Randomization on multiple metrics

[André Samayoa]

Hi Prof. Shawn,

As part of a postgraduate project, we are studying the biogeography of marine fishes using a set of taxonomic and phylogenetic metrics which at the moment are being examined with Biodiverse Release 3.99 from May 2022. We do have however the following questions to address to finish pulling out our methodology:

1. When using randomisations and multiple metrics are calculated using Biodiverse, are all metrics calculated from the same randomisation or is a new randomisation performed for each metric. i.e. for 1000 randomisations and two metrics are both metrics calculated from each randomisation or are 2000 randomisations performed 1000 for each metric?

 

1. We were hoping to examine Variation of Pairwise Distance (VPD, Clarke & Warwick, 2001) in combination with a range of metrics calculated in Biodiverse. Is there a way to calculate VPD using Biodiverse? Furthermore and related to our question above, would it be possible to export the randomisations from Biodiverse so that we might be able to calculate VPD outside of Biodiverse?

  Thank you for your time and help!

André 

[Shawn Laffan]

Hello André,

For the randomisations, all metrics are recalculated for each random realisation.   This means that all indices at each iteration are calculated using the same random samples. 


VPD was in version in version 3.0 but it was moved out of the main distribution in version 3.1.  The relevant ticket tracking the move is https://github.com/shawnlaffan/biodiverse/issues/751

If you want to use version 3.0 then you can access it through the http://biodiverse.unsw.edu.au/downloads/ 
Look for "Taxonomic/Phylogenetic distinctness" under the "Phylogenetic Indices". 

However, those indices do not run the resampling analysis to calculate significance.  You can instead run a normal randomisation where richness is held constant per group (this is actually the default randomisation).  The results are rank relative significance scores instead of z-scores but can still be used to identify scores that are more extreme than expected.  Z-scores of significance are in the 3.99 series, but those versions obviously do not have the VPD index.  (A caveat with this approach is that richness is only held constant on a per-group basis, so if your analysis neighbourhood includes additional groups (cells) then the richness of the pooled sample of labels (taxa) is unlikely to be the sum of the individual richness scores).


Another alternative if you are using version 3.1 or later is to look at the PMPD1_RMSD index.  This is the square root of the mean squared distances between pairs of tree tips in the sample.  It is the first part of the VPD calculation but lacks the subtraction of the mean.  In Biodiverse the mean is the PMPD1_MEAN index so it would not be difficult to also calculate the variance if it is needed.  (Actually it would not be strictly the same as VPD uses the sum of squared distances to the most recent common ancestor whereas the RMSD index uses the sum of pairwise distances passing through the most recent common ancestor.  It will be proportionally related, though). 
https://github.com/shawnlaffan/biodiverse/wiki/Indices#phylogenetic-and-nearest-taxon-distances-unweighted


More details about the MPD and related indices are http://biodiverse-analysis-software.blogspot.com/2021/09/faster-calculation-of-phylocom-indices.html

More details about the randomisations are in these blog posts:
http://biodiverse-analysis-software.blogspot.com/2020/12/biodiverse-now-includes-independent.html
http://biodiverse-analysis-software.blogspot.com/2020/11/randomisations-now-also-generate-z.html
http://biodiverse-analysis-software.blogspot.com/2020/11/randomisations-how-randstructured.html
http://biodiverse-analysis-software.blogspot.com/2020/11/randomisations-modelling-spatial.html
http://biodiverse-analysis-software.blogspot.com/2020/11/spatially-partition-your-randomisations.html

If you want all the posts related to randomisations then see http://biodiverse-analysis-software.blogspot.com/search/label/randomisations

Regards,
Shawn.

--
You received this message because you are subscribed to the Google Groups "Biodiverse Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to biodiverse-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/biodiverse-users/5cc9cbe7-6d15-4c9f-b1d5-c4b394ba0454n%40googlegroups.com.

