Monte Carlo Simulations 
Top Previous Next 
Monte Carlo randomization tests are only undertaken if selected from the Randomization dropdown menu. This menu allows you to select 100, 1000, and 10000 trials. It is a good idea to initially run only a hundred simulations as this procedure can take a long time if you have a large data set.
For each simulation the order of the samples in the samplespecies array is randomly shuffled and the Eigenvectors calculated as usual. By shuffling the order of the samples we are breaking the correlation between the environmental variables and the species. Thus there will not be any relationship. The probability measure given in the Monte Carlo output grid is the probability that an Eigenvalue as large as that observed could have occurred by chance. Also given in the table is the largest and smallest Eigenvalues observed and the mean Eigenvalue. The calculations are only undertaken for the 3 axes with the largest Eigenvalues as it is unlikely that you will have significant variation explained by subsequent axes.
If the observed probability for an axis is greater than 0.05 then there is a greater then 5% chance that the observed magnitude of the Eigenvalue could occur by random chance and the axis is unlikely to explain any significant part of the variability in the species data.
