Movatterモバイル変換

Type:	Package
Title:	Analysis of Plant Disease Epidemics
Version:	0.5.0
Description:	A toolbox to make it easy to analyze plant disease epidemics. It provides a common framework for plant disease intensity data recorded over time and/or space. Implemented statistical methods are currently mainly focused on spatial pattern analysis (e.g., aggregation indices, Taylor and binary power laws, distribution fitting, SADIE and 'mapcomp' methods). See Laurence V. Madden, Gareth Hughes, Franck van den Bosch (2007) <doi:10.1094/9780890545058> for further information on these methods. Several data sets that were mainly published in plant disease epidemiology literature are also included in this package.
License:	MIT + file LICENSE
URL:	https://github.com/chgigot/epiphy,https://chgigot.github.io/epiphy/
BugReports:	https://github.com/chgigot/epiphy/issues
Encoding:	UTF-8
LazyData:	true
RoxygenNote:	7.2.3
LinkingTo:	Rcpp
Depends:	R (≥ 4.0)
Imports:	stats, methods, utils, grDevices, ggplot2, transport, msm,pbapply, Rcpp
Suggests:	magrittr, dplyr, tidyr, spdep, emdist, vegan, MASS, emdbook,knitr, rmarkdown
Collate:	'RcppExports.R' 'betabinom.R' 'data.R' 'utils.R''mle-factory.R' 'epiphy.R' 'intensity-classes.R''distr-fitting.R' 'indices.R' 'mapcomp.R' 'power-law.R''sadie.R' 'spatial-hier.R'
VignetteBuilder:	knitr
NeedsCompilation:	yes
Packaged:	2023-11-15 07:43:36 UTC; tof
Author:	Christophe Gigot [aut, cre, cph], Adam H. Sparks [ctb], Katrin Leinweber [ctb]
Maintainer:	Christophe Gigot <ch.gigot@gmail.com>
Repository:	CRAN
Date/Publication:	2023-11-16 11:20:10 UTC

The beta-binomial distribution.

Description

Density, distribution function, quantile function and random generation forthe beta-binomial distribution with parameterssize,prob,theta,shape1,shape2. This distribution corresponds toan overdispersed binomial distribution.

Usage

dbetabinom(x, size, prob, theta, shape1, shape2, log = FALSE)pbetabinom(  q,  size,  prob,  theta,  shape1,  shape2,  lower.tail = TRUE,  log.p = FALSE)qbetabinom(  p,  size,  prob,  theta,  shape1,  shape2,  lower.tail = TRUE,  log.p = FALSE)rbetabinom(n, size, prob, theta, shape1, shape2)

Arguments

Details

Be aware that in this implementationtheta = 1 / (shape1 +shape2).prob andtheta, orshape1 andshape2 must be specified. iftheta = 0, use *binom familyinstead.

Value

dbetabinom gives the density,pbetabinom gives the distributionfunction,qbetabinom gives the quantile function andrbetabinomgenerates random deviates.

See Also

dbetabinom in the packageemdbookwhere the definition of theta is different.

Examples

# Compute P(25 < X < 50) for X following the Beta-Binomial distribution# with parameters size = 100, prob = 0.5 and theta = 0.35:sum(dbetabinom(25:50, size = 100, prob = 0.5, theta = 0.35))# When theta tends to 0, dbetabinom outputs tends to dbinom outputs:sum(dbetabinom(25:50, size = 100, prob = 0.5, theta = 1e-7))sum(dbetabinom(25:50, size = 100, shape1 = 1e7, shape2 = 1e7))sum(dbinom(25:50, size = 100, prob = 0.5))# Example of binomial and beta-binomial frequency distributions:n   <- 15q   <- 0:np1  <- dbinom(q, size = n, prob = 0.33)p2  <- dbetabinom(q, size = n, prob = 0.33, theta = 0.22)res <- rbind(p1, p2)dimnames(res) <- list(c("Binomial", "Beta-binomial"), q)barplot(res, beside = TRUE, legend.text = TRUE, ylab = "Frequency")# Effect of the aggregation parameter theta on probability density:thetas <- seq(0.001, 2.5, by = 0.001)density1 <- rep(sum(dbinom(25:50, size = 100, prob = 0.5)), length(thetas))density2 <- sapply(thetas, function(theta) {    sum(dbetabinom(25:50, size = 100, prob = 0.5, theta = theta))})plot(thetas, density2, type = "l",     xlab = expression("Aggregation parameter ("*theta*")"),     ylab = "Probability density between 25 and 50 (size = 100)")lines(thetas, density1, lty = 2)

Easily switch between different power law formulations.

Description

a2a was designed to avoid headaches that are likely to occur whenworking with different formulations of the binomial power law analysis.

Usage

a2a(x, ...)## S3 method for class 'numeric'a2a(  x,  slope,  n,  from = c("Ai", "ai", "AI", "aI"),  to = c("Ai", "ai", "AI", "aI"),  ...)## S3 method for class 'list'a2a(x, to = c("Ai", "ai", "AI", "aI"), ...)

Arguments

Details

The binomial power law can be expressed as:s_y^2 = (intercept)(s_{bin}^2)^b.But different forms of (intercept) are possible depending on the formulation of thebinomial power law.

Value

A numeric vector.

Examples

# Values from the power_law() example:Ai    <- 38.6245slope <- 1.9356n     <- 9# Usual function call syntax:a2a(Ai, slope, n, from = "Ai", to = "ai")# Other syntaxes:inputs <- list(Ai = Ai, slope = slope, n = n)a2a(inputs, "ai")require(magrittr)inputs %>% a2a("ai")

Several aggregation indices.

Description

This function can compute different aggregation indices. See "Details"section for more information about the available indices.

Usage

agg_index(  x,  method = c("fisher", "lloyd", "morisita"),  flavor = c("count", "incidence"),  n = NULL,  ...)

