Is heterosis in maize mediated through better water use?


  • José Luis Araus,

    1. International Maize and Wheat Improvement Center (CIMMYT), El Batán, Mexico
    2. Unitat de Fisiologia Vegetal, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    Search for more papers by this author
  • Ciro Sánchez,

    1. International Maize and Wheat Improvement Center (CIMMYT), El Batán, Mexico
    Search for more papers by this author
  • Llorenç Cabrera-Bosquet

    1. International Maize and Wheat Improvement Center (CIMMYT), El Batán, Mexico
    2. Unitat de Fisiologia Vegetal, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    Search for more papers by this author

Author for correspondence:
José Luis Araus
Tel: + 34 934021469


  • Heterosis increases yield potential and improves adaptation to stress in maize (Zea mays); however, the underlying mechanisms remain elusive.
  • A set of tropical inbred lines and their hybrids were grown in the field for 2 yr under three different water regimes. First-year plant water use was evaluated by measuring instantaneous traits (stomatal conductance (gs) and steady-state chlorophyll fluorescence (Fs)) in individual leaves together with time-integrative traits, which included mineral accumulation in the whole leaves of plants and oxygen isotope enrichment above source water (Δ18O) and carbon isotope discrimination (Δ13C) in the same pooled leaves and in mature kernels. Second-year water use was evaluated by measuring leaf temperature, gs and relative water content (RWC).
  • Within each growing condition, hybrids showed higher Fs, mineral accumulation, RWC, and lower leaf temperature, Δ18O and Δ13C than inbred lines. Therefore, hybrids had a better water status than inbred lines, regardless of the water conditions. Differences in grain yield across growing conditions were explained by differences in water-use traits, with hybrids and inbred lines following a common pattern. Within each growing condition, most variations in grain yield, between hybrids and inbred lines, were also explained by differences in plant water-use traits.
  • Heterosis in tropical maize seems to be mediated by improved water use, irrespective of the water conditions during growth.


The term heterosis refers to the superior performance of heterozygous hybrid plants over their homozygous parental inbred lines. In maize, heterosis was paramount in raising the yield ceiling seven decades ago, when the first commercial hybrids were introduced (Tollenaar & Lee, 2006). Heterosis has also been reported to help maize to better adapt under diverse stress conditions (Tollenaar et al., 2004; Tollenaar & Lee, 2006). Thus, differences in grain yield between hybrids and their parental inbred lines increased in plants grown under drought (Blum, 1997; Betrán et al., 2003a,b,c) and greater plant density (Liu & Tollenaar, 2009b).

Although heterosis in maize has been studied since the early 1900s and is of vital agronomical importance, the genetic and physiological basis of heterosis remains unsolved (Sinha & Khanna, 1975; Tollenaar et al., 2004). Understanding this phenomenon might open up oppor-tunities for increasing yield potential and stress adapta-tion. Plant physiologists have long attempted to explain the greater biomass and grain production in F1 hybrids (Sinha & Khanna, 1975; Tollenaar et al., 2004; Tollenaar & Lee, 2006; and references herein). There are reports that indicate that heterosis for growth is mediated by a single factor, such as embryo size (Ashby, 1930) or endogenous concentration of gibberellins (Rood et al., 1988). However, overall heterosis is considered to be the product of quantitative characters, which are the result of complex interactions between simpler processes throughout the crop cycle. Recent studies explain heterosis as a dynamic attribute affected by both the environment and the stage of development (Tollenaar et al., 2004; Tollenaar & Lee, 2006). Thus, Tollenaar et al. (2004) dissected the physiological mechanisms responsible for heterosis for grain yield in maize. They concluded that three main processes are involved: heterosis for DM accumulation before silking, which results mainly from greater light interception owing to increased leaf size; heterosis for DM accumulation during the grain-filling period, which results from greater light interception owing to greater maximum leaf area index and increased stay-green; and heterosis for harvest index. However, how heterosis mediates these processes remains elusive.

We postulate that heterosis confers better water use to plants during growth. The responses associated with heterosis are in some way opposed to what is considered the effect of water stress. For example, the larger total leaf area of the plant in hybrids is sustained basically through longer leaves rather than through variation in final leaf number or even leaf width (Sinha & Khanna, 1975; Tollenaar et al., 2004). When maize encounters water deficit, there is a decline in photosynthesis. This is caused by a reduction in light interception as leaf expansion is reduced or as leaves senesce, and the reduction in photosynthetic assimilation per unit leaf area as stomata close or as photoinhibition increases (Bruce et al., 2002). Overall, these symptoms may fall within what is considered inbred depression.

Alternatively, heterosis might confer better nutritional status, which would explain its effect on growth and yield. Hybrids have better tolerance to high sowing density than inbreds (Liu & Tollenaar, 2009b). In this context, lower nutrient acquisition by inbreds might affect either plant growth or maximum leaf area and its duration. Thus, leaf senescence might be accelerated as the demand for nitrogen (N) by grains exceeds capacity for uptake (Chapman & Edmeades, 1999).

Even though heterosis is most evident in adult traits such as plant biomass or yield, it is also apparent during embryo (Meyer et al., 2007) and early seedling development (Hoecker et al., 2006). Thus, differences in rooting pattern and performance of seedling between inbreds and hybrids have been reported (Hoecker et al., 2006, 2008). The effect of such differences, even if tiny at the beginning of the crop cycle, accumulate during the growth of the plant, following what is known as the ‘compound interest law of growth’ (Blackman, 1919; Sinha & Khanna, 1975).

Therefore, when evaluating the kind of physiological factors responsible for heterosis, it is necessary to use traits able to integrate in time (crop cycle) and scale (whole plant) the physiological response of the plant. We showed recently that the oxygen isotope enrichment above source water (Δ18O) of plant dry matter as well as the total accumulated mineral content in transpirative tissues (Cabrera-Bosquet et al., 2009a,b), provide an indication of stomatal conductance and transpiration and thus of water use during the crop cycle. These approaches are precise enough to trace genotypic differences in yield within a given water regime. In addition, carbon isotope discrimination (Δ13C) in plant tissues, even if it is not feasible for assessing genotypic differences in C4 species such as maize (Monneveux et al., 2007; Cabrera-Bosquet et al., 2009b), may still trace differences in photosynthetic gas exchange in response to changes in water status.