-- 
Prof Shawn Laffan, FMSSANZ
  Director, Earth and Sustainability Science Research Centre
  School of Biological, Earth and Environmental Sciences
  UNSW, Sydney 2052, Australia
  Tel +61 2 9065 5607
  https://www.bees.unsw.edu.au/our-people/shawn-laffan
  https://www.unsw.edu.au/research/essrc
  https://shawnlaffan.github.io/biodiverse (free diversity analysis software)
  
  International Journal of Geographical Information Science
  http://www.tandf.co.uk/journals/ijgis

[Shawn Laffan]

It turns out to not be difficult to add. 

Version 4 will include indices to calculate the variance of the pair-wise distances between tree tips. 
https://github.com/shawnlaffan/biodiverse/issues/811

This includes the resampling approach used for the NRI and NTI indices to generate the funnel plot data and significance scores.  Someone might one day develop exact metrics for this, as for the NRI and NTI, but for now it takes advantage of some of the recent optimisations put in place to speed up the MPD, MNTD, NRI and NTI calculations.  More details about those, and references for the exact methods, are at http://biodiverse-analysis-software.blogspot.com/2021/09/faster-calculation-of-phylocom-indices.html

André - if you need to use these indices in the short term then I will cut a new dev release.  The caveat is that this will need to be on Windows and Linux only as I do not have ready access to a sufficiently recent Mac at the moment. 

Regards,
Shawn.

To view this discussion on the web visit https://groups.google.com/d/msgid/biodiverse-users/3f4483ab-4a58-45df-e032-266b7a2990d6%40unsw.edu.au.

[Francisco F. Pedroche]

Hi Shawn:

We have already uploaded a map from Mexico and it seems to be ok, but we can not see it. Is there a instruction to display it in the screen?

Thanks

Francisco

To view this discussion on the web visit https://groups.google.com/d/msgid/biodiverse-users/3f4483ab-4a58-45df-e032-266b7a2990d6%40unsw.edu.au.

--

Dr. Francisco F. Pedroche

Depto. Ciencias Ambientales

División CBS

UAM-Lerma

http://www.ler.uam.mx/

http://www.uam.mx/

[Shawn Laffan]

Hello Francisco,

Can you clarify what you mean?  

If you have imported your data then you can use the View Labels tab to see the data and explore the distribution of labels (taxa).  This is opened via the Basedata menu or by clicking Show in the outputs tab. 

If you have run an analysis followed by a randomisation, and are looking for the randomisation results then they are stored in additional lists on the spatial output.  More details are at this URL:
https://github.com/shawnlaffan/biodiverse/wiki/AnalysisTypes#where-do-the-randomisation-results-go-and-what-do-they-mean

If you have not done so already it is worth checking the quick start guide. 
https://github.com/shawnlaffan/biodiverse/wiki/Downloads#quick-start-guide


And as a general note to all, the help system needs work.  Any assistance would be greatly appreciated.  This could range from noting errors or points that need to be clarified to helping write the documents. 


Regards,
Shawn.

To view this discussion on the web visit https://groups.google.com/d/msgid/biodiverse-users/CALf2nRzUM0sc%2BAAaGaMsnC7Lb_S5CAK9z3F6oyKiCwp9xyR82Q%40mail.gmail.com.

[André Samayoa]

Hi Shawn, thank you for your reply! 

It clarified more about the way randomizations work within Biodiverse. At the moment I am able to produce interesting data with the indices already included in the software. Regarding VPD and other indices, it would be interesting for our research group (and probably for other investigators in the field) to wait for your next release with those metrics so we can explore further phylogenetic patterns within our focal taxonomic groups.

Thanks for your time and clarifications,

André

[Shawn Laffan]

Hello André,

I have just released Biodiverse 3.99_004 which contains VPD.  I will post the full announcement separately. 

A Windows executable can be downloaded from https://github.com/shawnlaffan/biodiverse/releases/tag/r3.99_004

Regards,
Shawn.

To view this discussion on the web visit https://groups.google.com/d/msgid/biodiverse-users/18dbf860-d40b-4e0d-87b2-431fa99c33f5n%40googlegroups.com.