Arguments

Details

There are currently three implemented methods to compute indices ofaggregation.

fisher: Fisher's index of aggregation. In case of a count, this indexcorresponds to the ratio of the observed variance to the observed mean, andthis is why this index is also known as the variance to mean ratio. For abinary variable, a similar index can be calculated using instead the ratio ofthe observed variance to the theoretical variance if data follow a binomiallaw (i.e. a reference distribution for a random pattern of individuals withinsampling units).

lloyd: Lloyd's index of patchiness. The value of this index increaseswith the degree of aggregation. Note that Lloyd's mean crowding can also bereturned iftype = "mean-crowding" is provided as parameter.

morisita: Morisita's coefficient of dispersion. This index can becomputed for either count or incidence data, but its interpretation can beuncertain.

Values of Fisher's and Lloyd's indices can be interpreted as follows:

• index < 1: uniform pattern;

• index = 1: random pattern;

• index > 1: aggregated pattern.

The following table gives information about the applicability of the variousmethods.

where + means implemented, and -, not implemented (or not possible). At themoment, there is no index of aggregation for severity data.

Value

An object of classagg_index, which is a list containing the followingcomponents:

References

Fisher RA. 1925. Statistical methods for research workers. Oliver and Boyd,Edinburgh.

Lloyd M. 1967. Mean crowding. The Journal of Animal Ecology 36, 1–30.

Morisita M. 1962. I\delta-Index, a measure of dispersion ofindividuals. Researches on Population Ecology 4, 1–7.doi:10.1007/BF02533903

Madden LV, Hughes G. 1995. Plant disease incidence: Distributions,heterogeneity, and temporal analysis. Annual Review of Phytopathology 33(1):529–564.doi:10.1146/annurev.py.33.090195.002525

See Also

vegdist invegan package.

Examples

# Count flavor of Fisher's index:my_fisher_count <- agg_index(aphids$i)my_fisher_count# And incidence flavor of Fisher's index:my_fisher_incidence <- agg_index(tobacco_viruses$i, n = tobacco_viruses$n)my_fisher_incidence# Either standard R or epiphy idioms can be used:identical(my_fisher_count, agg_index(count(aphids)))identical(my_fisher_incidence, agg_index(incidence(tobacco_viruses)))# Lloyd's index (only for count data):agg_index(aphids$i, method = "lloyd")# Lloyd's mean crowding:agg_index(aphids$i, method = "lloyd", type = "mean-crowding")# Count flavor of Morisita's index:agg_index(aphids$i,  method = "morisita")# Incidence flavor of Morisita's index:agg_index(tobacco_viruses$i, n = tobacco_viruses$n, method = "morisita")

Counts of aphids.

Description

Counts of 554 aphids of the species Sitobion avenae, sampled on 28 June 1996in a 250 x 180-m field of winter wheat near Wimborne, Dorset, UK. The 63sampling units, made of the inspection of five tillers each, were located ona 9 x 7 rectangular grid at intervals of 30 m.

Usage

aphids

Format

A data frame with 63 rows and 3 variables:

Source

Perry JN, Winder L, Holland JM, Alston RD. 1999. Red-blue plots fordetecting clusters in count data. Ecology Letters 2, 106-13.doi:10.1046/j.1461-0248.1999.22057.x

Counts of arthropods.

Description

A sampling unit was made of a pitfall to collect arthropods in a field oforganic winter wheat, near Wimborne, Dorset, UK in 1996. The sampling unitswere located on a 9 x 7 rectangular grid at intervals of 30 m. There weresix sampling dates.

Usage

arthropods

Format

A data frame with 378 rows and 4 variables:

Source

Holland JM, Winder L, Perry JN. 1999. Arthropod prey of farmlandbirds: Their spatial distribution within a sprayed field with and withoutbuffer zones. Aspects of Applied Biology 54: 53-60.

Coerce to a data frame.

Description

Functions to coerce anintensity object to a data frame.

Usage

## S3 method for class 'intensity'as.data.frame(  x,  row.names = NULL,  optional = FALSE,  ...,  stringsAsFactors = FALSE)

Arguments

Value

A data frame.

Examples

my_data <- incidence(tomato_tswv$field_1929)head(as.data.frame(my_data))

C(alpha) test.

Description

The C(alpha) test is a test of the binomial distribution against thealternative of the beta-binomial distribution.

Usage

calpha.test(x, ...)## S3 method for class 'fisher'calpha.test(x, ...)

Arguments

Details

It is based on calculation of a test statistic, z, that has an asymptoticstandard normal distribution under the null hypothesis. It is one-sided (inthe way that the alternative is aggregation, not just "non-randomness"), thuswith a confidence level of 951.64. When all the sampling units contain the same total number ofindividuals, n, the test statistic is calculated from:

z = (n(N - 1)I - Nn)/(2Nn(n - 1))^(1/2)

where N is the number of sampling units, and I, Fisher's index of aggregationfor incidence data.

Value

Same kind of object as the one returns by the statschisq.test function for example.

References

Neyman J. 1959. Optimal asymptotic tests of composite statistical hypotheses.In: Probability and Statistics, 213-234. Wiley, New York.

Tarone RE. 1979. Testing the goodness of fit of the binomial distribution.Biometrika, 66(3): 585-590.

See Also

chisq.test,z.test

Examples

# For incidence data:my_incidence <- incidence(tobacco_viruses)my_fisher <- agg_index(my_incidence, method = "fisher")calpha.test(my_fisher)

Chi-squared test.

Description

Performs chi-squared tests for Fisher's aggregation indices (computed witheither count or incidence data). If another kind of data is provided, the Rstandardchisq.test function is called.

Usage

chisq.test(x, ...)## Default S3 method:chisq.test(x, ...)## S3 method for class 'fisher'chisq.test(x, ...)

Arguments

Details

Under the null hypothesis for Fisher's aggregation index (index = 1, i.e. arandom pattern is observed), (N - 1)*index follows a chi-squared distributionwith N - 1 degrees of freedom. N is the number of sampling units.

Value

Same kind of object as the one returns by the statschisq.test function.

