selection under favorable environments allows for the identification of genotypes well adapted to moderate Mediterranean water stress levels. The alternate use of optimum and stress conditions during or after selection (Calhoun et al., 1994), however, may be even better for identifying genotypes with high yield potential with low or high yield stability under more stressed conditions.
The genotypes identified here are endowed with genes for a wide range of adaptability under favorable and stress environments. These genotypes can be used to understand better which metabolic processes and morphophysiological traits are crucial to assure high yield performance under different environments.
|
ACKNOWLEDGMENTS
|
|---|
The authors wish to thank Renzo Alberici and Antonio Gallo for their excellent technical assistance and Dr. José Louis Araus for critical reading of the manuscript and useful suggestions. We also thank Dr. Giovanni Delogu for providing advice on statistical analyses.
|
NOTES
|
|---|
This research was funded by the MiPaF project "Risorse Genetiche Vegetali", the EU GENRES project CT-98-104, the special project MiToS of the Italian Ministry of Agriculture and the EU MABDE INCO-MED ICFP502A3PR03 project.
Received for publication June 30, 2003.
|
REFERENCES
|
|---|
- Abelado, L.G., F. Calderini, and G.A. Slafer. 2002. Physiological changes associated with genetic improvement of grain yield in barley. p. 361–385. In G. Slafer et al. (ed.) Barley science. Recent advances from molecular biology to agronomy of yield and quality. Food Product Press, Binghamton, NY.
- Amir, J., and T.R. Sinclair. 1991a. A model of the temperature and solar radiation effects on spring wheat growth and yield. Field Crops Res. 28:47–58.
- Amir, J., and T.R. Sinclair. 1991b. A model of water limitation on spring wheat growth and yield. Field Crops Res. 28:59–69.
- Araus, J.L., D. Villegas, N. Aparicio, L.F. Garcìa del Moral, S. El Hani, Y. Rharrabti, J.P. Ferrio, and C. Royo. 2003. Environmental factors determining carbon isotope discrimination and yield in durum wheat under Mediterranean conditions. Crop Sci. 43:170–180.[Abstract/Free Full Text]
- Bidinger, F.R., V. Mahalakshmi, and G.D.P. Rao. 1987. Assessment of drought resistance in pearl millet [Pennisetum americanum (L) Leeke]. II. Estimation of genotype response to stress. Aust. J. Agric. Sci. 38:49–59.
- Blum, A., and Y. Pnuel. 1990. Physiological attributes associated with drought resistance of wheat cultivars in a Mediterranean environment. Aust. J. Agric. Res. 41:799–810.
- Bolle, H.-J. 2003. Climate, climate variability, and impacts in the Mediterranean Area: An overview. p. 5–86. In H.-J. Bolle (ed.) Mediterranean climate: Variability and trends. Springer, Berlin.
- Borel, C., T. Simonneau, D. This, and F. Tardieu. 1997. Stomatal conductance and ABA concentration in the xylem sap of barley lines of contrasting genetic origins. Aust. J. Plant Physiol. 24:607–615.
- Braun, H.J., S. Rajaram, and M. van Ginkel. 1997. CIMMYT's approach to breeding for wide adaptation. p. 197–205. In Adaptation in plant breeding. PMA Tigerstedt, Kluwer Academic Publisher, Dordrecht, the Netherlands.
- Calhoun, D.S., G. Gebeyehu, A. Miranda, S. Rajaram, and M. van Ginkel. 1994. Choosing evaluation environments to increase wheat grain yield under drought conditions. Crop Sci. 34:673–678.[Abstract/Free Full Text]
- Cattivelli, L., G. Delogu, V. Terzi, and A.M. Stanca. 1994. Progress in barley breeding. p. 95–181. In A. Slafer (ed.) Genetic improvement of field crops. Marcel Dekker, Inc., New York.
- Ceccarelli, S., E. Acevedo, and S. Grando. 1991. Breeding for yield stability in unpredictable environments: Single traits interaction between traits architecture of genotypes. Euphytica 56:159–185.
- Ceccarelli, S., S. Grando, and A. Impiglia. 1998. Choice of selection strategy in breeding barley for stress environments. Euphytica 103:297–318.
- Falconer, D. S. 1990. Selection in different environments, effects on environmental sensitivity (reaction norm) and on mean performance. Genet. Res. (Cambridge) 56:57–70.
- Finlay, K.W., and G.N. Wilkinson. 1963. The analysis of adaptation in a plant breeding programme. Aust. J. Agric. Res. 14:742–754.[Web of Science]
- Fischer, R.A., and R. Maurer. 1978. Drought resistance in spring wheat cultivars. I. Grain yield response. Aust. J. Agric. Res. 29:897–912.[Web of Science]
- Fischer, R.A., and J.T. Wood. 1979. Drought resistance in spring wheat cultivars. III. Yield association with morpho-physiological traits. Aust. J. Agric. Res. 30:1001–1020.
- Fukai, S., G. Pantuwan, B. Jongdee, and M. Cooper. 1999. Screening for drought resistance in rainfed lowland rice. Field Crops Res. 64:61–74.
