[Vegan-commits] r1784 - in pkg/vegan: R inst man tests/Examples
noreply at r-forge.r-project.org
noreply at r-forge.r-project.org
Sat Sep 3 20:46:43 CEST 2011
Author: jarioksa
Date: 2011-09-03 20:46:42 +0200 (Sat, 03 Sep 2011)
New Revision: 1784
Added:
pkg/vegan/inst/ONEWS
Modified:
pkg/vegan/R/vegandocs.R
pkg/vegan/inst/ChangeLog
pkg/vegan/man/vegandocs.Rd
pkg/vegan/tests/Examples/vegan-Ex.Rout.save
Log:
get ONEWS (for vegan 1.*), and expand doc of vegandocs
Modified: pkg/vegan/R/vegandocs.R
===================================================================
--- pkg/vegan/R/vegandocs.R 2011-09-01 15:19:02 UTC (rev 1783)
+++ pkg/vegan/R/vegandocs.R 2011-09-03 18:46:42 UTC (rev 1784)
@@ -1,7 +1,7 @@
`vegandocs` <-
- function (doc = c("NEWS", "ChangeLog", "FAQ-vegan.pdf", "intro-vegan.pdf",
- "diversity-vegan.pdf", "decision-vegan.pdf",
- "partitioning.pdf", "permutations.pdf"))
+ function (doc = c("NEWS", "ONEWS", "ChangeLog", "FAQ-vegan.pdf",
+ "intro-vegan.pdf", "diversity-vegan.pdf",
+ "decision-vegan.pdf", "partitioning.pdf", "permutations.pdf"))
{
doc <- match.arg(doc)
if (length(grep(".pdf", doc)) > 0) {
Modified: pkg/vegan/inst/ChangeLog
===================================================================
--- pkg/vegan/inst/ChangeLog 2011-09-01 15:19:02 UTC (rev 1783)
+++ pkg/vegan/inst/ChangeLog 2011-09-03 18:46:42 UTC (rev 1784)
@@ -13,6 +13,9 @@
* meandist: warns if some dissimilarities are negative (small
negative values are tolerated).
+
+ * vegandocs: copied branches/1.17/inst/NEWS as ONEWS (old news),
+ and adapted vegandocs. Expanded documentation of vegandocs().
Version 1.92-1 (closed August 28, 2011)
Copied: pkg/vegan/inst/ONEWS (from rev 1783, branches/1.17/inst/NEWS)
===================================================================
--- pkg/vegan/inst/ONEWS (rev 0)
+++ pkg/vegan/inst/ONEWS 2011-09-03 18:46:42 UTC (rev 1784)
@@ -0,0 +1,2338 @@
+-*-Text-*-
+
+ VEGAN RELEASE VERSIONS
+ ======================
+
+ CHANGES IN VEGAN 1.17-12
+
+ - This is a maintenance release which improves robustness of
+ several functions. A major release is expected soon.
+
+ - tolerance.cca: a new function to find the species response
+ widths (a.k.a. tolerances) and sample heteregeneities from a
+ cca() result.
+
+ - adonis: much faster.
+
+ - betadiver: argument 'index' was renamed to 'method' so that the
+ function can be used similarly as dist(), vegdist() and other
+ distance functions. This allows using betadiver() as a distance
+ function in metaMDS().
+
+ - cca/rda/capscale: handling of aliased and other zero-rank
+ components changed in vegan 1.17-11, but not all support
+ functions were adapted to these changes in that release. Now the
+ following functions cope with the changes and are more robust:
+ capscale, anova.cca, bstick.cca, goodness.cca, predict.cca,
+ screeplot.cca, calibrate.cca, deviance.cca/rda, ordiplot3d,
+ ordiresids, ordirgl, ordixytplot.
+
+ - isomap: text or points are plotted more cleanly, and text uses
+ ordilabel().
+
+ - make.cepnames: a bit more flexible and robust, with a new
+ argument that allows selecting first and second name component
+ instead of first and last.
+
+ - metaMDSrotate: rotation could be slightly off with more than two
+ dimensions when there are NA scores.
