Demographic projections

Description

Make deterministic and stochastic demographic projections according to a transition matrix.

Usage

projectn(v0, mat, matsd = NULL, estamb = FALSE, estdem = FALSE,
         equalsign = TRUE,stmat=NULL, fecundity1=TRUE,
        nrep = 1, time = 10, management=NULL, round=TRUE)
project1(v0, mat, matsd=NULL, estamb=FALSE, estdem=FALSE,
         equalsign=TRUE, stmat=NULL, fecundity1=TRUE)
estambi(mat, matsd, equalsign)
estdemo(v0,mat,stmat=NULL, fecundity1=TRUE)
## S3 method for class 'rmas'
plot(x, sum = TRUE, mean=FALSE, type="l", harvest=FALSE, ...)
## S3 method for class 'rmas'
summary(object, stage=NULL, harvest=FALSE,...)
## S3 method for class 'summary.rmas'
plot(x, ylim=NULL, col=NULL, xlab=NULL, ylab=NULL, main=NULL,...)

Arguments

Details

The function projectn makes demographic proyections by repetedly calling (as long as the time argument) to project1. If no environmental or demographic stochasticity is required, project1 will multiply the transition matrix mat by the vector of stage abundances.

If demographic stochasticity is required, project1 will call to estdemo function, that for each time will assign abundances to each stage based in sampling from both a binomial and a poisson distribution. Sampling from rbinom with probability mat[i,j] will assign "survival" chances to each individual accounted for in the abundance vector, i.e. it would allow some of them to remain in the same stages (for transitions mat[i,i]) or to pass to another stage (for transitions mat[i,j]). Sampling from rpois with mean mat[i,j] will assign to each individual of stage[j] a random number of offspring of type stage[i].

In the current implementation there are 3 options to generate demographic stochasticity. By deafult ( stmat=NULL and fecundity1=TRUE) probabilities in the first row of the transition matrix (i.e mat[1,j]) are assumed to represent only fecundities, i.e., they would not account for "survival" transitions from stage[j] to stage[1], but only for newborns. This means that these probabilities will only be used as the mean for sampling from rpois. Transitions in rows others than first row will be assumed to represent "survival" transitions if its value is <=1 and accordingly will be employed to sample from rbinom. Transitions >1 will be assumed to represent fecundities and will be employed to sample from rpois.

If stmat=NULL and fecundity1=FALSE transition probabilities in all the rows of the matrix mat are treated in the same way, i.e. probabilities <=1 will be sampled from rbinom and probabilities >1 will be sampled from rpois.

If a stmat matrix (a matrix with values between 0 and 1) is provided, it will be used to divide transition probabilities mat[i,j] into fecundities (mat[i,j] * stmat[i,j]) and survival probabilities (mat[i,j] - (mat[i,j] * stmat[i,j])) and these matrices will be used to sample from rbinom and rpois respectively.

The current implementation of estdemo assumes that reproduction takes place before "survival sorting", so even for individuals that wouldn't survive (according to its sampled binomial probability) offspring is computed (if approppriate) and accounted for.

If environmental stochasticity is required (i.e., a matsd matrix is provided), project1 will call to estambi function, that for each time will change mat[i,j] probabilities sampling from rnorm with mean= mat[i,j] and sd = matsd[i,j]. If equalsign=TRUE the random changes in all cells of mat will have the same sign and the same magnitude (relative to each individual matsd[i,j]). If equalsign=FALSE every transition probability will change independently.

If both environmental and demographic stochasticity are required project1 will call first to estambi and using the modified mat will call to estdemo.

If management is required, a vector or matrix of magement actions should be provided. In the simplest case (i.e., a vector), each element in the vector will be interpreted as the management action that will be applied each time step to the corresponding stage. Positive and negative elements in the vector represent respectively the introduction or extraction of individuals from the corresponding stage. Elements whose absolute value is >= 1 will be interpreted as the introduction or extraction of exactly that number of individuals; absolute values < 1 will be interpreted as the introduction or extraction of that proportion of individuals from the existing individuals in the corresponding stage. If "management" is a matrix, each time step the management actions represented by each column of the matrix will be applied sequentially, from first to last.

