Chilling tolerance of Central European maize lines and their factorial crosses.

BACKGROUND AND AIMS
Chilling-stress tolerance is a prerequisite for maize production under cool climatic conditions. The main goal of this study was to evaluate the Central European dent and flint pools for chilling tolerance during heterotrophic and early autotrophic growth in field trials and growth chamber experiments.


METHODS
Five European flint and five dent inbreds and their 25 factorial crosses were evaluated in six natural environments, where chilling occurred, for chlorophyll concentration and plant height at the three-leaf stage, and plant height and fresh weight at the seven-leaf stage. In growth chambers, leaf 3 growth was analysed under cold and control conditions.


KEY RESULTS
Comparing the field and growth chamber data, the strongest association was found between leaf elongation rate during cold nights and plant height at the three-leaf stage, with a weaker association with the seven-leaf stage. In the field, moderate correlations were observed between plant height at the three-leaf stage, and plant height and fresh weight at the seven-leaf stage, respectively. Furthermore, mid-parent and hybrid performance were only moderately correlated.


CONCLUSIONS
The results suggest that heterotrophic and early autotrophic growth stages are controlled by different genetic factors or that maternal effects play a role. In addition, the findings showed that mid-parent performance is a poor predictor of hybrid performance. Consequently, test cross performance should be the target in quantitiative trait locus (QTL) mapping studies with the final goal of establishing marker-assisted breeding programmes for chilling-tolerant hybrids.


INTRODUCTION
Maize (Zea mays) is considered a chilling-sensitive species with a relatively high temperature optimum for germination, development and dry matter accumulation (Miedema, 1982). Under climates characterized by cool and humid springs, such as in Central Europe, adaptation has been partially successful due to late planting and breeding for early-maturing maize hybrids. These strategies are useful to minimize the risk of field losses due to chilling stress (Stamp, 1986). Improvement of chilling tolerance would support earlier spring planting and, consequently, lead to higher yielding maize hybrids (Lee et al., 2002). Furthermore, earlier soil coverage would help to reduce erosive processes.
At approximately the three-leaf stage of maize, seed reserves are exhausted and the seedling has to rely on its photosynthetic activity for carbon gain (Cooper and MacDonald, 1970). Heterotrophic growth (i.e. germination and growth relying partly on seed reserves) and autotrophic growth (i.e. photosynthesis-based growth after the exhaustion of seed reserves) appear to require different minimum limiting temperatures, which may be attributable to different genetic control (Brandolini et al., 2000). In agreement with this hypothesis, Hodges et al. (1997) and Revilla et al. (2000) reported low associations between chilling tolerance at heterotrophic and autotrophic growth stages. In contrast, Janowiak and Markowski (1987) observed a high correlation between the two stages, suggesting a similar genetic control. Hence, further studies comparing heterotrophic and autotrophic plant growth for chilling tolerance in maize are needed.
The correlation between hybrid and mid-parent performance is important for designing optimum breeding strategies and deciding whether selection of superior hybrids could be based on predictions obtained from parental performance. In addition, this correlation is crucial for answering the question of whether quantitative trait locus (QTL) mapping studies for chilling tolerance should concentrate on inbred lines or hybrids. Maryam and Jones (1983) observed a high association between mid-parent and hybrid performance for chilling tolerance in maize, suggesting that prediction of hybrid performance is feasible based on mid-parent performance. In contrast, Hodges et al. (1997) reported that reliable prediction of hybrid maize chilling tolerance was not possible from information about the parental inbreds. Hence, further research on the association between hybrid and mid-parent performance for chilling tolerance in maize is required.
Field trials depend largely on the combined effects of multiple environmental parameters changing simultaneously throughout the duration of the experiment. In contrast, plants can be grown in the laboratory under wellcontrolled conditions in growth chambers. This allows the evaluation of the effects of environmental factors on performance one at a time, which reduces the complexity and enhances the sensitivity of the analysis. Moreover, such experiments do not depend on outside conditions * For correspondence. E-mail: melchinger@uni-hohenheim.de and are therefore reproducible over time, and can be performed year-round, which will result in a reduction in the screening time of potentially valuable germplasm. The higher reproducibility and time reduction of the screening process would be of great advantage for future breeding programmes. However, growth chamber trials should reflect the aspects of stress situation occurring in field trials in order to accomplish this advantage.
The main goal of this study was to evaluate the Central European dent and flint pools for chilling tolerance during heterotrophic and early autotrophic growth in field trials and growth chambers experiments. In particular, we (a) examined the associations between mid-parent and hybrid performance for four measured traits; (b) analysed correlations between chilling tolerance-related traits determined at different developmental stages; (c) investigated the effects of low night temperature on seedling growth in growth chambers; and (d ) assessed to what extent growth chamber data predict maize performance in the field.

