BIOL2022 L22 nMDS and hypothesis testing

In a(n) nMDS ANOSIM PERMANOVA SIMPER Bray-Curtis dissimilarity matrix( nMDS, ANOSIM, PERMANOVA, SIMPER, Bray-Curtis dissimilarity matrix ), a Low High( Low, High ) stress value means the model is good.

ANOSIM is ANalysis Of SIMilarities.
Retains philosophy underpinning nMDS ANOVA PERMANOVA SIMPER( nMDS, ANOVA, PERMANOVA, SIMPER ), driven by Information theory Hypothesis testing( Information theory, Hypothesis testing ).
Hands Simple Complex( Simple, Complex ) designs best.
Non-parametric Parametric Metric Non-metric( Non-parametric, Parametric, Metric, Non-metric ) permutation test combined with randomisation approach to generate significance levels (Monte Carlo tests)
Null – no difference in assemblage composition Within Between( Within, Between ) sites.

Steps for ANOSIM:
1. Test statistic – reflects difference among sites to differences within sites. Based on rank similarities in the underlying triangular matrix, generates Global R Stress value P value Pearson's correlation( Global R, Stress value, P value, Pearson's correlation ).
2. Recompute [above] under permutations of sample labels. If null is true there should be little change to [above] if labels arbitrarily assigned.
3. Significance value – refer the observed value of [above] to permutation distribution. If it is unlikely (<5%) to fall in this distribution, reject null H