hnp (version 1.2-7)

fungi: Diaphorina citri mortality data

Description

Mortality of the Citrus psyllid, Diaphorina citri, a major pest of Citrus worldwide, when exposed to different concentrations of two fungi species, Beauveria bassiana and Isaria fumosorosea.

Usage

data(fungi)

Format

A data frame with 30 observations on the following 5 variables.

y	numeric	number of dead insects
m	numeric	total number of insects
conc	numeric	fungi concentration (in conidia/ml)
lconc	numeric	natural logarithm of fungi concentration
species	factor	levels isaria and beauveria, fungi species

Details

The Citrus psyllid Diaphorina citri is a vector of Huanglongbing, known as greening disease. An alternative to chemical control is to use solutions of fungi conidia as a biological control strategy. D'Alessandro (2014) conducted a completely randomized experiment to assess how different conidia concentrations ($10^4$, $10^5$, $10^6$, $10^7$ and $10^8$ conidia/ml) of two fungi species, Beauveria bassiana and Isaria fumosorosea, infected D. citri adults. Each experimental unit consisted of around 20 D. citri adults, which were placed on Citrus limonia plants. The insects were pulverized with the solutions and after 10 days the number of dead insects and dead insects due to fungus infection were observed. Note that in this case the conidia concentrations are obtained in successive dilutions and therefore small variations in the number of conidia per ml may contribute additional variability to the data. Such additional variability may be accounted for in the model by including an additive random effect in the linear predictor.

References

D'Alessandro (2014) Unpublished data, private communication.

Examples

data(fungi)

# Binomial fit
model1 <- glm(cbind(y, m-y) ~ lconc*species,
              family=binomial, data=fungi)
anova(model1, test="Chisq")
sum(resid(model1, ty="pearson")^2)
1 - pchisq(sum(resid(model1, ty="pearson")^2), 20)
hnp(model1)

# Quasi-binomial fit
model2 <- glm(cbind(y, m-y) ~ lconc*species,
              family=quasibinomial, data=fungi)
anova(model2, test="F")
hnp(model2)

if (FALSE) {
# Logistic-normal fit
require(lme4)
fungi$ind <- factor(1:nrow(fungi))
model3 <- glmer(cbind(y, m-y) ~ lconc*species + (1|ind),
                family=binomial, data=fungi)
summary(model3)
hnp(model3)
}

## for discussion on the analysis of this data set,
## see Demetrio et al. (2014)