Plant materials
Five European dent and five European flint inbred lines of Zea mays L. were used, as well as their 25 factorial crosses in the F 1 generation. Seeds were produced using the flint lines as pollinators and dent lines as seed parents. The lines were chosen as a representative sample of modern public elite inbreds developed during the 1990s at the University of Hohenheim. The flint lines were F039, F047, F048, F052 and L024. The dent lines were D23, P048, P087, S046 and S070.

Field trials
The ten parental inbred lines and their 25 factorial crosses were evaluated in field trials over 2 years (2005 at 2006) at three locations in South Germany [Hohenheim, 400 m above sea level (asl), silty loam soil texture; Ihingerhof, 500 m asl, loam soil texture; and Oberer Lindenhof, 700 m asl, loam soil texture]. The experiment was laid out as a randomized complete block design with two replicates. Each plot consisted of two rows, and each row was 5 m length with a distance of 0 . 75 m between rows. Plant density was 88 000 plants ha 21 . To analyse the chilling tolerance of maize genotypes, the following four traits were measured: (1) chlorophyll concentration at the fully expanded three-leaf stage for 20 plants per plot (measurements were taken at leaf 3 at three positions and values were averaged); (2) plant height at the three-leaf stage; (3) plant height at the seven-leaf stage immediately before harvest; and (4) fresh weight per plant at the sevenleaf stage. Chlorophyll concentration was estimated with a handheld portable SPAD-502 chlorophyll meter (Minolta Corporation, Ramsey, NJ, USA). This instrument nondestructively measures the relative amount of chlorophyll in plant leaves. High values observed for the four traits indicate chilling tolerance of the respective lines or hybrids. Soil temperatures were recorded throughout the growing period from planting till harvest using automatic data loggers installed at the three locations. Temperature data for Hohenheim in the year 2006 are incomplete, because of technical problems with the temperature sensor.

Growth chamber experiments
The maize lines were germinated in peat pellets (Jiffy International As, Norway) at 25 8C and 70 % humidity, with a 16 h photoperiod and light intensity of 200 mmol m 22 s 21 PAR. Thereafter, the seedlings were transferred to pots with a height of 20 cm and a diameter of 4 . 6 cm filled with soil (N80, Structural, Kaprijke, Belgium) and placed in a growth chamber (type vb1014, Vötsch industrietechnik, Balingen, Germany) at 70 % relative humidity, 400 mmol m 22 s 21 PAR at plant level provided by a combination of fluorescent tubes (Osram-77 and Osram-31-830, Osram, Munich, Germany). The diurnal cycle was 16/8 h (day/night) with a gradual decrease and increase of radiation intensity over 0 . 5 h. Temperature was kept at 25 8C during the photoperiod and decreased to 18 8C (control) or to 4 8C (treatment) during the last 6 h of the night.
To calculate leaf elongation rates (LERs) during the photoperiod and the night, the length of leaf 3 (ten plants per treatment) was measured at the beginning and end of the photoperiod from leaf emergence to maturity, using the soil level as a reference point.

