######################################################################
#####Supplementary material - R script:###############################
######################################################################


#Bacterial communities differ among Drosophila melanogaster populations and affect host resistance against parasitoids
#Mariia Chaplinska, Sylvia Gerritsma, Francisco Dini-Andreote, Joana Falcao Salles, &, Bregje Wertheim

#R script for statistical analyses of:
#Effect of antibiotic treatment on parasitoid resistance in D melanogaster
#Effect of antibiotic treatment on mortality rate

##################################################################################################
#Load packages:
library(lattice)
library(plotrix)
library(lme4)
library(multcomp)

#Data file: R_data.txt
#Parameters:
# Line
# Treatment: C=control, T=antibiotic treatment
# Sample: sample or replica (1-10)
# Pupae: counted pupae after parasitization
# Flies: total emerged flies (with and without capsule)
# FliesCaps: flies with one or more capsules
# Wasps: wasp count
#Samples removed (no data): GOTH_s4; STA_c4,r1,r7,s4; GRO_c4,c5,c6,s2,s3,s8; BRE_c3, r4, s3; BAY_s9

data<-read.table("R_data.txt",sep="\t", quote="",header=T)
names(data)

###########################################################################################################
##Encapsulation rate ######################################################################################
###########################################################################################################

#Encapsulation rate
#Calculate proportion flies that encapsulated wasp egg
ER<-data$FliesCaps/(data$FliesCaps+data$Wasps)
data$ER<-ER 
ER_mean<-tapply(data$ER,list(data$Line,data$Treatment),mean,na.rm=T) #calculate ER mean per line per treatment
ER_mean
#         C         T
#ARL  0.3055533 0.4016973
#BAY  0.6591876 0.7014547
#BRE  0.4463490 0.3603293
#GOTH 0.7889191 0.9102583
#GRO  0.3156563 0.4343155
#STA  0.5995743 0.3769421

#---------------------------------------------------------------------------------------------------------
#GLM: Is there an effect of Line and treatment on encapsulation rate?
#---------------------------------------------------------------------------------------------------------
#Calculate proportion capsule:wasps (success:failure)
y<-cbind(data$FliesCaps, data$Wasps)

M1<-glm(y~Line*Treatment, data=data, family=quasibinomial)
anova(M1, test="F") 
M2<-update(M1,~.-Line:Treatment)
anova( M1, M2,test="F") 
#Significant interaction: F=3.5629, df=152,5, p=0.004481. 

#---------------------------------------------------------------------------------------------------------
#Multiple comparisons between treatment and control within lines
#---------------------------------------------------------------------------------------------------------
#For the multiple comparison we combine the varaibles Line and Treament: variable Line_T in the dataset
M3<-glm(y~Line_T, data=data, family=quasibinomial)
anova(M3, test="F") 

#Comparison between antibiotic treament and control within lines
# Perform Tukey's Honest Significant Difference.
a <- summary(glht(model = M3, linfct = mcp(Line_T = c("GOTH_C - GOTH_T = 0", "STA_C - STA_T = 0","GRO_C - GRO_T = 0","ARL_C - ARL_T = 0","BRE_C - BRE_T = 0", "BAY_C - BAY_T = 0"))))

# Write test results to file 'multcomp_results.txt'.
sink(file = "multcomp_results.txt", type = c("output"))
a
sink()


#---------------------------------------------------------------------------------------------------------
#Plot encapsulation rate per treatment per Line. Black open (control) and closed (treatment) circles
#---------------------------------------------------------------------------------------------------------
par (mfrow = c(1,1))
Order_Lines<-factor(data$Line,levels=c("ARL","BRE","GRO","STA","BAY","GOTH"))
interaction.plot(Order_Lines, data$Treatment, data$ER , legend=F, type="b", ylim=c(0,1), las=1, ylab="Encapsulation rate",xlab=NA,cex.axis=1.2,cex.lab=1.2, lwd=2,pch=c(1,16),cex=1.5,lty="blank")
plotCI(rep(1:6, each=2),tapply(data$ER, list(data$Treatment, Order_Lines), mean), tapply(data$ER, list(data$Treatment, Order_Lines), std.error), add=TRUE, pch=c(1,16),cex=1.5)


###########################################################################################################
##Mortality ###############################################################################################
###########################################################################################################
#Exclude samples ARL T4 [92], ARL T8 [106] and Bay T10 [164] --> no data on pupae count available

EmergedInd<-data$Flies[-c(92,106,164)]+data$Wasps[-c(92,106,164)]
Mort<-data$Pupae[-c(92,106,164)]-EmergedInd
Z<-cbind(Mort,EmergedInd)

#---------------------------------------------------------------------------------------------------------
#GLM: Is there an effect of Line and treatment on mortality rate?
#---------------------------------------------------------------------------------------------------------

MM1<-glm(Z~Line*Treatment, data=data[-c(92,106,164),], family=quasibinomial)
summary(MM1)
anova(MM1, test="F")
MM2<-update(MM1,~.-Line:Treatment)
anova( MM1, MM2,test="F")
MM3<-update(MM2,~.-Treatment) 
anova( MM2, MM3,test="F") #No effect of Treatment: F=1.8541, DF=1,155, p=0.1753 


#Percentage mortality:
MortPerc<-Mort/data$Pupae[-c(92,106,164)]
MortPercL<-tapply(MortPerc, list(data$Line[-c(92,106,164)],data$Treatment[-c(92,106,164)]), mean)
MP<-tapply(MortPerc, data$Line[-c(92,106,164)], mean, na.rm=T)
OrderMP<-sort(MP)
Order_Lines<-factor(data$Line[-c(92,106,164)],levels=c("GOTH","GRO","BAY","ARL","STA","BRE"))

#Plot + SE
interaction.plot(Order_Lines, data$Treatment[-c(92,106,164)], MortPerc , legend=F, type="b", ylim=c(0,1), las=1, ylab="Mortality rate",xlab=NA,cex.axis=1.2,cex.lab=1.2, lwd=2,pch=c(1,16),lty="blank")
plotCI(rep(1:6, each=2),tapply(MortPerc, list(data$Treatment[-c(92,106,164)], Order_Lines), mean), tapply(MortPerc, list(data$Treatment[-c(92,106,164)], Order_Lines), std.error), add=TRUE, pch=c(1,16),cex=1.5)



#########THE END##########################################################