Over two consecutive years we examined the physiological expression of heterosis in maize using a set of inbreds and derived hybrids of tropical maize grown in the field under three contrasting water regimes. We followed the approach proposed by Tollenaar et al. (2004), studying, together with biomass and grain yield, informative traits about plant growth, stay-green during grain filling and partitioning. The water and N status of hybrids and inbreds were also measured. Water status was assessed (mostly as water use) using instantaneous and time-integrated traits. Among the former, stomatal conductance, chlorophyll fluorescence, relative water content (RWC) and leaf temperature were measured. Integrated traits included signatures of stable isotopes 18O and 13C in leaves and kernels, as well as total mineral content in leaves. Nitrogen status was inferred by analysing the N content in leaves and kernels. The final aim was to examine whether, regardless of the water conditions during growth, heterosis gives hybrids better water use. Indeed, such differences in water use may underlie the effect of heterosis on growth and yield.

Materials and Methods

Plant material and growing conditions

Fifteen tropical white-grained maize (Zea mays L.) inbred lines, derived from the La Posta Sequía (LPS) population, and 16 of their hybrid combinations were evaluated separately in trials under three different water regimes during 2007 and 18 inbreds and hybrids from the same population the year after (see the Supporting Information, Table S1). The LPS population is a white dent (Tuxpeño-related synthetic), with enhanced tolerance to drought and other stresses (Betrán et al., 2003a,b), produced by the International Maize and Wheat Improvement Center (CIMMYT). Selection schemes are detailed in Pandey et al. (1986) and Monneveux et al. (2008).

Trials were conducted at the CIMMYT experimental station in Tlaltizapán, Morelos, Mexico (18°41′ N, 99°07′ W, 940 m above sea level) during the 2007 (Fig. 1) and 2008 (Table S2) dry (winter) seasons. Full irrigation (WW) and two different levels of water stress, intermediate (IS) and severe (SS), were assayed. Each of the three water regimes was set up as a randomized complete block design with three replications per genotype. For each of the water regimes, lines and hybrids were evaluated separately in two trials planted side-by-side. Both hybrids and lines were planted in one-row plots. Entries were over-sown with two seeds per hill every 0.25 m in single rows (5 m long and spaced 75 cm apart) which were later thinned to a final plant density of 6.67 plants m−2.

Figure 1.

 (a) Maximum (Tmax) and minimum (Tmin) daily air temperatures and total daily radiation flux (Radtotal) and (b) daily precipitation and potential evapotranspiration (ETP) values, at the CIMMYT experimental station, Tlaltizapán, Morelos, Mexico, during the maize (Zea mays) growing season (December 2006 to May 2007). Irrigation times and amounts for each growing condition are indicated by vertical arrows. Dates in the x-axis are expressed as days after planting (DAP). Growing conditions were: well-watered (WW), intermediate water stress (IS) and severe water stress (SS). Sprinkler irrigation was used in all treatments after planting, to ensure homogeneous germination. In addition, the WW trials were irrigated by furrow irrigation every 2 wk throughout the cycle. The IS and SS treatments were also irrigated every 2 wk by furrow irrigation until water stress was imposed by deficit irrigation of the trials 1 month after planting and withholding irrigation c. 2–4 wk before anthesis, respectively, and then irrigating again 1 wk after anthesis. Water quantities were determined after calibrating the water delivered per unit time by furrow and sprinkler irrigation systems.

The average ratio between total water input (i.e. irrigation plus precipitation) and reference evapotranspiration (WI : ETP) under WW, IS and SS conditions were 1.5, 1.0 and 0.5, for the first year trial (Fig. 1) and 1.9, 0.9 and 0.6 for the second year trial, respectively (Table S2). For both years N fertilizer was applied before and 35 d after planting (DAP) at V6 stage (Ritchie et al., 1993), at a rate of 80 kg ha−1 of urea each time. All trials received 80 kg ha−1 of phosphorus as a triple calcium superphosphate (Ca (H2PO4)2 H2O), applied prior to planting. No potassium was applied because previous tests showed no response to this element in these soils. Experiments were kept free from weeds, insect pests and diseases by recommended chemical measures, as described in Monneveux et al. (2008).

Field measurements and samples were taken from plants inside the row with full competition. Except for grain yield (GY) and stomatal conductance (gs), which were measured both years, and RWC and leaf temperature, measured only the second year, all the traits detailed below were only measured during the first year.

Plant biomass and grain yield

The final effect of the different water regimes on growth was assessed by measuring the maximum height attained by the plant and the total aerial biomass 2 wk after anthesis; before the onset of senescence. Plant height was measured as the distance from the ground to the first tassel branch on 10 plants per plot. Total aerial plant biomass was measured in individual plants. One entire plant per plot was harvested. At maturity, GY and its main agronomical components, such as the number of ears per plant, kernel number per ha, kernel number per ear and the 100-kernel weight, were determined.

Crop growth and stay-green

Plant growth during the vegetative period, before the full establishment of the three water regimes, was assessed indirectly, in a nondestructive manner, by measuring the normalized difference vegetation index (NDVI) 35 and 55 DAP. Stay-green during middle grain filling (c. 20 d and 25 d after anthesis) was also assessed by NDVI. This index, as a proxy measurement of shoot biomass (Teal et al., 2006; Marti et al., 2007; Inman et al., 2008), was determined with a portable spectroradiometer (GreenSeeker; NTech Industries, Ukiah, CA, USA). The sensor head was held c. 0.5 m above and perpendicular to the canopy. Travel velocity through the 5-m row was at slow walking speed, enabling > 15 NDVI measurements per row to be taken.

On the same days as the NDVI measurements, the chlorophyll content of individual leaves was measured with a portable chlorophyll meter (SPAD 502; Minolta Camera Co., Osaka, Japan).

Stomatal conductance and chlorophyll fluorescence

Stomatal conductance was measured 55, 62 and 98 DAP on the abaxial surface of sun-exposed leaves from the upper part of the plant, using a Decagon Leaf Porometer SC-1 (Decagon Device Inc., Pullman, WA, USA). Measurements were taken in the 4 h around solar noon on sunny and windless days.

Chlorophyll fluorescence measurements were taken around anthesis on leaves of the same developmental age as those used to measure gs 98 DAP. A portable pulse-modulated chlorophyll fluorometer (FMS1; Hansatech Instruments Ltd, King’s Lynn, Norfolk, UK) was used. Light-adapted leaves were used for measuring steady-state chlorophyll fluorescence (Fs) under external irradiance conditions of c. 1400 μmol m−2 s−1 photosynthetic photon flux density (PPFD). The Fs value is reported as a water stress-sensitive parameter (Flexas et al., 1999, 2002). In addition, the maximum quantum yield of photosystem II (PSII) (Fv/Fm) was measured after leaves were adapted to darkness for 30 min. (Andrews et al., 1993).

