NMS, Environmental Vectors Not Displaying

[Terry Serres]

Hello, First a little background. I am revisiting analyses for a long-abandoned Master's thesis from ten years ago. So, please excuse any stupidities. You can cattle-prod me in the right direction.

I am running PC-ORD version 6. With the McCune & Grace book and Jeri's manual. 

I want to run NMS on environmental data, but at the outset at least without plant community data. (I have only two species.) I just want to see where my plots by treatment sort in environmental space, and I did this successfully before. I am attempting to rerun these analyses, eventually to add one new variable. 

My matrix has 12 variables, all transformed and standardized to normality.  A secondary matrix has the groups by plot. Consulting both the manual and my output / results file from ten years ago in order to verify parameters, I am able to reconstruct my NMS and run it. Output looks consistent with results from before. 

Graphs look fine with one critical exception. My group hulls look good in separating the treatments, I am able to explore separation of groups/treatments by individual environmental variables using box plots and using overlays in the graphing interface. I can run a useful MRPP to analyze the separation among my groups/treatments. 

Problem: I cannot get the vector of environmental variables to show up. I have tried everything in terms of double-checking my work. I can produce and append the Correlations output that essentially allows me to reproduce the set of vectors in R, but having no vectors displaying in PC-ORD is making it impossible to rotate my axes to explore the best presentation of data and to do the other things I need to do to optimize my output.

Thanks for any guidance you may be able to offer,

Terry Serres

University of Minnesota

Department of Forest Resources

[Bruce McCune]

Terry, since you have the environmental matrix in the main matrix position, if you want to show the environmental variables as a joint plot, you should open that as the second matrix at the same time. Then select the joint plot button or select Graph | Joint plot.

Bruce McCune

--
You received this message because you are subscribed to the Google Groups "PC-ORD" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pc-ord+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pc-ord/e39e2534-e16e-4f6e-8113-3eff48fb03f4n%40googlegroups.com.

[Terry Serres]

Hi Bruce, Thanks for the guidance here. 

I tried this before, and retried it again after your reply. Main matrix with 96 plots, 12 environmental variables with columns labeled Q. Second matrix the same, but with the column indicating my groups/treatments appended (labeled C). (I also did a run where both matrices were identical, in other words the main matrix included the groups/treatments.)

I was expecting the results to be similar to the many consistent runs I've done with the same main matrix, second matrix showing only the treatments. Those runs all provide a reasonable 3-dimensional solution with treatments sorting logically in a way that reflects my other analyses. I get logical group/treatment hulls and centroids, and using the main matrix overlays I can see how plots of various treatments sort in environmental space. The only thing missing from these ordinations is environmental vectors.

Instead, using the same matrix for the main and second, I get a high-stress one-dimensional solution with a "jittery" stress plot, which doesn't allow me to get vectors either. Am I doing something wrong, or is my NMS just whack? Thanks for your insights,

Terry

You received this message because you are subscribed to a topic in the Google Groups "PC-ORD" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pc-ord/tmwsj1aw1OE/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pc-ord+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pc-ord/CAK1WPPnOhoZ7saRBvRg4UnOxseUZNZ5Of12WOVXDxKAeDUJExA%40mail.gmail.com.

[Bruce McCune]

Terry, if the vectors SHOULD be there but they are not, then I suspect none of them are beating the cutoff value for strength of relationship with the axes. You can find out for sure by setting the cutoff r2 value to zero. (Click on the button showing the .200 default value). Then all of the vectors will display, regardless of their length.

It sounds like you are having stability issues with your NMS, or perhaps you are using different settings from last time.

Since you are using the environmental matrix in main matrix position, I would recommend relativizing by standard deviates of columns. Then use Euclidean distance. 

Or, if you saved your files from years ago, just reopen the graphrow.gph file (File | Open | Graph Row File).

If you saved your result files, you can figure out what settings you used for the NMS previously. The options that you selected are all listed at the top of the result file.

Bruce McCune

To view this discussion on the web visit https://groups.google.com/d/msgid/pc-ord/CANB8vzgzOLVRL7r14F7weT8FUhQHjTMysnCxXPXFwvnwRFQoqw%40mail.gmail.com.

[Terry Serres]

Hi Bruce, Thank you for the suggestions. Luckily, I had documented my 2012 work well enough to reconstruct. It did indeed work. And I have been able to make the ordinations work as expected with my current data configuration (I had swapped out one variable and corrected the data for another). 

I thought I had done everything correctly, including verifying my settings. So I don't quite know what I was doing wrong the past few days. Perhaps there was a mismatch between my main  and second matrices. This time I appended my class fields to the main matrix to construct my second matrix and verified the data against source documents (no errors found).

So I don't know what I did wrong before, but reconstructing my 2012 NMS step by step per your suggestion jiggered something. Thanks very much,

Terry

To view this discussion on the web visit https://groups.google.com/d/msgid/pc-ord/CAK1WPPkUrrkxWCZsk8_ny2AE0XNpDsvvUsrXahdcdJR713i9xA%40mail.gmail.com.

[Mona N. Högberg]

Terry´s question; yes some options are not unavailable under certain conditions. I am in the same situation, not really knowing why things don't work. 

The issue I am trying to solve today is how to add a successional vector (illustrating change in time and connecting my five time periods) as JE Peck suggests in her text book on p. 154 for her one-dimensional solution (the joint plot option is not available either, this is to understandable). Any tips? I have already used the many suggestions here regarding one-dimensional solutions and how to present the data but I find Peck´s suggestion as really good. I use PCord v. 7.11 

Many thanks,

Mona N. Högberg

Swedish University of Agricultural Sciences

Forest Ecology and Management