Euphytica 112: 23–31, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.
23
Genetics of wheat starch B-granule content F.L. Stoddard Quality Wheat CRC Ltd, Locked Bag No 1345, P.O. North Ryde, NSW 2113 and Plant Breeding Institute, Woolley Bldg A20, The University of Sydney, NSW 2006, Australia Received 3 December 1998; accepted 16 July 1999
Key words: A granules, B granules, quantitative analysis, starch quality, triploid endosperm, Triticum aestivum
Summary Two lines of hexaploid wheat were crossed and the basic generations of parent, F1, F2 and back-cross were sown in a controlled-environment chamber. Fresh F1 and back-cross grains were generated, so the material could be handled either as the standard set of basic generations on a whole-plant basis, or as an extended set on an embryo or endosperm basis. The experiment was repeated. Mature grains were harvested and the starch particle size distribution was analysed in 3284 grains from 111 plants. Means and variances were partitioned into additive, dominance and interaction components. Grains from cross-pollinations had B-granule contents between parental values, rather than of the maternal parent, indicating an involvement of the grain genotype. Quantitative models based on endosperm genotype gave a better fit to the data than those based on embryo genotype. The difference in starch B-granule content between the parents was largely due to additive genes. Dominant genes were also indicated, with the first dose in the triploid endosperm having a large effect while the second dose had little or none. Non-allelic interactions were significant in the second experiment where the use of more types of backcross made them more detectable. There were also small and significant residual effects of the maternal plant in the first experiment, attributed to the vigour of the F1 mother plant and to the cytoplasm of Sunco. Narrow-sense heritability was low, between 0.05 and 0.18 depending on the generation. Transgressive segregation was not found, suggesting that all alleles tending to increase the B-granule content were found in the Sunco parent and none in ME71. There was also no detectable heterosis in this character. The results show that breeding and selection for a low B-granule content should be possible but a further reduction will require new and complementary genes.
Introduction Starch in wheat and some of its relatives is deposited in two sizes of granule, whereas in most other plants there is only one size class of starch granule. Each amyloplast of the mature wheat endosperm contains a single large A-granule, initiated soon after anthesis, and a variable number of smaller B granules, initiated late in grain filling (Parker, 1985). The B granules are undesirable for several end uses. Most importantly, in a rapid wet process such as washing gluten and starch from flour, they do not precipitate quickly and are carried away in the flow, representing both a waste of resources and a cost for treatment of the effluent. Secondly, their higher surface:volume ratio makes them more absorbent of water than the A gran-
ules, so an excess content of B granules can depress loaf volume or require a longer baking time (Hoseney et al., 1971; Lelievre et al., 1987). Examination of ten bread wheat cultivars showed that they varied in the reduction of starch B-granule content following a heat shock of 35◦ during grain filling (Blumenthal et al., 1994) and the significance of this variation was confirmed in a later survey of 45 cultivars (Blumenthal et al., 1995). Cultivar ‘Sunco’ had a high B-granule content (33% of starch in nonstressed control conditions) and ‘ME71’ a low content (22%). These two varieties were therefore chosen for experiments on the genetics of B-granule content in hexaploid wheat. Analysis of the endosperm of the early generations following a cross is complicated by the fact that
24 the grain is a generation further advanced than the parent plant. In a segregating generation such as the F2 population, a small sample of grain could therefore bias the result and unequal samples from similar plants would also introduce a bias. For these and related reasons, quantitative geneticists have extended the models formulated by Mather in 1949 (Mather & Jinks, 1971, 1977) to triploid endosperms. The key difficulties have involved analysis of the allelic and non-allelic interactions that may be expressed in triploid tissues. The models have been developed to a high degree of refinement (Gale, 1976; Mo, 1987; Bogyo et al., 1988; Pooni et al., 1992), but the number of cases where these models have been applied is very small. Discrepancies between the scales used by the various authors, different nomenclatures and errors in the published tables have further confused the issues. This paper reports on the application of quantitative genetic analysis to the B-granule content of wheat endosperms.