Statistical analyses
Analyses of variance were computed for the hybrids and inbreds across the environments using PROC MIXED in SAS (Version 9 . 1, SAS Institute, 2004). Environments were treated as fixed effects, and genotypes and genotype Â environment interactions as random effects. A Wald test was used to test whether variances were significantly greater than zero. Heritability (h 2 ) on an entry mean basis was calculated for all traits, according to Hallauer and Miranda (1981). Best linear unbiased predictions (BLUPs) were estimated for all four traits. Differences between the mean performance of the dent and flint lines were tested with a t-test. Pearson's correlation coefficients (r) between the BLUP values of the traits measured in the field trials and the growth chambers were calculated for inbreds and hybrids. Significance tests of r were performed using tabulated values based on Fisher's (1921) z transformation.

Field trials
Mean daily soil temperatures from sowing to harvest ranged from 8 . Temperatures at Oberer Lindenhof, which is the location with the highest elevation, were consistently lower than at Ihingerhof and Hohenheim (on average 3 . 7 8C). Nevertheless, the temperature trends were similar across the three locations.
For all traits except chlorophyll concentration, mean performance of the hybrids was higher than for inbred lines (Table 1). Genotypic variances (s 2 G ) of the hybrids and inbreds were significant (P , 0 . 05) for the measured traits except for plant height at the three-leaf stage for the inbred lines. The magnitude of variances associated with genotype Â environment interactions (s 2 GÂE ) was similar to estimates of s 2 G for all traits. A more detailed analysis of the performance of the ten inbred lines and 25 hybrids within location Â year combinations revealed a complex clustering of environments (Supplementary Information 2, available online). This complex pattern cannot be explained by single factors such as temperature, rainfall or soil texture alone, but rather points to an interaction of various external factors. Heritability (h 2 ) estimates ranged from 0 . 65 to 0 . 91 for the inbred lines and from 0 . 72 to 0 . 83 for the hybrids.
For inbreds and hybrids, no significant phenotypic correlations were observed between chlorophyll concentration and the other three traits (Table 2). Plant height at the seven-leaf stage and fresh weight at the seven-leaf stage were tightly correlated (r ¼ 0 . 89) for both inbreds and hybrids. The phenotypic correlations between (a) plant height at the three-leaf stage and (b) plant height and fresh weight at the seven-leaf stage were higher for hybrids than for inbreds. Mid-parent performance was positively correlated with hybrid performance for all traits, with the highest correlation observed for fresh weight (r ¼ 0 . 71) and the lowest for plant height at the three-leaf stage (r ¼ 0 . 34).
For plant height and fresh weight, the mean performance of the flint lines was significantly (P , 0 . 01) higher than that of the dent lines (Table 3). For chlorophyll concentration, no significant difference was found between the mean of the flint and dent lines. D23 was the best performing dent inbred line for all traits. F047, F048 and F052 were the best performing flint lines. Hybrids produced from flint parents F039 and F052 and dent parents D23 and S046 showed higher plant height and fresh weight compared with other hybrids.

Phenotypic evaluation in growth chambers
In the growth chamber experiments, cold was only applied during the night, while daytime temperatures were the same for control and cold treatments (Rymen et al., 2007). For all inbred lines and hybrids, this treatment resulted in a reduced overall leaf growth (Fig. 2). Similar to the field data, the hybrids performed better than the inbred lines under both conditions. Moreover, both the mean reduction and the range of variation in mature length of leaf 3 were smaller for the hybrids compared with those of the inbred lines.
Correlation analyses of field and growth chamber data revealed that mature leaf length determined in the growth chamber trials was significantly associated with plant height of the inbred lines evaluated in the field at the threeleaf stage (Table 4). Interestingly, for performance of the hybrids, higher correlations were observed between field data and the difference of cold vs. control treatment than between field data and cold treatment. Analyses of correlations between growth chamber and single environmental   The measurements of day and night LER in combination with the cold treatment during the night allowed the growth dynamics during daytime and at night to be investigated separately, and for the general temperature effect and the induced cold stress effect to be estimated (cf. Rymen et al., 2007). This analysis revealed for all genotypes a typical growth curve with an initial steady-state LER, followed by a decline in LER (Rymen et al., 2007). The steady-state growth contributed most to the overall growth of the leaf and was used for the subsequent calculations. LER was affected by the cold during daytime and at night. The largest reduction occurred at night when the cold was applied, with an average LER reduction of 256% (range 238 % to 272 %). During the daytime, LER was also reduced for most genotypes, although some compensated their reduced growth at night by a faster growth during the day (e.g. the hybrids S070 Â F039 and D23 Â F047 had an increased LER during the day of 24 and 22 %, respectively).
Significant associations were found between the LER and the reduction in LER at night when the cold was applied and plant height measured in the field experiments (Table 4). In contrast, no significant associations were observed between the field and LER data during the daytime, indicating the general temperature response rather than the induced stress effect as the main contributor to the performance of the plants under field conditions.

