I think you used the wrong measure for the question of how litter inhabiting animals influence decomposition. To sample only animals from within the litterbags gives a much biased view of the abundance and diversity in litter and soil. And as you used the same mixture of leaves in all samples, you pushed to decrease differences. Actually you got one measure of decomposition rate (mean) under one condition of decomposer community (including the effect of fauna). Studies showing influence of soil fauna on decomposition rates most often use litterbags of a mesh size not allowing certain groups to enter (mesofauna, macrofauna) and analyse the effect. Correlations of decomposition rates with soil fauna abundance and diversity should be based on a broad/reliable sampling of litter and soil fauna under different soil/litter conditions. And still, all these studies have the problem to correlate measures from one point in time, because an abundant fauna will help decompose litter rapidly and when litter becomes sparse decreases due to the lack of space or food. Both variables are in a continuum over the time. I would only correlate measures of abundance and diversity summed or averaged over a longer period with decomposition rates calculated by regression.
Chapter Biodiversity and Ecosystem Functioning in Tropical Habitats ...