References

For count and incidence data:

Madden LV, Hughes G. 1995. Plant disease incidence: Distributions,heterogeneity, and temporal analysis. Annual Review of Phytopathology 33(1):529–564.doi:10.1146/annurev.py.33.090195.002525

Patil GP, Stiteler WM. 1973. Concepts of aggregation and theirquantification: a critical review with some new results and applications.Researches on Population Ecology, 15(1): 238-254.

See Also

calpha.test,z.test

Examples

# For incidence data:my_incidence <- incidence(tobacco_viruses)my_fisher <- agg_index(my_incidence, method = "fisher")chisq.test(my_fisher)

Incidence of citrus tristeza virus (CTV) disease in three fields.

Description

CTV incidence data for three orchards in eastern Spain reported forconsecutive years.

Usage

citrus_ctv

Format

There are three data frames:

• IVI3and4: A data frame with 864 rows and 5 variables.

• IVI6and7: A data frame with 648 rows and 5 variables.

• El_Realengo: A data frame with 2000 rows and 5 variables.

The structure is the same for all the data frames:

Details

BothIVI3and4 andIVI6and7 orchards consisted of 216 trees eachof Washington navel orange on Troyer citrange planted in 1978 on a 2 x 6-mspacing.El_Realengo orchard consisted of 400 Marsh seedlessgrapefruit on Troyer citrange planted in 1973 on a 5.5 x 5.5-m spacing.

Source

Gottwald TR, Cambra M, Moreno P, Camarasa E, Piquer J. 1996. Spatialand temporal analyses of citrus tristeza virus in eastern Spain.Phytopathology 86, 45–55.

Gibson GJ. 1997. Investigating mechanisms of spatiotemporal epidemicspread using stochastic models. Phytopathology 87, 139-46.doi:10.1094/PHYTO.1997.87.2.139

Regroup observational data into even clumps of individuals.

Description

This function provides a easy way to regroup recorded data into groups ofsame number of individuals.

Usage

clump(object, ...)## S3 method for class 'intensity'clump(object, unit_size, fun = sum, inclusive_unspecified = FALSE, ...)

Arguments

Value

Anintensity object.

Examples

my_incidence <- incidence(tomato_tswv$field_1929)plot(my_incidence, type = "all")# Different spatial size units:my_incidence_clumped_1 <- clump(my_incidence, unit_size = c(x = 3, y = 3))plot(my_incidence_clumped_1, type = "all")my_incidence_clumped_2 <- clump(my_incidence, unit_size = c(x = 4, y = 5))plot(my_incidence_clumped_2, type = "all")# To get mean disease incidence for each plant over the 3 scoring dates:my_incidence_clumped_3 <- clump(my_incidence, unit_size = c(t = 3), fun = mean)plot(my_incidence_clumped_3)

Count of codling moth larvae.

Description

Codling moth diapausing larvae were collected in an apple orchard insouth-eastern France. Larvae were caught on strip traps wrapped around treetrunks in July 2008 and collected the following October. 30 traps were used.

Usage

codling_moths

Format

A data frame with 30 rows and 3 variables:

Source

Lavigne C, Ricci B, Franck P, Senoussi R. 2010. Spatial analyses ofecological count data: A density map comparison approach. Basic andApplied Ecology 11: 734-42.doi:10.1016/j.baae.2010.08.011

Extract Model Coefficients

Description

coef is a generic function which extracts model coefficientsfrom objects returned by modeling functions.coefficients isanalias for it.

Usage

## S3 method for class 'smle'coef(object, ...)

Arguments

Value

Coefficients extracted from the model objectobject.

For standard model fitting classes this will be a named numeric vector.For"maov" objects (produced byaov) it will be a matrix.

Incidence of dogwood anthracnose.

Description

Incidence data from the Dogwood Anthracnose Impact Assessment Program for1990 and 1991, in the Southeast of the USA, reported by Zarnoch et al (1995).Only plots with exactly n = 10 dogwood trees are present in the data set (168and 161 plots in 1990 and 1991, respectively).

Usage

dogwood_anthracnose

Format

A data frame with 329 rows and 3 variables:

Source

Zarnoch SJ, Anderson RL, Sheffield RM. 1995. Using the\beta-binomialdistribution to characterize forest health. Canadian journal of forestresearch 25, 462–469.

epiphy: An R package to analyze plant disease epidemics.

Description

epiphy makes it easy to analyze plant disease epidemics. It providesa common framework for plant disease intensity data recorded over time and/orspace. Implemented statistical methods are currently mainly focused onspatial pattern analysis (e.g., aggregation indices, Taylor and binary powerlaws, distribution fitting, SADIE and mapcomp methods). Several data setsthat were mainly published in plant disease epidemiology literature are alsoincluded in this package

Author(s)

Maintainer: Christophe Gigotch.gigot@gmail.com

See Also

Useful references:

Gosme M. 2008. Comment analyser la structure spatiale et modéliser ledéveloppement spatio-temporel des épiphyties? Canadian Journal of PlantPathology, 30:4-23.

Madden LV, Hughes G, van den Bosch F. 2007. Spatial aspects of epidemics -III: Patterns of plant disease. In: The study of plant disease epidemics,235–278. American Phytopathological Society, St Paul, MN.

Maximum likelihood fitting of two distributions and goodness-of-fitcomparison.

Description

Different distributions may be used depending on the kind of provided data.By default, the Poisson and negative binomial distributions are fitted tocount data, whereas the binomial and beta-binomial distributions are usedwith incidence data. Either Randomness assumption (Poisson or binomialdistributions) or aggregation assumption (negative binomial or beta-binomial)are made, and then, a goodness-of-fit comparison of both distributions ismade using a log-likelihood ratio test.

Usage

fit_two_distr(data, ...)## Default S3 method:fit_two_distr(data, random, aggregated, ...)## S3 method for class 'count'fit_two_distr(  data,  random = smle_pois,  aggregated = smle_nbinom,  n_est = c(random = 1, aggregated = 2),  ...)## S3 method for class 'incidence'fit_two_distr(  data,  random = smle_binom,  aggregated = smle_betabinom,  n_est = c(random = 1, aggregated = 2),  ...)

