[R-sig-Geo] Testing for differences in Niche Breadth using Levins' B calculated from Maxent SDMs

Grant, Alannie-Grace agrant11 at vols.utk.edu
Wed Apr 13 19:19:48 CEST 2016

Good Day R-Sig-Geo Friends,

I am interested in testing whether sister species pairs (species with trait
A vs species with trait B) have significant differences in niche breadth.

To do this, I have generated for each species pair 100 Maxent bootstrap
replicates. For each of these replicates I have calculated the niche
breadth using Levins' B from ENMTools.

Now I would like to use these values for a significance test. Is it
incorrect to use a t-test to look for difference between species pairs due
to the lack of independence of the replicates? Is it better to use a
nonparametric test? If so, why and what should the sample size be and which

Finally, I am also interested to know whether groups of species with trait
A and groups of species with trait B have differences in niche breadth.
Using the same data above, what is the most appropriate statistical test
and appropriate sample size?

Sample data:

species A   species B0.168581128 0.0924769420.138468258
0.0975361750.16412658  0.1076619820.16371685  0.1032603730.162732326
0.1217574970.129050784 0.1130964890.126583781 0.1149817320.176855441
0.0993640140.257605323 0.1037768860.182453389 0.1082939280.175081388

Thanks for reading and I am very interested to know your suggestions.

Alannie-Grace Grant
Graduate Student
Kalisz Lab
University of Tennessee - Knoxville


	[[alternative HTML version deleted]]

More information about the R-sig-Geo mailing list