Leaf temperature and relative water content

Leaf temperature was measured at anthesis in the 4 h around solar noon in sun-exposed leaves above the female flower. One leaf per plant from three to five plants were measured per plot. The RWC of leaf blades was also determined in leaves similar to those used for leaf temperature determination around midday for one leaf per plot. Leaf blade segments were weighed (FW), floated on distilled water at 4°C overnight and weighed again (TW). They were then dried at 80°C for 48 h. After this, the DW was determined. The relative water content was then calculated as:

image(Eqn 1)

Leaf ash content and leaf and kernel carbon, oxygen and nitrogen isotope analyses

Ash content was analysed in the leaves sampled for total plant biomass. In addition, the same leaves and mature kernels were used for total N concentration and stable C and O isotope analyses. Samples were oven-dried at 60°C for 48 h and ground. To determine the ash content, c. 2 g of leaf dry mass was placed in preweighed porcelain crucibles. Samples were burnt in a muffle furnace for 6–8 h at 600°C. The mineral residue was then weighed and the results expressed as percentage of dry mass.

The total N concentration and the stable C (13C : 12C) isotope ratio (R) of leaves and kernels were determined by using an elemental analyser (2100; Carlo Erba, Milan, Italy) interfaced with an isotope ratio mass spectrometer (IRMS) (Deltaplus Advantage; Thermo-Finnigan, Bremen, Germany) at the Colorado Plateau Stable Isotope Laboratory (CPSIL). The N concentration was expressed as a percentage of the DW. Carbon isotope signatures were expressed as δ13C values, using secondary standards calibrated against Vienna Pee Dee Belemnite calcium carbonate (VPDB), with analytical precision at c. 0.1‰:

image(Eqn 2)

The carbon isotope discrimination (Δ13C) was then calculated as:

image(Eqn 3)

where δ13Ca and δ13Cp refer to air and plant carbon isotope compositions, respectively. The δ13C of free atmospheric CO2 was taken as −8‰ (Farquhar et al., 1989).

The 18O : 16O ratios of irrigation water were determined by the CO2 : H2O equilibration technique, using an IRMS (Delta S; Finnigan MAT, Bremen, Germany) at the Scientific Facilities of the University of Barcelona, Spain. The 18O : 16O ratios of leaf and kernel samples were analysed at the CPSIL by pyrolysis over glassy carbon at 1350°C, using a Thermo-Electron TC/EA (Thermo-Chemical Elemental Analyzer) interfaced via a CONFLO-II to a Thermo-Electron Delta Plus XL gas IRMS (Finnigan MAT, Bremen, Germany). Results were expressed as δ18O values, using secondary standards calibrated against the Vienna Standard Mean Oceanic Water (VSMOW). Analytical precision was c. 0.3‰ for dry matter and 0.2‰ for irrigation water:

image(Eqn 4)

Then, 18O enrichment in kernels (Δ18O) was calculated as:

image(Eqn 5)

where δ18Op and δ18Oiw refer to the oxygen isotope compositions of plant sample and irrigation water, respectively (δ18Oiw was c.−10.78‰).

Evaluation of heterosis

For each water regime, heterosis was calculated for grain yield, agronomical yield components and all secondary traits measured in this study. Because one of the two testers (CML–449) used as a male parent did not germinate properly, we calculated heterosis in all cases by comparing with the parent used as a female. We considered it the better parent, using the following formula (Liang et al., 1972):

image(Eqn 6)

where F1 and the better parent are the means of the F1 hybrid and the maternal inbred line within each of the three water regimes assayed.

Statistical analysis

To evaluate differences between hybrids and inbred lines within each water treatment (WW, IS and SS), means for the traits measured were compared by an unpaired t-test at 5% significance level, using the SPSS 15.0 statistical package (SPSS Inc., Chicago, IL, USA). Multiple linear regression analysis (stepwise) was used to analyse the relationships between the water status traits and grain yield. Linear stepwise models were built by combining values of hybrid and inbred lines of the parameters studied (GY and water-use traits) within each water regime. Nonlinear regressions were fitted to the relationships between the traits measured across the three growing conditions assayed.


Crop growth and N status

Only slight differences in NDVI between WW and drought conditions (IS and SS) were recorded during the vegetative stage before the full establishment of the three water regimes. However, in all cases heterosis was positive (Table 1). Water stress led to a 35% and near-50% decrease from WW to SS in plant height and aerial biomass, respectively. Regardless of the water status, values for these traits were higher in hybrids than inbreds, and heterosis was higher under well-watered conditions (Table 1).

Table 1.   Mean values for crop growth and stay-green parameters of maize (Zea mays) hybrids and inbred lines exposed to different water regimes (WW, full irrigation; IS, intermediate water stress; SS, severe water stress) and estimates of mean heterosis (H) for the parameters studied.
HybridsInbredsH (%)HybridsInbredsH (%)HybridsInbredsH (%)
  1. Plants were grown at the CIMMYT experimental station at Tlaltizapan, Morelos, Mexico during the 2007 dry season. DAP , days after planting; DAA, days after anthesis (determined when 50% of the plants had shed pollen). Anthesis occurred c. 84 DAP and 91 DAP for the hybrids and inbred lines, respectively.

  2. *Significant differences between hybrids and inbred lines within each water treatment (WW, IS and SS) according to unpaired t-test (< 0.05); ns, not significant.

  3. Data are the means of 16 (hybrids) and 15 (inbred lines) genotypes ± SE. Each individual genotype value was the average of three single (i.e. plot) replications. Parameters studied included: NDVI, normalized difference vegetation index; plant biomass, shoot DW of an individual plant 2 wk after anthesis; plant height, maximum plant height; leaf N, nitrogen concentration in the whole leaves of a plant 2 wk after anthesis; kernel N, nitrogen concentration in mature kernels; SPAD, leaf chlorophyll content measured with a portable device; Fv/Fm. the maximum quantum yield of photosystem II (PSII).