Arguments

Details

Under the hood,distr_fit relies on thesmle utilitywhich is a wrapped around theoptim procedure.

Note that there may appear warnings about chi-squared goodness-of-fit testsif any expected count is less than 5 (Cochran's rule of thumb).

Value

An object of classfit_two_distr, which is a list containing at leastthe following components:

Other components can be present such as:

References

Madden LV, Hughes G. 1995. Plant disease incidence: Distributions,heterogeneity, and temporal analysis. Annual Review of Phytopathology 33(1):529–564.doi:10.1146/annurev.py.33.090195.002525

Examples

# Simple workflow for incidence data:my_data <- count(arthropods)my_data <- split(my_data, by = "t")[[3]]my_res  <- fit_two_distr(my_data)summary(my_res)plot(my_res)# Simple workflow for incidence data:my_data <- incidence(tobacco_viruses)my_res  <- fit_two_distr(my_data)summary(my_res)plot(my_res)# Note that there are other methods to fit some common distributions.# For example for the Poisson distribution, one can use glm:my_arthropods <- arthropods[arthropods$t == 3, ]my_model <- glm(my_arthropods$i ~ 1, family = poisson)lambda <- exp(coef(my_model)[[1]]) # unique(my_model$fitted.values) works also.lambda# ... or the fitdistr function in MASS package:require(MASS)fitdistr(my_arthropods$i, "poisson")# For the binomial distribution, glm still works:my_model <- with(tobacco_viruses, glm(i/n ~ 1, family = binomial, weights = n))prob <- logit(coef(my_model)[[1]], rev = TRUE)prob# ... but the binomial distribution is not yet recognized by MASS::fitdistr.# Examples featured in Madden et al. (2007).# p. 242-243my_data <- incidence(dogwood_anthracnose)my_data <- split(my_data, by = "t")my_fit_two_distr <- lapply(my_data, fit_two_distr)lapply(my_fit_two_distr, function(x) x$param$aggregated[c("prob", "theta"), ])lapply(my_fit_two_distr, plot)my_agg_index <- lapply(my_data, agg_index)lapply(my_agg_index, function(x) x$index)lapply(my_agg_index, chisq.test)

Incidence of three viruses in an Australian hop garden.

Description

Three viruses, i.e. Hop latent virus (HpLV), Hop mosaic virus (HpMV), andApple mosaic virus (ApMV), were monitored in an Australian hop garden for twoconsecutive years (1996 and 1997). The hop garden was established in 1989with the variety Victoria in a commercial hop farm at Bushy Park, Tasmania,Australia. It consisted of 25 rows containing 51 plants each, so that therewere 1275 hop plants in total. There were 2.1 m between rows, and 1.8 mbetween plants within rows.

Usage

hop_viruses

Format

There are three data frames, one for each virus (HpLV,HpMV andApMV). Each data frame consists of 2550 rows and 7 variables:

Source

Pethybridge SJ, Madden LV. 2003. Analysis of spatiotemporal dynamicsof virus spread in an Australian hop garden by stochastic modeling. PlantDisease 87:56-62.doi:10.1094/PDIS.2003.87.1.56

Retrieve vector or array indices

Description

ind2sub is just an alias forarrayInd.sub2ind is the reverse ofind2sub.

Usage

ind2sub(ind, .dim, .dimnames = NULL, useNames = FALSE)sub2ind(sub, .dim)

Arguments

Value

SeearrayInd.

Examples

set.seed(12345)mat <- matrix(round(runif(6, min = 0, max = 10)), nrow = 2, ncol = 3)ind2sub(4, dim(mat))sub2ind(c(2, 2), dim(mat))subs <- as.matrix(expand.grid(1:2,2:3))sub2ind(subs, dim(mat))

Construct count, incidence and severity objects.

Description

count(),incidence() andseverity() create eponymobjects. All of these classes inherit from the base classintensity.The choice of the class depends on the nature of the data set.

Usage

count(data, mapping, keep_only_std = TRUE)incidence(data, mapping, keep_only_std = TRUE)severity(data, mapping, keep_only_std = TRUE)

Arguments

Details

incidence reads disease incidence data from a data frame and return anincidence object. All of these classes inherit fromintensity class.

• count: Each sampling unit contains from 0 to theoreticaly an infinity of data.Number are positive integers.

• incidence: Each sampling unit contains an number of diseased plants,ranging from 0 ton which is the total amount of plants per samplingunit.

• severity: Each sampling unit contain a percentage of disease, a positivereal number ranging from 0.0 to 1.0.

Class intensity and inherited classes

All the classes recording disease intensity measurements inherit from thisclass. The classintensity is virtual which means that no object of aclassintensity can be constructed. This class only describes commonfeatures of all the different disease intensity measurements implemented inthis package (count,incidence andseverity). You should call one of these inherited classesinstead, depending on the nature of your data.

By convention, the first columns of the different data frames of each slotshave names, but the spatial, temporal or even disease information do not needto fit to these conventions or may be less straightforward and need morecolumns to record correctly all the information. In such unusual situations,the automatic options of the analysis tools would need to be overridden to beable to work in the desired way.

The differences between the different inherited classes regard only theobs slot. In the case ofcount, the data expected foreach record are positive integers (N+). Forincidence, the datasets are supposed to be two information set per records, the number ofdiseased unit per sampling unit (r) and the total number of units persampling unit (n). Note that in its current implementation, n is supposed tobe the same for a whole data set. Unequal sampling units are not implementedyet. Finally, forseverity, r is positive real ranging from 0to 1 and depecting a percentage.

space A data frame containing only spatial information. Each rowcorresponds to a sampling unit. By convention, the first 3 columns arenamesx,y,z.

time A data frame containing temporal information. By convention, thefirst column is namedt.

obs A data frame containing disease observations themselves. The nameof the columns may differ between the sub-class chosed to record the data.

Note that it is possible to create a "severity" object but no statisticaltools are currently implemented to deal with such an object.