Correlation between mid-parent and hybrid performance
The range of correlations between mid-parent and hybrid performance observed in the present study was similar to estimates obtained for various complex agronomic traits (for a review, see Hallauer and Miranda, 1981). The observed differences between mid-parent and hybrid performance can be fully explained by a simple model with only additive and dominance genetic effects (Smith, 1986). In addition, linkage and/or epistasis can cause deviations of mid-parent from hybrid performance. In summary, the observed magnitude of the correlation for the four investigated traits suggests that mid-parent performance is a poor predictor of hybrid performance. In addition, test cross performance should be the target in QTL mapping studies with the final goal of establishing marker-assisted breeding programmes for chilling-tolerant hybrids.
The involvement of the photosynthethic apparatus in chilling tolerance Marocco et al. (2005) proposed to group physiological response to chilling stress into effects due to: (a) mild chilling stress (12 -17 8C) in the light, with reduced photosynthesis and growth as well as the induction of photoprotective mechanisms in response to excess light energy captured by chlorophyll; (b) strong chilling stress (2 -10 8C) in the light with cold-induced water stress, because the rate of transpiration exceeds the rate of water uptake by roots due to inhibition of root hydraulic conductivity; and (c) chilling stress in the dark, not associated with oxidative stress in the chloroplast or with water stress, but rather with changes in gene expression (cf. Rymen et al., 2007). In the growth chamber experiments, the focus was on the third class, with only application of low temperature during the night period. The significant association between the LER measured in the growth chambers and the field data indicated that the effects of chilling stress occurring  under field conditions are similar to the third class of chilling stress suggested by Marocco et al. (2005). Moreover, it suggests that the observed growth reductions can only be partially explained by photoinhibition due to a combination of high light and cold during the day. The findings are in accordance with previous studies of Fracheboud et al. (2004) and Jompuk et al. (2005) reporting that the photosynthetic apparatus did not noticeably affect biomass accumulation, but are in contrast to the results of Stirling et al. (1993) and Andrews et al. (1995), who observed a significant effect of photoinhibition on CO 2 assimilation. The conclusion that growth reduction in the field can only partially be explained by photoinhibition is consistent with the low correlation observed between chlorophyll concentration and growth in the field experiment (Table 2). Nevertheless, a decrease in chlorophyll concentration is only one possible factor for reduced photosynthesis. Further analyses based on the ratio of the level of variable to maximal fluorescence of dark-adapted leaves (cf. Andrews et al., 1995) are warranted to exclude photoinhibition as a main factor for growth reduction.