Crop growth
 NDVI35DAP0.36 ± 0.010.26 ± 0.01*40.30.33 ± 0.010.25 ± 0.01*32.10.32 ± 0.010.25 ± 0.01*28.8
 NDVI55DAP0.84 ± 0.010.68 ± 0.01*24.20.80 ± 0.010.67 ± 0.01*19.50.77 ± 0.010.67 ± 0.01*16.3
 Plant biomass (g)149.2 ± 5.191.7 ± 4.2*67.294.9 ± 3.666.6 ± 3.1*44.078.5 ± 2.854.5 ± 2.7*47.6
 Plant height (cm)198.6 ± 2.0110.8 ± 2.5*81.1143.3 ± 1.793.2 ± 2.5*55.0124.7 ± 2.590.1 ± 1.5*38.5
 Leaf N (%)1.29 ± 0.041.54 ± 0.04*−16.01.60 ± 0.021.71 ± 0.02*−6.61.59 ± 0.031.72 ± 0.03*−6.7
 Kernel N (%)1.49 ± 0.021.74 ± 0.03*−14.11.88 ± 0.022.04 ± 0.04*−7.61.93 ± 0.022.02 ± 0.05ns−3.7
 SPADanthesis46.7 ± 0.840.2 ± 1.0*17.937.3 ± 0.836.3 ± 1.1ns3.734.1 ± 0.933.4 ± 1.0ns3.9
 Fv/Fm anthesis0.81 ± 0.010.75 ± 0.01*6.80.72 ± 0.010.63 ± 0.03*16.80.58 ± 0.010.56 ± 0.02ns5.2
 NDVI20DAA0.74 ± 0.010.71 ± 0.01*4.70.63 ± 0.010.37 ± 0.01*70.60.49 ± 0.010.37 ± 0.01*31.8
 NDVI25DAA0.77 ± 0.010.54 ± 0.01*43.10.62 ± 0.010.19 ± 0.01*224.90.40 ± 0.010.19 ± 0.01*109.6
 SPAD20DAA46.6 ± 0.836.0 ± 1.3*32.935.0 ± 0.822.8 ± 1.2*54.425.8 ± 0.817.2 ± 1.2*53.7
 SPAD25DAA47.2 ± 1.432.3 ± 1.4*51.929.1 ± 1.313.4 ± 1.5*154.416.8 ± 0.98.7 ± 1.0*112.5

For both leaves sampled 2 wk after anthesis and kernels, water stress increased N concentration. Moreover, it was slightly higher in leaves and kernels of inbreds than of hybrids, and heterosis for N concentration was most negative under well-watered conditions.


Leaf chlorophyll content at anthesis decreased in both hybrids and inbreds in response to water stress. Hybrids showed higher values than inbreds under well-watered conditions, while the differences were not significant under drought (Table 1). Thus, heterosis was at its highest under WW. The Fv/Fm measured at anthesis also decreased steadily in response to water stress, with hybrids showing higher values than inbreds, regardless of the water conditions (Table 1). During grain filling differences in leaf chlorophyll content across water regimes increased, with values steadily diminishing from 20 d to 25 d after anthesis in response to water stress. Moreover, the leaf chlorophyll content of inbreds decreased more rapidly than in hybrids, particularly under drought. As a consequence, heterosis for leaf chlorophyll increased throughout grain filling, particularly under drought compared with WW. The NDVI also decreased from 20 d to 25 d after anthesis, in response to drought conditions, more so for inbreds than for hybrids. As a consequence, heterosis for the NDVI was higher under drought than under well-watered conditions and increased during grain filling.

Grain yield and its agronomical components

Yield was strongly affected by growing conditions, with hybrids yielding more than inbreds in all growing conditions in the first (Table 2) and second (Table S3) year. Both kernels per ha and kernels per ear also decreased strongly in response to water stress and differences between hybrids and inbreds were present in the three growing conditions (Table 2). The number of ears per plant decreased in response to water stress, but inbreds showed higher values than hybrids under WW and SS conditions, and thus heterosis was negative in these environments. The 100-kernel weight decreased in response to drought stress and hybrids tended to show higher values, however, differences were only significant in WW (Table 2).

Table 2.   Mean values for maize (Zea mays) grain yield and water status parameters of hybrids and inbred lines grown under different water regimes (WW, full irrigation; IS, intermediate water stress; SS, severe water stress) and estimates of mean heterosis (H) for the parameters studied
HybridsInbredsH (%)HybridsInbredsH (%)HybridsInbredsH (%)
  1. Plants were grown at the CIMMYT experimental station at Tlaltizapan, Morelos, Mexico during the 2007 dry season.

  2. DAP, days after planting.

  3. *Significant differences between hybrids and inbred lines within each water treatment (WW, IS and SS) according to unpaired t-test (P < 0.05); ns, not significant.

  4. Data are the mean of 16 (hybrids) and 15 (inbred lines) genotypes ± SE. Each individual genotype value was the average of three single (i.e. plot) replications. Parameters studied included: GY, grain yield; HKW, 100-kernel weight; gs, stomatal conductance; Fs, steady-state chlorophyll fluorescence; leaf ash, leaf ash concentration in the leaves of a plant 2 wk after anthesis; total leaf ash, total leaf ash content in the whole leaves of a plant 2 wk after anthesis; leaf Δ18O, oxygen isotope enrichment in the same leaves used for ash content; kernel Δ18O, oxygen isotope enrichment in mature kernels; leaf Δ13C, carbon isotope discrimination in the same leaves used for ash content; kernel D13C, carbon isotope discrimination in mature kernels.

