ABSTRACT

Sowing date is an important determinant of yield in canola. However, the magnitude of the decline in yield with delayed sowing has been poorly defined for different environments. We show from collated published studies and sowing date studies carried out on canola and Indian mustard in the northern cropping region in 1998, that the decline in yield with delayed sowing is highly variable (from –10% to +4% per week delay in sowing). The mean value (5% per week delay in sowing date) agrees well with that already published for wheat, and may form the basis for a rough rule-of-thumb. Simulations studies for one location are used to highlight the risks associated with the tradeoff between sowing early to avoid end-of-season high temperatures and water deficit, which depresses seed yield and oil concentration, and sowing later to lessen the risk of damaging frosts during early grain-filling.

INTRODUCTION

Sowing date is an important determinant of yield in canola. In dryland Australian environments there is a trade-off between sowing early to avoid end-of-season high temperatures and water deficit, which depresses seed yield and oil concentration, and sowing later to lessen the risk of damaging frosts during early grain-filling. Knowledge of the consequences of delaying sowing for yield, oil concentration and frost risk can be used to define optimum sowing dates for different varieties, climates and farming systems (Hodgson, 1979). We examine previously published sowing date studies conducted in a range of Australian environments, and compare these with sowing date studies carried out on canola and Indian mustard in the northern cropping region, which is a new region for canola production. Simulation studies are used to analyses the variability in yield response to sowing date accounting for frost risk.

METHODS

Published sowing date studies

Grain yield was collated from studies where grain yield was measured in response to sowing date (Table 1). Yields were hand-harvested and corrected to oven-dry basis.

Experiments in northern Australia

Four sowing date studies were conducted in the northern grains region in 1998 (Table 2) at Tamworth, Moree, Lawes and Dalby. The Lawes study was irrigated, while the others were dryland. Crops were grown at 40-60 plants m^-2 with 20-30 cm row spacing, and supplied with 100-120 kgN ha^-1 at sowing. No pests, disease or weed limitations were recorded. The dates were recorded of open flowers on 50% of plants and at least 95% brown pods on all plants. Grain yield and above-ground biomass at maturity was measured from 0.9–2.7 m² quadrats in 2 or 3 replications.

Simulation studies

Simulations using APSIM-Canola (Robertson et al., 1999) were conducted for Roma in south-west Queensland (29.50^oS 149.9^oE), a potential new area for canola production. Sowings of cultivar Hyola 42 were simulated for 15^th of April, May, June and July. The crops were sown at 40 plants m^-2, supplied with 100 kgN ha^-1 and starting soil nitrate of 58 kgN ha^-1. Initial soil conditions were reset each year at sowing. The risk of frost during grain-filling damaging yield potential was quantified by the occurrence of at least one day when the minimum temperature reached a threshold temperature during simulated early grain-fill.

The rate of decline in yield (oven-dry basis) with delayed date of sowing was calculated using linear regression between sowing date and yield from the published studies, the 1998 experiments and the simulation studies. All regressions were statistically significant. The rate of decline was expressed both in absolute terms (kg/ha/d) and relative to the yield of the earliest sowing date (% per week) in order to account for differences in yield levels across studies. This method of expressing the response to sowing date has been used previously for wheat in Australia by Kohn and Storrier (1970), Doyle and Marcellos (1974), and McDonald et al. (1983).

RESULTS AND DISCUSSION

Published studies on the response to sowing date were under dryland conditions with the exception of that of Taylor and Smith (1992) and Thurling (1974) (Table 1). One spring sowing study from Canada (Degenhardt and Kondra, 1981) was included for comparison with Australian data. The published studies exhibited a wide range in yield response to sowing date. In all but 3 cases, yield declined with delay in sowing date (i.e. the linear regression slope coefficient between sowing date and grain yield was negative). The response ranged from a decline of 36.8 kg/ha/d to an increase of 11 kg/ha/d. Relative yield loss ranged from –10.8 to 4.2 % per week, with a mean of –5.1 % per week. Expressing the yield loss in terms relative to the yield potential, reduced the variability across studies, as found for wheat by Doyle and Marcellos (1974). The variability in response to sowing date possible at just one location (Condobolin: -10.7 to 4.2 % per week) due to differences in season and nitrogen supply, was notable. The strongest response to sowing date of –10.8 to –6.7 % per week was recorded in the study of Richards and Thurling (1978), under dryland conditions in Western Australia. The study of Thurling (1974), also in Western Australia, but under irrigation, had a weaker response to sowing date. The Western Australian Mediterranean-type environment is known for the rapid onset of high temperatures and water deficit in spring, which results in large yield penalties to late-sown annual crops (Fischer, 1979), and may explain the strong response to sowing date in the study of Richards and Thurling (1978). Interestingly, the response in the spring sowing study in Canada (-5.1% per week) was similar to the mean of the Australian studies.

