Test Metric and Null Model Statistical Performance
Explore the behavior and statistical performance of 13 pre-defined
phylogenetic metrics and 11 null models, and of user-defined metrics
and null models, as detailed in Miller et al. (2017) .
Very minor update to fix an issue with distMRCA that became apparent with the new version of ape. This issue affected the first flavor of that metric described in the documentation. The most recent common ancestor was being defined based on the tip numbers, not their names. If you used the first flavor of this metric, you may want to confirm your numbers come out the same now as they did before. Since this was an experimental metric to begin with, I suspect no one was using it.
Minor update to maintain functionality with the new dplyr package, which will be released soon. The only meaningful edits were to the sesField function. Changed a call from summarise_each to summarise_all in that function. Ran some quick tests to see whether the output was still the same. The tests came back successful, but if you encounter an issue in the new sesField, please report it.
Added two new functions. A simple utility function to create the road map used for the FDis-related functions, and a new null model. The new null model is a modified version of the regional null model described in our 2016 Ecography paper. It samples from a regional abundance vector, but rather than maintaining the total number of individuals per plot (and summarizing by species richness later to maintain that), it only maintains the number of species per plot.
Minor update to documentation to make it clearer that the summaries function needs the first metric calculated to be "richness". This is a design flaw and will be fixed in future versions of the package.
Finished major revisions to internal package checks, inputs. Arguments are much easier to specify, and functions are better equipped to deal with outputs of varying sizes. Many functions were changed, and it is possible some things may have broken. Please report any issues on the GitHub site.
Continuing under the hood improvements: almost all downstream functions can now handle outputs from the simulation functions of variable length and dimensions with regards to which spatial simulations, null models and metrics were included. The functions metricPerformance and nullPerformance still need revision.
Major improvements in the way metrics, null models and spatial simulations can be specified. If only a subset of pre-defined options are desired, rather than needing to pass a named list of functions, these functions now accept character vectors corresponding to the options as defined in defineMetrics(), defineNulls(), and defineSimulations(). Can still accommodate new metrics, nulls and simulations defined on the fly.
Adding varLandscape. Also changed color scaling functions for examples from those in colorRamps to those in plotrix.
Replaced the previous simulateComm function with a newer version that always returns a community data matrix with all species from the input tree. This should avoid any errors in the examples.
Added a new null model, synthComm, that creates a synthetic community based on species attributes. It will create synthetic total community niche spaces based on the dispersion in trait space of the input communities. Particularly relevant to looking at niche overlap either between the species from multiple sites, or the individuals from multiple species. This is an experimental null model, so confirm it is working before using it extensively in your analyses. Report any issues on the GitHub site.
Added functional dispersion functions, sensu Laliberte and Legendre (2010). The two new relevant functions are FDis and centers. They take similar inputs to dbFD from the FD package, but can be calculated across a variety of ordination spaces (PCA, PCoA, NMDS, etc), instead of only with a PCoA. These functions and the theoretical background are still experimental, and should be treated as such.
Added prepFieldData, calcField, and sesField. Calculation of phylogenetic and trait fields is now a two-step process, where the data are prepped, then the metrics and nulls of interest are calculated over the prepped object. phyloField, traitField, sesPhyloField, and sesTraitField will be deprecated in future versions of the package.
Added mean root distance (MRD) and mean distance to the most recent common ancestor of a sample (distMRCA). For MRD, added both presence-absence and abundance-weighted forms, and for distMRCA, added ability to calculate both the distance to the overall MRCA of a sample, and the distance to the average of the pairwise MRCAs.
Added more informative error message for instances where the specified simulation parameters yield community data matrices with insufficient individuals/species.
Added code to run all parallel functions sequentially by default. User must now explicitly invoke parallel processing, to avoid issues on Windows machines.
This is the first submission of metricTester to CRAN. No other news to report.