for crossing without emasculation: distal florets and spikelets were removed, awns cut from glumes and a crossing bag placed over the heads, to provide a proper control for comparison with the manually pollinated grains. The control and three treatments (prepared for emasculation, pollinated with ME71, pollinated with Sunco) were applied in random order to the parent and F1 plants. Mature heads were harvested from each plant separately. To reduce non-genetic variance, the bottom three and top five spikelets were excluded from analysis and only the grains from the proximal two florets in each of the remaining spikelets were included (Stoddard, 1999). This selection coincided with the florets that were manually pollinated in the crossed heads. Twenty to twenty-five grains were analysed from each of the F2 and back-cross plants. The F2 grains from the F1 plants as well as the grains from all of the manual cross-pollinations were cut in half and the endosperm end was analysed, leaving the embryo half for planting. In total, 2339 grains were examined.
Materials and methods
Experiment 2
Plant materials and growing conditions
In the second experiment, four grains of each parent, F1 hybrid and reciprocal F1 (RF1) hybrid, harvested from the above plants, were sown in identical conditions. Each of the four genotypes was crossed with all of the others, so all possible F1 and backcross hybrids were generated. In addition, the 13 F2 grains with the highest B-granule contents and the 13 with the lowest were planted, and ten grains were analysed from each of these plants. In total, 945 grains were examined. Thirty-six grains of each parent were cut in half and starch particle size distribution was determined in both halves.
Plants of hexaploid wheat (T. aestivum L. emend Thell.) cv ‘Sunco’ and a landrace known as ‘ME71’ were grown in a glasshouse and crossed in both directions. The next season, parents and F1 hybrids were again grown and fresh F1 grain prepared, together with back-crosses to each parent, and all grains were harvested. Particle size analysis was undertaken on some grains, as described below. Where appropriate, grains were cut in half so the endosperm half could be analysed and the embryo half planted. The main experiment was conducted in a 90 cm × 180 cm × 180 cm controlled environment growth chamber (Thermoline, Smithfield, NSW, Australia) with 18◦ days, 13◦ nights and a twelve hour photo/thermoperiod. Pots (15-cm) were filled with a soil-free potting mix (30 L washed sand, 20 L peat, 1 kg dolomite, 400 g lime, 300 g fertilizer mix) and one grain planted per pot. Experiment 1 Four pots of each parent, nine F1 hybrids, 25 F2 grains and 15 of each of the two back-crosses were sown. Further F1 hybrids and back-crosses of F1 to parents were prepared. In addition, selected heads of the parent and F1 plants were subjected to preparation
Starch preparation Each grain was weighed to the nearest 0.1 mg, cut in half if appropriate, crushed with smooth-jawed pliers, put into a 2-mL Eppendorf microfuge tube with 0.5 mL of 0.5 M NaCl and soaked overnight at 4◦ C. The next day this mixture was ground with an Eppendorf pestle attached to an electric drill press until the gluten formed a tight ball, the starch slurry was decanted through a 0.2 mm mesh sieve into a fresh microfuge tube and the residue was ground in a fresh 0.5 mL of NaCl solution. The grinding was repeated, the slurry added to the first, and a third cycle was conducted as well (Stoddard, 1999). For experiment 1, the starch was precipitated by 2 min centrifugation at
