When plant community totals only were analyzed, bare ground retained its high canonical loading.
Bare ground was combined with diversity index variables rather than cover or live standing crop variables, however, as the primary indicators of treatment effect.
The contrast between total species diversity and seeded species diversity also received highest canonical loadings in the second canonical variate.
These findings suggest that herbicide treatment response for the plant community as a whole may be more accurately determined by augmenting cover and live standing crop measurements with estimates of plant diversity.