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


     

Published online 18 May 2006
Published in Crop Sci 46:1479-1487 (2006)
© 2006 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Duggan, B. L.
Right arrow Articles by Fowler, D. B.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Duggan, B. L.
Right arrow Articles by Fowler, D. B.
Agricola
Right arrow Articles by Duggan, B. L.
Right arrow Articles by Fowler, D. B.
Related Collections
Right arrow Crop Growth and Development
Right arrow Wheat
Right arrow Plant and Environment Interactions

CROP PHYSIOLOGY & METABOLISM

Yield Structure and Kernel Potential of Winter Wheat on the Canadian Prairies

B. L. Duggana and D. B. Fowlerb,*

a Oregon State University, Central Oregon Agricultural Research Center, 850 N.W. Dogwood Lane, Madras, OR 97741
b Dep. of Plant Sciences, Univ. of Saskatchewan, Saskatoon, SK, S7N 5A8, Canada

* Corresponding author (Brian.Fowler{at}usask.ca)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
Improvements in agronomic practices and cultivars have allowed for expanded production of winter wheat (Triticum aestivum L.) on the Canadian prairies. In this study, yield and yield components were measured in dry land and irrigation trials to identify the factors determining yield potential and sample uniformity. Although genotype x environment interactions were important contributors to variation in the yield determining factors, genotype and position of the kernel in the spike had the major influence on kernel weight. Large differences in kernels spikelet–1 and kernel weight indicated that these two variables were responsible for yield adjustments to stress during the spikelet and kernel development phase. Kernels from the lower and middle section of the spike and the proximal (A and B) floret positions were heavier than those from the upper spike section and the distal (C and D) floret positions. Artificially reducing spikelet numbers increased weight of the remaining kernels but only under dry land conditions. Weight of kernels in the C position was increased 22% when ovules of the proximal florets were removed during head initiation. These observations indicate that because growing season moisture availability is extremely variable on the Canadian prairies, an ability to compensate for limitations or excesses in sink size as the season progresses must be bred into cultivars if grain yield is to be maximized. This means that successful genotypes tend to have higher than average values for all yield components rather than one exceptionally high component and uniform seed size is difficult to achieve.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
GROWING SEASON soil moisture availability and evapotranspiration are primary environmental factors limiting the grain yield potential of crops grown in semiarid climates. This is particularly so on the Canadian prairies where natural variation in weather among locations and years produces a wide range in the timing and intensity of drought stress (Lehane and Staple, 1965; Robertson, 1974; Williams et al., 1975; Campbell et al., 1988; Entz and Fowler, 1989) with the result that dry land crops are exposed to both intermittent and terminal drought stress (Domitruk et al., 2000). Stored soil water is usually not sufficient to support the crop throughout the growing season in this region and the amount of growth is generally reflective of the pattern of water availability. Consequently, because growing season precipitation is variable in both timing and amounts, it is difficult to optimize plant breeding and crop management strategies for this region.

While the wheat plant is more sensitive to stress during certain developmental stages (Baier and Robertson, 1967; Day and Intalap, 1970; Fischer, 1973; Passioura, 1977; Warrington et al., 1977; Musick and Dusek, 1980; Entz and Fowler, 1988), drought stress retards the formation of the yield component that is most actively developing at the time of stress (Aspinall, 1984; Entz and Fowler, 1988), and without yield component compensation a crop will be unresponsive to improved water conditions later in the growing season (Duggan et al., 2000). Consequently, the ability of winter wheat plants to respond to changes in environmental conditions can be characterized by the limitations imposed by stress at different stages of growth and is reflected in (i) plants m–2, (ii) spikes plant–1, (iii) spikelets spike–1, (iv) kernels spikelet–1, and (v) kernel size.

Early season moisture is required to set up a high winter wheat grain yield potential and ample precipitation during grain filling is necessary to secure yield. On the Canadian prairies, a rapid early season response to moisture normally results in prolific tillering and a large proportion of the total above ground biomass of winter wheat being accumulated by anthesis (Darroch and Fowler, 1990). Rapid growth under cool temperatures during April and early May often consumes much of the plant available stored water with the result that differences in drought stress are related to the volume and distribution of precipitation throughout the remainder of the growing season (Entz and Fowler, 1989; Domitruk et al., 2001). Winter wheat has the ability to produce large numbers of tillers and tiller maintenance determines the plants' ability to retain high early season yield potential (Johnston and Fowler, 1992; Duggan et al., 2000). Early season drought stress causes cereal plants to lose potential yield through tiller death and prolific tillering immediately before the onset of drought has been positively associated with tiller death that has variable consequences for grain yield (Darwinkel, 1978; Innes et al., 1981; Johnston and Fowler, 1992; Domitruk et al., 2000). Excessive tiller die-back results in reduced ability to produce assimilate, smaller sink capacity and limited ability to respond to subsequent improvements in growing season weather conditions.

