Find the proportion of variance captured by the first two PC. hint
Function eigenvals() may be used to extract the eigenvalues.
Use Kaiser-Guttman kriterion and broken stick model to find the number of axes to keep. hint
Check functions screeplot() and bstick() in the vegan package.
Draw an ordination biplot focused on distances among samples. Use different symbols to distinguish groups from exercise 4 clustering. Eclose these groups into envelopes. In the diagram, keep only a reasonable number of species (about 20) with best fit. hint
To be able to control for the symbol type, you may need first to extract proper site scores using scores() and then plot them with regular plot() function.
Make also sure the correct aspect ratio (asp = 1) is set.
Envelopes may be drawn using vegan's ordihull(). Check ?ordihull for other possibilities to distinguish groups.
Linear fit of variables (and of species as well) may be calculated using vegan's envfit(). You can also use a simple function reduce.pca(), which is based on envfit(), too. It takes a PCA result, calculates the species fit and return the same PCA with just the desired number of best-fitted species.