25 Table 1. Coefficients used to calculate components of means from triploid endosperm data. Coefficients [c], [am ] and [dm ] were also used for components of means based on diploid embryos Generation
m crossing
[a]
[d1]
[d2l
[aa]
[ad1]
[ad2]
[d1d1]
[d2d2]
[d1d2]
[c]
[am ] [dm ]
P1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 –1 –1 1/3 –1/3 0 0 0 0 0 1/3 1/3 2/3 –1/3 –1/3 –2/3 0 0 1/2 –1/2
0 0 0 0 1 0 1/4 1/4 1/4 1/4 1/4 0 0 1/2 1/2 1/2 0 1/8 1/8 1/8 1/8
0 0 0 0 0 1 1/4 1/4 1/4 1/4 1/4 1/2 1/2 0 0 0 1/2 1/8 1/8 1/8 1/8
1 1 1 1 1/9 1/9 0 0 0 0 0 1/9 1/9 4/9 1/9 1/9 4/9 0 0 1/4 1/4
0 0 0 0 1/3 0 0 0 0 0 0 0 0 1/3 –1/6 –1/6 0 0 0 1/16 –1/16
0 0 0 0 0 –1/3 0 0 0 0 0 1/6 1/6 0 0 0 –1/3 0 0 1/16 –1/16
0 0 0 0 1 0 1/16 1/16 1/16 1/16 1/16 0 0 1/4 1/4 1/4 0 1/64 1/64 1/64 1/64
0 0 0 0 0 1 1/16 1/16 1/16 1/16 1/16 1/4 1/4 0 0 0 1/4 1/64 1/64 1/64 1/64
0 0 0 0 0 0 1/16 1/16 1/16 1/16 1/16 0 0 0 0 0 0 1/64 1/64 1/64 1/64
1 1 –1 –1 1 –1 1 1 –1 –1 –1 1 –1 1 1 –1 –1 1 –1 –1 1
1 1 –1 –1 1 –1 0 0 0 0 0 0 0 1 0 0 –1 0 0 0 0
(Sunco) thinned, bagged head P2 (ME71) thinned, bagged head F1: P1 × P2 P2 × P1 (RF1) F2: F1 selfed F1 × RF1 RF1 selfed RF1 × F1 thinned, bagged head B1: F1 × P1 RF1 × P1 P1 × F1, P1 × RF1 B2: F1 × P2 RF1 × P2 P2 × F1, P2 × RF1 F3: F2 selfed RF2 selfed B1F2 (RF1 × P1, selfed) B2F2 (P1 × F2, selfed)
0 1 0 1 1 1 0 1 0 1 1 1 1 1 1 1 1 0 0 0 0
5500 × g, resuspended in 2% sodium dodecyl sulphate (SDS), centrifuged again, washed twice in water and once in ethanol, and then frozen in water until analysis. For experiment 2, an additional step of washing through 1 mL of 50% (w/w) CsCl was included before the SDS wash, and samples were dried over silica gel instead of being frozen in water. Particle size analysis Particle size distribution was determined on subsamples from each starch suspension in experiment 1 using the stirred small-sample cell in a Malvern 2600C laser-diffraction analyser (Malvern Instruments Ltd, Malvern, UK) connected to an 8086 desktop computer. Data were retyped into a spreadsheet for manipulation and statistical analysis. Particle size analysis in experiment 2 was done in a Malvern Mastersizer S laser-diffraction analyser (Malvern Instruments Ltd, Malvern, UK) (connected to a Pentium desktop computer) using the entire sample in the flow through small-volume sampler, the 300 mm Reverse Fourier lens, 2.4 mm beam length and optim-
0 0 0 0 0 0 1 1 1 1 1 1 1 0 1 1 0 1/2 1/2 1/2 1/2
