Introduction to {epicrop}
{epicrop} provides an R package of the ‘EPIRICE’ model as described
in (Savary et al. 2012), the modified
EPIRICE model as described in (Kim et al.
2015), the ‘EPIWHEAT’ model as described in (Savary et al. 2015) and a generic SEIR model
function, seir(), for modelling crop disease epidemics. The
model uses daily weather data to estimate disease intensity. A function,
get_wth(), is provided to simplify downloading weather data
via the {nasapower}
package (Sparks 2018) and predict disease
intensity of five rice diseases using a generic SEIR model (Zadoks 1971) function, seir().
For ‘EPIRICE’, default values derived from the literature suitable
for modelling unmanaged disease intensity of five rice diseases,
bacterial blight (bacterial_blight()); brown spot
(brown_spot()); leaf blast (leaf_blast());
sheath blight (sheath_blight()) and tungro
(tungro()) and two modified by Kim et al. (2015), (modified_kim_leaf_blast()
and helper_modified_kim_sheath_blight()), are provided. The
modified Kim versions provide leaf blast and sheath blight models that
include additional weather parameters and modified equations to better
reflect disease progress under certain environmental conditions on the
Korean Peninsula. The ‘EPIWHEAT’ model includes two wheat diseases, leaf
rust (leaf_rust()) and septoria tritici blotch
(s_tritici_blotch()), with default values derived from the
literature suitable for modelling unmanaged disease intensity of these
diseases.
Using the package functions is designed to be straightforward for
modelling rice disease risks, but flexible enough to accommodate other
pathosystems using the seir() function. If you are
interested in modelling other pathosystems, please refer to (2012) for the development of the parameters
that were used for the rice diseases as derived from the existing
literature and are implemented in the individual disease model
functions.
Get weather data
The most simple way to use the model is to download weather data from
NASA POWER using get_wth(), which provides the data in a
format suitable for use in the model and is freely available. See the
help file for naspower::get_power() for more details of
this functionality and details on the data (Sparks 2018).
# Fetch weather for year 2000 season at the IRRI Zeigler Experiment Station
wth <- get_wth(
lonlat = c(121.25562, 14.6774),
dates = c("2000-01-01", "2000-12-31")
)
wth## Key: <YYYYMMDD>
## YYYYMMDD DOY TEMP TMIN TMAX RHUM RAIN LAT LON
## <IDat> <int> <num> <num> <num> <num> <num> <num> <num>
## 1: 2000-01-01 1 24.38 22.85 27.46 91.25 14.93 14.6774 121.2556
## 2: 2000-01-02 2 24.28 22.68 27.42 90.88 6.96 14.6774 121.2556
## 3: 2000-01-03 3 23.82 22.17 26.96 88.36 2.28 14.6774 121.2556
## 4: 2000-01-04 4 23.68 21.90 27.14 88.00 0.87 14.6774 121.2556
## 5: 2000-01-05 5 24.11 21.54 28.18 88.12 0.43 14.6774 121.2556
## ---
## 362: 2000-12-27 362 24.46 22.90 26.28 92.49 21.15 14.6774 121.2556
## 363: 2000-12-28 363 24.64 23.32 27.28 91.92 6.01 14.6774 121.2556
## 364: 2000-12-29 364 24.58 22.51 27.90 90.79 5.49 14.6774 121.2556
## 365: 2000-12-30 365 25.31 22.84 28.84 86.25 2.07 14.6774 121.2556
## 366: 2000-12-31 366 24.47 21.63 28.63 87.83 3.45 14.6774 121.2556Predict bacterial blight
All of the helper family of functions work in exactly the same
manner. You provide them with weather data and an emergence date, that
falls within the weather data provided, and they will return a data
frame of disease intensity over the season and other values associated
with the model. See the help file for seir() for more on
the values returned.
# Predict bacterial blight intensity for the year 2000 wet season at IRRI
bb_wet <- bacterial_blight(wth, emergence = "2000-07-01")
summary(bb_wet)## simday dates sites latent
## Min. : 1.00 Min. :2000-07-01 Min. : 0.00 Min. : 0.000
## 1st Qu.: 30.75 1st Qu.:2000-07-30 1st Qu.: 16.28 1st Qu.: 0.000
## Median : 60.50 Median :2000-08-29 Median : 64.21 Median : 1.000
## Mean : 60.50 Mean :2000-08-29 Mean : 57.02 Mean : 3.744
## 3rd Qu.: 90.25 3rd Qu.:2000-09-28 3rd Qu.: 98.22 3rd Qu.: 5.080
## Max. :120.00 Max. :2000-10-28 Max. :100.00 Max. :18.014
## infectious removed senesced rateinf rlex
## Min. : 0.000 Min. : 0.00 Min. : 0.00 Min. :0.0000 Min. :0
## 1st Qu.: 1.000 1st Qu.: 0.00 1st Qu.: 0.00 1st Qu.:0.0000 1st Qu.:0
## Median : 6.838 Median : 1.00 Median : 1.00 Median :0.0000 Median :0
## Mean :17.062 Mean :12.88 Mean :13.39 Mean :0.5348 Mean :0
## 3rd Qu.:34.146 3rd Qu.:19.56 3rd Qu.:21.82 3rd Qu.:0.6962 3rd Qu.:0
## Max. :53.168 Max. :61.49 Max. :62.04 Max. :4.0939 Max. :0
## rtransfer rremoved rgrowth rsenesced diseased
## Min. :0.0000 Min. :0.0000 Min. :0 Min. :0.0000 Min. : 0.00
## 1st Qu.:0.0000 1st Qu.:0.0000 1st Qu.:0 1st Qu.:0.0000 1st Qu.: 1.78
## Median :0.0000 Median :0.0000 Median :0 Median :0.0000 Median :37.14
## Mean :0.5348 Mean :0.5125 Mean :0 Mean :0.5170 Mean :33.68
## 3rd Qu.:0.6962 3rd Qu.:0.6962 3rd Qu.:0 3rd Qu.:0.6962 3rd Qu.:64.18
## Max. :4.0939 Max. :4.0939 Max. :0 Max. :4.0939 Max. :64.69
## intensity lat lon
## Min. :0.0000 Min. :14.68 Min. :121.3
## 1st Qu.:0.0178 1st Qu.:14.68 1st Qu.:121.3
## Median :0.3602 Median :14.68 Median :121.3
## Mean :0.4158 Mean :14.68 Mean :121.3
## 3rd Qu.:0.7351 3rd Qu.:14.68 3rd Qu.:121.3
## Max. :1.0000 Max. :14.68 Max. :121.3Plotting using {ggplot2}
The data are in a wide format by default and need to be converted to long format for use in {ggplot2} if you wish to plot more than one variable at a time.
