Supplement 1. The Corrigan and the Nile Creek data sets and the S-PLUS functions for fitting the piecewise-regression models used in the paper.
2016-08-05T00:09:13Z (GMT) by
<h2>File List</h2><p><i>All files at once</i></p> <blockquote> <p><a href="pw_regression.zip">pw_regression.zip</a></p> </blockquote> <p><i>Data files</i></p> <blockquote> <p><a href="corrigan_data.txt">corrigan_data.txt</a> - tab-delimited ASCII file of presence/absence data at the Corrigan site.</p> <p><a href="nile_creek_data.txt">nile_creek_data.txt</a> - tab-delimited ASCII file of presence/absence data at the Nile Creek site.</p> </blockquote> <p><i>S-PLUS files</i></p> <blockquote> <p><a href="pwreg_fns.txt">pwreg_fns.txt</a> - ASCII file of S-PLUS functions for fitting the piecewise regression models</p> <p><a href="fn_calls.txt">fn_calls.txt</a> - ASCII file of example function calls (will recreate most of the results in the paper)</p> </blockquote><h2>Description</h2>The files corrigan_data.txt and nile_creek_data.txt are tab-delimited ASCII files of the data used. Tree species are split into height categories (height >1.3 m, or <u><</u>1.3 m). In both files, columns 2–13 are from transect #1, columns 14–25 are from transect #2, and columns 26–37 are from transect #3. S-PLUS functions for fitting all piecewise-regression models in the paper (sharp, hyperbolic-tangent, bent-hyperbola, and bent-cable piecewise-regression models) are available in pwreg_fns.txt. The file also contains additional functions for deriving and plotting confidence intervals for an estimate of the breakpoint. Import these functions into a S-PLUS session using the internal S-PLUS function source. The functions are known to work in S-PLUS version 4.5 and 6.1 for Windows. The results for the case study can then be recreated using the commands listed in fn_calls.txt. As written, the function calls assume that the first PCA axis for the Corrigan data has been extracted and called corpapca.df, that the first PCA axis for the Nile Creek data has been extracted and called nilpapca.df, and that a vector of distances from edge has been created and called distance.