+
+ - ordiellipse, ordihull, ordispider, orditorp: accept NA scores.
+
+ - ordilabel: gained argument 'select'.
+
+ - ordiplot: better handling of graphical arguments, arg 'cex' can
+ be set by the user.
+
+ - swan: gained new argument 'maxit' that allows setting maximum
+ number of iterations. Defaults 'maxit = Inf' or the current
+ behaviour of continuing as long as there are zeros that can be
+ filled.
+
+ CHANGES IN VEGAN 1.17-11
+
+ - This is a maintenance release which improves the robustness of
+ several functions but introduces no new important features.
+
+ - cca/rda/capscale: handling of aliased and other zero-rank
+ components changed. They are no longer made NULL and left
+ undisplayed, but now they are shown as zero rank. The general
+ output also changed so that proportions of inertia are only
+ shown if there are conditions or constraints, but not for
+ unconstrained analysis. The capscale() function no longer shows
+ species scores if these are unavailable. Several support
+ functions more robust.
+
+ - nobs: R 2.13.0 introduced generic function nobs() to find the
+ number of observations, and this release provides nobs() for
+ several vegan results. (This does not make vegan dependent on R
+ 2.13.0, but vegan works with older R as well.)
+
+ - prc: allows anova(..., by = "axis") and other 'by' cases for
+ prc() results. Some rda() support functions still fail, but now
+ they stop informatively.
+
+ - specaccum: gained argument 'groups' which can be used to find
+ the number of species in subsets of the data.
+
+ CHANGES IN VEGAN 1.17-10
+
+ - This is minor revision that mainly fixes vegan with respect to
+ changes in the currently released R 2.13.0. Most importantly,
+ cmdscale() output changed in R 2.13.0 and because of this
+ capscale() could fail in some rare situations with argument 'add
+ = TRUE'. This vegan bug made BiodiversityR package fail its
+ tests in R 2.13.0.
+
+ - metaMDSrotate: gained argument na.rm = TRUE.
+
+ CHANGES IN VEGAN 1.17-9
+
+ - anova of cca/rda/capscale results gave wrong results in partial
+ models. The bug was introduced in vegan 1.17-7.
+
+ - diversity and related functions rarefy, rrarefy and specnumber
+ now accept vector input. Earlier a single site had to be
+ analysed either as a single-row matrix or using the non-default
+ setting MARGIN = 2.
+
+ - drarefy: new function that returns a matrix of probabilities
+ that a species occurs in a rarefied sample of a given size.
+
+ - metaMDS: it is possible to supply a starting configuration with
+ argument 'previous.best'. A previous metaMDS or isoMDS result can
+ also be given as a starting configuration. If the starting
+ configuration has a higher number of dimensions than requested,
+ the extra ones are dropped, and if the starting configuration
+ has fewer dimensions, random scores for extra dimensions will be
+ added. This may help in running metaMDS over a range of
+ dimensionalities.
+
+ - metaMDSrotate: can now rotate metaMDS solutions with any number
+ of dimensions so that the first axis is parallel to a fitted
+ environmental vector. Previously, only two dimensional solutions
+ worked.
+
+ - ordilabel: gained argument 'xpd' that allows labels outside the
+ plot area. This allows labels above axes, for instance.
+
+ - ordisurf: gained several new arguments to control the mgcv::gam
+ fitting. Also gained an argument to suppress plotting, and a new
+ plot method. The fitted model can be specified with a formula.
+
+ - prestonfit and friends: default is now 'tiesplit = TRUE' (which
+ was a new feature introduced in vegan 1.17-8).
+
+ CHANGES IN VEGAN 1.17-8
+
+ - prestonfit: fixed a bug in as.preston(): the largest octave
+ could be missing with 'tiesplit = TRUE'.
+
+ - decorana: spurious axes scores and eigenvalues could be reported
+ when the eigenvalues actually were zero. This was rarely a
+ problem with real data, but occurred only in arbitrary examples.
+
+ - procrustes: checks input.
+
+ CHANGES IN VEGAN 1.17-7
+
+ - anova.cca: more robust when models were fitted without 'data'
+ argument, or when 'na.action' or 'subset' was used.
