clResults: Congreve and Lamsdell tree distances

In CongreveLamsdell2016: Distance Metrics for Trees Generated by Congreve and Lamsdell

Description

Distance of CL trees from generative tree.

Usage

clBremQuartets

clBremPartitions

clMkvPartitions

clMkvQuartets

clBootFreqPartitions

clBootFreqQuartets

clJackFreqPartitions

clJackFreqQuartets

clBootGcPartitions

clBootGcQuartets

clJackGcPartitions

clJackGcQuartets

Format

An object of class list of length 7.

Details

For each of the 100 matrices generated by Congreve & Lamsdell (2016), I conducted phylogenetic analysis under different methods:

• Mkv: using the Markov K model in MrBayes;

• eq: using equal weights in TNT;

• k1, k2, k3, k5, kX: using implied weights in TNT, with the concavity constant (k) set to 1, 2, 3, 5, or 10;

• kC: by taking the strict consensus of all trees recovered by implied weights parsimony analysis under the k values 2, 3, 5 and 10 (but not 1).

For each analysis, I recorded the strict consensus of all optimal trees, and also the consensus of trees that were suboptimal by a specified degree.

I then calculated, of the total number of quartets or partitions that were resolved in the reference tree, how many were the same or different in the tree that resulted from the phylogenetic analysis, and how many were not resolved in this tree (r2).

The data object contains a list whose elements are named after the methods, as listed above.

Each list entry is a three-dimensional array, whose dimensions are:

1. The suboptimality of the tree. Different measures of node support are employed:

   • Mkv: Posterior probabilities, at 2.5% intervals (50%, 52.5%, ... 97.5%, 100%).

   • Brem: Bremer supports: the consensus of all trees that are (equal weights) 0, 1, .... 19, 20 steps less optimal than the optimal tree (implied weights: the consensus of all trees that are 0.73^(19:0) less optimal than the optimal tree).

   • Boot: Bootstrap supports (symmetric resampling, p = 0.33).

   • Jack: Jackknife supports (p = 0.36).

   Boot and Jack results are reported both as the frequency of splits among replicates, and using the gc (Groups Present / Contradicted) measure (Goloboff et al. 2003); frequency columns correspond to 100%, 97.5%, 95% ... 0% support; gc columns correspond to 100%, 95%, ... 0% present, 5%, 10%, ... 100% contradicted.

2. Counts of the condition of each quartet or partition:

   • Q: The total number of quartets defined on 22 taxa.

   • N: The total number of partitions present, counting each tree separately.

   • P1: The number of partitions in tree 1 (the reconstructed tree).

   • P2: The number of partitions in tree 2 (the generative tree).

   • s: The number of quartets or partitions resolved identically in each tree. * d: The number of quartets resolved differently in each tree.

   • d1: The number of partitions resolved in tree 1, but contradicted by tree 2.

   • d2: The number of partitions resolved in tree 2, but contradicted by tree 1.

   • r1: The number of partitions or quartets resolved in tree 1 that are neither present in nor contradicted by tree 2.

   • r2: The number of partitions or quartets resolved in tree 2 that are neither present in nor contradicted by tree 1.

   • u: The number of quartets that are not resolved in either tree.

3. The number of the matrix, from 1 to 100.

Source

Congreve, C. R. & Lamsdell, J. C. (2016). Implied weighting and its utility in palaeontological datasets: a study using modelled phylogenetic matrices. Palaeontology 59(3), 447–465. doi: 10.1111/pala.12236.

References

Goloboff, P. A., J. S. Farris, M. Källersjö, B. Oxelman, M. J. Ramírez, and C. A. Szumik. 2003. Improvements to resampling measures of group support. Cladistics 19, 324–332. doi: 10.1016/S0748-3007(03)00060-4.

See Also

clMatrices, clReferenceTree.

CongreveLamsdell2016 documentation built on Jan. 7, 2020, 5:07 p.m.