indpower: Indicator Power of Species
In pattakosn/Rworkshop: Community Ecology Package

Description Usage Arguments Details Value Author(s) References See Also Examples

Indicator power calculation of Halme et al. (2009) or the congruence between indicator and target species.

1	indpower(x, type = 0)

`x`	Community data frame or matrix.
`type`	The type of statistic to be returned. See Details for explanation.

Halme et al. (2009) described an index of indicator power defined as IP_I = sqrt(a*b), where a = S / O_I and b = 1 - (O_T - S) / (N - O_I). N is the number of sites, S is the number of shared occurrences of the indicator (I) and the target (T) species. O_I and O_T are number of occurrences of the indicator and target species. The type argument in the function call enables to choose which statistic to return. type = 0 returns IP_I, type = 1 returns a, type = 2 returns b. Total indicator power (TIP) of an indicator species is the column mean (without its own value, see examples). Halme et al. (2009) explain how to calculate confidence intervals for these statistics, see Examples.

A matrix with indicator species as rows and target species as columns (this is indicated by the first letters of the row/column names).

Peter Solymos

Halme, P., Mönkkönen, M., Kotiaho, J. S, Ylisirniö, A-L. 2009. Quantifying the indicator power of an indicator species. Conservation Biology 23: 1008–1016.

indval (package labdsv) for the indicator species analysis of Dufrêne & Legendre. Function beals estimates individual cell probabilities of species occurrences.

data(dune)
## IP values
ip <- indpower(dune)
## and TIP values
diag(ip) <- NA
(TIP <- rowMeans(ip, na.rm=TRUE))

## p value calculation for a species
## from Halme et al. 2009
## i is ID for the species
i <- 1
fun <- function(x, i) indpower(x)[i,-i]
## 'c0' randomizes species occurrences
os <- oecosimu(dune, fun, "c0", i=i, nsimul=99)
## get z values from oecosimu output
z <- os$oecosimu$z
## p-value
(p <- sum(z) / sqrt(length(z)))
## 'heterogeneity' measure
(chi2 <- sum((z - mean(z))^2))
pchisq(chi2, df=length(z)-1)
## Halme et al.'s suggested output
out <- c(TIP=TIP[i], 
    significance=p,
    heterogeneity=chi2,
    minIP=min(fun(dune, i=i)),
    varIP=sd(fun(dune, i=i)^2))
out