Anintensity object contains at very least the "pure" intensityrecords (columnr) which is a so-called observational variable.Another observational variable, the number of individuals in a sampling unit(n), is present in the case of aincidence object. Very oftenin addition to observational variables, there are spatial (columnsx,y and/orz) and/or temporal (columnt) variables.

Note that theseverity class and thez variable (the 3rdspatial dimension) are implemented but no statistical methods use them atthis point.

Value

Anintensity object.

When printed, difference information are available:

• The number of sampling units.

• The time.

• Is it georeferenced (TRUE/FALSE)

• Are there any NA data (TRUE/FALSE)

• Is it a complet array (TRUE/FALSE)? A complete array means that all the recorded values allow todisplay an array (even if some data are not available), but this was explicitelly specified. Tocomplete a dataset, just usecomplete(data). You can also remove NA, which is necessary to usesome analysis technics, usingreplaceNA(data) orreplace.na(data). Note that using bothcommands will results in modifying the original data sets which will be specified.

Examples

## Create intensity objects# Implicite call: The variable mapping does not need to be specified if the# column names of the input data frame follow the default names.colnames(tomato_tswv$field_1929) # Returns c("x", "y", "t", "i", "n")my_incidence_1 <- incidence(tomato_tswv$field_1929)my_incidence_1my_incidence_2 <- incidence(tomato_tswv$field_1929,                            mapping(x = x, y = y, t = t, i = i, n = n))identical(my_incidence_1, my_incidence_2)# Explicite call: Otherwise, the variable mapping need to be specified, at# least for column names that do not correspond to default names.colnames(aphids) # Returns c("xm", "ym", "i")my_count_1 <- count(aphids, mapping(x = xm, y = ym, i = i))my_count_1# We can drop the "i = i" in the mapping.my_count_2 <- count(aphids, mapping(x = xm, y = ym))identical(my_count_1, my_count_2)# It is possible to change the variable mapping after the creation of an# intensity object:another_incidence <- incidence(hop_viruses$HpLV)another_incidenceremap(another_incidence, mapping(x = xm, y = ym))## Plotting dataplot(my_incidence_1) # Same as: plot(my_incidence_1, type = "spatial")plot(my_incidence_1, type = "temporal")plot(my_count_1, tile = FALSE, size = 5)plot(my_count_1, type = "temporal") # Not possible: there is only 1 date.# Using grayscale:plot(my_count_1, grayscale = TRUE)plot(my_count_1, grayscale = TRUE, tile = FALSE, size = 5)

Test if an object is of classintensity or one of its subclasses.

Description

Test if an object is of classintensity or one of its subclasses(i.e.count,incidence orseverity).

Usage

is.intensity(x)is.count(x)is.incidence(x)is.severity(x)

Arguments

Value

TRUE if its argument's value is the correspondingintensityobject and FALSE otherwise.

Some link functions.

Description

Logit, probit and cloglog functions are available.The logit and the logistic (with rev = TRUE), i.e. the inverse-logit functions.Probit is a wrapper aroundqnorm (forprobit) andpnorm (forprobit^{-1})Complementary log-log transformation.

Usage

logit(x, rev = FALSE)probit(x, rev = FALSE)cloglog(x, rev = FALSE)

Arguments

Value

A numeric vector.

Extract log-likelihood

Description

This function returns the maximal log-likelihood estimated withsmle, iff returned log-likelihood value.

Usage

## S3 method for class 'smle'logLik(object, ...)

Arguments

Value

Returns an object of classlogLik. This is a number with atleast one attribute,"df" (degrees offreedom),giving the number of (estimated) parameters in the model.

There is a simpleprint method for"logLik" objects.

There may be other attributes depending on the method used: see theappropriate documentation. One that is used by several methods is"nobs", the number of observations used in estimation (afterthe restrictions ifREML = TRUE).

Map Comparison procedure.

Description

mapcomp performs a spatial pattern analysis based on the calculationof a formal distance (the Hellinger distance) between the density map ofcount or incidence data, and the density map of sampling effort. Statisticaltests of spatial homogeneity are based on permutations across sampling sitesand on valuable properties of the Hellinger distance.

Usage

mapcomp(data, ...)## S3 method for class 'data.frame'mapcomp(  data,  delta,  bandwidth,  nperm = 100,  edge_correction = FALSE,  threads = 1,  verbose = TRUE,  ...)## S3 method for class 'matrix'mapcomp(  data,  delta,  bandwidth,  nperm = 100,  edge_correction = FALSE,  threads = 1,  verbose = TRUE,  ...)## S3 method for class 'count'mapcomp(  data,  delta,  bandwidth,  nperm = 100,  edge_correction = FALSE,  threads = 1,  verbose = TRUE,  ...)## S3 method for class 'incidence'mapcomp(  data,  delta,  bandwidth,  nperm = 100,  edge_correction = FALSE,  threads = 1,  verbose = TRUE,  ...)

Arguments

Value

An object of classmapcomp, which is a list containing the followingcomponents:

References

Lavigne C, Ricci B, Franck P, Senoussi R. 2010. Spatial analyses ofecological count data: A density map comparison approach. Basic and AppliedEcology. 11:734–742.

Examples

set.seed(123)my_res <- mapcomp(codling_moths, delta = 1, bandwidth = 11,                  edge_correction = FALSE, nperm = 20)my_resplot(my_res)set.seed(123)my_count <- count(codling_moths, mapping(x = xm, y = ym))my_res <- mapcomp(my_count, delta = 1, bandwidth = 11,                  edge_correction = FALSE, nperm = 20)my_resplot(my_res, bins = 10)

Existing variable mappings.

Description

Get or set existing variable mappings.

Usage

mapped_var(x)mapped_var(x, keep = TRUE) <- value

Arguments

Value

mapped_var returns the list of current mapped names of the objectx.

See Also

mapping

Examples

my_data <- count(aphids)my_datamapped_var(my_data)mapped_var(my_data) <- mapping(x = X, y = Y)mapped_var(my_data)mapped_var(my_data) <- mapping(x = x, r = r, keep = FALSE)mapped_var(my_data)

Construct data mappings.