The experiments conducted in northern region in 1998 produced a range of yield loss values from –3.1% per week at Tamworth to –10.5% per week at Moree (Table 2). The decline at Moree and Dalby was steeper than that at Tamworth and Lawes because heavy rain resulted in waterlogging during early growth in the later sowing, exaggerating the decline. There were no frosts recorded that would be expected to reduce yield. It is notable that the winter growing season in Tamworth was one of the wettest on record so that the response to sowing date would have been largely a function of crop phenology, rather than the occurrence of a more severe water deficit in the later sowings. Thurling (1974) found that under irrigated conditions the decline in seed yield was associated primarily with a reduction in the total dry weight of the plant at final harvest which, in turn, was most closely correlated with the duration of vegetative phase of growth.

Overall, the relative yield decline from published studies and the 1998 experiments are similar to those published previously for wheat, in those situations where frost was not a complicating factor. Doyle and Marcellos (1974) measured a 5-7% reduction in relative grain yield of wheat for each week that sowing was delayed beyond the end of June. Kohn and Storrier (1970) arrived at a value of 3.7% for the Wagga Wagga district measured over 5 seasons, while McDonald et al. (1983) recorded 6% per week from June for irrigated wheat in the Namoi Valley of NSW. All studies showed evidence of seasonal variation around the mean value, as in the current study.

Crop growth simulation offers a powerful tool to examine the tradeoffs between sowing time, yield potential, frost risk and high temperatures during grain-filling. Before being able to use simulation with some confidence it is necessary to validate the simulation of the response to sowing date. Figure 1 shows that the model was able to capture the variation in yield with sowing date at Lawes in 1998. The shorter period of sowing to flowering and flowering to physiological maturity was also simulated well, although the model consistently under-predicted the date of maturity. This may have been an artifact of the accurate assessment of physiological maturity in the field.

Traditional practice in the northern grain belt is to sow winter grain crops in mid-May to early June, a time which in some years predisposes the crop to damage by spring frosts through flowering being too early (Single 1961). The yield of the crop therefore becomes a compromise between yield loss due to frost if it flowers too early, or from water deficit if too late. In addition, with canola, high temperatures during grain-filling are known to depress seed oil concentration and hence crop profitability. Table 3 illustrates the difficulty in optimising sowing time, using simulations for Roma in Queensland. For this scenario, mean grain yield declined by 3.6 % per week from 1686 to 1030 kg ha^-1 with sowing date delayed from 15^th April to 15^th July. With June and July sowing, 25% of years would yield less than 600 kg ha^-1 versus 1200 kgha^-1 for April sowing. Mean daily air temperature during grain-filling also increased sharply with delayed sowing, which would translate to seed oil concentrations (@8.5% moisture) of 42.6, 38.6, 37.0 and 35.6 % for the four sowing dates, using the function of Walton (pers. comm.) derived for WA. For canola in Australia, seed oil concentrations below 40% are subject to a price penalty, indicating that delayed sowing beyond early May in Roma is likely to lead to reduced grain prices. Frost risk for this scenario is presented for a range of frost thresholds from 2⁰C to –4^oC, because of lack of a well-defined threshold for canola. Also, the relationship between screen temperature and crop temperature will be a function of paddock aspect, topography, and the extent of crop canopy development, so that the air temperature causing yield loss will vary from situation to situation. These simulations suggest that there are strong reasons to delay the sowing of canola beyond the end of April, if the avoidance of significant frost risk is an important priority of the grower. However, this delay will come at the cost of a lowered yield potential, higher temperatures during grain-filling, and possible conflict with the sowing time of other winter crops on the farm. Clearly, the arrival at a optimum sowing time will depend on the risk attitude of the grower. Information, such as that in Table 3, will allow the various risks to be assessed.

CONCLUSIONS

The analysis here suggests that the yield response of canola to sowing date is highly variable, however the mean value across all studies in similar to that already published for wheat (around 5% per week). In the northern region frost risk from early sowing needs to be balanced with lower yield potential and high temperatures during grain-filling from later sowings. Simulation analysis can be used to evaluate the risks associated with different sowing dates.

