The edaphic fauna plays an important role in the organic matter dynamics of the soil. Millipeds (Diplopoda),
centipeds (Chilopoda) and isopods (Isopoda) are among the main components of the soil fauna. Vegetation
structure represents one of the main factors influencing composition and diversity of these species that inhabit
leaf litter and soil, what is particularly important in the Amazon Forest, where there are several singular ve-
getation types. Savannah and forest areas stand out in ferruginous hills, which represents one of the most poorly
known and highly threatened ecosystems in the Amazon. However, sampling the edaphic fauna in high diversity
sites represents a challenging task. Studies have shown that sampling method may influence observed patterns of
diversity. From this perspective, the present study aimed to (i) investigate whether assemblages of Chilopoda,
Diplopoda and Isopoda differ between vegetation types in the Amazon, (ii) indicate how biological parameters
behave in such communities, (iii) identify which method or combination of methods is better suited to sample
species richness and (iv) quantify the efficiency of each method regarding time of work and cost. Sampling was
conducted by means of hand collection, leaf litter collection, hay-bait traps and soil sampling in four patches,
two in the savannah (also called canga) and two in the forest. Richness was higher in forest habitats for
Chilopoda and Diplopoda, while for Isopoda there was no significant difference. Species composition also dif-
fered between vegetation types for Chilopoda and Isopoda, but not for Diplopoda. Beta diversity between ve-
getation types was mainly determined by nestedness and balancing, while turnover was the main beta diversity
component responsible for differences among methods. Soil sampling and hand collection added more species to
the samples, therefore, this combination was the most efficient to access diversity. Our results support the
hypothesis that differences in the soil cover promoted by distinct vegetation types influence the species diversity
and composition of edaphic fauna. The differences were mainly determined by nestedness and balancing, in-
dicating that the fauna sampled in the canga represents a subset of those present in the surrounding forest. In this
sense, even in ecosystems with distinct vegetation types, the combination of soil and surface sampling methods is
recommended. Lastly, the present study proposes a set of decision rules and a key to select the most appropriate
method to sample the studied taxonomic groups.
