PERMANOVA vs. dbRDA for unbalanced microbiome data with nested study design

[alegiacomini]

Hi everyone,

I'm working on a 16S faecal microbiome datasets from 135 horses (71 semi-feral and 64 more intensively managed individuals).
The semi-feral horses were sampled from 5 areas (n per area: min = 5, median/mean = 15, max = 23), while the managed horses were sampled from 12 areas (n per area: min = 1, median/mean = 5, max = 15). All horses were sampled only once.
Sampling occurred over multiple months. While semi-feral horses areas were sampled more than once, all but one managed horses areas were sampled only once. Below is a table showing the distibution of samples by sampling area (rows; red = semi-feral, blue = managed) and month (columns):

We're interested in testing the effects of sampling month and sex on microbiome beta diversity using adonis2().
We first ran ANOSIM and found that, for most sampling areas, microbiomes were more similar within vs among areas. Therefore, we decided to restrict permutations within sampling area.
However, I'm uncertain if and how it works in areas with only 1 or very few samples. I'm aware that restricting permutations within strata assumes sufficient group-level variability, which is likely not present here due to the study design.

Additionally, for a subset of 37 managed horses (from 9 areas; n per area: min = 1, median = 2, mean = 4, max = 12), we have metadata on management variables. We'd like to explore the effect of age (continuous), diet (categorical, 4 levels), time spent grazing (categorical, 5 levels), time after last anthelmintic treatment (continuous).
Again, my first idea was to use adonis2() with permutation restricted within area. However, because of the unbalanced sampling effort, and the fact that many areas have onlu one or two diet categories, I'm unsure whether this is appropriate.

I've been considering running dbRDA instead, with sampling area included as conditioning variable, to account for its effect directly.
Do you think it could be a better idea, compared to PERMANOVA with stratification?

Thank you very much in advance, and please let me know if any additional information would be helpful.