I am looking for a clear explanation of how to correctly interpret ordination ellipses on an NMDS plot and how that interpretation changes with different user-selected confidence levels and ellipse 'kinds' (eg. standard error vs. standard deviation). Specifically I am using the ordiellipse() function within the vegan package.
For example, in my NMDS plot my data is grouped by a factor with three levels. I can plot three separate ordination ellipses using ordiellipse() specificying kind='sd' and conf=.60 and they encompass most but not all of my points, giving me an idea of the centroid and spread of the points within each group. But I'm struggling to articulate EXACTLY what these ellipses represent. Do they mean we have 60% confidence the true mean of each group lies within its ellipse? Or that we are 60% confident that the mean + 1 Standard Deviation of each group lies within its ellipse? Or something completely different? How does this interpretation change if I alter the confidence level conf within ordiellipse()? Or the ellipse kind from SD to SE?