+
+ - capscale: implemented 'subset' in model definition. Additive
+ constant with 'add = TRUE' is taken into account in predict()
+ and fitted(). Implemented simulate() which returns a
+ dissimilarity matrix with random error about predicted values.
+
+ - eigenvals: can now extract eigenvalues of some ade4 and labdsv
+ result objects.
+
+ - nestednodf: did not use binary data when weighted = FALSE was
+ used together with order = FALSE. Reported by Daniel Spitale.
+
+ - prestonfit: gained option to split tied frequencies (1, 2, 4, 8,
+ etc.) between adjacent octaves.
+
+ - specaccum: implemented choice of using either "individuals" or
+ "sites" as x-axis in plot(). Corrected a typo in the result
+ object: now returns indeed "individuals".
+
+
+ CHANGES IN VEGAN 1.17-6
+
+ - capscale: vegan 1.17-5 defined total inertia as the sum of
+ absolute values of eigenvalues which changed the total inertia
+ from previous versions. This is changed so that the total
+ inertia is the sum of all eigenvalues, i.e., negative
+ eigenvalues are subtracted from the total which was the
+ definition in 1.17-4 and earlier. The proportions of inertia
+ components are now expressed for non-negative eigenvalues only,
+ and a new item "Real Total" (sum of positive eigenvalues) is
+ added to the output if there are negative eigenvalues
+ ("Imaginary" component). The function no longer returns the
+ negative eigenvalues.
+
+ - CCorA: scaling of object scores changed: they are no longer
+ divided with sqrt(n-1). Minor fixes for improved robustness.
+
+ - metaMDS: automatically changes some arguments to non-default
+ values if input data contains negative entries. Help page gives
+ advice to do so for non-community data.
+
+ - wascores: checks that input weights are non-negative.
+
+ NEW FEATURES AND BUG FIXES IN VEGAN 1.17-5
+ - anova.cca: Empty models with no constrained component have
+ correct degrees of freedom. Tests 'by = "axis"' are really
+ marginal for all axes (first axis used to be non-marginal to
+ latter axes).
+
+ - as.mlm.cca/rda, intersetcor, vif.cca: avoid bug in qr.X in R
+ 2.12.0 which use wrong variable names in aliased models (fixed
+ also in R 2.12.1).
+
+ - betadisper: plot did not allow selecting axes. Reported by Sarah
+ Goslee.
+
+ - capscale: failed with NA action (reported by Nevil Amos). Total
+ inertia is defined as the sum of absolute values of all
+ eigenvalues consistently with cmdscale (from R 2.12.1) and other
+ vegan functions.
+
+ - cca/rda/capscale: print proportions of inertia components.
+
+ - decorana: Fortran code prints warnings of failed convergence.
+ Default plot() uses points if items have no name labels to be
+ plotted.
+
+ - deviance.cca/rda: return 0 (instead of NA) for over-paremetrized
+ models with no unconstrained component.
+
+ - eigenvals: uses sum of absolute values of eigenvalues in models
+ that may have negative eigenvalues (cmdscale, wcmdscale,
+ capscale). Can extract eigenvalues of pcnm() and cmdscale() of R
+ 2.12.1. Uses zapsmall() to print near-zero
+ eigenvalues.
+
+ - mantel.correlog: P-values could slightly off. The function uses
+ internally mantel() to evaluate the statistic and its
+ significance. The mantel() function adds one both to the
+ denominator and numerator, but mantel.correlog() did not notice
+ this and made the addition for the second time. When the
+ mantel.correlog() reported a positive value of the statistic, it
+ reversed the direction of the one-sided test of mantel(), but
+ did not handle tied values correctly in this reversal.
+
+ - nestednodf: Gustavo Carvalho's new version that also implements
+ a new quantitative method of Almeida-Neto & Ulrich (Env Mod
+ Software, 26, 173-178; 2011).
+
+ - oecosimu: takes care that the statistic is evaluated with binary
+ data when null models are binary (regression introduced in vegan
+ 1.17-0). Can handle NA values in permutations (as na.rm = TRUE).
