[Primer-E] Doubts on Global test vs. Pairwise test for ANOSIM analyses

#1
Hello,
If anyone could help me with my doubt I would appreciate it.

I'll describe briefly my work.
The objective of this work is to study the fungal community in different grape varieties (5 different grape varieties) and in different types of pruning (2 different types of pruning).

I've done a Two-way ANOSIM test with Primer-E software with factor A belonging to the grape variety and factor B belonging to the type of pruning.

Here are the results:

Tests for differences between Grape Variety GROUPS (across all Pruning groups)
Global Test
Sample statistic (Global R): 0,325
Significance level of sample statistic: 3,4%
Number of permutations: 999 (Random sample from 893025)
Number of permuted statistics greater than or equal to Global R: 33

We consider two things statistically different when p<5%
In this case we obtained a p-value of 3,4%, which means that the grape variety contributed to differences in the fungal community. In order to see which(s) one(s), the program did Pairwise Tests and here are the results:



As you can see, all of the pairwise tests for the grape variety are not statistically significant (p>5%).

And here are the results for the Factor B:
Tests for differences between Pruning GROUPS (across all Grape Variety groups)
Global Test
Sample statistic (Global R): -0,05
Significance level of sample statistic: 61,7%
Number of permutations: 243 (All possible permutations)
Number of permuted statistics greater than or equal to Global R: 150


So my doubt is, if the global test for grape variety were statistically different, shouldn't I expect to see a significant value for some grape varieties at the pairwise test?

Is there any explanation for this?