Grain yield
 GY (Kg ha−1)7615 ± 2223085 ± 208*158.81687 ± 132277 ± 55*716.5438 ± 47177 ± 27*189.4
 Kernels ha−1 (×107)2.5 ± 0.071.5 ± 0.09*76.61.1 ± 0.090.2 ± 0.04*742.50.3 ± 0.040.1 ± 0.02*163.0
 HKW (g)30.2 ± 0.521.0 ± 0.6*45.215.2 ± 0.414.8 ± 0.7ns5.814.7 ± 0.413.4 ± 0.6ns12.9
 Ears per plant1.1 ± 0.031.4 ± 0.1*−21.00.8 ± 0.030.8 ± 0.1ns10.00.4 ± 0.020.6 ± 0.1*−23.7
 Kernel weight per ear (g)109.3 ± 2.633.8 ± 2.2*240.334.9 ± 1.95.5 ± 0.7*684.320.0 ± 2.35.0 ± 0.4*332.5
Water status
 gs 55DAP (mmol m−2 s−1) 288.8 ± 16.9230.5 ± 13.9*28.0311.5 ± 12.3265.9 ± 7.3*19.6226.4 ± 11.7251.7 ± 8.5ns−8.8
 gs 62DAP (mmol m−2 s−1)165.4 ± 9.5153.8 ± 9.6ns21.2170.8 ± 8.5203.6 ± 10.1*−13.0251.5 ± 8.1154.4 ± 9.6*66.8
 gs 98DAP (mmol m−2 s−1) 117.2 ± 5.6118.9 ± 6.2ns3.486.5 ± 6.346.9 ± 3.5*105.156.6 ± 4.545.9 ± 6.0ns63.4
 Fs anthesis444.8 ± 7.8305.1 ± 7.4*46.4326.7 ± 12.6269.9 ± 7.6*21.2342.6 ± 16.8210.4 ± 9.8*65.4
 Leaf ash (%)18.7 ± 0.216.7 ± 0.3*12.313.7 ± 0.212.3 ± 0.2*11.212.4 ± 0.211.5 ± 0.3*8.2
 Total leaf ash (g)27.9 ± 1.015.3 ± 0.6*86.213.1 ± 0.58.3 ± 0.4*60.39.7 ± 0.46.3 ± 0.3*59.7
 Leaf Δ18O (‰)42.0 ± 0.242.8 ± 0.2*−2.143.3 ± 0.144.0 ± 0.2*−1.643.3 ± 0.244.6 ± 0.2*−2.9
 Kernel Δ18O (‰)40.3 ± 0.142.6 ± 0.1*−5.341.3 ± 0.143.1 ± 0.2*−4.242.6 ± 0.143.6 ± 0.2*−2.2
 Leaf Δ13C (‰)5.30 ± 0.035.64 ± 0.04*−5.85.64 ± 0.035.82 ± 0.03*−3.05.79 ± 0.035.97 ± 0.05*−2.7
 Kernel Δ13C (‰)3.82 ± 0.024.01 ± 0.04*−4.44.61 ± 0.024.30 ± 0.06*7.54.65 ± 0.024.39 ± 0.03*5.9

Water use

No consistent differences across trials were observed in gs, measured c. 8 wk and 9 wk after planting, before water regimes were fully established (Table 2). Differences between hybrids and inbreds were diverse. Thus, 8 wk after planting, hybrids showed significantly higher gs than inbreds under WW and IS conditions but no significant differences occurred under SS. One week later, while no significant differences in gs were observed in WW, hybrids showed higher values than inbreds in SS. However, the opposite pattern was observed in IS. Stomatal conductance was measured again 14 wk after planting. By this time the water regime was fully established and differences across treatments were evident. However, gs was only significantly higher in hybrids than inbreds in IS; differences under WW and SS were not significant.

Fs measured at anthesis was higher in WW than under the stress treatments. Hybrids showed consistently higher values than inbreds and heterosis was highest in SS followed by WW and IS conditions (Table 2).

Ash concentration in leaves 2 wk after anthesis decreased steadily from WW to SS. Moreover, hybrids had higher values than inbreds, regardless of the growing conditions. Positive heterosis was present for all three water regimes even when values tended to be higher in WW (Table 2). When ash content was expressed as total content per plant, differences across treatments and between hybrids and inbreds were even clearer and heterosis values were higher. However, for ash concentration, heterosis was slightly higher in WW than under the water stress treatments.

In the same leaves where ash content was determined, and in mature grains, Δ18O significantly increased in response to water stress during growth. Moreover, as hybrids had lower values than inbreds under the three water regimes, heterosis for Δ18O of leaves and kernels was negative in all cases. However, differences between hybrids and inbreds increased in Δ18O of grains more than in Δ18O of leaves under WW and IS conditions (Table 2). The Δ13C of the same leaves and kernel samples increased in response to water stress. Moreover, leaf Δ13C was consistently lower in hybrids than in inbreds, regardless of the water regime. However, while kernel Δ13C was also lower in hybrids than in inbreds under WW, the opposite trend occurred under IS and SS.

Second-year trials also showed hybrids to have better water status than inbreds. Thus at anthesis, hybrids exhibited higher RWC and gs, and lower leaf temperature, than inbreds regardless of growing water conditions. Two weeks after anthesis, RWC was again higher in hybrids for each of the three water regimes (Table S3).

Relationships between traits in hybrids and inbreds across environments

Relationships were examined by combining the set of hybrids and inbreds grown under the three different water regimes during the first year (= 93). Among growth traits, total aerial biomass per plant correlated strongly and positively (R2 = 0.80) with GY (Fig. 2a). The correlation between the stay-green trait NDVI, 20 d after anthesis, and GY was greater (R2 = 0.87; Fig. 2b) and the agronomical yield component, kernel number per ha, correlated with GY even better (R2 = 0.97; Fig. 2c). The 100-kernel weight also correlated positively and significantly (< 0.0001) with GY, but its determination coefficient was somewhat lower (R2 = 0.54).

Figure 2.

 Relationship between maize (Zea mays) grain yield (GY) and (a) total aerial biomass per plant c. 2 wk after anthesis, (b) the normalized difference vegetation index (NDVI) measured c. 2 wk after anthesis and (c) the kernel number per ha. Data from hybrids (open circles) and inbred lines (closed circles) were plotted together (= 93), each point representing a mean value for three plots of a single genotype grown under a particular water regime.

Integrated indicators of water status correlated strongly with GY and total biomass. The best trait was leaf ash concentration, which positively correlated with biomass (R2 = 0.60; Fig. 3a) and even more so with GY (R2 = 0.82; Fig. 4a). The Δ18O of kernels also correlated, although negatively, with biomass (R2 = 0.68; Fig. 3b) and GY (R2 = 0.53). Conversely, the instantaneous measurements Fs, correlated positively with GY (R2 = 0.54; data not shown). The other traits (either integrative or instan-taneous) concerning water status also correlated with GY but to a lesser extent (data not shown). Ash content correlated positively and strongly with the two agronomical determinants of GY: kernels per ha (R2 = 0.79) and 100-kernel weight (R2 = 0.73) (Fig. 4b,c). In addition, 100-kernel weight correlated negatively with N concentration in kernels (R2 = 0.60; Fig. 5).

Figure 3.

 Relationship between maize (Zea mays) biomass per plant c. 2 wk after anthesis and (a) ash concentration in leaves c. 2 wk after anthesis and (b) the oxygen isotope enrichment in mature kernels (18O). Data from hybrids (open circles) and inbred lines (closed circles) were plotted together (= 93), each point representing a mean value for three plots of a single genotype grown under a particular water regime.