Description

Data mappings describe how variables in the data are mapped to standard namesused throughoutepiphy.

Usage

mapping(...)mapping_(x)remap(data, mapping, keep_only_std = TRUE)

Arguments

Details

Standard names arex,y andz for the three spatialdimensions, andt for the time.r corresponds to the recordsof (disease) intensity, andn, the number of individuals in a samplingunit (if applicable).

mapping() works with expressions, andmapping_(), with a vectorof characters.

Value

A list of mapped names.

See Also

mapped_var

Examples

mapping(x = col1, y = col2)mapping_(c("x = col1", "y = col2"))

Offspring survival of rats experiencing different diets.

Description

Results of an experiment where two groups of 16 female rats were feddifferent diets during pregnancy and lactation periods. One group's dietcontained a chemical under review, and the other one was a control. For eachlitter, the number of pups alive at 4 days, and the number of pups weaned(i.e. that survived the 21-day lactation period) were recorded.

Usage

offspring_survival

Format

A data frame with 32 rows and 3 variables:

Source

Weil CS. 1970. Selection of the valid number of sampling units and aconsideration of their combination in toxicological studies involvingreproduction, teratogenesis or carcinogenesis. Food and CosmeticsToxicology 8: 177-182.

Incidence of bacterial blight of onion.

Description

Assessments of bacterial blight of onion at two dates. The experimental plotwas sown with naturally X. axonopodis pv. allii-contaminated onion (A. cepaL. cv. Chateau-vieux) seed lot, with a contamination rate of about 0.04%.

Usage

onion_bacterial_blight

Format

A data frame with 1134 rows and 5 variables:

Source

Roumagnac P, Pruvost O, Chiroleu F, Hughes G. 2004. Spatial andtemporal analyses of bacterial blight of onion caused by Xanthomonasaxonopodis pv. allii. Phytopathology 94, 138–146.doi:10.1094/PHYTO.2004.94.2.138

Taylor's and binary power laws.

Description

Assesses the overall degree of heterogeneity in a collection of data sets atthe sampling-unit scale.

Usage

power_law(data, log_base = exp(1), ...)

Arguments

Details

The power law describes the relationship between the observed variance ofindividuals within a data set (s^2) and the corresponding varianceunder the assumption of no aggregation (s\'^2). It can be expressedunder its logarithmic form as:log(s^2) = log(a) + b log(Y), with:

• Y = p in the case of count data (Taylor's power law).

• Y = p(1 - p) in the case of incidence data (binary power law).

p corresponds to the mean proportion of recorded individuals in caseof incidence data, and the absolute value in case of count data.

Value

Apower_law object.

References

Taylor LR. 1961. Aggregation, variance and the mean. Nature 189: 732–35.

Hughes G, Madden LV. 1992. Aggregation and incidence of disease. PlantPathology 41 (6): 657–660.doi:10.1111/j.1365-3059.1992.tb02549.x

Madden LV, Hughes G, van den Bosch F. 2007. Spatial aspects of epidemics -III: Patterns of plant disease. In: The study of plant disease epidemics,235–278. American Phytopathological Society, St Paul, MN.

Examples

require(magrittr)my_data <- do.call(c, lapply(citrus_ctv, function(citrus_field) {   incidence(citrus_field) %>%       clump(unit_size = c(x = 3, y = 3)) %>%       split(by = "t")}))# my_data is a list of incidence object, each one corresponding to a given# time at a given location.my_power_law <- power_law(my_data)my_power_lawsummary(my_power_law)plot(my_power_law) # Same as: plot(my_power_law, scale = "log")plot(my_power_law, scale = "lin")

Incidence of ray blight disease of pyrethrum.

Description

An assessment of the incidence of ray blight disease of pyrethrum in 62sampling units, containing 6 plants each.

Usage

pyrethrum_ray_blight

Format

A data frame with 62 rows and 2 variables:

Source

Pethybridge SJ, Esker P, Hay F, Wilson C, Nutter FW. 2005.Spatiotemporal description of epidemics caused by Phoma ligulicola inTasmanian pyrethrum fields.Phytopathology 95, 648–658.doi:10.1094/PHYTO-95-0648

Spatial Analysis by Distance IndicEs (SADIE).

Description

sadie performs the SADIE procedure. It computes different indices andprobabilities based on the distance to regularity for the observed spatialpattern and a specified number of random permutations of this pattern. Bothkind of clustering indices described by Perry et al. (1999) and Li et al.(2012) can be computed.

Usage

sadie(data, ...)## S3 method for class 'data.frame'sadie(  data,  index = c("Perry", "Li-Madden-Xu", "all"),  nperm = 100,  seed = NULL,  threads = 1,  ...,  method = "shortsimplex",  verbose = TRUE)## S3 method for class 'matrix'sadie(  data,  index = c("Perry", "Li-Madden-Xu", "all"),  nperm = 100,  seed = NULL,  threads = 1,  ...,  method = "shortsimplex",  verbose = TRUE)## S3 method for class 'count'sadie(  data,  index = c("Perry", "Li-Madden-Xu", "all"),  nperm = 100,  seed = NULL,  threads = 1,  ...,  method = "shortsimplex",  verbose = TRUE)## S3 method for class 'incidence'sadie(  data,  index = c("Perry", "Li-Madden-Xu", "all"),  nperm = 100,  seed = NULL,  threads = 1,  ...,  method = "shortsimplex",  verbose = TRUE)

Arguments

Details

By convention in the SADIE procedure, clustering indices for a donor unit(outflow) and a receiver unit (inflow) are positive and negative in sign,respectively.

Value

Asadie object.

