Skip to content

Error with "raup" distance, in metaMDS

3 messages · Brian Inouye, Chris Mcowen, Manuel Spínola

Original
Brian Inouye
# 
There are other dissimilarity metrics that could also be used, each with 
different assumptions of course.
Raup-Crick is a reasonable choice of distance metric, but I'm not sure 
if the current version of vegdist() has been updated to reflect 
discussion in Chase et al 2011 in Ecosphere 
(http://www.esajournals.org/doi/abs/10.1890/ES10-00117.1).

-Brian
On 3/28/2011 12:00 PM,"Manuel Sp?nola"<mspinola10 at gmail.com> wrote:
Chris Mcowen
# 
Dear list,

I am modelling the distribution of a variable FD, FD is the functional diversity contained within ecoregions i have various predictors temperature, rain fall, population density etc etc, I have ( through AIC and stepwise) arrived at the most likely model the predicts the distribution of FD. However when i look at the community composition of the ecoregions that deviate from the model they have community compositions that are heavily skewed to certain families.

So i would like to know the effect on the family composition on FD in relation to the other factors so i can say something like: anthropogenic and environmental factors explained 10 percent of the observed variance in FD however the composition of the community explained.. %

I have done 

explanation_mod_FD <- lm(Residuals_FD~Temperature+Agriculture+Population) which had a r sqaured of 0.10 and all correlates were significant

now i tried 

explanation_mod_FD <- lm(Residuals_FD~Temperature+Agriculture+Population + FAMILY, correlation = exponential)

However, as expected i guess, i get

(Intercept)             4.944e-04  5.219e-03   0.095   0.9245    
Temperature             2.311e-04  1.716e-05  13.470  < 2e-16 ***
Agriculture            -2.637e-09  3.567e-10  -7.393 1.84e-13 ***
Population              3.389e-06  6.380e-07   5.311 1.16e-07 ***
FAMILYAlliaceae        -1.598e-03  5.432e-03  -0.294   0.7686    
FAMILYAlstroemeriaceae -6.986e-03  5.419e-03  -1.289   0.1974    
FAMILYAmaryllidaceae   -3.045e-03  5.344e-03  -0.570   0.5689    
etc etc

Where each family is broken down.

Multiple R-squared: 0.13

Does this suggest that adding FAMILY explains 3% more of the variance than not adding it? 

What i want to know is A) is the compostion of family within the region significant and by how much.

Any ideas or suggestion would be gratefully received, thanks in advance.

Chris