ised for the refractive index of starch in water. This analyser provided four times faster sample throughput than the older model, three times as many size classes, an extra digit of accuracy within each size class and automated transfer of data to a useful spreadsheet format. Residual samples from the previous part of the experiment (616 in total, from P1, P2, F1, F2, B1 and B2 generations) were also put through this analyser to provide a basis for comparison of the two machines. The correlation between the two data sets was r = 0.80, the slope was 0.900 and the coefficient of variance from the Mastersizer S averaged 69% of that of the 2600C. Particles less than 9.8 µm in diameter were considered as B granules and those between 9.8 and 49.8 µm were considered as A granules (Blumenthal et al., 1994, 1995, Stoddard, 1999). Statistical analyses Means and variances were partitioned according to the weighted least-squares methods of Mather & Jinks (1971, 1977) using their coefficients for diploid tis-
26 Table 2. Coefficients used to calculate components of variances from triploid endosperm data Generation
E1
E2
A
D1
D2
AD1
AD2
D1D2
F
F0
F00
P1 P2 F1 F2 B1
1 1 0 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2
0 0 1 1/2 1/2 1/2 1/2 1/2 1/2 1/2 1/2
0 0 0 5/9 4/9 1/9 4/9 1/9 7/9 19/36 19/36
0 0 0 3/16 0 1/4 1/4 0 7/64 7/64 7/64
0 0 0 3/16 1/4 0 0 1/4 7/64 7/64 7/64
0 0 0 1/6 0 –1/3 2/3 0 1/12 –1/24 5/24
0 0 0 –1/6 –2/3 0 0 1/3 –1/12 –5/24 1/24
0 0 0 –1/8 0 0 0 0 –1/32 –1/32 –1/32
0 0 0 1/6 0 0 0 0 0 0 0
0 0 0 0 0 –1/3 2/3 0 0 0 0
0 0 0 0 –2/3 0 0 1/3 0 0 0
B2 F3 B1F2 B2F2
F1 × P1, RF1 × P1 P1 × F1, P1 × RF1 F1 × P2, RF1 × P2 P2 × F1, P2 × RF1 (RF1 × P1, selfed) (F1 × P2, selfed)
sues, those of Pooni et al. (1992) for means of triploid endosperms (Table 1) and those of Mo (1987) and Bogyo et al. (1988) for their variances (Table 2). Fisher’s original notation of ‘a’ and ‘d’ for additive and dominance effects (Bogyo et al., 1988; Kearsey and Pooni, 1996) has been followed instead of ‘d’ and ‘h’ (Mather and Jinks 1971, 1977 and many other authors). Similarly, the interaction components have been indicated as ‘aa’ for additive × additive, ‘ad’ for additive × dominance and ‘dd’ for dominance × dominance. The coefficients followed Pooni et al. (1992) where the value for a BBB genotype was m + a, for BBb m + (1/3)a + d1, for Bbb m – (1/3)a + d2 and for bbb m – a. Recalculation showed that the coefficients for [d1d1] and [d2d2] should be 1/16 for the F2 generation and 1/64 for the F3 generation, rather than the 0 given for both by Pooni et al. (1992). Coefficient [c] represented the cytoplasmic effect, [am ] the residual effect of the inbred mother plant and [dm ] the residual effect of the F1 hybrid mother plant (Pooni et al., 1992). These three coefficients were used with the embryo models as well as the endosperm ones for which they were derived, because of the critical importance of the mother plant for grain development. An additional term, ‘crossing’, was introduced to allow for the treatment of heads in preparation for emasculation and cross-pollination. Narrow-sense heritability was estimated using standard methods (Kearsey & Pooni, 1996). In the first experiment, the grains for planting were chosen at random following particle size analysis. The mean of the progeny grains from a plant was plotted against the value for the parent grain and narrow-sense heritability was estimated from the slope of the regression. In the
second experiment, the F2 grains had been selected on the basis of their B-granule content, so the narrowsense heritability was estimated from the ratio of the selection response to the selection differential.
