Fixing some documentation formatting issues for CRAN
Vignettes work on a subset of data for speed.
Arguments to autoplot() functions have changed so first is now object rather than x to meet CRAN rules. This probably does not effect anybody unless they had named the first argument.
CRAN compliance version re CITATION
New vignette showing how to use a red-noise null model for randomTF().
Changed h-block cross-validation to use sf package for geographic data (rather than sp).
rne() is now several orders of magnitude faster.
In randomTF(), species and fossil data were automatically transformed to matrices so this was done once, not in every iteration.
This sometimes failed with MAT.
Now this step is recommended in the help file.
Sometimes great-circle distances from a point to itself, calculated with fields::rdist.earth(), are non-zero due to rounding errors.
This lead to them not being excluded in the h-block resampling in rne() when the threshold distance was zero.
Now the diagonal of the distance matrix is set to zero, removing this problem.
In randomTF(), if argument col was missing, all columns of the reconstruction would be used, because lazy evaluation was not done before use in a formula. Code now check col is set when needed.
Maintenance release to fix CRAN check error.
MAT shortcut now works when autosim is set
Lots of changes in this version
p.val becomes p_val and is not the p-value rather than the percentile of the null distributionmodelMaker and randomTFmm (which can be used when a calibration set is used for many cores) is moved within randomTFrandomTF and obs.corrandomTF now accepts a vector for the environmemtal datarandomTF and obs.cor have a permute argument that permutes the environmental variable rather than simulating one from a uniform distribution. This is only possible with a single environmental variable.autoplot functions for the output of randomTF and obs.cormat.h Deleted as rioja::MAT can now do h-block cross-validationmodelMaker and randomTFmm - this functionality is now implemented in randomTFagelme, predict.agelme and plot.fittedAgelme age-depth modelling functions from Heegaard et al (2005)coverage.plot Diagnostic plot for reconstructionscentipede.plot Plot WA optima and tolerancesmake.env, species, make.env Simulate species-environment relationships based on Minchin's (1983) compas.rotate No longer needed as gstat now takes geodesic distance.simulate.spatial Was a wrapper for krige(), now easier to use krige() directly.randomTF and randomTFmm bug when partialling reconstructions out fixedobs.cor bug when species not in same order in calibration set and fossil data fixed.ModelMaker and randomTFmm allows models for randomTF to be fitted once and used for several reconstructions. This can be much faster, but does not work with MAT.plot.palaeoSig p-value highlighted can be selected.plot.obscor p-value highlighted can be selected.randomTF Modified to accept predictions in a vector.obs.cor bug when species not in same order in calibration set and fossil data fixed.identify.obscor allows species names to be added to plots interactively.obs.cor Now includes correlations with several different species weightsplot.obscor upgraded to allow the a choice of which abundance weighting is used, and the code for scaling points has been improved.randomTF bug when partialling out reconstruction other than MAT now fixed.RNE replaces function mat.rne, mat.he and mat.rd. RNE allows any of the transfer function methods in \pkg{rioja} to be used to find the dependence of transfer function performance on spatially close observations.plot.RNE replaces plot.rne to work with the output of RNE.obs.cor replaces the previous version of obs.cor, sim.cor and obscor.sig to make the function easier to use and more similar to randomTFplot.obscor replaces the previous version of plot.obscor and plot.simcor to use the new output of obs.cor. The two plot types can be selected with the argument which.jointsig: tests if two environmental variables have joint control on fossil assemblage composition.plot.js: a plot function for jointsigHill.N2.core now calculates the Minimum, first quartile and median effective number of species for all fossil observations. This makes much more sense than the previous version, which conflated diversity in individual levels and turnover between levels.plot.palaeoSig has been improved to give neater figures.wajack was removed as WA is now implemented in \pkg{rioja}.mat.h will be removed in the next release, when code for h-block resampling should be implemented in \pkg{rioja}.