- Gavuzzi, P., G. Delogu, G. Boggini, N. Di Fonzo, and B. Borghi. 1993. Identification of bread wheat, durum wheat and barley cultivars adapted to dry areas of southern Italy. Euphytica 68:131–145.
- Gollan, T., N.C. Turner, and E.D. Schulze. 1985. The response of stomata and leaf gas exchange to vapour pressure deficits and soil water content. III. In the sclerophyllous woody species Nerium oleander. Oecologia 65:356–362.
- Gorny, A.J. 2001. Variation in utilization efficiency and tolerance to reduced water and nitrogen supply among wild and cultivated barleys. Euphytica 117:59–66.
- Jamieson, P.D., G.S. Francis, D.R. Wilson, and R.J. Martin. 1995. Effects of water deficits on evapotranspiration from barley. Agric. For. Meteorol. 76:41–58.
- Karamanos, A.J., and A.Y. Papatheohari. 1999. Assessment of drought resistance of crop genotypes by means of the water potential index. Crop Sci. 39:1792–1797.[Abstract/Free Full Text]
- Lin, C.S., and M.R. Binn. 1988. A superiority measure of cultivar performance for cultivar x location data. Can. J. Plant Sci. 68:193–198.
- Lopez-Castaneda, C., and R.A. Richards. 1994. Variation in temperate cereals in rainfed environments. I. Grain yield, biomass and agronomic characteristics. Field Crops Res. 37:51–62.
- Motzo, R., F. Giunta, and M. Deidda. 2001. Factors affecting the genotype x environment interaction in spring triticale grown in a Mediterranean environment. Euphytica 121:317–324.
- Nilsen, D.C., and R.L. Anderson. 1989. Infrared thermometry to measure single leaf temperatures for quantification of water stress in sunflower. Agron. J. 81:840–842.[Abstract/Free Full Text]
- Panthuwan, G., S. Fukai, M. Cooper, S. Rajatasereekul, and J.C. O'Toole. 2002. Yield response of rice (Oryza sativa L.) genotypes to different types of drought under rainfed lowlands. Part 1. Grain yield and yield components. Field Crop Res. 73:153–168.
- Sayre, K.D., E. Acevedo, and R.B. Austin. 1995. Carbon isotopic discrimination and grain yield for three bread wheat germplasm groups grown at different levels of water stress. Field Crops Res. 41:45–54.
- Scheffer, F., and P. Schachtschabel. 1998. Lehrbuch der Bodenkunde, 14 ed. Enke, Stuttgart, Germany.
- Slafer, G. A., and J. L. Araus. 1998. Improving wheat responses to abiotic stresses. p. 201–213. In A.E. Slinkard (ed.) Proceedings of the 9th International Wheat Genetics Symposium, Vol. 1, Saskatoon, Saskatchewan, Canada. 2–7 Aug. 1998. University Extension Press, University of Saskatchewan, Saskatoon.
- Slafer, G.A., E.H. Satorre, and H. Andrade. 1994. Increases in grain yield in bread wheat from breeding and associated physiological changes. In A. Slafer (ed.) Genetic improvement of field crops. Marcel Dekker, Inc., New York.
- Soika, R.E., L.H. Stolzy, and R.A. Fischer. 1981. Seasonal drought response of selected wheat cultivars. Agron. J. 73:838–845.[Abstract/Free Full Text]
- Thornthwaite, C.W. 1948. An approach toward a rational classification of climate. Geogr. Rev. 38:55–94.
- Turner, N.C., E.D. Schulze, and T. Gollan. 1985. The response of stomata and leaf gas exchange to vapour pressure deficits and soil water content. II. In the mesophyllic herbaceous species Helianthus annuus. Oecologia 65:348–355.
- Voltas, J., I. Romagosa, A. Lafarga, A.P. Armesto, A. Sombrero, and J.L. Araus. 1999. Genotype by environment interaction for grain yield and carbon isotope discrimination of barley in Mediterranean Spain. Aust. J. Agric. Res. 50:1263–1271.
- Voltas, J., F. van Eeuwijk, E. Igartua, L.F. Garcia del Moral, J.L. Molina Cano, and I. Romagosa. 2002. Genotype by environment interaction and adaptation in barley breeding: Basic concepts and methods of analysis. p. 205–242. In G. Slafer et al. (ed.) Barley science. Recent advances from molecular biology to agronomy of yield and quality. Food Product Press, Binghamton, NY.
- Yadav, O.P., and S.K. Bhatnagar. 2001. Evaluation of indices for identification of pearl millet cultivars adapted to stress and non-stress conditions. Field Crops Res. 70:201–208.
Related articles in Crop Science:
- THIS ISSUE IN CROP SCIENCE
Crop Science 2004 44: 1889-1892.
[Full Text]
This article has been cited by other articles:
|
|
|
|
 
P. Langridge, N. Paltridge, and G. Fincher
Functional genomics of abiotic stress tolerance in cereals
Brief Funct Genomic Proteomic,
February 1, 2006;
4(4):
343 - 354.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