Value

project1, estdemo and estambi return a vector of stage abundances with the same length that v0.

projectn return an object of class rmas, basically a list with four elements: vn, with length = (nrep) where each of its elements represents a replicate simulation and consist of a matrix of dimensions [lengtth(v0), time] representing the abundance of each stage at each time. If the simulation included management actions, harvest will be a list of length (nrep) where each of its elements represents the trajectory of harvest in a replicate simulation and consist of a matrix of dimensions [lengtth(v0), time] representing the number of individuals extracted of each stage at each time. The other two elements, mat and management, are respectively the transition matrix and the managenet matrix employed in the simulations. The plot method will draw the demographic trajectory of the population. By default (sum = TRUE, mean=FALSE) it will plot the abundance of the whole population (i.e. the the sum of abundances in each stage) vs. time. If nrep >1 it will plot together the trajectory of each replicated population. If ( sum = TRUE, mean=TRUE) it will plot the mean of all repplicated populations. If (sum = FALSE, mean=TRUE) it wil plot the abundance (or the mean abundance in all the replications) of each stage vs. time.

The summary and plot.summary methods will print a table and draw a plot respectively with the maximun, mean + 1 sd, mean, mean - 1 sd and minimum values of population abundance in all the simulations.

Author(s)

Marcelino de la Cruz marcelino.delacruz@upm.es

References

Akcakaya, H. R., Burgman, M. A. and Ginzburg L.R. 1999. Applied Population Ecology. Sinauer. Caswell, H. 2003. Matrix Population Models: Construction, Analysis, and Interpretation . Sinauer.

Examples

   ## Not run: 
     data(coryphanthaA)
     coryphanthaA <- as.tmatrix(coryphanthaA)
  
     # run a deterministic simulation of 20 years from an initial state of
     # 100 small juveniles:
     v0 <- c(100,0,0)
     simu20 <- projectn(v0=v0, mat=coryphanthaA, time = 20)
     plot(simu20, sum=FALSE)
     summary(simu20)
  
     # run 100 simulations of 20 years with  demographic stochasticity:
     simu20.ds <- projectn(v0=v0, mat=coryphanthaA, time = 20, estdem=TRUE, nrep=100)
     plot(simu20.ds)
     summary(simu20.ds)
     
     # run 100 simulations of 20 years with  demographic stochasticity but 
     # assuming that the first row of the transition matrix represent both
     # fecundity and survival, each with a 50
     
     # first generate the stmat matrix:
     stmat <- (coryphanthaA >0)
     stmat <- stmat*c(0.5,0,0)
     stmat
     
     simu20.ds2 <- projectn(v0=v0, mat=coryphanthaA, time = 20, estdem=TRUE,
                             stmat=stmat, nrep=100)
     plot(simu20.ds2)
     summary(simu20.ds2)
          
  
     # run 100 simulations of 20 years with  both demographic and environmental
     # stochasticity:
     # first generate a sd matrix to describe environmental stochasticity:
     sdenv <- coryphanthaA/20 
     sdenv
     
     simu20.eds <- projectn(v0=v0, mat=coryphanthaA, matsd =sdenv,  time = 20,
                            estdem=TRUE,estamb=TRUE, nrep=100)
     plot(simu20.eds)
     summary(simu20.eds)
     
     # Example of management actions
     #    each time step, 10 individuals will be added to the first stage ,10 individuals  
     #    will be added to the second stage, and 50 percent of the individuals in the 
     #    third stage will be extracted
     
        man <- c(10, 10, -0.5)
        p1 <- projectn(v0 = c(100, 100,100), mat= coryphanthaA, management=man)
	
        # summarize and plot population trajectory
        summary(p1) 
	
	# summarizes and plots harvest history
	summary(p1, harvest=T)  
           
  
## End(Not run)

Results