In BastilleRousseau/IndRSA: An R package for individual-level Resource Selection Analysis

``` {r eval=T, echo=F, message=F, warning=F}

library(devtools)

install_github("BastilleRousseau/IndRSA")

library(IndRSA) library(survival) simu_coefs<-function(coef, se, n=1000) { out<-array(NA, c(nrow(coef), ncol(coef), n)) for (i in 1:nrow(coef)) { for (j in 2:(ncol(coef)-2)) { out[i,j,]<-rnorm(n,coef[i,j], se[i,j]) }} return(out) } simu_avg<-function(simu) { tt<-(matrix(unlist(lapply(1:dim(simu)[3], function(x) colMeans(simu[,,x], na.rm=T))), nrow=dim(simu)[3], ncol=dim(simu)[2], byrow=T)) }

simu_spe<-function(simu) { return(matrix(unlist(lapply(1:dim(simu)[3], function(x) apply(simu[,,x], 2, function(y) mean(abs(y), na.rm=T)))), nrow=dim(simu)[3], ncol=dim(simu)[2], byrow=T)) }

simu_sd<-function(simu) { return(matrix(unlist(lapply(1:dim(simu)[3], function(x) apply(simu[,,x], 2, function(y) sd(y, na.rm=T)))), nrow=dim(simu)[3], ncol=dim(simu)[2], byrow=T)) }

simu_cons2<-function(simu, col1, col2) { return(unlist(lapply(1:dim(simu)[3], function(x) mean(abs(simu[,col1,x]-simu[,col2,x]), na.rm=T) ))) }

simu_cons3<-function(simu, col1, col2, col3) { return(unlist(lapply(1:dim(simu)[3], function(x) mean(abs(simu[,col1,x]-simu[,col2,x])+abs(simu[,col1,x]-simu[,col3,x])+abs(simu[,col3,x]-simu[,col2,x])/3, na.rm=T) ))) }

simu_rev2<-function(simu, col1, col2) { return(unlist(lapply(1:dim(simu)[3], function(x) mean(ifelse(sign(simu[,col1,x])!= sign(simu[,col2,x]), 1, 0), na.rm=T)))) }

simu_rev3<-function(simu, col1, col2, col3) { return(unlist(lapply(1:dim(simu)[3], function(x) mean(c(ifelse(sign(simu[,col1,x])!= sign(simu[,col2,x]), 1, 0), ifelse(sign(simu[,col1,x])== sign(simu[,col3,x]), 1, 0), ifelse(sign(simu[,col3,x])== sign(simu[,col2,x]), 1, 0)),na.rm=T)))) }

# Individual and population-level RSF - *rsf_ind* and *pop_avg*

The function *rsf_ind* performs individual-level RSFs. The function takes a list of candidate models (using *as.formula()*). The function *aictab_ind* performs population-level AIC model selection (by adding up individual model loglikelihood). *pop_avg* performs population-level averaging. Two methods are available for calculating confidence intervals, the default (and recommended) is based on Murtaugh (2007). A bootstrap approach (based on Prokopenko 2016) is also available. *ind_coef* and *ind_se* also extract individual-level coefficients and standard errors.<br> <br>
Murtaugh, P. Simplicity and complexity in ecological data anlysis. Ecology 88, 56–62 (2007).<br> <br>
Prokopenko, C. M., Boyce, M. S. & Avgar, T. Characterizing wildlife behavioural responses to roads using integrated step selection analysis. J. Appl. Ecol. 1, (2016).<br> <br>
\newline

``` {r eval=T}
data(goats)
ls1<-list()
ls1[[1]]<-as.formula(STATUS~ELEVATION+SLOPE+ET+ASPECT+HLI+TASP)
ls1[[2]]<-as.formula(STATUS~ET+ASPECT+HLI+TASP)
out<-rsf_ind(goats$ID, data=goats, form_ls=ls1)
aictab_ind(out) # Model 1 is the best 

#Calculate population average for the first model of the list (m=1)
pop<-pop_avg(m=1, out, method="murtaugh") 
pop[[1]] #Population average summary 
pop[[2]] #Individual level coefficients 
ind_coef(m=1, out) #Another way of getting individual coefficients 
ind_se(m=1, out) #Individual level standard errors 
kfold_ind(m=1, out, ls=ls1, grph=F) #kfold cross validation for each individual  

