ABSTRACT

The relationships between yield loss and incidence or severity of stem canker in winter oilseed rape were analysed using data from experiments at Rothamsted in 1991/92, Withington in 1992/93, Boxworth in 1993/94 and Rothamsted in 1997/98. Critical point models and area under disease progress curve (AUDPC) models were better than multiple point models for describing relationships between yield (t ha^-1) and incidence or severity of stem canker. Since yield in t ha^-1 is influenced by many factors other than disease, % yield loss was calculated. Critical point models and AUDPC models relating % yield loss to stem canker were constructed and models for different experiments were compared. A critical point model relating % yield loss to stem canker incidence (GS 6.3-6.4), and AUDPC models for incidence or severity from growth stages 4.8 to 6.4 were derived from the combined data sets for the three experiments. The relationships between % yield loss and % plants with different stem canker severity scores at different growth stages, and between % yield loss and incidence or severity of both basal stem cankers and upper stem lesions were also analysed.

INTRODUCTION

Stem canker, caused by Leptosphaeria maculans (Desm.) Ces. & de Not., is a damaging disease of oilseed rape in the UK, France, Germany, Australia and Canada. The pathogen attacks cotyledons, leaves, stems, roots and pods of oilseed rape and can cause yield loss of 10-50% (Gugel and Petrie, 1992). Most previous work indicated that relationships between yield or yield loss and stem canker incidence or severity can be described by a linear arithmetic equation (McGee and Emmett,1977; Rempel et al., 1991; Church and Fitt, 1995; Sansford et al., 1996; Pierre et al., 1982). However, Hall et al. (1993) suggested that the relationship between yield losses and stem canker severity in Canada was best described by a logarithmic equation. These relationships between yield or yield loss and stem canker were all single point models based on assessment of the disease at a single growth stage, although the growth stage and method of assessment differed. Thus these models differed from season to season and site to site. This paper reports detailed examinations of the effects of severity and timing of stem canker symptoms on yield of winter oilseed rape in the UK, using critical point, multiple point and AUDPC models.

METERIALS AND METHODS

UK data sets that were used to analyse and establish relationships between winter oilseed rape yield loss and incidence or severity of stem canker were from field experiments at Rothamsted in 1991/92, Withington in 1993/94, Boxworth in 1993/94 and Rothamsted in 1997/98. In these experiments, light leaf spot and other oilseed rape diseases were absent or present at a low severity. The experiments were arranged in randomised block designs, with two replicate blocks of 22 plots at Rothamsted in 1991/92, Withington in 1993/94 and Boxworth in 1993/94, and with four replicate blocks of five main plots, each split into two sub-plots at Rothamsted in 1997/98. The plot areas were 75m² in the 1991/92 experiment, 96m² in 1992/93, 108m² in 1993/94 and 45m² in 1997/98. The winter oilseed rape cultivar Envol was grown in experiments at Rothamsted in 1991/92, at Withington in 1992/93, at Boxworth in 1993/94, and the cultivars Capitol and Lipton were grown at Rothamsted in 1997/98. Different disease epidemic patterns in different experiment plots were obtained by using fungicides, with 10 - 22 different treatment regimes in each experiment.

Ten or twenty-five plants were sampled at approximately monthly intervals from sampling areas at the edges and ends of each plot. Incidence (% plants with stems affected) and severity (mean severity score) of stem canker (including basal stem cankers and upper stem lesions) were recorded. The disease severity assessment used a 0 - 4 scale (Hardwick et al., 1989). Growth stages (GS) of the crop were recorded using the identification key of Sylvester-Bradley (1985). Plots were combine harvested directly and yields adjusted to 90% dry matter.

Correlation and linear regression analyses were used to examine the relationships between yield or yield loss and stem canker and to establish CP (Critical Point) models, MP (Multiple Point) models and AUDPC (Area Under Disease Progress Curve) models between yield loss and stem canker for the each experiment. Differences between CP models and AUDPC models in different experiments were compared by using linear regression analyses of position and parallelism. The statistical package Genstat (Payne et al., 1993) was used for all analyses.