Kernel weight is usually negatively correlated with the number of kernels spike–1 (Fischer et al., 1977). Kernel weight can also be reduced when the length of the grain-filling period is restricted by drought and heat stress after anthesis (Warrington et al., 1977). Conversely, cereals possess a degree of developmental plasticity that allows for increases in kernels spike–1 or kernel weight to compensate for losses in tiller number if environmental conditions and assimilate availability improves after the stem elongation stage (Kirby and Jones, 1977). This yield component compensation is often responsible for a lack of grain yield response to changes in environmental conditions over the growing season. However, the scope for compensation, particularly through increases in kernel weight, is generally not sufficient to compensate for losses in kernel number resulting from tiller mortality (Johnson and Kanemasu, 1982; Entz and Fowler, 1988).

Water stress interferes with both the production of photoassimilate and the import of assimilated material into the developing grains. Yield potential in cereals is limited by the capacities of both the sink (Fischer et al., 1977; Slafer and Savin, 1994) and the assimilatory source (Fischer and HilleRisLambers, 1978; Duggan et al., 2000). However, there is debate over whether independent manipulation of either source or sink capacity is a plausible approach toward developing cultivars with higher grain yields for water limited conditions (Fischer et al., 1977; Blum et al., 1988). Several workers have reported that kernel growth increased in response to both artificial reductions in sink capacity or artificially increased assimilatory capacity (Bremner and Rawson, 1978; Simmons et al., 1982; Blum et al., 1988; Ma et al., 1990; Blade and Baker, 1991). Conversely, kernel growth has been shown to be restricted by reductions in assimilation near anthesis (Jedel and Hunt, 1990) indicating that assimilate supply controls expression of kernel weight. Prevailing environmental conditions, in particular the availability of moisture, during grain filling is therefore a significant consideration when assessing the source–sink relationship in cereal yield composition.

In dry land trials in eastern Colorado (Shanahan et al., 1984) and Saskatchewan (Entz and Fowler, 1990), grain yields of winter wheat were reported to be positively associated with kernel number. This positive association indicates that grain yield is predominantly sink limited (Fischer et al., 1977; Borrás et al., 2004) and increasing the number of kernels that comprise the sink even at the expense of kernel weight can theoretically increase grain yield. However, Blade and Baker (1991) expressed the view that sink-related limitations to spring wheat yield are not important in a semiarid environment where drought stress typically limits photoassimilation. Entz and Fowler (1989) also suggested that, if winter wheat yield is limited by postanthesis assimilate supply, it is due to drought related reductions in photoassimilation rather than low photosynthetic capacity. Furthermore, van Herwaarden et al. (1998) demonstrated that wheat yields can be source limited in postanthesis drought stress conditions, especially under high nitrogen fertilizer conditions, primarily via the production of shriveled kernels.

The increased grain yield of modern day wheat genotypes has usually been achieved by increasing kernel number, which has been associated with a greater contribution by the more distal (C and D) kernels (Fig. 1) in the spikelet (Miralles and Slafer, 1995) and reduced kernel weight (Siddique et al., 1989; Slafer and Andrade, 1989). However, because distal kernels are typically smaller and lighter than the more proximal (A and B) kernels (Bremner and Rawson, 1978; Stoddard, 1999; Calderini and Ortiz-Monasterio, 2003), they reduce the average kernel weight of a genotype and make a disproportional contribution to grain lost during harvest, cleaning, and subsequent dockage.


Figure 1
View larger version (66K):
[in this window]
[in a new window]
  		Fig. 1. Wheat spikelet displaying the location of the various floret/kernel positions.