Wet season sites
The model records the number of sites for each bin daily; this can be graphed as follows.
dat <- pivot_longer(
bb_wet,
cols = c("diseased", "removed", "latent", "infectious"),
names_to = "site",
values_to = "value"
)
ggplot(data = dat,
aes(
x = dates,
y = value,
shape = site,
linetype = site
)) +
labs(y = "Sites",
x = "Date") +
geom_line(aes(group = site, colour = site)) +
geom_point(aes(colour = site)) +
theme_classic()
Wet season intensity
Plotting intensity over time does not require any data manipulation.
ggplot(data = bb_wet,
aes(x = dates,
y = intensity * 100)) +
labs(y = "Intensity (%)",
x = "Date") +
geom_line() +
geom_point() +
theme_classic()
Comparing epidemics
The most common way to compare disease epidemics in botanical
epidemiology is to use the area under the disease progress curve (AUDPC)
(Shaner and Finney 1977). The AUDPC value
for a given simulated season is returned as a part of the output from
any of the disease simulations offered in {epicrop}. You can find the
value in the AUDPC column. We can compare the dry season
with the wet season by looking at the AUDPC values for both seasons.
bb_dry <- bacterial_blight(wth = wth, emergence = "2000-01-05")
summary(bb_dry)## simday dates sites latent
## Min. : 1.00 Min. :2000-01-05 Min. : 0.00 Min. : 0.00
## 1st Qu.: 30.75 1st Qu.:2000-02-03 1st Qu.: 25.50 1st Qu.: 0.00
## Median : 60.50 Median :2000-03-04 Median : 64.45 Median : 1.00
## Mean : 60.50 Mean :2000-03-04 Mean : 58.57 Mean : 3.13
## 3rd Qu.: 90.25 3rd Qu.:2000-04-03 3rd Qu.: 97.56 3rd Qu.: 4.81
## Max. :120.00 Max. :2000-05-03 Max. :100.00 Max. :21.80
## infectious removed senesced rateinf rlex
## Min. : 0.000 Min. : 0.00 Min. : 0.00 Min. :0.0000 Min. :0
## 1st Qu.: 1.000 1st Qu.: 0.00 1st Qu.: 0.00 1st Qu.:0.0000 1st Qu.:0
## Median : 8.491 Median : 1.00 Median : 1.00 Median :0.0000 Median :0
## Mean :14.181 Mean :12.42 Mean :12.85 Mean :0.4471 Mean :0
## 3rd Qu.:28.857 3rd Qu.:21.71 3rd Qu.:21.71 3rd Qu.:0.3144 3rd Qu.:0
## Max. :40.572 Max. :51.11 Max. :51.11 Max. :7.0252 Max. :0
## rtransfer rremoved rgrowth rsenesced diseased
## Min. :0.0000 Min. :0.000 Min. :0 Min. :0.000 Min. : 0.000
## 1st Qu.:0.0000 1st Qu.:0.000 1st Qu.:0 1st Qu.:0.000 1st Qu.: 2.452
## Median :0.0000 Median :0.000 Median :0 Median :0.000 Median :34.553
## Mean :0.4471 Mean :0.426 Mean :0 Mean :0.426 Mean :29.734
## 3rd Qu.:0.3144 3rd Qu.:0.000 3rd Qu.:0 3rd Qu.:0.000 3rd Qu.:53.122
## Max. :7.0252 Max. :7.025 Max. :0 Max. :7.025 Max. :53.933
## intensity lat lon
## Min. :0.00000 Min. :14.68 Min. :121.3
## 1st Qu.:0.02447 1st Qu.:14.68 1st Qu.:121.3
## Median :0.34238 Median :14.68 Median :121.3
## Mean :0.36664 Mean :14.68 Mean :121.3
## 3rd Qu.:0.55330 3rd Qu.:14.68 3rd Qu.:121.3
## Max. :1.00000 Max. :14.68 Max. :121.3Dry season intensity
Check the disease progress curve for the dry season.
ggplot(data = bb_dry,
aes(x = dates,
y = intensity * 100)) +
labs(y = "Intensity (%)",
x = "Date") +
geom_line() +
geom_point() +
theme_classic()
The AUDPC values can be viewed directly from the attributes of the
data.table outputs above, but we can also create a small
helper function to extract them.
# Dry season
get_audpc(bb_dry)## [1] 43.49663
# Wet season
get_audpc(bb_wet)## [1] 49.39145The AUDPC of the wet season is greater than that of the dry season. Checking the data and referring to the curves, the wet season intensity reaches a peak value of 100% and the dry season tops out at 100%. So, this meets the expectations that the wet season AUDPC is higher than the dry season, which was predicted to have less disease intensity.