Random presence points were drawn by bg.similarity.test() over the entire space
defined by argument env, i.e. the bioclimatic predictors, which in most cases is
maybe not a reasonable choice. Now you can use the new argument study.area.y = "mcp"
to sample random presence points of the second species (species Y of Warren et al. 2008)
from minimum spanning polygon laid around the observed presence points of Y. Thanks to Frank
Burbrink for bringing this issue up!
Only minor changes.
Only minor changes.
Only minor changes.
Only minor changes.
Due to an unconsistent behavior of sapply some data
produced meaningless p-values in age.range.correlation().
It has been replaced by lapply to be robust.
(thanks to Dan Warren for bug report)
There was a mistake in the calculation of p-values in
niche.equivalency.test() and the print method for both
niche.quivalency.test() and bg.similarity.test() stated the null
hypothesis instead of the alternative hyposthesis. (thanks to
Guarino Colli for bug report)
The confidence level for CIs in bg.similarity.test() can now be
set via the option conf.level.
niche.equivalency.test() and bg.similarity.test() are now able to use
categorical environmental covariates.
Calculation of nonparametric confidence intervals in
bg.similarity.test().
The code of functions niche.equivalency.test() and bg.similarity.test()
has been completely rewritten to match exactly the definition in
Warren et al. (2011). Previous versions had relied entirely on the
SWD-format (see MAXENT tutorial), which was quicker and easy to im-
plement at the cost of altering the definition of the back-
ground similarity test.
niche.equivalency.test() and bg.similarity.test() now produce output
objects of class 'ntest'. For class 'ntest' a print and a plot
method have been included, the latter replacing the former
non-generic functions plotNicheEquivalency() and
plotBackgroundSimilarity().
Plotting of tiplabels in function plotAncClim() has been completey
reworked and is now robust regarding the scale of the phylogenetic
tree. Have a look at ?plotAncClim and note the new arguments tipmode,
tipspace, cladespace, and nchar. The functionality of arguments
respectively. Note that cladespace is still in a somewhat
show.tips and xspace has been included in tipmode and
cladespace, experimental state.
Code of function pno() was reworked and is now ca. 20 percent faster
than in version 0.8.3. This is due to the replacement of the
dependency on adehabitat for raster map import by pno's own import
device for raster layers.
A check for correct path_model argument is included in function pno().
Improved documentation of plotAncClim().
Some users obtained PNOs containing NA values because the
distibution of NODATA values (e.g., -9999) in their bioclimatic data
and in their ENM models was different. This problem might arise when
raster maps differ slightly in their coastlines. The function pno()
will now issue a warning in these cases and treat the corresponding
cells as having zero probability in the ENM (e.g., the MAXENT
distribution).
plotPNO() is now capable of handling PNO matrixes with only one
species.
Only minor changes in the documentation of PHYLOCLIM.
The formula of Fitzpatrick and Turelli (2006) was incorrectly
implemented in the function age.range.correlation(). First, it
used 'n' (instead of n - 1) to weight the overlap statistic 'o'.
Second, the function was not summing 'o' correctly over all
species in both clades. (thanks to Pedro Abellán and
Ben Fitzpatrick for bug report and suggestions for a fix)
There was mistake in the formala for the calculation of
Hellingers distance in the original publication of Warren,
Glor and Turelli (2008, Evolution 62:2868-2883; see also
http://enmtools.blogspot.com/2010/09/error-in-i-metric-but-
dont-panic.html). The corrected formula is now implemented
in getDI(). As stressed by Dan Warren the error is only a
scaling issue and does not affect the p-values of the tests
based upon it. (thanks to Pascal Title for bug report)
Function bg.similarity.test() was mislead by different
choices of column names in the SWD files, e.g. "species"
instead of "spec". The function can now handle any column
name. (thanks to Pascal Title for bug report)
The functions niche.equivalency.test() and bg.similarity.test() have
a new argument mx, which controls the number of memory that is given
to JAVA during the calls to MAXENT.
The functions niche.equivalency.test() and bg.similarity.test() were
not fully Window compatible. (thanks to Alastirs Potts for the fix)
The function plotPNO() has now an argument legend.pos to control
the position of the legend. Alternatively, the plotting of
the legend can now be suppressed.