RESULTS

Relationships between yield and stem canker at different assessment dates

Coefficients of correlation between crop yield and incidence or severity of stem canker at different assessment dates (Table 1) showed that yield was inversely related to the stem canker incidence or severity from the middle of May (GS 4.8) to early July (GS 6.5), but not in April (GS 3.7). At Boxworth in 1993/94 and Rothamsted in 1997/98, correlation coefficient values were similar on assessment dates in May, June and July (GS 4.8-6.4), but at Withington in 1992/93 correlation coefficient values for incidence and severity were greater on 12 July (GS 6.4) than in May or June.

Models of relationships between yield and stem canker

Critical point (CP) models, multiple point (MP) models and area under disease progress curve (AUDPC) models relating yield (t ha^-1) to stem canker incidence or severity were constructed by simple or multiple linear regression. The linear regressions were statistically significant (p<0.05) for all models. The CP models for the experiment at Rothamsted in 1991/92 accounted for <15% of the variance. The CP, AUDPC and MP models for the experiments at Withington in 1992/93 and Boxworth in 1993/94 all generally accounted for 45-55% of the variance. The CP, AUDPC and MP models for the experiment at Rothamsted in 1997/98 all accounted for 22-27% of the variance. Analyses of model collinearity indicated that there were strong correlations between variables used in MP models. This produced some positive values for slopes in MP models, which are very difficult to interpret biologically, even though the values of negative slopes in MP models did not really represent the contribution of the variables to yield loss. These results demonstrated that CP models and AUDPC models were generally better than MP models.

Table 1. Coefficients of correlation ( r) between yield (t ha^-1) of winter oilseed rape and incidence or severity of stem canker at different growth stages in different seasons at different sites

CP and AUDPC models relating % yield loss to stem canker

Since yield in t ha^-1 is influenced by many factors specific to the crop other than disease, % yield losses were calculated. CP models and AUDPC models relating % yield loss to stem canker incidence or severity in different experiments were constructed using linear regression (Table 2). The differences between experiments for these models were compared using analyses of position and parallelism. The results showed that CP models for stem canker incidence or severity at Rothamsted in 1991/92 were significantly different to those for the other experiments, but that there were no differences between models for the other three experiments, except between the models for severity at Withington in 1992/93 and Boxworth in 1993/94. Thus a CP model for incidence based on combined data sets from Withington in 1992/93, Boxworth in 1993/94 and Rothamsted in 1997/98 was constructed. Comparison of AUDPC models showed no differences between Withington in 1992/93, Boxworth in 1993/94 and Rothamsted in 1997/98 in models for both incidence and severity, and AUDPC models for incidence and severity were derived from the combined data sets for these three experiments (Table 2).

Relationships between % yield loss and % plants with different severity scores

The relationships between % yield loss and % plants with different severity scores were analysed by linear regression. The results demonstrated that the greatest yield losses were generally associated with the largest severity scores for plants assessed at the same crop growth stages. However, there were some differences between experiments. For example, the results suggested that before harvest only severity scores of 3 or 4 at GS 6.5 (12 July) in 1991/92 and score 4 at GS 6.4 (1 July) in 1997/98 affected yield, but that severity scores of 1 or 2 at GS 6.3 (28 June) in 1993/94 and score 2 at GS 6.4 (12 July) in 1992/93 also affected yield. Nevertheless, yield losses were also associated with severity scores of 1 and 2 in May (GS 4.9-5.5) and June (GS 5.9-6.3), when assessments were made in 1992/93, 1993/94 and 1997/98. Thus, the results suggested that the earlier that lesions with a particular severity score were recorded on plants, the greater was the effect on yield, because lesions with a small severity score in early spring continued to become more severe.

Table 2. Equations describing relationships between % yield loss (L) of winter oilseed rape and incidence (X) or severity (S) of stem canker at Rothamsted in 1991/92, 1997/98, Withington in 1992/93 and Boxworth in 1993/94

Relationships between % yield loss and incidence or severity for basal stem cankers and upper stem lesions

The relationships between % yield loss and % plants with basal stem cankers or upper stem lesions or between % yield loss and mean severity score for basal stem cankers or upper stem lesions at Boxworth in 1993/94 and Rothamsted in 1997/98 were also analysed by linear regression. The results showed that yield losses were significantly related to incidence or severity of both basal stem cankers and upper stem lesions in both seasons in May, June and July, except for basal stem cankers at Boxworth on 13 May 1994. The different values of the slopes of the regression lines suggested that effects of basal stem cankers on yield were different from those of upper stem lesions.

