Following

is it, Multidimensional Scaling (MDS) is a multivariate statistical technique first used in geography. The main goal of MDS it is to plot multivariate data points in two dimensions, thus revealing the structure of the dataset by visualizing the relative distance of the observations

if it is yes download the R s/w you can use that s/w you can get the results

You need first to compute the mean per site, and then you can run the nmds on this new data frame. To this, run

data$site

Hello, thank you Facundo. Yes I am trying to find a multivariate summary of each site - from my reading this is called the centroid (the multivariate mean taking into consideration all variables - ie all species).

Does the above script find the mean of the quadrats from a site in multidimensional species space (that is what I am looking for when I say centroid). Or does it just take all the averages of all variables across all the quadrats individually for each species? I feel like it might do the latter?

I recommend you enter https://stackoverflow.com/ and leave me the question with the code, in order to analyze what you want to do or even publish your code at: https://www.researchgate.net/project/R -for-biologists-Implementation-in-research. Is that the code that comes in the R packages is very basic to do those other things you have to modify it. regards

Hello dear Meredith Karcz first you run the nmds analysis, then you could use envfit function to test if your variables have an effect on this multivariate distribution. Particularly I use ellipse package to plot the ellipses on your initial NMDS biplot. (I developed all process in R with R-studio).

This is an example of both packages.

Regards,

Cristian

follow

following

Hello,

You could estimate centroids and vectors values from envit function.

You would need to use your MDS value create from metaMDS function and your original dataset, only with factors and variables column.

In my case, I use this code:

MDS1