+
+ - ordilabel: gained new argument 'col' to set the text colour of
+ the labels separately from 'border'. Function ordiellipse() uses
+ this in filled polygons to set the colour of labels to that of
+ borders or foreground instead of the colour of the fill.
+
+ - ordiR2step: new function to model selection in rda() and
+ capscale() based on adjusted R2 following the recommendations of
+ Blanchet, Legendre & Borcard (Ecology 89, 2623-2632; 2008).
+
+ - ordistep: correct a name clash when the fitted model had term or
+ item called 'mod'. Reported by Richard Telford (Bergen, Norway).
+
+ - pcnm: returns the truncated distanced matrix with argument
+ 'dist.ret'.
+
+ - prc: rewritten by Cajo ter Braak with a fix to scaling of
+ coefficients for full compatibility with Canoco.
+
+ - procrustes: Fixed translation following the report and fix by
+ Christian Dudel (Bochum, Germany). New predict() method to
+ rotate 'newdata' points to the target. Graphics with plot()
+ select the direction of arrows with new argument 'to.target',
+ and use ordilabel() to display the labels with 'type = "t"'.
+
+ - rankindex: can take a list of dissimilarity functions as the
+ argument to allow the use of other dissimilarity functions than
+ vegdist() of vegan.
+
+ - rda/capscale: the scores() can take two scaling constants
+ ('const'). One is used for species, second for WA and LC scores
+ of sites. This allows using scalings of prcomp(), princomp() or
+ Canoco (like documented in the design decisions vignette).
+
+ - swan: disconnected data caused infinite loop. Now zeros are left
+ to the result with disconnected data.
+
+ - treedist: can handle empty or one-node trees. Gained new
+ argument 'relative' (defaults TRUE): if FALSE, finds raw
+ dissimilarities of tree heights. Help page tells that relative
+ tree distances are in range 0..2 instead of 0..1, since
+ combining two trees may add a new common root.
+
+ - treedive: handles trivial cases of zero (diversity NA) or one
+ species (diversity 0).
+
+ - vegdist: help page gives now binary variants of the indices
+ using designdist() notation.
+
+ - vif.cca: can handle models with aliased terms.
+
+ - wcmdscale: returns goodness of fit statistic (item GOF) and
+ handles negative eigenvalues consistently with cmdscale() and
+ capscale().
+
+ NEW FEATURES AND BUG FIXES IN VEGAN 1.17-4
+
+ - MOStest: a new set of functions to implement generalized
+ Mitchell-Olds & Shaw test for the location of the quadratic
+ extreme in a given interval. The test can also be used for the
+ location of the optimum of the Gaussian response model. In
+ addition to the basic test, there are Fieller and deviance
+ profile methods for the confidence interval of the location of
+ the quadratic extreme or Gaussian optimum.
+
+ - mrpp & meandist: mrpp() does not evaluate the Classification
+ Strength (CS) of Van Sickle & Hughes (J. N. Am. Benthol. Soc.,
+ 19: 370-384; 2000) any longer. The old version misinterpreted
+ the weighting scheme of CS, and with correct scheme there is no
+ exact relationship between CS and the corresponding MRPP
+ statistic and therefore permutation tests are not available. CS
+ is still evaluated in meandist(), but with corrected weighting
+ scheme, and no tests are performed. Reported by Dr John Van
+ Sickle (Corvallis OR).
+
+ - msoplot: legend text was in a wrong order. Reported by Daniel
+ Borcard.
+
+ - vegdist: Marti Anderson variant of the Gower distance (method =
+ "altGower") should be without range standardization. Reported by
+ Sergio Garcia.
+
+ - ordiellipse: labels disappeared in their background with 'draw =
+ "polygon"'.
+
+ - predict.cca and predict.rda: gained argument 'newdata' for 'type
+ = "response"' and 'type = "working"'. These return estimates
+ of the response data, or dissimilarities in capscale().
+
+
+ NEW FEATURES AND BUG FIXES IN VEGAN 1.17-3
+
+ - adonis: handles ties correctly.
