Arthropod communities recovered by metabarcoding and sampling blocks
The estimation of species richness in any ecological setting, and especially in forested environments, can be challenging due to the rarity of some species, differences in detection probabilities, and the field effort necessary to collect enough samples or species to ensure meaningful coverage (Andújar et al., 2017; Arribas et al., 2016; Creedy et al., 2019). In our study, we used pitfall traps in sampling blocks, maximizing the probability of detecting arthropod species by sampling intensively at multiple sites in one mountain, covering eight orders of arthropods from haplotypes to communities. Our sampling method and size sorting step allowed the recovery of eight major arthropod orders, congruently with other metabarcoding analyses (Elbrecht et al., 2017; Elbrecht et al., 2018; Creedy et al., 2019). According to our rarefaction curves, our sampling detected different taxa per sample (Figure S3), demonstrating the utility of cMBC and our sampling design to study a region of high biological diversity and ecological complexity. It is difficult to compare our results against morphological studies in Nevado de Toluca because there are no complete checklists of arthropods for Mexican highlands. For instance, out of 29 dung beetle species found by a recent survey of four sky-islands, more than 10% were new species (Arriaga-Jiménez et al., 2018). Comparing our results against other cMBC studies shows that more OTUs (913 at 3% CL) of different orders were found at a tropical forest canopy (Creedy et al., 2019) than what was found here for a tropical conifer forest floor (476 OTUs at 3% CL).
Additionally, the metabarcoding pipeline we followed (Arribas et al., 2020) allowed us to analyze diversity patterns for each order separately and allowed us to consider multi-hierarchical levels into community assembly, from haplotypes to 7.5% CL lineages. Our metabarcoding approach coupled with pitfall traps allowed for the automated identification of 1,277 ASV from 42 bulk samples, which contained whole organisms (or part of them). This allowed to calculate community composition and turnover without the bias introduced by traditional taxonomy (Creedy et al.,2019), including small taxa (< 0.5 mm), such as Collembola, and specimens that break easily, such as Diptera (Figure 3). Thus, combining metabarcoding and pitfall traps sampling, allows to perform large scale community composition analyses in tropical mountains.