Chilling tolerance during heterotrophic and autotrophic growth
During the heterotrophic growth phase, from sowing until the end of May, chilling stress was more severe than during the autotrophic growth phasw (Fig. 1). Nevertheless, severe chilling stress also occurred during the early autotrophic growth phase. Thus, the environments were suitable to identify genotypes with increased chilling tolerance at both growth phases, which was confirmed by significant s 2 G observed for inbreds and hybrids (Table 1). The focus of the present experiments was on plant height at the three-leaf stage in the field and a detailed trajectory analysis of the growth of leaf 3 in the laboratory as an indicator of chilling tolerance during heterotrophic growth, and on plant height and fresh weight at the seven-leaf stage in the field as an indirect indication of chilling tolerance during autotrophic growth. The response to chilling stress of inbred lines and hybrids investigated in this study changed from heterotrophic to autotrophic plant growth (Tables 2 and 3). Consistently, we found a lower association between the growth of leaf 2 in the laboratory and plant height at the seven-leaf stage than at the three-leaf stage in the field experiments ( Table 4). The latter can be interpreted as an indicator for differences in chilling tolerance during heterotrophic and autotrophic growth, but may also be explained by the different temperature conditions during the two growth phases (Fig. 1).
The differences between the effects on the autotrophic and heterotrophic growth are in accordance with previous studies in maize analysing hybrids from Canada (Hodges et al., 1997) and Europe (Revilla et al., 2000). The different response to chilling stress during heterotrophic and autotrophic plant growth was more pronounced for inbreds (Tables 2 and 3), which is due to the higher capacity of individual buffering for hybrids compared with inbreds (Hallauer et al., 1988). The differences in response to chilling stress at heterotrophic and autotrophic developmentalstages can be explained by: (a) a control by different genetic factors (Hodges et al., 1997); (b) an overriding importance of maternal effects during heterotrophic but not during autotrophic growth; and/or (c) reduced chilling stress during autotrophic growth. Significant maternal effects associated with heterotrophic growth potential were reported by several authors (Eagles and Hardacre, 1979;Maryam and Jones, 1983), but they are in contrast to the findings of Aidun et al. (1991) reporting the absence of maternal effects among six reciprocal maize hybrids analysed at early growth stages. A significant (P , 0 . 05) correlation was observed between seed weight and plant height at the three-leaf stage of the hybrids (r ¼ 0 . 48; data not shown), which points to the importance of maternal effects during heterotrophic growth. Nevertheless, further analyses including reciprocal crosses are required to verify whether reciprocal effects are present.
Flint inbred lines were on average more chilling tolerant than dent inbred lines (Table 3). This can be explained by the history of the Central European heterotic groups (Reif et al., 2005). Hybrid breeding was started in the 1950s and as a promising heterotic pattern, non-adapted highyielding US lines were crossed with flint lines. The flint inbreds were developed by selfing from European openpollinated varieties, which were adapted to the cool climatic conditions of Central Europe. However, the results for line per se performance for fresh weight also indicate that the dent line D23 possesses a high level of chilling tolerance. Therefore, D23 is an excellent source to improve the chilling tolerance of the European dent pool further.
For inbreds and hybrids, plant height at the seven-leaf stage was highly correlated with fresh weight (Table 2). This indicates that plant height is suitable to predict fresh weight of the genotypes under chilling stress to a considerably high level. Consequently, plant height may serve as a non-destructive measurement for biomass at early developmental stages. This non-destructive determination of chilling tolerance would facilitate investigations on associations of chilling tolerance at early developmental stages and grain yield at harvest. Alternatively, canopy reflectance may be used as a non-invasive method to determine biomass at early developmental stages of maize (cf. Montes et al., 2007).

Associations between field trials and growth chamber experiments
The use of laboratory conditions for evaluating plant performance in breeding could be advantageous. Laboratory conditions allow plants to be grown under welldefined conditions, independent of the climate. This could reduce the cost of the breeding of chilling-tolerant hybrids, due to reduction in the time needed and higher reproducibility of the screening process. Unfortunately, it is known that plants under laboratory conditions often perform differently from under field conditions, especially under stress conditions (Mittler, 2006). One possible reason is that plants in the field are constantly exposed to several stresses at the same time. Therefore, it is important to tune the laboratory conditions to the field conditions of interest. In this study it was shown that the effect of low temperature on specific aspects of seedling growth in growth chambers is significantly associated with the growth observed in the field obtained with the same genotypes at the same stage of development, but that it is more difficult to extend this to later stages (Table 4). These findings are in accordance with a study of Hodges et al. (1997) evaluating 12 Canadian hybrids in laboratory and field experiments. This indicates that growth chambers can be used for screenings for tolerant lines, followed by field trials for the final selection.

SUPPLEMENTARY INFORMATION
Supplementary information is available online at http://aob. oxfordjournals.org/ and provides the following data.
(1) Pairwise correlation coefficients between field trial data generated in six environments and growth chamber data.
(2) Principal co-ordinate analyses of field data for five flint and five dent lines for chlorophyll concentration, plant height, plant fresh weight and site location.
(3) Daily minimum soil temperatures at the three locations during the course of the study.