Figure 4.

 Relations between maize (Zea mays) (a) grain yield (GY) with ash content of leaves c. 2 wk after anthesis. (b) the kernel number per ha with ash content of leaves and (c) the 100-kernel weight (HKW) with the ash content of leaves. Data from hybrids (open circles) and inbred lines (closed circles) were plotted together (= 93), each point representing a mean value for three plots of a genotype grown under a particular water regime.

Figure 5.

 Relationship between maize (Zea mays) 100-kernel weight (HKW) and kernel nitrogen (N) concentration. Data from hybrids (open circles) and inbred lines (closed circles) were plotted together (= 93), each point representing a mean value for three plots of a single genotype grown under a particular water regime.

During the second year, leaf temperature (R2 = 0.69) as well as RWC (R2 = 0.59) measured around anthesis also correlated with GY (= 108; see the Supporting Information, Fig. S1) with both hybrids and lines following the same pattern.

Relationships between GY and water status traits in hybrids and inbreds within environments

A stepwise analysis was performed, with GY for the set of hybrids and inbreds growing under each of the three growing conditions as a dependent variable and the different water status traits displayed in Table 3 (leaf and kernel Δ13C, Δ18O and δ15N, leaf ash content, gs and Fs) as independent variables. Analyses considered either individual plot values or genotypic (i.e. mean of three plots) values. In both cases the combination of water status traits explained most of the variability in GY. Thus, for plot values, water status traits explained from 50% (SS) to 72% (WW) of variability in GY, while for genotype values they explained from 52% (SS) to 88% (WW).

Table 3.   Multiple linear regressions (stepwise) explaining maize (Zea mays) grain yield (GY) variation from water status traits: stable isotopes (leaf and kernel Δ13C and Δ18O), leaf ash content measured 2 wk after anthesis, gs measured 55, 62 and 98 d after planting and Fs measured at anthesis
Water regimeInitial variableInitial R2Initial MSEFinal stepwise modelFinal R2Final MSE
  1. ***, < 0.001; WW, full irrigation; IS, intermediate water stress; SS, severe water stress.

  2. Plants were grown during the 2007 dry season. Linear stepwise models were built in two different manners: upper, combining hybrid and inbred line individual (i.e. plot) genotype values for the parameters studied (GY and water status traits) within each water regime; and lower, combining hybrid and inbred line genotype means (i.e. averaged across the three plots per genotype) for the parameters studied (GY and water status traits) within each water regime. Initial variable, first variable entering the model; initial coefficient of determination (R2) and mean square error (MSE), adjusted R2 and MSE after including the first variable in the model; final R2 and MSE, adjusted R2 and MSE obtained with the final stepwise model. Parameters studied included: Δ13Ck, kernel carbon isotope discrimination; Δ13Cl, leaf carbon isotope discrimination; Δ18Ok, oxygen isotope enrichment in kernels; Δ18Ol, oxygen isotope enrichment in leaves; Fs, steady-state chlorophyll fluorescence measured at anthesis; gs55, gs62 and gs98, stomatal conductance measured at 55, 62 and 98 d after planting.

= 93
SSΔ13Ck0.297***0.192GY = 0.47Δ13Ck–0.23Δ13Cl-0.38Ashl+0.01gs62–0.240.503***0.164
ISΔ18Ok0.345***0.693GY = −0.12Δ18Ok–1.56Δ13Cl +0.09gs98 + 0.87Δ13Ck+10.680.642***0.509
WWΔ18Ok0.589***1.650GY = −9.2Δ18Ok–2.69Δ13Cl +0.07Fs+55.550.720***1.376
n = 31
SSgs620.521***0.139GY = 0.02gs62–0.180.521***0.139
ISgs980.627***0.506GY = 0.11gs98+1.22Δ13Ck–1.51Δ13Cl+0.05gs55+0.04gs62+2.900.865***0.321
WWFs0.794***1.107GY = 0.08Fs–1.03Δ18Ok-2.70Δ13Cl +59.620.884***0.876

Within each growing condition leaf Δ18O of the 15 inbred lines and the corresponding single hybrids were positively related, while for kernel Δ18O the relationships were not significant (Fig. 6). Positive relationships were found for GY under SS (R2 = 0.22) and IS (R2 = 0.25), but not in WW (data not shown).

Figure 6.

 Relationship between the 15 inbreds and the derived single crosses within each of the three growing water regimes for (a) the oxygen isotope enrichment (Δ18O) of the leaves and (b) of kernels of maize (Zea mays). The two points encircled were not used to calculate the coefficient of regression. Growing conditions were: well-watered (WW), intermediate water stress (IS) and severe water stress (SS). Within each plot, dashed lines show a 1 : 1 correlation. *, P < 0.05; ***, P < 0.001.


Our range of yields for hybrids and inbreds under well-watered conditions is comparable to that reported previously by Betrán et al. (2003a,b), using a set of tropical lines and the derived single hybrids from the same LPS population grown in the absence of water stress during a winter season at the CIMMYT station, Tlaltizapán. The other two growing conditions assayed may be considered as intermediate–severe and very severe drought stress (Betrán et al., 2003a,b). In this previous study with the LPS population, hybrids gave higher yield and had taller plants, slower leaf senescence and higher leaf chlorophyll content than inbreds, under both drought stress and optimal conditions (Betrán et al., 2003a,b). Moreover, heterosis for GY increased with the intensity of drought stress (Betrán et al., 2003b). Overall, our results agree with previous findings in the LPS population.

In agreement with Tollenaar et al. (2004), we found that heterosis for biomass and grain yield was accompanied by heterosis for growth, stay-green and partitioning traits. Moreover, growth traits such as the total aerial biomass or the NDVI values around middle grain filling correlated positively with GY when all hybrids and inbreds were combined under the three growing conditions. In both cases, hybrids and inbreds followed a common relationship pattern (Fig. 2), which further indicates that heterosis for growth and stay-green traits is responsible for the higher GY attained by hybrids.

The main agronomical yield component responsible for differences in grain yield across environments combining hybrids and inbred lines was the number of kernels per ha (Fig. 2). Moreover, agronomical traits influencing kernel number showed consistent heterosis, regardless of water conditions (Table 2). Liu & Tollenaar (2009a) reported that heterosis for grain yield was closely associated with heterosis for kernel number, with this trait being associated with the amount of dry matter partitioning to the kernels during the sensitive period for kernel establishment (Echarte et al., 2004). Furthermore, Liu & Tollenaar (2009a) conclude that heterosis for kernel set was attributable, in part, to the relationship between kernel number and plant growth rate during the period bracketing flowering. In this context, our results suggest that hybrids show consistently higher partitioning to kernels than inbreds, irrespective of the water conditions during growth. Nevertheless, short anthesis-to-silking interval (ASI) in hybrids and subsequently better pollination should be not discarded as explanation of heterosis in grain number (Bolaños & Edmeades, 1996).