References

Perry JN. 1995. Spatial analysis by distance indices. Journal of AnimalEcology 64, 303–314.doi:10.2307/5892

Perry JN, Winder L, Holland JM, Alston RD. 1999. Red–blue plots for detectingclusters in count data. Ecology Letters 2, 106–113.doi:10.1046/j.1461-0248.1999.22057.x

Li B, Madden LV, Xu X. 2012. Spatial analysis by distance indices: analternative local clustering index for studying spatial patterns. Methods inEcology and Evolution 3, 368–377.doi:10.1111/j.2041-210X.2011.00165.x

Examples

set.seed(123)# Create an intensity object:my_count <- count(aphids, mapping(x = xm, y = ym))# Only compute Perry's indices:my_res <- sadie(my_count)my_ressummary(my_res)plot(my_res)plot(my_res, isoclines = TRUE)set.seed(123)# Compute both Perry's and Li-Madden-Xu's indices (using multithreading):my_res <- sadie(my_count, index = "all", threads = 2, nperm = 20)my_ressummary(my_res)plot(my_res) # Identical to: plot(my_res, index = "Perry")plot(my_res, index = "Li-Madden-Xu")set.seed(123)# Using usual data frames instead of intensity objects:my_df <- aphids[, c("xm", "ym", "i")]sadie(my_df)

Examples of simulated epidemic data.

Description

Epidemics were generated using the stochastic simulator from Xu and Madden(2004). The data consist of the numbers of diseased plants per samplingunit (out of a total of n = 100 plants in each sampling unit). N = 144sampling units, and different values for the parameterspattern andmu were used for the simulations.

Usage

simulated_epidemics

Format

A data frame with 864 rows and 6 variables:

Source

Xu XM, Madden LV. 2004. Use of SADIE statistics to study spatialdynamics of plant disease epidemics. Plant Pathology 53, 38–49.doi:10.1111/j.1365-3059.2004.00949.x

Simple maximum likelihood estimation

Description

By default, this function performs a maximum likelihood estimation for one orseveral parameters, but it can be used for any other optimization problems.The interface is intented to be rather simple while allowing more advancedparametrizations.

Usage

smle(data, ...)## Default S3 method:smle(data, f, param_init, max = TRUE, ...)## S3 method for class 'intensity'smle(data, f, param_init, max = TRUE, ...)

Arguments

Details

Theoptim tool does the hard work under the hood. Extraarguments (e.g. method of optimization to be used) can be passed tooptim through the... argument. Note thatcontrary to the defaultoptim arguments,smletries to solve a maximization problem using the method "L-BFGS-B" by default(seeoptim documentation for more information).

Value

An object of classsmle, which is a list containing the followingcomponents:

Examples

set.seed(12345)data <- rlogis(100, location = 5, scale = 2)ll_logis <- function(data, param = c(location = 0, scale = 1)) {    sum(dlogis(x = data, location = param[["location"]],               scale = param[["scale"]], log = TRUE))}res <- smle(data, ll_logis)ressummary(res)# Using the magrittr syntax:require(magrittr)data %>% smle(ll_logis)# Comparision with the output of fitdistr (MASS package), which works for a# limited number of predefined distributions:require(MASS)fitdistr(data, "logistic")# Example with an intensity object:require(magrittr)require(dplyr)data <- tomato_tswv$field_1929 %>%    filter(t == 1) %>%    incidence() %>%    clump(unit_size = c(x = 3, y = 3))ll_betabinom <- function(data, param) {    sum(dbetabinom(x = data[["i"]], size = data[["n"]],                   prob = param[["prob"]], theta = param[["theta"]],                   log = TRUE))}epsilon <- 1e-7res <- smle(data, ll_betabinom, param_init = c(prob = 0.5, theta = 0.5),            lower = c(prob  = 0 + epsilon,                      theta = 0 + epsilon),            upper = c(prob = 1 - epsilon,                      theta = Inf))ressummary(res)param_init <- data.frame(lower = c(0 + epsilon, 0 + epsilon),                         start = c(0.5, 0.5),                         upper = c(1 - epsilon, Inf))rownames(param_init) <- c("prob", "theta")res <- smle(data, ll_betabinom, param_init)ressummary(res)

Wrappers using maximum likelihood estimation for some distributions

Description

These functions are the core of the fitting processes performed infit_two_distr.

Usage

smle_pois(data)smle_nbinom(data)smle_binom(data)smle_betabinom(data)

Arguments

Value

Seesmle

Examples

set.seed(12345)data <- rpois(100, lambda = 5)res <- smle_pois(data)ressummary(res)data <- count(aphids)res <- smle_pois(data)ressummary(res)

Spatial hierarchy analysis.

Description

The manner in which the data are collected provides information aboutaggregation of disease at different levels in a spatial hierarchy (Hughes etal. 1997). For example, a sampling unit (upper level) can be reported as"healthy", if no diseased leaves (lower level) were found within the samplingunit.

Usage

spatial_hier(low, high)

Arguments

Details

In a pairwise comparison between levels, the probability that an individualat the lower hierarchical level is diseased is denoted plow, and phigh refersto the probability of disease at the higher level. The relationship betweenthese two probabilities can be written as

phigh = 1 - (1 - plow)^nu

where n is a parameter ranging from 0 to the corresponding number ofindividuals at the hierarchical level referenced by plow. If the value of nis equal to the number of individuals at the lower hierarchical levelcontained in a unit of the higher level (n low ), this suggests that there isno aggregation of disease incidence at the lower level. Conversely, a valueof n less than n low is indicative of aggregation at that level. The value ofn can be interpreted as an effective sample size (Hughes and Gottwald 1999;Madden and Hughes 1999) in the statistical sense that its value indicates thenumber of independent pieces of information at the lower level. Here, theeffective sample size concerns the equating of the zero-term of the binomialdistribution with the zero-term of an overdispersed distribution, asdescribed in Madden and Hughes (1999). Using the complementary log-logtransformation, CLL(x) = ln(-ln(1-x)), one can rewrite the Equation 5 asfollows (Madden et al. 2007):

CLL(phigh) = ln(nu) + CLL(plow)

from which the value of ln(n) can be obtained as the intercept of a linearregression when the slope is constrained to 1.

Value

Aspatial_hier object.