 
The objectives of this study were: 1) To determine the effects of different source-sink relationships on the kernel weight of genotypes with widely differing yield component structure under irrigated and water stress conditions and identify the major factors responsible for variation in kernel size of grain samples. 2) To investigate the kernel weight architecture of the spike. 3) To identify the role played by kernels from various spike positions in determining the plant's ability to adjust to variable environmental stresses and determine the contribution of these responses to variability in final kernel weight.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Yield, Yield Components, and Contribution to Yield from Different Spikelet and Floret Positions
Six winter wheat genotypes (‘CDC Falcon’, ‘Norstar’, ‘CDC Osprey’, ‘CDC Ptarmigan’, ‘Oro Blanco’ and the breeding line TW96516) were selected for evaluation in these experiments on the basis of adaptability and differences in yield component structure identified in earlier trials conducted by the University of Saskatchewan winter wheat project. Oro Blanco reaches anthesis first, followed 2 d later by CDC Osprey and CDC Falcon. CDC Ptarmigan and TW96516 reach anthesis the next day followed 2 d later by Norstar (data not shown). These differences in anthesis were maintained so that cultivars reached physiological maturity after a similar grain filling duration, both in calendar and thermal time. The trials were grown on irrigation (Dark brown Chernozem, clay loam), partial irrigation and dry land (Dark brown Chernozem, clay) at Saskatoon in 1998–1999 and 1999–2000 and on dry land at Yorkton (Black Chernozem, loam), Saskatchewan in 1998–1999. The irrigation trial received 7.5 and 13.5 cm added water before heading in 1999 and 2000, respectively, and 10 and 11.5 cm added water after heading in 1999 and 2000, respectively. The partial irrigation trial received 2.0 and 4.0 cm added water before heading in 1999 and 2000, respectively, and 2.0 cm added water after heading in both 1999 and 2000.

The experimental design of all trials was a randomized complete block with three replicates. Plot size was 1.2 x 6 m with row spacing of 20 cm and a seeding rate of 250 seeds m–2. All trials were direct seeded in late August or early September into standing canola or cereal stubble. Phosphate (11–51–0) was applied with the seed and nitrogen (34–0-0) was top-dressed as soon as equipment was able to travel on the field in the spring at rates based on soil test recommendations for maximum grain yield. Broadleaf weeds were controlled by recommended post-emergence herbicides applied at the recommended rates and times.

Yield components were determined for each genotype from samples collected from a 25- x 100-cm area of each plot in each of two replicates in each trial. Five spikes were also selected at random from each plot to determine kernel size and contribution to yield by different spikelet and floret positions. Each spike was divided into upper, middle, and lower third sections on the basis of the number of spikelets spike–1. Kernels were removed and segregated according to their spikelet and floret position, i.e., A, B, C, or D (Fig. 1). Kernels for each position were then counted, weighed, and kernel weight for each position determined. In addition to the yield component sample area, approximately 30 cm was trimmed from each end at maturity and the plots were then direct cut with a self-propelled small plot combine for grain yield and average kernel weight determinations. The outside two rows of each plot were not harvested. Exact plot lengths were recorded before harvest.

Effect of Spikelet and Floret Removal
The winter wheat genotypes ‘AC Tempest’, CDC Falcon, Norstar and ‘Norwin’ were grown on partial irrigation to alleviate drought stress at Saskatoon in 1998–1999 and 1999–2000 and dry land at Yorkton in 1998–1999 and Saskatoon in 1999–2000 with the same plot size and management practices as outlined above. AC Tempest and CDC Falcon reached anthesis first, followed 2 d later by Norwin and another 2 d later by Norstar. The experimental design was a randomized complete block with two replicates. Two methods were used to vary assimilate availability to the developing kernels in these trials.

  1. Three spikelet reduction treatments (a check and 25 and 50% spikelet removal) were imposed at heading (Zadoks' growth stage 60; Zadoks et al., 1974) on each of 10 spikes from each genotype in each replicate. Forceps were used to remove selected spikelets starting at the base of the spike.
  2. The A and B ovules were removed from each of 10 spikes of each genotype in each replicate at the head initiation phase (Zadoks' growth stage 50) and at anthesis (Zadoks' growth stage 60) in the 1999–2000 trials to increase the assimilate availability to kernels in the C and D positions. Kernels from the mature spikes were hand-sorted into A, B, C, and D kernel positions in the spikelet (Fig. 1), counted, and weighed. The plots were harvested for grain yield determinations as outlined above.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
Yield and Yield Components
The dry land trials and the opportunity to provide supplementary irrigation on selected sites provided a wide range of environmental conditions in these studies. The differences in environmental potential were reflected in the average grain yield, which was 5.2 Mg ha–1 on dry land at Yorkton (Table 1) and 6.0, 5.5, and 3.5 Mg ha–1 under irrigation, partial irrigation, and dry land, respectively, at Saskatoon. Significant differences among genotypes were observed in all trials (not shown) with CDC Falcon producing the highest yield under irrigation and CDC Ptarmigan most commonly producing the highest yield on dry land and partial irrigation. Norstar is quite tall and suffered a large yield penalty because of lodging under irrigation in both years. These differences in levels of grain yield expression in the different environments resulted in a significant trial x genotype interaction (Table 2).