Results Basic statistics The mode of the A granule distribution was in the next larger size class in ME71 than in Sunco in both experiments. The total B-granule content of Sunco averaged 51% and that of ME71 averaged 34% in the first experiment, and 40% and 27% respectively in the second experiment (Table 3). In both experiments, the mean B-granule content of the F1 hybrid was below that of the reciprocal F1 (Table 3 and Figure 1). This result had also been noted in glasshouse-grown grains in the preparatory experiments. The B1 and B2 means were both clearly skewed toward the recurrent parent, rather than being similar to those of the F2 generation from the same F1 plants. These results clearly showed that the genotype of the grain, either the embryo or the endosperm, determined the starch B granule content of the endosperm. Once the effect of preparation of heads for crossing was taken into account, there was very little difference between the F1, F2 and F3 means, indicating a lack of heterosis in this trait. The frequency distributions in the F2 (Figure 1) and F3 generations were approximately normal. The B1 and B1F2 means were very near each other, as were the B2 and B2F2 means, and the
27 Table 3. B-granule content (volume % of starch) means and variances from the cross Sunco × ME71 Generation
First experiment Grains Mean examined
SE
Population variance )
Second experiment Grains Mean examined
SE
Population variance
P1
(Sunco) thinned, bagged head
186 94
51.1 48.5
0.31 0.58
P2
(ME71) thinned, bagged head
173 88
35.7 33.8
0.31 0.51
F1:
P1 × P2
54
41.1
0.69
59
25.7
0.30
48
44.5
0.86
F2:
P2 × P1 (RF1) F1 selfed F1 × RF1
48 48 38
30.0 34.7 30.1
0.55 0.59 0.65
200
45.1
0.42
38 39
35.8 30.4
0.55 0.55
91
44.8
0.62
B1:
RF1 selfed RF1 × F1 thinned, bagged head F1 × P1
33
33.9
0.78
45
B2:
RF1 × P1 P1 × F1, P1 × RF1 F1 × P2
40 91 37
32.2 29.4 26.6
0.57 0.39 0.60
82
39.0
0.41
49 109
25.4 26.1
0.38 0.34
F3:
RF1 × P2 P2 × F1, P2 × RF1 F2 selfed
227
43.7
0.32
B1F2 B2F2
RF2 selfed (RF1 × P1, selfed) (F1 × P2, selfed)
359 349 343
43.9 48.0 39.4
0.33 0.28 0.29
)
45
40.4
0.40
7.35
40
26.9
0.38
5.70
}24.1 }19.3 }32.2
}14.03
}20.34 }35.0
}16.34 48.3
0.62
17.1
13.74 }9.99
14.1
12.32
}32.6
Table 4. Results of the joint scaling tests on grain B-granule data Scaling test
value
SE
t
A (2B1-P1-F1) B (2B2-P2-F1) C (4F2-2F1-P2-P1) D (4F3-2F2-P1-P2)
5.017 –1.417 8.149 0.683
1.445 1.345 2.226 1.263
3.471 –1.054 3.661 0.540
p < 0.01 ns p < 0.01 ns
distributions were skewed towards the recurrent parent. Segregation patterns were not clearly shown by any generation. Models of generation means The joint scaling tests (Table 4) indicated that a simple additive-dominance model was not adequate to explain the results. Models based on the diploid embryo genotype did not provide an adequate fit to the means or variances in the first experiment (Table 5) but endo-
26.5 27.9
sperm models were acceptable (Table 6). Both types provided adequate fits in the second experiment. Dropping the advanced generations (F3, B1F2 and B2F2) from the first experiment did not improve the fit of the embryo model. Preparation of heads for crossing reduced Bgranule contents in both experiments, as previously found (Stoddard, 1999), probably because of the damage to the head and shading by the paper crossing bag. In the second experiment, the endosperm ends of the grains had lower B-granule contents than the embryo ends (Table 7). The additive genetic effects were quite large. Dominance component [d1] was negative and relatively small in the first experiment and non-significant in the second. Dominance component [d2] was large and positive in both experiments. Digenic interactions [ad1], [d1d1] and [d2d2] were significant in the second experiment, when the greater number of back-cross generations allowed their detection, and all acted to reduce B-granule contents. In the first ex-
