Beyond central-tendency: If we agree discrete vegetation communities do not exist, should we investigate other methods of clustering?

There was a weak trend towards reduced interconnectivity across the data as the number of samples increased, with the full dataset showing a slightly larger skew in the number of weakly interconnected samples than any of the subsets (Figure 6 ). 4 DISCUSSION 4.1 Is interconnectivity a better indicator of instability?

also in 244 other papers

However, cluster solutions generated using a neighbourhood size proportional to the number of samples (prop) in combination with symmetric link (SL) edge weighting had marginally higher classification stability on average (Table S1 ), as reflected in a marginal significance test for the interaction term in analysis A (Table S2 ).

also in 974 other papers