View this table:
[in this window]
[in a new window]
  		Table 1. Grain yield (Mg ha–1) of six wheat genotypes grown in seven trials in 1998–1999 and 1999–2000.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Percentage of total sums of squares due to differences among trials, genotypes, and other sources (residual) for grain yield (Mg ha–1) and yield components of six winter wheat genotypes grown in seven trials in 1998–1999 and 1999–2000.

 
Analyses of variance indicated that differences due to trials and genotypes were significant for kernels m–2, kernel weight, kernels spike–1, and spikes m–2 (Table 2). Differences in spikelets spike–1 due to trials were nonsignificant (large replicates within trials effect), while differences due to genotypes were significant. Yield components of the genotypes displayed considerable variation, ranging from Oro Blanco, which produced a large number of spikes m–2 but a low number of small kernels spike–1, to TW96516, which produced few spikes but compensated by producing a larger spike with significantly heavier kernels (Table 3). The high yield potentials of CDC Falcon, CDC Ptarmigan, and TW96516 were achieved through different combinations of yield components indicating that a strong ability for compensation should be expected when selection pressure is focused on individual components and final grain yield.


View this table:
[in this window]
[in a new window]
  		Table 3. Mean values for yield components of six wheat genotypes grown in seven trials.

 
The trial x genotype interaction was not significant for kernel weight, kernels spike–1, and spikelets spike–1 (Table 2), indicating the genotype responses were similar for these characters when grown under a wide range of environments. In contrast, the trial x genotype interaction was significant for kernels m–2 and spikes m–2, indicating that environmental differences did not have a uniform effect on the expression of genotypic differences for these two yield components. Differences due to trials accounted for the largest percentage of total sums of squares for spikes m–2, indicating that this yield component had a large influence on the plants' ability to respond to variation in the growing season environments experienced in these studies.

In the analyses of variance, the residual component accounted for a large proportion of the total sums of squares for all the yield components except kernels m–2 (Table 2). This large residual variability could be due to any number of factors such as systematic and random measurement errors, sample heterogeneity, other experimental errors, and their interactions. Because repeatability of measurements is necessary for effective and efficient selection (Youden, 1963; Baker and Campbell, 1971; de la Roche and Fowler, 1975), further analyses were conducted to identify the reasons for the large residual sums of squares associated with measures of kernel weight and kernels spike–1.

Contribution to Yield from Different Spikelet and Floret Positions
Detailed analyses of variance for kernel numbers and weights indicated that, in addition to the significant differences due to trials and genotypes noted above, there were significant differences for these characters among the upper, middle, and lower sections of the spike and the A, B, C, and D kernel positions within the spikelet (Table 4). There were also a number of highly significant interactions among genotypes, trials, and kernel position for number of kernels in the different sections of the spike and for kernel weight and number within the spikelet.


View this table:
[in this window]
[in a new window]
  		Table 4. Percentage of total sums of squares due to differences among trials, genotypes, kernel position, and other sources for kernel weight (mg) and number of kernels in the upper, middle, and lower section of the spike and in the A, B, C, and D floret positions of six wheat genotypes grown in seven trials.

 
Kernel weights varied considerably along the length of the spike and kernels from the lower and middle section of the spike were heavier than those from the upper section (Table 5). The upper section of the spike also produced fewer kernels in the C and D spikelet positions with the result that the total number of kernels was higher in the lower and middle sections and the kernels from the upper section only contributed 24% of the total grain yield. Both the weight and number of kernels in the C and D floret positions was much lower with the result that, while they only contributed 18 and 2% of the total grain yield, they introduced a great deal of variability into the sample kernel weight.