28 Table 5. Components of means and variances of wheat starch B-granule content using embryo models Coefficient
First experiment Value SE
Second experiment Value SE
m crossing [a] [d] [aa] [ad] [dd] [am ] [dm ] [c]
41.57 –1.60 8.96 10.24 1.56
0.69 0.34 0.30 2.65 0.74
33.59 –5.78 8.83
0.24 0.27 0.36
–2.84
0.93
–7.30 –1.36
2.11 0.32
–2.10 1.77
0.22 0.25
E1 E2 A D AD
χ 2 7 = 14.1, p < 0.05 21.9 1.29 27.9 3.16 10.6 2.52 –8.06 3.56
χ 2 8 = 11.64, p > 0.05 6.92 0.97 12.93 1.64 10.19 2.02
χ 2 5 = 24.88, p < 0.01
χ 2 5 = 7.09, p > 0.10
Models of variances Two estimates of the environmental variance were necessary (Tables 5 and 6), because the F1 variance was higher than the parent variances in both experiments (Table 3). The additive variance was not significant in the first experiment (endosperm model) but was quite important in the second. The significant genetic variances in the first experiment were the D1, AD1 and AD2 components while in the second experiment, the remaining contribution came from the A(D1-D2) covariance. Estimates of heritability Figure 1. Frequency distributions of B-granule content (volume% of starch) in (a) parents Sunco (broken line) and ME71 (solid line), (b) F1 hybrids (broken line) and reciprocal F1 grains (solid line), (c) F2 grains selfed (broken line) and manually cross-pollinated (solid line), and (d) B1 grains (broken line) and B2 grains (solid line).
periment, [dm ] was significant and positive, showing that F1 hybrid plants contributed about 1% greater Bgranule contents to their grains, but this effect was not confirmed in the second experiment. Finally, the cytoplasm component was small and negative in the first experiment.
In the first experiment, three parent-progeny correlations showed low narrow sense heritability, 0.18 for the 25 F2 plants, 0.15 for the 15 B1 plants and 0.05 for the 15 B2 plants (Figure 2). The pooled estimate for these three generations was much higher, 0.28 with a standard error of 0.08. In the second experiment, the mean B-granule contents of the two selected populations of 13 F2 grains were 32.6% and 57.3% (determined on the 2600C analyser) while the means of the F3 progeny grains were 32.5% and 35.5% respectively (Mastersizer S analyser). In both cases the selection differential was 12.3% and the response to
29 Table 6. Components of means and variances of wheat starch B-granule content using endosperm models Coefficient
First experiment Value SE
m crossing [a] [d1] [d2] [aa] [ad1] [ad2] [d1d1] [d2d2] [d1d2] [am ] [dm ] [c]
43.05 –1.43 7.97 –2.92 5.21
E1 E2 A D1 D2 AD1 AD2 D1D2 F (A(D1+D2)) F0 (A(D1-D2)) F00 (A(D2-D1))
0.17 0.28 0.18 0.72 0.80
Second experiment Value SE 33.61 0.24 –5.94 0.34 6.78 0.25 8.49 1.33 –6.10
1.11
–2.23 –4.06
0.61 1.44
0.96 0.32 –0.31 0.12 χ 2 7 = 8.88, p > 0.10 χ 2 7 = 13.38, p > 0.05 21.7 1.30 8.13 0.95 31.4 3.75 13.54 1.85 17.13 3.98 81.0 18.6 –18.9 40.2
than the others. Nevertheless, the estimates are fairly consistent.
4.22 8.56
Correlations with grain mass –12.51
2.86
χ 2 4 = 7.13, p > 0.10 χ 2 4 = 5.33, p > 0.10
selection was 1.5%; applying the regression between analysers of 0.900 suggests that the response to selection in old terms was 1.6%. This gives estimates of narrow-sense heritability of 0.13 in both directions. The change of analyser makes this estimate less robust Table 7. Starch B-granule contents of embryo halves and endosperm halves of grains of ME71 and Sunco Generation Number of grains B-granule content, volume % of starch Embryo end Endosperm end Sunco ME71 SE
36 36
Figure 2. Starch B-granule content (volume%) of F3 (M), B1F2 (#) and B2F2 () grains plotted against that of the parent F2, B1 and B2 grains. Line shows overall regression, y = 0.276∗x + 36.4, r = 0.451. Each point shows the results from an individual plant.