+
+ - anova.cca and permutest.cca: permutations of cca result is
+ faster and now nearly equally fast (or slow) in rda and cca.
+
+ - betadisper: 'type = "median"' (the default) was not computing
+ the spatial median on the real and imaginary axes separately.
+ Reported by Marek Omelka.
+
+ - cca, rda and capscale failed when Condition() was a factor, but
+ the constraints had only continuous variables.
+
+ - envfit: defaults to use 999 permutations instead of skipping
+ permutations.
+
+ - mantel and mantel.partial: faster permutation.
+
+ - mantel.correlog: upgraded and faster.
+
+ - ordiarrows, ordisegments and ordispider gained argument 'label'
+ to label the groups corresponding to drawn arrows or lines.
+
+ - ordiresids: de-weights residuals and fitted values in CCA so
+ that these are identical to the values shuffled in simulate.cca.
+
+ - permutest.cca: adds observed statistic among permutations when
+ printing the result (does not influence anova.cca or
+ calculations).
+
+ - RsquareAdj: The 'rda' method used wrong number of degrees of
+ freedom in rank deficit models (number of dependent variables
+ was lower than the rank of constraints and conditions). Default
+ method handles vector input.
+
+ - scores: default method works with one row of scores. The scores
+ of "cca" and "rda" methods always have names, even if there are
+ no names in the input data.
+
+ - Permutation tests: The available permutation tests are described
+ in a new help files accessed via ?permutations.
+
+ NEW FEATURES IN VEGAN 1.17-2
+
+ - permutest.betadisper: printed P-values in a wrong order in
+ displaying the the matrix of pairwise tests (the values were
+ correct, but formatting failed). Reported by Dan O'Shea.
+
+ - nesteddisc: failed if the most species rich sites were tied. The
+ function is now much faster (though still slow), but the price
+ is that it does not try as hard to find the optimal ordering.
+
+ - screeplot.cca & friends: new argument 'legend' for all methods
+ (except 'decorana') to draw a legend if the observed and broken
+ stick distribution are both plotted.
+
+ - ordistep: adds an 'anova' item to the final model similarly as
+ the standard step(). You can suppress the tracing ('trace =
+ FALSE'), and find the model build history save in 'anova' in a
+ compact form.
+
+ - densityplot.oecosimu: gave warnings when there was only one
+ statistic and hence one lattice panel.
+
+ - predict functions for cca and rda objects match 'newdata' by
+ dimension names instead of index.
+
+ - simulate.rda and simulate.cca have new argument 'rank' which
+ allows using lower rank presentation of fitted values
+ (including rank = 0).
+
+ - treedist: new function to find dissimilarities of species
+ property trees of communities. The property trees can be, e.g.,
+ functional diversity trees, taxonomies or phylogenies. Similar
+ in spirit to UniFrac distance (C. Lozupone & R. Knight, Appl
+ Environ Microbiol 71:8225-8235; 2005), but completely different
+ in design and works only with binary data.
+
+ NEW FEATURES AND FIXES IN VEGAN 1.17-1
+
+ - multipart: new functions for multiplicative partionining of
+ gamma diversity into alpha and beta diversity components.
+
+ - CCorA: Fixed choice of scores in biplots -- Used PC scores
+ instead of observed scores in the right-hand-side biplot
+ panel. The biplot function has several new and enhanched
+ alternatives of plots.
+
+ - envfit: ignored weights in cca() results for factors or a single
+ continuous variable. The bug was introduced with NA handling
+ upgrade in vegan 1.17-0. The problem with fitted vectors was
+ reported by Richard Telford (Bergen, Norway).
+
+
+ CHANGES AND NEW FEATURES IN VEGAN 1.17-0
+
+ - Guillaume Blanchet joined the vegan team.
+
+ - New function to partition data-set diversity (gamma) into
+ within-plot (alpha) and between-plot (beta) diversity
+ components. Function adipart performs additive
+ partitioning (gamma = alpha + beta). Function hiersimu performs
+ hierarchical null model testing similar to adipart but by using
+ custom function to calculate statistics for levels of a hierarchy.