View this table:
[in this window]
[in a new window]
  		Table 5. Average kernel weight, kernel number, and yield contribution for the lower, middle, and upper sections and the A, B, C, and D florets of the spike of six wheat genotypes grown in seven trials.

 
Nonsignificant trial x genotype and genotype x spike position interactions (Table 4) indicated that changes in kernel weight followed similar patterns for the different spike positions in the wide range of environments and genotypes evaluated. In contrast, significant interactions indicated that the genotypic differences in kernel numbers at the different spike and floret positions and kernel weight at the different floret positions were not as consistent or predictable in different environments. However, the variability due to these interactions was small compared with the main genotype and floret position effects, where the average weight of kernels for the A, B, C, and D florets ranged from 34, 33, 27, and 13 mg, respectively, for Oro Blanco to 45, 45, 36, and 24 mg, respectively, for TW96516. A large residual variance component also indicated that there was still considerable unexplained variability for kernel weight in the different sections of the spike.

The variation in weight of kernels from the different sections of the spike and the percentage contribution toward grain yield reflect the availability of assimilate to the developing kernels throughout grain filling. The numbers of kernels spikelet–1 are determined between the time of the head initiation phase and shortly after anthesis (Kirby and Appleby, 1987; Slafer and Rawson, 1994). Increased assimilate availability during this phase increases potential grain yield though increased kernel set. Because spikelets in the upper third of the spike have access to assimilate only after those in the lower two thirds of the spike, florets in the upper third are unable to set as many kernels as those lower in the spike. The ability to fill kernels is also influenced by assimilate availability, although generally from the period shortly after anthesis to maturity (Fischer, 1985). Increasing assimilate availability during this phase is expected to increase grain yield through greater weight of kernels from the upper spikelets on the spike and the distal positions of the spikelet.

Effect of Spikelet and Floret Removal
There were significant differences in the grain yield and kernel weight of the genotypes selected to evaluate the effect of assimilate availability on kernel size (Table 6). As in the first set of experiments, the high yield potential of CDC Falcon was evident. There was also considerable variation in the yield structure of these genotypes, ranging from Norwin, which typically produces a large number of spikes per m2 and a small number of kernels spike–1 (Duggan et al., 2000), to AC Tempest, which produces fewer spikes per m2 but compensates by producing heavier kernels than the other genotypes.


View this table:
[in this window]
[in a new window]
  		Table 6. Mean grain yield and kernel weight of four wheat genotypes grown in four different trials. Measurements made on machine-harvested samples after cleaning, i.e., smaller kernels removed as clean-out.

 
There were fewer kernels in the D floret position and the kernels in the C position were much smaller than the A and B kernels (Table 5), which agrees with the observations reported by Miralles and Slafer (1995, 1996). Miralles and Slafer (1995) also showed that, under dry land conditions, reducing spikelet numbers 7 d after anthesis increased the kernel weight in dwarf but not semi-dwarf and standard height wheat lines. In the present study, artificially reducing spikelet numbers by 25 and 50% to increase assimilate supply to the remaining kernels increased kernel size under dry land conditions but did not have a significant (P > 0.05) influence on kernel size under irrigation (Fig. 2) indicating that genetic restrictions imposed by kernel size may limit grain yield potential when growing conditions are favorable. The removal of spikelets took place at the commencement of flowering (Zadoks growth stage 60; Zadoks et al., 1974), and it may be that irrigation allowed the control treatment to assimilate more carbon while the dry land crops were unable to fill the extra kernels of the control to the same extent as the 25 and 50% spikelet removal treatments. These observations indicate that the period between heading and anthesis is important in determining potential kernel weight (Fig. 2 and 3) and increases in kernel number under even a slight moisture stress after anthesis begins to reduce kernel weight.


Figure 2
View larger version (41K):
[in this window]
[in a new window]
  		Fig. 2. Average kernel weight of four winter wheat genotypes grown under partial irrigation and dry land conditions when 0 (control), 25, and 50% of the spikelets were removed at anthesis.

 

Figure 3
View larger version (16K):
[in this window]
[in a new window]
  		Fig. 3. Average weight of kernels in the C floret position for four winter wheat genotypes grown under partial irrigation and dry land conditions in 1999–2000 when the A and B floret ovules were removed at the initiation of head phase and anthesis and when 25, and 50% of the spikelets were remove at anthesis.

