Summarizes individual iteration performance of each sim/null/metric combination across all iterations.
sesIndiv(raw.results, direction)A list of lists of lists of dataframes; the results of a call to readIn.
Character vector that needs to be provided if spatial simulations beyond the standard "random", "filtering", and "competition" simulations are run. The character vector must be the same length as the number of spatial simulations that were run, and can take the possible values of "two.sided" (for a two-tailed test when the SES scores are expected to be centered around 0), "less" (for when the observed SES scores are expected to be less than 0), and "greater" (for when the observed SES scores are expected to be greater than 0). For instance, habitat filtering would be set to "less". The relevant simulation to which these directional tests will be applied can be determined by calling names(raw.results[[1]]). Note that currently, even if a direction vector is supplied, if the standard simulations were run then the supplied direction vector will be replaced with the standard expectations.
A data frame summarizing the total number of runs per spatial simulation, null, metric, concat.by combination, as well as the type I and II error rates of each such unique combination.
This function takes a raw list of results from multiple iterations from multiLinker, and runs a non-exported function, sesSingle, over each element (iteration) in that list. This sesSingle function runs a Wilcoxon signed rank test over each iteration to determine whether the plots from a spatial simulation/null/metric differ from expectations. Assuming there are three spatial simulations named random, filtering, and competition, this function will use two.sided, lesser and greater Wilcoxon tests, respectively. If additional (or a limited set of) spatial simulations are included, requiring modified expectations, these can be passed along with the "direction" argument. It then summarizes the results of those single run tests as the number of sim/null/metrics that deviated beyond expectations and the number that were within expectations. A single run from a given unique metric + null approach is considered as throwing a type I error only if p is less than or equal to 0.05 for the random spatial simulation. It would be possible to also assess whether such unique combinations throw the opposite signal than expected for habitat filtering and competitive exclusion. A unique combination iteration is considered to throw a type II error if the p value from either the filtering or the exclusion simulation is greater than 0.05.
Miller, E. T., D. R. Farine, and C. H. Trisos. 2016. Phylogenetic community structure metrics and null models: a review with new methods and software. Ecography DOI: 10.1111/ecog.02070
# NOT RUN {
#not run
#results <- readIn()
#summ <- sesIndiv(results)
# }
Run the code above in your browser using DataLab