Fig 1.
Late blight of potato is caused by the oomycete pathogen Phytophthora infestans.
A: sporangia are aerially dispersed and deposited on the surface of potato leaves. B: Sporangia release motile zoospores in cool wet conditions. C: penetration of the leaf epidermis occurs via production of appressoria. D (close up) and E: disease symptoms including: water soaked and sporulating necrotic lesions in susceptible cultivar Binjte grown under field conditions, Borgeby, Sweden 2015. Scale bars A and C, 2 μm; B, 10 μm. All photographs: Laura Grenville-Briggs.
Fig 2.
Graphical representation of the SIMCAST lookup table.
The number of blight units to be added for each moist period (relative humidity ≥ 90%) depends on the mean temperature (in degree Celsius) during the period and the length of the period in hours.
Table 1.
Covariates used in the statistical analysis.
All variables except Location are time dependent.
Fig 3.
Continuous representation of SIMCAST based on a beta function along the temperature gradient and a sigmoidal function along the hour gradient.
The number of blight units to be added for each wet period (relative humidity≥ 90%) depends on the mean temperature (in degree Celsius) during the period and the length of the period in hours.
Fig 4.
Occurrence of late blight over the investigation time in Finland (triangles) and Sweden (circles).
Fig 5.
Proportion of non-infected potato trials separated in three 10 year periods.
Fig 6.
Predictability of occurrence of late blight in Finland (triangles) and Sweden (circles) based on a moving window of 10 years expressed in the p-value of the model containing all parameters listed in Table 1.
Please note that the data series for Finland starts in 1990 and ends already in 2011.
Table 2.
Predictive power and coefficient of the covariates without combinations.
Acronyms: temp.: temperature; rel.hum.: relative humidity; min.: minimum. An exp(coeff) value larger unity indicates a positive relationship.
Fig 7.
Example prediction of incidence of late blight based on weather data in the year 2010 at a single site.
The central cross-line indicates the estimated hazard rate while the upper and the lower confidence intervals are given as dotted or broken lines respectively.
Fig 8.
Evaluation of parameter robustness.
The coefficient is estimated from a proportion of the data sets excluding ca 40% of the data using a bootstrap method.