 
Average kernel weight for the four genotypes considered appeared to reach a maximum between 34 and 36 mg when 50% of the spikelets were removed (Fig. 2), further suggesting that there was an upper limit on kernel size. The absence of a significant treatment by floret/kernel position interaction (P > 0.05) for kernel weight indicated that the size of the kernels in the C floret positions did not change relative that of the A and B positions (data not shown) suggesting that the weight of C kernels was limited by the size of the A and B kernels or there was some other factor limiting kernel size. Reduction of the spikelet number by 25 and 50% produced small, but nonsignificant (P > 0.05), increases in the weight of the C kernels (Fig. 3). Calderini and Reynolds (2000) showed an increase in the weight of distal (i.e., C and D) kernels of synthetic hexaploid wheat genotypes when the proximal kernels were removed. In the present study, removal of the A and B ovules at the initiation of heading phase significantly (P < 0.05) increased the mature kernel weight of the C kernels by 22% but this advantage was not observed when removal of the A and B florets was delayed until anthesis.

Yield Structure and Kernel Potential
High yield was achieved by the genotypes considered in this study through different combinations of yield components (Tables 1 and 3). Significant negative genotypic correlations (calculated on the basis of genotypic averages over all trials) between spikes m–2 and both kernels spike–1 (r = –0.85, P = 0.03) and kernel weight (r = –0.80, P = 0.05) indicated that there was a strong ability for compensation among the individual components. While spikes m–2, kernels spike–1, and kernel weight were all influenced by both trial and genotype, only spikes m–2 was affected by their interaction. This did not come as a surprise because a significant environmental correlation (calculated using trial averages) between spikes m–2 and grain yield (r = 0.85, P = 0.02) indicated that changes in spikes m–2 was the primary method by which the plants responded to differences in environmental stress. These observations also suggest that kernel weight and kernels spike–1 are grain yield components that can be manipulated without compromising adaptation as long as tiller production remains high enough for the plant to respond to the variable growing season drought stress experienced on the Canadian prairies.

Kirby and Jones (1977) found that barley (Hordeum vulgare L.) plants detillered before spike elongation produced spikes with more kernels of greater weight than plants that were allowed to tiller freely indicating that a higher number of spikelets spike–1 is one way in which cereal genotypes with reduced tillering ability are able to effectively compensate. An ability to compensate may be the result of genotypes with lower tillering capabilities having relatively greater assimilate available for the spike apex shortly after the terminal spikelet stage. Conversely, favorable growing conditions immediately after the terminal spikelet stage may result in a reduction in average number of spikelets spike–1 as tillers that would have failed under dry conditions survive and produce a spike with fewer spikelets spike–1 than tillers that were initiated early in the plant's development. However, the number of spikelets spike–1 remained relatively constant in the present studies (Table 3) indicating that this was not a primary mechanism for yield compensation in wheat. In contrast, there were large differences in kernel weight and kernels m–2 for the dry land and irrigation environments sampled.

A major difference between wheat and barley species is that barley only has one floret spikelet–1, while a spikelet of wheat contains several florets (Kirby and Appleby, 1987). Each spikelet primordium in wheat can produce eight to 10 floret primordia of which a maximum of four to six will set kernels. The numbers of kernels spikelet–1 are determined between the time of the head initiation phase and shortly after anthesis (Kirby and Appleby, 1987; Slafer and Rawson, 1994). The plant is especially sensitive to water stress during pollen meiosis and increased assimilate availability during this phase increases potential grain yield though increased kernel set (Kirby and Appleby, 1987). The large differences in kernels spikelet–1 and kernel weight for trials in the present study (Table 3) indicate that that these two variables were responsible for grain yield adjustments to stress during the spikelet and floret development phase.

Changes in kernel weight followed similar patterns for the different spike positions for the wide range of environments and genotypes evaluated and were similar to the responses shown by Slafer and Savin (1994). Significant differences among the upper, middle, and lower sections of the spike and the A, B, C, and D kernel positions within the spikelet were associated with a large amount of variability in kernel weight and kernels spike–1 (Table 4). Kernel weight varied depending on both its position within the spikelet and along the length of the spike with heavier kernels located at the basal positions of spikelet and the middle sections of the spike (Table 5). The weight of kernels in the C and D floret positions was much lower and, while they only contributed 18 and 2% of the total grain yield (Table 5), they were responsible for a great deal of variability in the sample kernel weight. Because the C and D kernels are typically lighter than the more proximal (A and B) kernels, they make a considerable contribution to "clean-out" during harvest and dockage at the time of grain delivery.

