I am running a PERMANOVA using adonis2 from the vegan package. I have 52 vegetation plots within 4 study areas. I am interested in how exclosure, fire, and bear presence is impacting these vegetation plots. I also want to include a random effect for study area. All of the plots with bears are in Study Area A or B, and all of the plots without bears are in Study Area C or D. The other treatments (exclosure, fire) were applied randomly across al study areas. I understand adonis2 will not permit random effects explicitly, so I put it first in the formula list since order matters in the adonis2 function.
The following code returns an output with the full expected list of independent variables:
> adonis2(formula = comp[ ,10:22] ~ FenceStatus + WolfStatus + FireStatus,
+ data = comp,
+ method = "euc")
Permutation test for adonis under reduced model
Terms added sequentially (first to last)
Permutation: free
Number of permutations: 999
adonis2(formula = comp[, 10:22] ~ FenceStatus + WolfStatus + FireStatus, data = comp, method = "euc")
Df SumOfSqs R2 F Pr(>F)
FenceStatus 1 4435 0.00812 6.3237 0.003 **
BearStatus 1 26352 0.04823 37.5756 0.001 ***
FireStatus 1 299575 0.54831 427.1694 0.001 ***
Residual 308 216001 0.39534
Total 311 546362 1.00000
---
However, adding in the StudyArea term causes the Fire variable to disappear from the results list:
> adonis2(formula = comp[ ,10:22] ~ StudyArea + FenceStatus + BearStatus + FireStatus,
+ data = comp,
+ method = "euc")
Permutation test for adonis under reduced model
Terms added sequentially (first to last)
Permutation: free
Number of permutations: 999
adonis2(formula = comp[, 10:22] ~ StudyArea + FenceStatus + BearStatus + FireStatus, data = comp, method = "euc")
Df SumOfSqs R2 F Pr(>F)
StudyArea 3 137664 0.25196 68.271 0.001 ***
FenceStatus 1 4435 0.00812 6.598 0.003 **
FireStatus 1 198587 0.36347 295.452 0.001 ***
Residual 306 205677 0.37645
Total 311 546362 1.00000
---
Could this be because of the nested structure of BearStatus within Study Area? Since all of the plots with bears are in Study Area A or B, and all of the plots without bears are in Study Area C or D A or B, and all of the plots without bears are in Study Area C or D.
Any help is appreciated. I can provide a reprod dataset if needed, but I thought maybe it would be obvious to someone with just seeing my output.
