ClineFit uses the methods published in Evolution
(Porter et al. 1997), which in turn follows Barton's methods very
closely. Its main difference is that it fits genotypes to the
cline shape directly, rather than fitting transformed data to a
linearized model; this isn't such a big difference.
Support for data of (almost) all inheritance types:
Codominant markersDisequilibrium estimation:
Sex-linked/haplodiploid codominant markers
Sex-linked/haplodiploid dominant markers
Cytoplasmic markers inherited through females (e.g., mtDNA, cpDNA)
Cytoplasmic markers inherited through males (e.g., Y chromosomes)
Threshold traits (e.g., color morphs)
X/Y and W/Z sex determination
Uses all the data on the transect, instead of populations in a user-defined central region. This gives more statistical power.Cline orientation:
Supports estimation within and between genetic markers of all inheritance types.
Traits with higher values at either the left or right ends of the cline are supported.Considerable power to customize:
Parameters such as width and tail shape can be co-estimated within and between traits of either orientation.
Turn parameters on or off to include or exclude them from the model.Comparison among transects:
Co-estimate vs. separately estimate parameters for arbitrary sets of traits (and use model selection to see if clines or parts of clines differ among them).
Data from multiple transects can be estimated togetherModel selection:
Among locationsConstruct customized models to compare parameter combinations shared within and among transects simultaneously.
Within or among groups of individuals (e.g., life stages)
Among years (e.g., look for evidence of movement)
Not yet supported: models that include disequilibrium
Reports AIC as well as log-likelihood scores, to compare customized models properly.Improved MCMC algorithm:
Estimate only the best fitting model under your settings, or
choose to include 2-unit support limits.
Modified Gibbs sampler makes fitting and sampling for support much more efficient with large numbers of parameters.More output files:
Better able to escape blind alleys that appear when large numbers of parameters are estimated under customized models.
Better user control over the rate of approach to the maximum-likelihood cline shape.
Control over the number of cycles between saved parameter sets during sampling for support.
Output files suitable for making graphics.User's manual & sample files:
Pairwise disequilibrium tables for all populations.
Instructions for creating the data files.Graphics:
Step by step instructions for using each option.
Background on the models used in the program.
Explanations of each of the parameters.
Implications of the assumptions made in implementing the models and interpreting the results.
Background on maximum-likelihood estimation and model selection.
Ideas to try if your results look funky.
A Mathematica notebook for creating graphics (you still have to get your hands on a copy of Mathematica).
(No R package is available.)
Join (or lurk).
Post your questions and insights. I'll answer and others can too.
Help: Report glitches ahead of my release of the full version.
Others depend on you.
August 29, 2013
- Threshold traits (those with 2 alternative states, for example 'red' and 'black') are now supported.Bug fixes
- This version should run on Mac OS 10.7 and 10.8.
- Information on formatting data for threshold traits.
- Commentary on subtlties in the interpretation of cline widths and associated dispersal estimates for clines in continuous and threshold traits. In the section on Limits ot Interpretation.
August 20, 2013
This is the first release
ClineFit v2.0 beta-02 (.zip format)At the next window, double-click the name of the file you want and it will download.
ClineFit Manual (pdf)
ClineFit Graphics.nb (.zip format. - 107MB. A Mathematica notebook with some graphing functions)
These were used to generate the example graphics images:
ClineFit v2.0-beta-01 (.zip format) -- originally released: 20 August 2013
ClineFit Manual (pdf) -- originally released: 20 August 2013
The user's group might be the best place to post.