The number of grain-bearing ears per plant was the only agronomical yield component with negative heterosis. A similar pattern has also been reported in temperate maize (Tollenaar et al., 2004).

In summary, all the effects of heterosis on growth, stay-green, biomass, retranslocation and grain yield are consistent with the findings of Tollenaar and coworkers (Tollenaar et al., 2004; Tollenaar & Lee, 2006). Heterosis may be the result of improved N use efficiency or may result from the better water status of hybrids. We now discuss these two options.

Heterosis for N status

Heterosis in the uptake of N and other mineral nutrients was reported for many years, when data were expressed on a plant basis (Sinha & Khanna, 1975). However, as early as 1934, Smith (1934) reported that no heterosis for N accumulation existed if data were calculated on a unit weight basis. In our results, the higher N concentration in leaves and kernels of inbred lines, compared with hybrids, not only precludes a lower intrinsic capacity of inbreds for N uptake and use, but is also coherent with a somewhat lower water status of inbreds than of hybrids. Thus, it is widely reported that, under well-fertilized conditions, plants exposed to water stress usually increase their leaf N concentration on a dry matter basis, because of the limitations placed by drought on plant growth and carbon assimilation (Bänziger et al., 2000). In addition, when data from hybrids and inbreds under the three growing conditions were combined, the N concentration of grains correlated negatively with 100-kernel weight and both hybrids and inbreds followed a similar relationship pattern (Fig. 5). This kind of negative relationship is widely documented in maize and other cereals, when plants that are grown under different water conditions, but with good N fertilization, are compared (Bänziger & Lafitte, 1997; Bänziger et al., 2000). This does not support differences in N metabolism as a primary cause of heterosis. In fact, Betrán et al. (2003b), when working with inbreds and hybrids derived from the LPS population, concluded that the expression of heterosis was smaller under low N environments than under nonstress environments. In addition the higher values in the ratio total shoot N/total kernel N in inbreds than hybrids (data not shown), also precludes a limitation in N acquisition as a cause of heterosis.

Water status and heterosis for growth and stay-green

In our study, heterosis for plant growth and stay-green traits was present, irrespective of the growing conditions. However, there are some differential patterns between them. Thus, heterosis for growth traits showed the maximal differences under well-watered conditions. It has long been known that cell expansion is one of the first physiological changes caused by dehydration (Hsiao, 1973). This may occur even under conditions considered well-watered or in the transition to mild water stress (Kramer & Boyer, 1995; Passioura, 1996). Under our well-watered conditions, better water use of hybrids than inbreds, even if mild, might be responsible for these differences in growth. Moreover, the possible effect of a higher endogenous concentration of gibberellins for heterosis for shoot growth (Rood et al., 1988) may be better expressed under well-watered conditions.

Differences across water regimes for leaf chlorophyll content and Fv/Fm were already present at anthesis (Table 1), suggesting that permanent photoinhibition and photodamage were already present under drought conditions (Valentinuz & Tollenaar, 2004; Ding et al., 2006). Heterosis for stay-green traits was already evident at anthesis and steadily increased under the three growing conditions during grain filling. However, unlike during growth, heterosis for stay-green was higher under water stress than with well-watered conditions. Higher heterosis under drought may be a consequence of more severe stress causing accelerated senescence in inbreds (Hsiao, 1973; Passioura, 1996).

Water status and heterosis for partitioning and grain yield

Heterosis for grain yield as well as the number of kernels per ha and per plant were higher under drought than under WW conditions (Table 2). Betrán et al. (2003a,b) also reported increases in heterosis for GY and kernel number under drought stress, indicating that hybrids may be more tolerant to drought stress than inbred lines. However, the effects of drought (Bruce et al., 2002) and heterosis on GY are both sustained by decreases in kernel number (Fig. 2c).

Other stresses, such as increasing plant density, which may have a similar effect to increasing water stress, also increased heterosis for GY, while not affecting heterosis for dry matter at maturity (Liu & Tollenaar, 2009b). By contrast, stresses very different from drought, such as shading during the grain-filling period, did not increase heterosis for GY (Liu & Tollenaar, 2009a).

Heterosis for HKW was only important under fully irrigated conditions, which suggests that, compared with hybrids, inbreds suffered a shortage of photoassimilates during grain filling owing some degree of water stress, together with less stay-green behavior, (i.e. shorter grain filling) while the number of kernels to fill was still relatively high. Deeper roots of hybrids under well-watered conditions may offer an advantage in plant water status, particularly when plants are subjected to high vapour pressure deficit (VPD) such in our growing conditions (Fig. 1, Table S2), by preventing atmospheric water deficit.

Heterosis for water-use traits

Stomatal conductance has been proposed as a key trait indicative of the effect of water stress on photosynthetic leaf performance (Medrano et al., 2002; Jiang et al., 2006). However, in our first-year results, heterosis for gs measured at different times during the plant cycle did not show a coherent pattern throughout any of the growing conditions (Table 2). Results illustrate the inherent heterogeneity associated with the characterization of water status by use of an instantaneous trait such as gs measured in single leaves and affected by daily patterns. The available literature does not conclusively support or discard the effect of heterosis on gas exchange of leaves. Thus, the recent studies do not provide values of gs or transpiration that might reflect water status during growth (Ahmadzadeh et al., 2004; Tollenaar et al., 2004; Tollenaar & Lee, 2006).

