Sequential Univariate vs. True Multivariate Approaches
- Layering Boolean combinations of multiple variables considered sequentially leads to errors
- Determining ranges independently, variable by variable, and then combining using "AND" leads to overestimation of the actually occupied combinations
- Such overestimation errors are compounded as dimensionality increases
- Not unrealistic - some species can withstand hotter temperatures with more precipitation
- Multivariate clustering simultaneously considers all variables - better estimates the actual occupied combinations
