Summarizes raw output from TITAN's bootstrap procedure

Share:

Description

A function to take output from TITAN's bootstrap procedure and process it for summary output. The default is to perform this processing entirely within active memory, but in the event of overflowing system capacity, an optional program writes temporary files to a scratch directory to circumvent memory limits.

Usage

1
2
3
4
5
small.boot(ivz.bt.list, bSeq, sppmax, obs1, obs2, nBoot, numClass, numUnit,
  ncpus, pur.cut, rel.cut, minSplt)

big.boot(ivz.bt.list, bSeq, sppmax, obs1, obs2, nBoot, numClass, numUnit, ncpus,
  pur.cut, rel.cut, minSplt)

Arguments

ivz.bt.list

A list of output from each bootstrap replicate passed from boot.titan.

bSeq

An index of the sequence of bootstrap replicates.

sppmax

A taxon-specific summary output table for TITAN.

obs1

A binary vector indicating membership in the decreasing group of taxa.

obs2

A binary vector indicating membership in the increasing group of taxa.

nBoot

An argument specifying the number of bootstrap replicates. The default is to use the value specified in the original TITAN function call.

numClass

An argument specifying the number of candidate partitions along the environmental gradient.

numUnit

An argument specifying the number of values along the environmental gradient.

ncpus

An argument specifying the number of processing cores used by the TITAN function call. If ncpus>1 then parallel processing is indicated. The default is to use the value specified in the original TITAN function call.

pur.cut

An argument specifying the cutoff value for determining purity. The default is to use the value specified in the original TITAN function call.

rel.cut

An argument specifying the cutoff value for determining reliability. The default is to use the value specified in the original TITAN function call.

minSplt

An argument specifying minimum split size of partitioning along the environmental gradient. The default is to use the value specified in the original TITAN function call.

Details

Use of 'small.boot' versus 'big.boot' is controlled by the argument 'memory' in the original TITAN function call and passed to the wrapper function 'titan'. The two progams have identical functionality, but they accomplish those functions differently to deal with memory limitations.

For sequential processing of the bootsrtap, the index 'bSeq' is simply a sequence from 1:nBoot that is printed to the screen. For parallel processing, 'bSeq' is a list of length equal to 'ncpus', where each item is a segment of the sequence allocated to each processing core. Thus, depending on whether 'ncpus'>1, the value of 'bSeq' is used differently to extract values from the bootstrap output list.

The first part of each function consists of defining output matrices, the second involves extraction of output from the bootstrap list, the third part involves calculating purity, reliability, the median z score, and quantiles of the bootstrapped change points for each taxon. These values are used to complete the 'sppmax' output table and to identify the taxa that meet purity and reliability criteria. The final portion of each function finds the maximum sum(z-), sum(z+), f.sum(z-), and f.sum(z+) for each bootstrap replicate for later estimation of confidence intervals. The final portion of the summary involves calculating the filtered and unfiltered sum(z) scores for each bootstrap replicate from the matrix of z scores and response directions passed from the function boot.titan() within ivz.bt.list

Value

A list of six items:

  • sppSub1 A vector of taxon index numbers for pure and reliable decreasers

  • sppSub2 A vector of taxon index numbers for pure and reliable increasers

  • sppmax The completed taxon-specific summary output table for TITAN

  • maxSumz A 2-column matrix of environmental values at sum(z-) and sum(z+) maxima across all bootstrap replicates

  • maxFsumz A 2-column matrix of environmental values at filtered sum(z-) and sum(z+) maxima across all bootstrap replicates

  • metricArrayAn array of group membership, env change points, z scores, and p values for passing to 'plot.IVecdf'

Author(s)

M. Baker and R. King

References

Baker, ME and RS King. 2010. A new method for detecting and interpreting biodiversity and ecological community thresholds. Methods in Ecology and Evolution 1(1): 25:37.

Baker ME and RS King. 2013. Of TITAN and straw men: an appeal for greater understanding of community data. Freshwater Science 32(2):489-506.

See Also

boot.titan, tboot, titan