ACKNOWLEDGMENTS

This work was funded in part by the Grains Research and Development Corporation. Our thanks go to Brett Cocks for assistance in the field.

REFERENCES

1. Degenhardt, D.F. and Kondra, Z.P. 1981. The influence of seeding date and seeding rate on seed yield and yield components of five genotypes of Brassica napus. Can. J. Plant Sci. 61:175-83.

2. Doyle, A. D. and Marcellos, H. 1974. Time of sowing and wheat yield in northern New South Wales. Aust. J. Exp. Agric. Anim. Husb.. 14:93-102.

3. Fischer, R.A. 1979. Growth and water limitation to dryland wheat yield in Australia:a physiological framework. J. Aust. Inst. Agric. Sci. 45:83-94.

4. Fischer, R. A. and Kohn, G. D. 1966. The relationship of grain yield to vegetative growth and post-flowering leaf area in the wheat crop under conditions of limited soil moisture. Aust. J. Agric. Res. 17:281-295.

5. Hocking, P.J., Kirkegaard, J.A., Angus, J.F., Gibson, A.H., and Koetz, E.A. 1997. Comparison of canola, Indian mustard and Linola in two contrasting environments. I. Effects of nitrogen fertilizer on dry-matter production, seed yield and seed quality. Field Crop Res. 49:107-25.

6. Hodgson, A.S. 1979. Rapeseed adaptation in Northern New South Wales II Predicting plant development of Brassica campestris L. and Brassica napus L. and its implications for planting time, designed to avoid water deficit and frost. Aust. J. Agric. Res. 29:711-26.

7. Hodgson, A.S. 1979. Rapeseed adaptation in Northern New South Wales III Yield, yield components and grain quality of Brassica campestris and Brassica napus in relation to planting date. Aust. J. Agric. Res. 30:19-27.

8. Kohn, G. D. and Storrier, R. R. 1970. Time of sowing and wheat production in southern New South Wales. Aust. J. Exp. Agric. Anim. Husb. 10:604-609.

9. McDonald, G. K.; Sutton, B. G., and Ellison, F. W. 1983. The effect of time of sowing on the grain yield of irrigated wheat in the Namoi Valley, New South Wales. Aust. J. Agric. Res. 34:229-240.

10. Richards, R.A. and Thurling, N. 1978. Variation between and within species of rapeseed (Brassica campestris and B. napus) in response to drought stress. II. Growth and development under natural drought stress. Aust. J. Agric. Res. 29:479-90.

11. Robertson, M.J., Holland, J.F., Kirkegaard, J.A., Smith, C.J. (1999). Simulating growth and development of canola in Australia. These proceedings.

12. Salisbury, P., Potter, T., Castleman, G., Robson, D., and Hyett, J. 1989. Potential for a wider maturity range in rapeseed. Pp. 33-41 in Australian Rapeseed Agronomists and Breeders, 7th workshop (Toowoomba, Queensland, Australia, Sept 1989).

13. Single, W. V. 1961. Studies on frost injury in wheat.I. Laboratory freezing tests in relation to the behaviour of varieties in the field. Aust. J. Agric. Res. 12: 767.

14. Taylor, A.J. and Smith, C.J. 1992. Effect of sowing date and seeding rate on yield and yield components of irrigated canola (Brassica napus L.) grown on a red-brown earth in south-eastern Australia. Aust. J. Agric. Res. 43:1629-41.

15. Thurling, N. 1974. Morphophysiological determinants of yield in rapeseed (Brassica campestris and Brassica napus). I. Growth and morphological characters. Aust. J. Agric. Res. 25:697-710.

Table 1: Published studies of the response of Brassica napus or B. Juncea to sowing date. Maximum yield is that measured at the earliest sowing date in each study. Yield loss is the slope of grain yield on sowing date, expressed either as kg/ha/d or as a percent of the maximum yield per week. A negative yield loss indicates that yield declined with delay in sowing date.

Table 2: Response of canola and Indian mustard (cv. 887.2.6.1) to sowing date recorded in the northern region of Australia in 1998 (JF Holland and MJ Robertson, unpublished).

Table 3: Simulated response to sowing date at Roma, Queensland. Values are the mean of 105 years of simulations. Frost risk is defined as the proportion of years in which a temperature less than the stated value occurs during early to mid grain-filling

Figure 1: Simulated and observed response to sowing date for cv. Hyola 42 at Lawes, Queensland in 1998, for (a) grain yield, and (b) days from sowing to flowering (lower line) and physiological maturity (upper line). Points are observations and lines are simulated values.