boot.transitions: Bootstrap observed census transitions

Description

Calculate bootstrap distributions of population growth rates (lambda), stage vectors, and projection matrix elements by randomly sampling with replacement from a stage-fate data frame of observed transitions

Usage

boot.transitions(transitions, iterations, by.stage.counts = FALSE, ...)

Arguments

transitions

a stage-fate data frame with stage or age class in the current census, fate in the subsequent census, and one or more fertility columns

iterations

Number of bootstrap iterations

by.stage.counts

Resample transitions with equal probability (default) or by subsets of initial stage counts

...

additional options passed to projection.matrix

Value

A list with 3 items
lambdaA vector containing bootstrap values for lambda
matrixA matrix containing bootstrap transtion matrices with one projection matrix per row.
vectorA matrix containing bootstrap stage vectors with one stage vector per row.

References

see Morris and Doak 2005 in http://esapubs.org/Archive/mono/M075/004/appendix-A.htm for resampling by stage class counts

Examples

Run this code

data(test.census)

## create stage-fate dataframe using merge and subset
test.trans <- subset(
             merge(test.census, test.census, by="plant", sort=FALSE),  
                     year.x==2001 & year.y==2002)

## format column and row names
test.trans<-test.trans[,c(1:4,6)]
colnames(test.trans)[2:5] <- c("year", "stage", "fruits", "fate")
rownames(test.trans) <- 1:nrow(test.trans)
# order stage columns corresponding to matrix
test.trans$stage <- ordered(test.trans$stage, levels = c("seedling", "vegetative", "reproductive"))


## add individual fertilities using prebreeding census with no seed bank 
##  based on the proportional reproductive outputs of flowering plants 
## and the total number of seedlings at the end of the projection interval

seedlings<-nrow(subset(test.census, year==2002 & stage=="seedling"))
test.trans$seedling<-test.trans$fruits/sum(test.trans$fruits) * seedlings

test.trans

## STEP by step instructions for bootstrapping dataframe

n<-nrow(test.trans)
n

x <- sample(n, replace=TRUE)
x

test.trans[x,]


## or respample by stage class counts 

 lapply(split(test.trans, test.trans$stage, drop=TRUE), 
      function(x) x[sample(nrow(x), replace=TRUE),])

## using boot.transitions

boot.transitions(test.trans, 5)

boot.transitions(test.trans, 5, by.stage=TRUE)


## Aquilegia example


data(aq.trans)

x<-subset(aq.trans, year==1996)

# calculate lamda, seed survival and recruitment rate using aq.matrix
rec<-nrow(subset(aq.trans, year==1997 & stage == "recruit"))
aq.96<- aq.matrix(x, rec)

# add  individual fertilities to data frame only
aq.96.trans<-aq.matrix(x, rec, summary=FALSE)

# pass estimated transitions in aq.96 to projection matrix 
aq.96.boot<-boot.transitions(aq.96.trans, 200, 
            add=c(1,1, aq.96$seed.survival, 2,1, aq.96$recruitment.rate) )

# calculate confidence intervals using quantile()
ci<- quantile(aq.96.boot$lambda, c(0.025,0.975) )

aq.96$lambda
ci

# plot histogram
hist(aq.96.boot$lambda, col="green", xlab="Lambda", 
        main=paste('Bootstrap estimates of population<ngrowth>rate from 1996-1997'))
abline(v=ci, lty=3)</ngrowth>
<keyword>survey</keyword>