33.9 24.7 0.34
29.2 18.7
Starch B-granule content and grain mass were not significantly correlated, either within or between generations (Figure 3). Grain mass of Sunco and ME71 overlapped each other and the heterosis of the F1 plants was clearly shown in the greater grain mass.
Discussion The experiment was initially designed to provide the normal basic set of genetic populations, parents, F1 hybrids, F2 population and back-cross of F1 to both parents. It rapidly became clear that the particle size distribution of a grain was determined by the endosperm’s genotype, not that of the embryo nor the mother plant. Additive, dominance and epistatic genetic action all contributed to differences in starch particle size distribution in segregating generations of the cross of Sunco × ME71. The B-granule content of the parent genotypes was almost half again as great as that mentioned by Blumenthal et al. (1994) for the same two cultivars grown under identical temperatures (18◦/13◦ ), but our reports differ in several important aspects including
30
Figure 3. Starch B-granule content (volume%) versus grain mass (mg) for Sunco ( ), ME71 (), RF2 grains on RF1 plants (N), F3 and RF3 grains on F2 and RF2 plants (M), B1F2 grains on B1 plants (#) and B2F2 grains on B2 plants (). Each point shows the results from an individual plant.
the number of plants per pot and the nature of the growing medium. Nevertheless, the ratio of the two B-granule contents in both experiments (and also in glasshouse-grown materials during the development of the generations for this experiment) was about 1.5:1 (Sunco: ME71) and furthermore, Blumenthal et al. (1994) found that both cultivars expressed similar reductions of B-granule content following heat shock. Preliminary experiments in glasshouses also showed the same ratio of 1.5:1. The robustness of this difference between the cultivars indicates that genetic differences should be easier to identify than might otherwise be the case for other crosses. Preliminary studies on hybrid grain generated in glasshouses during the preparation of these experimental materials showed that the F1 grains had lower B granule contents than the reciprocal F1 grains. This was a surprising result and was considered attributable to small sample sizes (10–20 grains on each of two occasions) or possibly cytoplasmic effects. Nevertheless the result was confirmed in the much more stable environment of the growth chamber, indicating that it was real, not an experimental artifact, and attributable to dominance and epistasis. While these results may help in the design of experiments to understand the physiological basis for the genetic differences, they
are of lesser importance in the development of low B-granule wheats, since wheat is conventionally an inbreeding crop. Even from an F1 hybrid variety, the commercially harvested grains are the F2 generation, so it would be difficult to gain any benefit from a significant and negative [d1] effect. There was no sign of transgressive segregation for B-granule content in this cross, indicating that all of the alleles for increased B-granule content were associated in the Sunco parent and none in the ME71 parent. This is a useful outcome in that it validates the methods used for the genetic analysis, which are predicated on that association (Bogyo et al., 1987; Pooni et al., 1992). In terms of developing wheats with a still lower B-granule content, it indicates that we need to find a better source than ME71. Such sources have been found and are currently under investigation. The breadth of variation within inbred plants was somewhat discouraging and was likely to obscure Mendelian segregation ratios in the first experiment. The greater resolution obtained with the improved methods in the second experiment was still not good enough to reveal clear Mendelian segregation. The complex set of dominance and epistasis effects further obscured segregation patterns and implied that several genes were involved. The question of gene number will have to await QTL analysis of an appropriate population. Narrow-sense heritability was low, because of the importance of the dominance and epistasis effects. In particular, the ability to predict F3 values from an F2 grain was disappointly low. The prospects for using single-grain selection in early generation breeding are therefore poor. A doubled-haploid population should provide suitable material for selection purposes as all genes will be fixed. Low heritability due to non-additive gene effects will not then be a problem. Correlations between grain mass and starch Bgranule content have been noted before. In a germplasm survey of 59 wheats, it was noted that largegrained cultivars had larger A granules than other cultivars (Dengate & Meredith, 1984). Within cultivars, distal grains had both lower masses and lower B-granule contents than proximal grains (Stoddard, 1999). It might therefore be expected that grain mass might co-segregate with starch B-granule content. In this model, increased grain filling could be attributable to greater or continued production of B granules. Nevertheless, no correlation was found between grain mass and starch B-granule content in well filled grains of either Sunco or ME71. This correlation was also not
31 significant when calculated over the entire population. The controlling genotypes for the two traits have now been shown to be different: for grain mass it is that of the maternal plant (Millet, 1986) and for B-granule content it is the endosperm. We may therefore expect to be able to produce any reasonable combination of grain mass and B-granule content.