+
+ - Subsets and missing value handling added to constrained
+ ordination methods cca(), rda() and capscale(). The missing
+ values (NA) can be handled with setting na.action (defaults
+ na.fail). With na.action = na.omit, observations with missing
+ values are removed, and with na.action = na.exclude they are
+ kept, but scores may be NA. However, the WA scores for sites are
+ available in non-partial models with na.action = na.exclude.
+ The 'subset' can be defined using any variable in the
+ constraining data set or species in the dependent data.
+
+ - Functions for fitting environmental variables onto ordination
+ (envfit, ordisurf) are aware of missing values in constraints or
+ NA values in scores. The ordination plot functions also can
+ handle NA scores.
+
+ - New and upgraded quantitative null models. In particular,
+ quantitative swap models (function abuswap) allows generetating
+ null matrices where marginal totals and fill are fixed, or row
+ and column fills are fixed, or row and column fills and either
+ row or column totals are fixed.
+
+ - oecosimu: rewritten to handle quantitative null models. Gained
+ keyword 'alternative' for "two-sided", or "less" and "greater"
+ one-sided tests. More robust with degenerate solutions. The
+ 'method' can now be a user-supplied function. New support
+ functions as.ts() and as.mcmc which transform sequential models
+ into form that can be analysed using tools for time-series
+ (as.ts()) or MCMC sequences of the 'coda' package (as.mcmc).
+
+ - calibrate: calibrate is now a generic function with a new method
+ calibrate.ordisurf() in addition to the old calibrate.cca. These
+ find the estimates of environmental variables from ordination.
+
+ - ordistep: stepwise selection of terms in constrained ordination
+ (cca, rda and capscale) using permutation tests instead of
+ pseudo-AIC that is used by the standard step() function.
+
+ - pcnm: new function to find weighted principal coordinates of
+ neighbour matrix (PCNM) from distances between points. These are
+ typically used for spatial filtering in constrained
+ ordination. The function uses weighted analysis and can
+ therefore produce PCNM for correspondence analysis in addition
+ to PCA and RDA.
+
+ - betadisper: can use spatial medians which are now the default
+ method. Preliminary tests indicate that spatial medians correct
+ the anti-conservative Type I errors reported by Stewart
+ Schultz.
+
+ - decostand & vegdist: new transformation 'log' which implements
+ Marti Anderson's scaling log(x) + 1 for x>0 (which is not at all
+ the same as log(x+1)), and vegdist has alternative Gower
+ function that skips double zeros. Together these implement the
+ "modified Gower" distance of Anderson et al. (Ecology Letters 9,
+ 683-693; 2006). Feature request #473 by Etienne Laliberte in
+ vegan.r-forge.r-project.org.
+
+ - model.matrix.cca & model.frame.cca: new functions to reconstruct
+ the model frame and model matrix (model matrices in partial
+ models) of constraints used in ordination methods cca(), rda()
+ and capscale().
+
+ - simulate.cca & simulate.cca: simulate the community (response)
+ data from the result of cca() or rda() under alternative
+ hypothesis. Error is added to the fitted values from
+ ordination. The function uses either Gaussian error or permutes
+ residuals and adds these to fitted values.
+
+ - spandepth: a new function to find the depth of each node in the
+ minimum spanning tree produced by spantree(). Feature request by
+ W. E. Sharp.
+
+ - alias.cca: gained argument names.only.
+
+ - metaMDSrotate: a new function to rotate metaMDS so that first
+ axis is parallel to an environmental variable.
+
+ - msoplot: uses standard legend.
+
+ - nesteddisc: new method that tries to find the smallest possible
+ value of the statistic in tied data. Slow, but fixes the
+ problems of the published method. The problem with tied values
+ in nesteddisc was found with Carsten Dormann.
+
+ - contribdiv: gained a plot method.
+
+ - rarefy: failed with one site and many sample sizes like
+ rarefy(rpois(10, 2), sample=2:3).
+
+ - meandist: plot can draw histograms as an alternative to
+ dendrograms.
