Gives a weight to each pseudospecies, so that the rare ones weigh less
in the ordination.
The original TWINSPAN downweights them before the correspondence
analysis, and twinspan() does the same by default.
The weights are used only in the ordination:
the preference of the pseudospecies is counted on the raw occurrences.
Usage
tw_downweight(
y,
method = c("hill", "decorana"),
fraction = 5,
frq_lim = 0.2,
w_min = 0.01,
rw = NULL
)Arguments
- y
A binary matrix of stands by pseudospecies.
- method
A string, "hill" or "decorana".
- fraction
A numeric of the downweighting fraction of "decorana".
- frq_lim
A numeric of the frequency above which "hill" does not downweight.
- w_min
A numeric of the smallest weight of "hill".
- rw
A numeric vector of stand weights, or NULL.
Details
Two ways are available.
"hill" is the WEIGHT subroutine of the original TWINSPAN:
a pseudospecies occurring in a smaller proportion of the stands than
frq_lim is weighted in proportion to that shortfall, and no weight
falls below w_min.
"decorana" is the downweighting of decorana() and of
vegan::downweight(), where the frequencies are compared with the
most frequent pseudospecies instead of a fixed proportion.
Examples
# \donttest{
data(dune, package = "vegan")
psp <- pseudospecies(dune)
summary(tw_downweight(psp))
#> Min. 1st Qu. Median Mean 3rd Qu. Max.
#> 0.2575 0.6288 1.0000 0.8020 1.0000 1.0000
summary(tw_downweight(psp, method = "decorana"))
#> Min. 1st Qu. Median Mean 3rd Qu. Max.
#> 0.2778 0.6944 1.0000 0.8207 1.0000 1.0000
# }