Importantly, most of the integrative water-use traits measured in our study showed consistent differences between hybrids and inbreds across the different growing conditions. Thus, mineral accumulation in transpirative tissues was lower and Δ18O values in leaves and kernels and Δ13C values in leaves were higher in inbreds. The Δ13C of grains also increased in inbreds, but only under well-watered conditions. Ash content in leaves has been reported in maize (Cabrera-Bosquet et al., 2009b) and other cereals (Araus et al., 1998, 2002a) to be positively associated with higher transpiration because of better water status and in turn with higher grain yield. The Δ18O from plant tissues may also be considered a proxy for time-integrated stomatal conductance and transpiration (Barbour et al., 2000; Barbour, 2007; Farquhar et al., 2007). An increase in Δ18O of both leaves and kernels as a result of water stress has been reported recently in maize hybrids (Cabrera-Bosquet et al., 2009a). Previous results with other cereals also support an increase in Δ18O in response to increased soil water stress (Ferrio et al., 2007; Cabrera-Bosquet et al., 2009c). The small, yet significant, increase in Δ13C in inbreds compared with hybrids also reflects poorer water status, in this case affecting photosynthetic carbon assimilation and water-use efficiency (Dercon et al., 2006; Monneveux et al., 2007; Cabrera-Bosquet et al., 2009b). While leaf ash content and Δ18O for both leaves and kernels are traits precise enough to show genotypic differences in maize, Δ13C seems less precise and to be only affected by growing conditions (Cabrera-Bosquet et al., 2009b). Even so, the differences in water status between hybrids and inbred lines are strong enough to be reflected by the Δ13C in leaves.

The strength of the integrated traits used in our study is that they may provide an overall indication of the transpirative and photosynthetic record experienced by the plant. Moreover, as these traits have a different mechanistic basis, they provide independent lines of evidence on the poorer water status of inbreds. In addition, the Δ18O and Δ13C in kernels suggests that differences in water status during grain filling between hybrids and inbreds were maximal under well-watered conditions.

Nevertheless, even instantaneous traits measured in individual leaves reflected different water statuses of hybrids and inbreds during flowering – a critical stage for GY. Thus, during the first year Fs measured at anthesis was higher in hybrids than in inbreds, regardless of the water treatment. Fs it is a faster and less intrusive parameter to measure than gs (Flexas et al., 1999, 2002) which allowed for more replications per plot than for gs. During the second year, gs (this time greatly increasing the measurements per plot compared with the first year) together with leaf temperature and RWC of individual leaves also indicated a better water status for hybrids at anthesis. Moreover, these traits were strongly correlated with GY when inbreds and hybrids under three growing conditions were combined (Fig. S1). Kernel number (and thus grain yield) in maize is greatly affected by stress in the period between 2-wk-before and 3-wk-after silking (Andrade et al., 2000 and references therein).

In conclusion, the traits discussed clearly show that hybrids experienced better water use and status than inbreds, regardless of the water conditions during growth.

Potential mechanisms and implications for breeding

There are potential mechanisms responsible for better water use in hybrids, such as the ability of the plant to capture water from a drying soil through deeper or more thorough soil exploration or by osmotic adjustment. Some reports link osmotic adjustment to stable GY of hybrids under conditions of transient stresses (Lemcoff et al., 1998). However, osmotic adjustment seems to not be as relevant in maize (Tardieu, 2006; but see Chimenti et al., 2006). Moreover, heterosis for biomass and GY is always expressed regardless of the water conditions during growth.

Heterosis in root growth was being investigated in the 1930s (Ashby, 1932). Apparently, relative root size is not the target trait, thus no heterosis (Sinha & Khanna, 1975) and positive heterosis (Ahmadzadeh et al., 2004) have been reported for the shoot : root ratio. Therefore, compared with inbreds, hybrids seem to allocate a smaller proportion of their total dry matter to the root, which may be interpreted as an indirect indication that hybrids maintain a constitutively better water use than inbreds.

Hoecker et al. (2006) demonstrated in maize that heterosis can already be seen during the very early stages of root development, a few days after germination. In that study lateral root density showed the highest degree of heterosis. Chun et al. (2005) reported heterosis for total length of lateral roots together with a higher ratio of lateral to axial root length in 20-d maize seedlings growing under different N conditions. Later in development, lateral roots become dominant and are responsible, together with the post-embryonic shoot-borne root system, for the major portion of water and nutrient uptake (Hochholdinger et al., 2004). In addition Li et al. (2008) concluded that seedlings of hybrids develop a greater number of fine roots than their parents. Furthermore heterosis exists for water uptake ability at the root cell level under well-watered conditions (Liu et al., 2009). Moreover, Hoecker et al. (2008) identified nonadditive gene expression in primary roots of maize hybrids, compared with the average expression levels of their parental inbred lines. Among those genes, a superoxide dismutase 2 was expressed significantly above the mid-parent value in all hybrids studied. This may play a protective role in heterosis-related anti-oxidative defense in the primary root of maize hybrids. Paschold et al. (2010) state that a subtle regulation of particular biochemical pathways such as the phenylpropanoid pathway in hybrids might contribute to the manifestation of heterosis in maize primary roots.

Our study may also provide some clues for breeding. It is widely accepted that, while yield potential is important in determining yield under moderate water stress (Araus et al., 2002b, 2008), it becomes much less so if yields fall below 50–60% of the potential (Bänziger & Lafitte, 1997). This is when stress-adaptive traits acquire real significance. Thus, constitutive whole-plant traits play a major role in affecting plant water use and plant dehydration avoidance under stress. These largely determine some of the negative relations between yield potential and the ability to sustain yield under severe water shortage (Blum, 2005). However, heterosis, even if differentially affected by water conditions during growth, may be considered an example of nonexistence of significant interactions between the genotype and the environment. Thus heterosis, while associated with higher yield potential, also confers adaptation under a full range of stress conditions. Therefore, there is the possibility of a common constitutive physiological trait (responsible for heterosis) which confers both higher yield potential and stress adaptation. To introduce some degree of heterosis in maize open-pollinated varieties may confer better performance under drought stress (Betrán et al., 2004).

The data of the present study on physiological mechanisms of heterosis also illustrate that the effective use of water rather than water-use efficiency is the target for crop improvement under drought stress (Blum, 2009). However, the strength of the relationships between lines and derived hybrids for water-use traits (Fig. 6) and GY is low and comparable to that previously reported for GY (Betrán et al., 2003b). This illustrates the potential difficulty in breeding for a quantitative trait such as heterosis.

We conclude that, in our study, regardless of the growing conditions, hybrids have better water use than inbreds. Heterosis for growth and yield in maize may (at least in part) be mediated by such accumulated differences in water use and status. Further studies comparing root development and soil moisture extraction of inbreds and hybrids are necessary.


DTMA (Bill and Melinda Gates Foundation) and the 08.7860.3-001.00 project (BMZ, Germany) are acknowledged. L.C-B. was the recipient of a research grant (AP2005-4965) from the Spanish MEC. We also acknowledge the useful comments raised by the Editor and the four anonymous Referees. We thank D. Mullan for English editing.