Acknowledgements I thank Toni Swain, Simone Cunneen, Shakir Shah, Shiranee Gunasekera and Ranjana Sarker for technical assistance. I also thank Ian Batey, Grain Quality Research Laboratories, CSIRO and Hak-Kim Chan, Dept of Pharmacy, The University of Sydney for use of their particle size analysers. This project has operated under the auspices of the Quality Wheat CRC Ltd.
References Blumenthal, C., C.W. Wrigley, I.L. Batey & E.W.R. Barlow, 1994. The heat-shock response relevant to molecular and structural changes in wheat yield and quality. Austr J Plant Physiol 21: 901–909. Blumenthal, C., F. Bekes, P.W. Gras, E.W.R. Barlow & C.W. Wrigley, 1995. Identification of wheat genotypes tolerant to the effects of heat stress on grain quality. Cereal Chem 72: 539–544. Bogyo, T.P., R.C.M. Lance, P. Chevalier & R.A. Nilan, 1988. Genetic models for quantitatively inherited endosperm characters. Heredity 60: 61–67.
Dengate, H. & P. Meredith, 1984. Variation in size distribution of starch granules from wheat grain. J Cereal Sci 2: 83–90. Gale, M.D., 1976. High alpha-amylase – breeding and genetical aspects of the problem. Cereal Res Comm 4: 231–243. Hoseney, R.C., K.F. Finney, Y. Pomeranz & M.D. Shogren, 1971. Functional (breadmaking) and biochemical properties of wheat flour components. VIII. Starch. Cereal Chem 48: 191–201. Kearsey, M.J. & H.S. Pooni, 1996. The Genetical Analysis of Quantitative Traits. London, UK: Chapman and Hall. Lelievre, J., K. Lorenz, P. Meredith & D.W. Barch, 1987. Effects of starch particle size and protein concentration on breadmaking performance. Starch/Stärke 39: 347–352. Mather, K. & J.L. Jinks, 1971. Biometrical Genetics. Second ed. London, UK: Chapman and Hall Ltd. Mather, K. & J.L. Jinks, 1977. Introduction to Biometrical Genetics. London, UK: Chapman and Hall. Millet, E., 1986. Relationships between grain weight and the size of floret cavity in the wheat spike. Ann Botany 58: 417–423. Mo, H., 1987. Genetic expression for endosperm traits. In: B.S. Weir, E.J. Eisen, M.M. Goodman & G. Namkoong (Eds), Proceedings of the Second International Conference on Quantitative Genetics. pp. 478–487. Sinauer Associates Inc. Sunderland, MA, USA. Parker, M.L., 1985. The relationship between A-type and B-type starch granules in the developing endosperm of wheat. J Cereal Sci 3: 271–278. Pooni, H.S., I. Kumar & G.S. Khush, 1992. A comprehensive model for disomically inherited metrical traits expressed in triploid tissues. Heredity 69: 166–174. Stoddard, F.L., 1999. Variation in grain size, grain nitrogen and starch B-granule content in wheat heads. Cereal Chem 76: 139–144.