[R-sig-eco] K function

[marcelino.delacruz at upm.es]

Hi Yong,

You are talking about two different things.

Usually, analyzing the spatial pattern of tree mortality involves testing
against the random labeling  hypothesis (i.e., testing against random
mortality). For this you should go through the following steps:

1) build a marked point pattern, e.g.,

trees <- ppp(x= yourXcoordinates, y=yourYcoordinates,
marks=yourmortalitydata, window=yourwindow)

The mortality data should be a factor with two levels like yes/no,
dead/alive, etc.

2)Compute a summary statistics (e.g. the K-cross function, although you
could use the simple K-function with more scripting) and compute
envelopes simulating from the random labelling null model:

mortality.test <- envelope(trees, Kcross, i="death", j="alive",
nsim=99, simulation=expression(rlabel(trees)))

The important thing for your test here is the simulation that you get
with the rlabel function

3) plot the results, e.g.

plot(mortality.test, sqrt(./pi)-r~r, ylabel=expression(L[d,a](r)))

I published some "new" approaches for testing spatial patterns of
mortality in a paper in Ecography (2008) and included some functions for
this  in the R package "ecespa" (for example functions K012 and K1K2)
that you may also consider.

To "test" the patterns of tree recruitment, it depends on the
hypotehsis that you want to test. The old one was the CSR hipothesis,
just

envelope (recruitedtrees, Kest)

but this is not much informative.

HTH,

Marcelino



Con fecha 12/10/2011, "ÉòÓÂ" <winnie56100 at 163.com> escribió:

>Hi all,
>  I have a question for using the K  function (package "spatstat", spatial pattern), I want to test the spatial pattern of mortality tree using K function, is it the same as testing the spatial pattern of tree? Random simulation by removing the points which number is equal to the mortality tree from the original data set, are there any functions can implement this analysis? Or I should write a programming to perform it? And if I want to test the spatial pattern of recruitment tree, how to perform that? Thank you very much!
>Best wishes!
>                                                                                                                                                         Yong
>	[[alternative HTML version deleted]]
>
>_______________________________________________
>R-sig-ecology mailing list
>R-sig-ecology at r-project.org
>https://stat.ethz.ch/mailman/listinfo/r-sig-ecology