+
+ - plot functions for 'decorana' and 'cca' and friends and
+ 'ordiplot' use 'linestack' if only one axis was requested.
+
+
+ CHANGES IN VEGAN 1.15-4
+
+ - Changed package dependence: vegan does not depend on 'ellipse'.
+
+ - anosim: user interface identical to 'mrpp'. Accepts now data
+ matrix and finds dissimilarities internally.
+
+ - betadisper: fix removal of zero eigenvalues with non-Euclidean
+ distances. This may change the results slightly, but in most
+ cases the effects are minor or none.
+
+ - capscale: has now 'fitted' and 'residual' methods, and 'predict'
+ works with 'type = "response"'. These return dissimilarities
+ that produce the same ordination as the original data.
+
+ - indpower: new function to find the indicator power of species to
+ predict presence of other species (Halme et al., Conservation
+ Biology 23, 1008-1016; 2009). Closely related to the the power
+ to predict probabilities in beals().
+
+ - mantel.correlog: new functions to produce multivariate Mantel
+ correlograms (Legendre & Legendre, Numerical Ecology, section
+ 13.1.5; 1998).
+
+ - metaMDS: accepts user-supplied dissimilarities. Species scores,
+ data transformation, step-across and half-change scaling are
+ unavailable with user-supplied dissimilarities, but random
+ starts, PC rotation and scaling to original range of input
+ dissimilarities work.
+
+ - nestedtemp: row and column labels can be turned on/off
+ independently in plots.
+
+ - ordihull & ordiellipse: new argument 'label' to plot group names
+ for hulls or ellipses. Return (invisibly) the data for plotted
+ convex hulls or ellipses. In 'ordihull' this is a list of hull
+ vertices, and in 'ordiellipse' a list of (scaled) covariance and
+ centre data for ellipses. New 'summary' methods find the centres
+ and areas of plotted hulls or ellipses. Argument 'draw' has new
+ option "none" that suppresses all drawing so that only the data
+ for summary can be extracted without plotting.
+
+ - orditorp: works with reversed axes, like with 'xlim = c(3, -3)'.
+
+ - ordixyplot: has a panel function for arrows: a lattice variant
+ of ordiarrows.
+
+ - poolaccum, estaccumR: new functions to find the specpool() or
+ estimateR() estimates of extrapolated species richness in random
+ accumulations of sites. These have 'plot' and 'summary' methods.
+
+ - scores: functions biplot, points, text, ordilabel, ordiplot3d,
+ ordixyplot and spantreee did not pass all arguments to scores()
+ function. In particular, this concerned rda where 'scaling' and
+ 'const' arguments could not be used within these functions.
+
+ - radfit: new function 'radlattice' for a lattice plots of fitted
+ models for a single site.
+
+ - rrarefy: a new function to generate random rarefied communities.
+
+ - RsquareAdj: now a generic function to find adjusted R squared
+ with special cases to 'rda', 'cca', 'lm' and 'glm'.
+
+ - summary.cca: cleaner output.
+
+ - wcmdscale: returns scaled scores for axes with negative
+ eigenvalues.
+
+
+ CHANGES IN VEGAN 1.15-3
+
+ - anosim, mantel, mrpp, protest, envfit: did not include the
+ observed statistic among permutations. Functions adonis and
+ permutest.cca did this correctly, but did not print results
+ neatly.
+
+ - anova.cca: name clash if data were indexed with 'i'.
+
+ - capscale: fixed handling of negative eigenvalues with
+ non-Euclidean distances. The total inertia is the sum of all
+ eigenvalues so that negative eigenvalues are subtracted from the
+ total. The total inertia of negative components and their rank
+ (number) is given as 'Imaginary' component, and the negative
+ eigenvalues are listed after unconstrained positive eigenvalues.
+ The procedure is based on Gower, Linear Algebra and its
+ Applications 67, 81-97 (1985). New argument 'sqrt.dist' takes
+ square root of the internally calculated dissimilarities and
+ avoids negative eigenvalues with some indices, such as vegan
+ Jaccard and Bray-Curtis. The adjustment is corrected for indices
+ with upper limit of one, and the reported eigenvalues and
+ inertia components are reduced by a factor of sqrt(n-1) and are
+ similar to those reported by 'cmdscale' or 'wcmdscale'.
