Plotting indices with divergent colour schemes

Many diversity indices have numerical distributions that are divergent, i.e. they are centred on some value and the interesting bit is the magnitude of the differences away from that value.  A simple example is z-scores, where the data are centre on a value of zero and the values indicate how many standard deviations above or below the expected value the input data are.   These have been plotted using a divergent scheme since version 4.1, as described here.

However, one can also have indices that are simple differences, and also ratios where 1 is the centre of the distribution, and values of 1/2 and 2 are the same magnitude difference from the centre.  The relative phylogenetic diversity and endemism indices are examples of the latter.  

From version 5, Biodiverse plots difference and ratio indices using a divergent colour scheme.    These use the same colour range as the z-scores but plotted along a continuous scale instead of as ordinal classes.  

The colouring happens automatically based on metadata stored with the indices (incidentally, the much of GUI is built using this metadata).  

Colours are also scaled so the most extreme "high" colour is equivalent to the most extreme "low" colour, i.e. if the range of difference values is -5 to 1 then the colours are assigned to the range -5 to 5, and the same for -1 to 5.  This is also accounted for when the data are log scaled or percentile trimmed to de-emphasise extreme values.  

A useful point to note is that the colour schemes can be flipped, so if one prefers blue as extreme positive values then this can be done under the Map menu at the left of the display.  

An example is below to compare the old behaviour with the new.  

Prior to version 5, ratio data were plotted using the same colour scheme as any other data, making it difficult to interpret the relative magnitude of the index values across cells.  These are the Relative Phylogenetic Diversity results for the Acacia data set of Mishler et al. (2014), scaled to emphasise the inner 90% of the distribution (i.e. the upper 5% are assigned the same colour, so too the lower 5%).  This is the interval [0.406, 0.896], which means red cells include ratios <1 which is not ideal.  Compare with the next figure.    

The same data as in the previous figure, but now using a divergent colour scheme.  Biodiverse knows this is a ratio index, so assigns colours accordingly.  Red cells have ratios exceeding 1, blue cells less than 1.  Ratios close to 1 are in yellow.  The colours are assigned to the interval [0.406,2.463], where 2.463=1/0.406.  This means one can be sure red cells have ratios exceeding 1, and there is less chance of misinterpreting the results.  

It is not shown here, but the metadata is also stored for tree-based indices so divergent colours are assigned to the tree branches where appropriate.  More details about that process are in this post.  

----

Shawn Laffan

04-Nov-2024

For more details about Biodiverse, see http://shawnlaffan.github.io/biodiverse/  

For a list of some of the analyses Biodiverse has been used for, see https://github.com/shawnlaffan/biodiverse/wiki/PublicationsList 

You can also join the Biodiverse-users mailing list at https://groups.google.com/group/Biodiverse-users or start a discussion at https://github.com/shawnlaffan/biodiverse/discussions