CONCLUSIONS

Correlation analyses indicated that yield losses from stem canker were related best to disease incidence or severity at GS 6.3 or 6.4, which is the seed development stage and a key stage for oilseed rape yield production; thus it was concluded that GS 6.3 - 6.4 was the critical point for relating stem canker to yield loss. Comparison of CP, AUDPC and MP models for three experiments at Withington in 1992/93, Boxworth in 1993/94 and Rothamsted in 1997/98 suggested that accuracy of yield loss prediction by CP models and AUDPC models was better than that of MP models. Based on the results of comparison of CP or AUDPC models between different experiments, CP models for incidence (GS 6.3-6.4), and AUDPC models for incidence or severity from growth stages 4.8 to 6.4 were derived from the combined data sets for the three experiments at Withington in 1992/93, Boxworth in 1993/94 and Rothamsted in 1997/98.

The results of analyses of relationships between % yield loss and % plants with different stem canker severity scores indicated that early stem canker lesions in May and June, even when the disease score was slight, affected yield more than later stem canker because lesions with a small severity score in early spring became more severe later. Further analysis showed that incidences of both basal stem canker and upper stem lesions were related to yield loss of oilseed rape. However, the timing of the first appearance of lesions and the severity of basal stem cankers and upper stem lesions differed between seasons and sites. These results suggested that it is important for farmers to control early stem canker, and that both basal stem cankers and upper stem lesions need to be controlled. Ultimately it is necessary to estimate risks that early, severe basal stem cankers or upper stem lesions will develop to guide decision making for stem canker control during the period from autumn to early spring.

ACKNOWLEDGEMENTS

We thank the Royal Society, the Ministry of Agriculture, Fisheries and Food, the European Union (FAIR CT96-1669 coordinated by MH Balesdent), the Biotechnology and Biological Sciences Research Council and the Home-Grown Cereals Authority for supporting this work, AD Todd and WJ Stevenson for analysis of field experiment data.

REFERENCES

1. Church VJ and Fitt BDL (1995) Incidence and effects of diseases on seven winter oilseed rape cultivars in 1991/92 and 1992/93. International Organization for Biological Control Bulletin 18: 62-68.

2. Gugel RK and Petrie GA (1992) History, occurrence, impact, and control of blackleg of rapeseed. Canadian Journal of Plant Pathology 14:36-45.

3. Hall R, Peters RD and Assabgui RA (1993) Occurrence and impact of blackleg on oilseed rape in Ontario. Canadian Journal of Plant Pathology 15:305-313.

4. Hardwick NV, Culshaw FA, Davies JML, Gladders P, Hawkings JH and Slawson DD (1989) Incidence and severity of fungal diseases of winter oilseed rape in England and Wales, 1986-1988. Aspects of Applied Biology, Production and Protection of Oilseed Rape and other Brassica Crops, 23:393-392.

5. McGee DC and Emmett RW (1977) Blackleg (Leptosphaeria maculans (Desm.) Ces. et de Not.) of rapeseed in Victoria: crop losses and factors which affect disease severity. Australian Journal of Agricultural Research 28:47-51.

6. Payne RW, Lane PW, Baird DB, Gilmour AR, Harding SA, Morgan GW, Murray DA, Thompson R, Todd AD, Tunnicliffe Wilson G, Webster R, Welham SJ (1993) Genstat 5, Release 3, Reference Manual. Clarendon Press, Oxford.

7. Rempel CB, Lisieczko Z and Hall R (1991) Epidemiological aspects of resistance in rapeseed to blackleg. Proceedings of the Eighth International Rapeseed Congress, Saskatoon, Saskatchewan 2:460-464.

8. Pierre JG and Regnault Y, Peres A, Penot P, Bonnet T, Thoral N and Tiberghien (1982) Contribution à la mise au point d’une méthode de plein champ destinée à mesurer la sensibilité des variétés de colza au Phoma. Informations Techniques, Centre Technique Interprofessionnel des Oléagineux Métropolitains 81:3-18.

9. Sansford CE, Fitt BDL, Gladders P, Lockley KD and Sutherland KG (1996) Oilseed rape: disease development, forecasting and yield loss relationships. Home-Grown Cereals Authority Project Report OS17. 185pp.

10. Sylvester-Bradley R. (1985) Revision of a code for stages of development in oilseed rape (Brassica napus L.). Aspects of Applied Biology, Field Methods and Data Handling, 10:395-400.