+
+ - eigenvals: a new function to extract eigenvalues from rda, cca,
+ capscale, wcmdscale, prcomp, princomp, svd or eigen. If the
+ result object contains squareroots of eigenvalues, they are
+ squared. The summary method also finds proportions and
+ cumulative proportions explained. Function summary.cca now uses
+ this to display eigenvalues.
+
+ - kendall.global: could get wrong counts of ties in large data
+ sets.
+
+ - meandist: new sister function to 'mrpp' that finds mean
+ within-group and between-group dissimilarities. The summary
+ function finds overall averages of these, and returns all three
+ MRPP variants plus classification strength. The plot method
+ draws a dendrogram based on the mean dissimilarity matrix, with
+ leaves hanging to within-groups dissimilarity. The functions
+ follow Sickle, J. Agric. Biol. Envir. Stat. 2, 370-388 (1997).
+
+ - ordisurf: fits now linear or quadratic trend surfaces if
+ 'knots' argument is set to 0, 1, or 2.
+
+ - orditkplot: copes with NA and NaN scores.
+
+ - ordixyplot: mixed x and y axes for biplot arrows and class
+ centroids. Function failed in constrained ordination.
+
+ - tsallis: gained argument 'hill' to find results analogous to
+ Hill numbers in renyi(), or the species number equivalents of
+ indices.
+
+ - wcmdscale: removes now zero eigenvalues instead of the last
+ eigenvalue. The bug was copied from cmdscale(), which still has
+ the bug in R 2.9.0 (plus another that was not copied to
+ wcmdscale).
+
+
+ CHANGES IN VEGAN 1.15-2
+
+ - adonis: adds one to numerator and denominator of permutation
+ tests, the default number of iterations was raised from 5 to
+ 999, and the result object got a 'terms' component. Uses much
+ less memory allowing analysis of larger problems.
+
+ - anova.cca: anova(..., by = "axis") gained new keyword 'cutoff'
+ to stop permutation tests after exceeding the given cut off
+ level of significance. The second term of anova(..., by =
+ "margin") used different random numbers than other terms.
+
+ - beals: Completely rewritten by Miquel de Caceres. Knows now
+ also the cross validated version of Beals smoothing and other
+ choices described by De Caceres & Legendre, Oecologia 156,
+ 657-669; 2008.
+
+ - betadisper: handles missing values both in dissimilarities and
+ grouping.
+
+ - cca/rda: cleaner output of summary() in unconstrained models.
+
+ - commsimulator: simulated result retains original row and column
+ names.
+
+ - contribdiv: new function for contribution diversity (Lu et al.,
+ Basic and Applied Ecology 8, 1-12; 2007).
+
+ - decostand: gained dots in argument list, so that
+ stressplot(metaMDS(x, dist = "gower", trymax = 40)) works.
+
+ - dispindmorisita: new function for the Morisita index of
+ dispersion. See Krebs, Ecological Methodology; 1999.
+
+ - kendall.global and kendall.post: new functions for Kendall's
+ coefficient of concordance. In ecology these can be used to
+ identify significant species associations (Legendre, J Agric
+ Biol Environm Stat 10, 226-245; 2005).
+
+ - metaMDS: more robust with distances like Euclidean and Manhattan
+ which have no upper limit. The stepacross works correctly for
+ these, but gives a warning that its use may not be sensible.
+ There is a better heuristics to avoid half-change scaling with
+ these indices. The 'halfchange' argument is now honoured when
+ given in metaMDS() call.
+
+ - mrpp: returns within-class dissimilarities. Evaluates
+ classification strength (Van Sickle, Biological and
+ Environmental Statistics, 2, 370-388; 1997) if weight.type = 3
+ was used.
+
+ - nestednodf: new nestedness function of overlap and decreasing
[TRUNCATED]
To get the complete diff run:
svnlook diff /svnroot/vegan -r 1784
More information about the Vegan-commits
mailing list