Maximum kernel number is determined by the length of time before anthesis (Fischer, 1985), and maximum kernel weight is fixed by the period shortly after anthesis since this is when endosperm cell number is determined (Cochrane and Duffus, 1983). A significant increase in the mature weight of the C kernels when the A and B ovules were removed at the head initiation phase (Fig. 3) and the loss of this advantage when removal of the A and B florets was delayed until anthesis supports the conclusion that assimilate supply before anthesis determines maximum potential kernel weight, possibly by limiting endosperm cell numbers. There were fewer kernels formed in the D floret position, and the kernels in the C position were smaller than the A and B kernels regardless of the level of stress experienced in the environments sampled. Reduction of sink strength by removing 25 and 50% of the spikelets at anthesis to increase assimilate supply to the remaining kernels demonstrated that kernel weight does not currently reach its potential under dry land conditions on the Canadian prairies (Fig. 2). In contrast, artificially reducing spikelet numbers did not influence kernel weight under irrigated conditions indicating that restrictions imposed by kernel weight, or sink size, may limit grain yield when growing conditions in this region are favorable.

The wheat plants' ability to compensate among yield components and the need to ensure that the plant can adjust to changes in moisture availability throughout the growing season presents a dilemma to the plant breeder trying to select for uniformity in kernel size while maximizing grain yield potential. The results of this study suggest that a winter wheat plant architecture that has increased spikelets spike–1, larger kernels, and fewer kernels in the C and D spikelet positions could increase grain yield potential while producing a more uniform grain sample in variable semiarid environments such as western Canada. However, weight is usually negatively correlated with the number of kernels spike–1 (Fischer et al., 1977). Also, spikes m–2 is the primary method by which the plants respond to differences in environmental stress and the inverse relationship between spikes m–2 and both kernels spike–1 and kernel weight suggests that too much emphasis on any one yield component may compromise adaptation. Support for this observation can be drawn from the present study where the highest yielding genotypes across environments tended to have higher than average values for all yield components rather than one exceptionally high component.

The wide range in the kernel size of genotypes considered in this study (10.4 mg, Table 3) indicates that there is considerable genetic variability for this character available for selection in breeding programs. However, we have shown that winter wheat genotypes grown on the Canadian prairies are currently not achieving their maximum potential kernel weight under dry land conditions with the result that differences due to environmental factors are very large (7.5 mg, Table 3). Also, selection for increased kernel weight would be expected to result in a proportional increase in weight of the C and D kernels. This may reduce "clean-out" caused by small kernels but the available data suggests that the C and D kernels will only be about 76 and 45% of the weight of A and B kernels (Table 5), and if they remain in greater numbers, they may produce a less uniform grain sample. Selection for increased spikelets spike–1 may shift assimilate from the C and D kernels to A and B kernels, but kernels in the upper section of the spike would be similar in size to those produced in the C floret position (Table 5).

These observations emphasize the challenges faced by breeding programs targeting high grain yield potential and uniform seed size in wheat production regions that experience both intermittent and terminal drought stress. Because assimilate availability dictates the formation of the most actively developing yield component and growing season moisture availability is extremely variable on the Canadian prairies, an ability of later developing yield components to compensate for the limitations or excesses in sink size as the season progresses must be bred into cultivars if grain yield is to be maximized. This means that the genetic potential of seed size, which is the last yield component to form, must be as large as or larger than that required for the target market when premiums exist. However, if uniform kernel size is a major economic consideration, the most practical alternative available to ensure sample uniformity may only be through seed size segregation at the time of grain cleaning for market.


   ACKNOWLEDGMENTS
 
The excellent technical assistance of Ray Hankey is greatly appreciated.

Received for publication June 13, 2005.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Duggan, B. L.
Right arrow Articles by Fowler, D. B.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Duggan, B. L.
Right arrow Articles by Fowler, D. B.
Agricola
Right arrow Articles by Duggan, B. L.
Right arrow Articles by Fowler, D. B.
Related Collections
Right arrow Crop Growth and Development
Right arrow Wheat
Right arrow Plant and Environment Interactions

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