Examples

my_data_low <- incidence(tomato_tswv$field_1929)my_data_low <- clump(my_data_low, c(x = 3, y = 3))my_data_high <- my_data_lowmy_data_high$data$n <- 1my_data_high$data$i <- ifelse(my_data_high$data$i > 0, 1, 0)my_data_low  <- split(my_data_low, by = "t")my_data_high <- split(my_data_high, by = "t")res <- spatial_hier(my_data_low, my_data_high)ressummary(res)plot(res)

Divide into groups and reassemble.

Description

Divide into groups and reassemble.

Usage

## S3 method for class 'intensity'split(x, f, drop = FALSE, ..., by, unit_size)

Arguments

Value

A list ofintensity objects.

Examples

my_incidence <- incidence(tomato_tswv$field_1929)plot(my_incidence, type = "all")my_incidence_spl1 <- split(my_incidence, by = "t")my_incidence_spl2 <- split(my_incidence, unit_size = c(x = 8, y = 20, t = 1))

To go to higher level in the hierarchy.

Description

This function transforms the current numeric vector orintensity dataset into a "simplified black and white image" of this same data set: everyvalue of disease intensity below and above a given threshold is given thevalue 0 and 1, respectively.

Usage

threshold(data, value, ...)

Arguments

Details

By default, everything above 0 is given 1, and 0 stays at 0.thresholdis thus useful to report a whole sampling unit as "healthy" (0), if nodiseased individual at all was found within the sampling unit, or "diseased"(1) if at least one diseased individual was found.

Value

A numeric vector or anintensity object.

Examples

my_incidence <- incidence(tomato_tswv$field_1929)plot(my_incidence, type = "all")my_incidence_clumped_1 <- clump(my_incidence, unit_size = c(x = 3, y = 3))plot(my_incidence_clumped_1, type = "all")my_incidence_thr <- threshold(my_incidence_clumped_1, value = 4)plot(my_incidence_thr, type = "all")

Incidence of tobacco plants infected with viruses.

Description

Experimental plot consisted of 75 sampling units with 40 tobacco plants ineach one.

Usage

tobacco_viruses

Format

A data frame with 75 rows and 2 variables:

Source

Madden LV, Pirone TP, Raccah B. 1987. Analysis of spatial patterns ofvirus-diseased tobacco plants. Phytopathology 77, 1409–1417.

Incidence of tomato spotted wilt virus (TSWV) disease in field trials.

Description

Intensively mapped TSWV incidence data reported by Cochran (1936) and Bald(1937). The disease assessments were performed in field trials at the WaiteInstitute (Australia) in 1928 and 1929. TSWV is a virus disease spread bythrips.

Usage

tomato_tswv

Format

There are two data frames:

field_1928: A data frame with 11088 rows and 8 variables:

field_1929: A data frame with 4320 rows and 5 variables:

Details

The data setfield_1928, reported by Bald (1937), was a set of fourplots. Each plot consisted of 14 rows containing 33 plants each, so thatthere were 462 plants in each plot. The tomato variety Early Dwarf Red wasused in two plots, and the variety Burwood Prize in the other two. Thetomatoes were planted out on 15th October 1928. The two plots dedicated to agiven variety experienced different irrigation practices, using eitheroverhead sprays or trenches. Otherwise, all were treated alike. Weeklyrecords of TSWV incidence were performed from 6th November to 12th December.

The data setfield_1929, reported by Cochran (1936), was a field of 24rows containing 60 plants each, so that there were 1440 plants. The tomatoeswere planted out in 26th November 1929. TSWV incidence records made on 18thDecember 1929, 31st December 1929 and 22nd January 1930 are reported in thisdata set.

Source

Cochran WG. 1936. The statistical analysis of field counts ofdiseased plants. Supplement to the Journal of the Royal StatisticalSociety 3, 49–67.doi:10.2307/2983677

Bald JG. 1937. Investigations on "spotted wilt" of tomatoes. III.Infection in field plots. Bulletin 106. Melbourne, Australia: Council forScientific and Industrial Research.

Calculate Variance-Covariance Matrix for a Fitted Model Object

Description

Returns the variance-covariance matrix of the main parameters ofa fitted model object. The “main” parameters of modelcorrespond to those returned bycoef, and typically donot contain a nuisance scale parameter (sigma).

Usage

## S3 method for class 'smle'vcov(object, ...)

Arguments

Value

A matrix of the estimated covariances between the parameter estimatesin the linear or non-linear predictor of the model. This should haverow and column names corresponding to the parameter names given by thecoef method.

When some coefficients of the (linear) model are undetermined andhenceNA because of linearly dependent terms (or an“over specified” model), also called“aliased”, seealias, then sinceR version 3.5.0,vcov() (iffcomplete = TRUE, i.e., by default forlm etc, but not foraov) contains corresponding rows andcolumns ofNAs, wherevercoef() has alwayscontained suchNAs.

Z-test.

Description

Performs z-tests for Fisher's aggregation indices (computed with either countor incidence data).

Usage

z.test(x, ...)## Default S3 method:z.test(x, ...)## S3 method for class 'fisher'z.test(  x,  alternative = c("two.sided", "less", "greater"),  conf.level = 0.95,  ...)

Arguments

Details

For two-sided tests with a confidence level of 95the spatial pattern would be random. If z < -1.96 or z > 1.96, it would beuniform or aggregated, respectively.

Value

Same kind of object as the one returns by the statschisq.test function for example.

References

For count and incidence data:

Moradi-Vajargah M, Golizadeh A, Rafiee-Dastjerdi H, Zalucki MP, Hassanpour M,Naseri B. 2011. Population density and spatial distribution pattern of Hyperapostica (Coleoptera: Curculionidae) in Ardabil, Iran. Notulae Botanicae HortiAgrobotanici Cluj-Napoca, 39(2): 42-48.

Sun P, Madden LV. 1997. Using a normal approximation to test for the binomialdistribution. Biometrical journal, 39(5): 533-544.

See Also

calpha.test,chisq.test

Examples

# For incidence data:my_incidence <- incidence(tobacco_viruses)my_fisher <- agg_index(my_incidence, method = "fisher")z.test(my_fisher)