Individual and population-level SSF - ssf_ind and pop_avg

The function ssf_ind performs individual-level SSFs similar to rsf_ind. The function takes a list of candidate models (using as.formula()). The function aictab_ind, pop_avg, ind_coef, and ind_se works similarly than in example above. For the example below, we created a new column strata in the goat dataset that assigns two random points to each used points to create a conditional design similar to an SSF

\newline

``` {r eval=T} data(goats) goats<-goats[order(goats$ID),] goats_use<-goats[goats$STATUS==1,] goats_use$strata<-1:nrow(goats_use) goats_rnd<-goats[goats$STATUS==0,] goats_rnd$strata<-rep(1:nrow(goats_use), each=2) goats_ssf<-rbind(goats_use, goats_rnd) ls1<-list() ls1[[1]]<-as.formula(STATUS~ELEVATION+SLOPE+ET+ASPECT+HLI+TASP+strata(strata)) ls1[[2]]<-as.formula(STATUS~ET+ASPECT+HLI+TASP+strata(strata)) out<-ssf_ind(goats_ssf$ID, data=goats_ssf, form_ls=ls1) aictab_ind(out) pop<-pop_avg(m=1, out, method="murtaugh") pop[[1]] #Population average summary pop[[2]] #Individual level coefficients coef<-ind_coef(m=1, out) se<-ind_se(m=1, out)

# Metrics of variation in resource selection behavior (specialization, heterogeneity, consistency and reversal)  - *simu_spe*, *simu_sd*, *simu_consX*, and *simu_revX*
The functions *simu_spe* and *simu_sd* calculates specialization and heterogeneity metrics as described in Bastille-Rousseau & Wittemyer (in review). To propagate uncertainty associated to individual coefficients, *simu_coefs* is first used to simulate individual coefficients based on the coefficient value and its standard error. <br> <br>
\newline

``` {r eval=T}
data(goats)
ls1<-list()
ls1[[1]]<-as.formula(STATUS~ELEVATION+SLOPE+ET+ASPECT+HLI+TASP)
out<-rsf_ind(goats$ID, data=goats, form_ls=ls1)
coef<-ind_coef(m=1, out)
se<-ind_se(m=1, out)
simu<-simu_coefs(coef, se, n=100)

#Specialization 
spe<-simu_spe(simu)
colnames(spe)<-names(coef)
head(spe)
apply(spe, 2, quantile, na.rm=T) #Show variation around estimate of each covariate
colMeans(spe) #Calculate average specialization for each covariate

#Heterogeneity
sd<-simu_sd(simu)
colnames(sd)<-names(coef)
head(sd)
apply(sd, 2, quantile, na.rm=T) #Show variation around estimate of each covariate
colMeans(sd) #Calculate average heterogeneity for each covariate

For consistency and reversal, calculations are done one covariate at a time. simu_cons2 and simu_rev2 are used for calculations when there are two time periods and simu_cons3 and simu_rev3 are used when there is three temporal periods. For the example below, we created an artificial Season column to the goat dataset to estimate temporal consistency and reversal.

\newline

``` {r eval=T} data(goats) goats$Season<-c("1", "2") #Adding a fake season column ls1<-list() ls1[[1]]<-as.formula(STATUS~(ELEVATION+SLOPE+ET+ASPECT+HLI+TASP):Season)
out<-rsf_ind(goats$ID, data=goats, form_ls=ls1) coef<-ind_coef(m=1, out) se<-ind_se(m=1, out) simu<-simu_coefs(coef, se, n=100)

Consistency for elevation

head(coef)

Calculate specialization for elevation covariate, column 3 and 4 contains

coefficients for elevation for each season

cons_elevation<-simu_cons2(simu, 3, 4) quantile(cons_elevation) #Show variation around estimate of elevation covariate mean(cons_elevation) #Calculate average consistency for elevation covariate

Reversal for elevation

head(coef) rev_elevation<-simu_rev2(simu, 3, 4) #Calculate specialization for elevation covariate quantile(rev_elevation) #Show variation around estimate of elevation covariate mean(rev_elevation) #Calculate average reversal for elevation covariate

```

BastilleRousseau/IndRSA documentation built on July 1, 2024, 7:48 a.m.