Diversity
In general, considering not only richness, but also evenness and dominance, resulted in more significant differences between treatments at all taxonomic levels, except for Order.
Using COI, we found significant differences in MOTU diversity among treatments when considering all 72 trees (𝜒2 = 10.493; p = 0.005). Pairwise analysis showed that the diversity under one-year-exclusion trees was marginally significantly higher than under control trees (estimate = -0.467; T = -3.183; p = 0.063), while no differences were found between control trees and those with long-term livestock exclusion (estimate = -0.303; T = -2.111; p = 0.095), although it was slightly higher in the latter (Figure 3, bottom left panel A). Restricting the analysis to just the three farms with control and one-year exclosures, Treatment had significant effects on the MOTU diversity (𝜒2 = 10.188; p = 0.001), but not Farm (𝜒2 = 1.104; p = 0.576). Post hoc tests indicated that it was significantly higher at short-term livestock exclosures at the three farms (Figure 3, bottom left panel B).
Using 16S, the results were very similar to COI. We found significant differences due to Treatment in the analysis involving the three of them (𝜒2 = 14.417; p < 0.001), although Tukey tests indicated that only in one-year exclosures MOTU diversity was higher than at the controls (estimate = -0.439; T = 3.684; p = 0.001) (Figure 3, bottom right panel A). Considering only the farms with control and one-year exclosure trees, both Farm (𝜒2 = 30.058; p < 0.001) and Treatment (𝜒2 = 13.677; p < 0.001) had significant effects on the diversity of the arthropod community. Post hoc analysis showed that the diversity beneath the canopies of one-year-exclusion trees was higher in all three farms in comparison with their control pairs (Figure 3, bottom right panel B).
At family level we found significant differences due to Treatment (𝜒2 = 11.099; p = 0.004) in the analysis that included the three categories. Interestingly, in this case the post hoc analysis showed that not only the diversity under the one-year exclusion trees was significantly higher than under the control trees (estimate = -0.442; T = -3.183; p = 0.006), the same happened when control trees were compared with those at abandoned farms (estimate = -0.336; T = -2.414; p = 0.048) (Figure 4C). Removing the long-term-exclusion trees, we found differences due to Treatment (𝜒2 = 10.687; p = 0.001) but not Farm (𝜒2 = 1.127; p =0.568). Arthropod diversity at Family level was higher at short-term exclosures compared to controls in all the three farms (Figure 4D).
At Order level, similarly to the results obtained looking only at richness, we found no significant differences in any case.