Capturing arthropod diversity in complex cave systems

Aim: Identify the optimal combination of sampling techniques to maximize the detection of diversity of cave-dwelling arthropods. Location: Central-western New Mexico; north-western Arizona; Rapa Nui, Chile. Methods: From 26 caves across three geographically distinct areas in the Western Hemisphere, arthropods were sampled using opportunistic collecting, timed searches, and baited pitfall trapping in all caves, and direct intuitive searches and bait sampling at select caves. To elucidate the techniques or combination of techniques for maximizing sampling completeness and efficiency, we examined our sampling results using nonmetric multidimensional scaling (NMDS), analysis of similarity (ANOSIM), Wilcoxon signed-rank tests, species richness estimators and species accumulation curves. Results: To maximize the detection of cave-dwelling arthropod species, one must apply multiple sampling techniques and specifically sample unique microhabitats. For example, by sampling cave deep zones and nutrient resource sites, we identified several undescribed cave-adapted and/or cave-restricted taxa in the south-western United States and eight new species of presumed cave-restricted arthropods on Rapa Nui that would otherwise have been missed. Sampling techniques differed in their detection of both management concern species (e.g., newly discovered cave-adapted/restricted species, range expansions of cave-restricted species and newly confirmed alien species) and specific taxonomic groups. Spiders were detected primarily with visual search techniques (direct intuitive searches, opportunistic collecting and timed searches), while most beetles were detected using pitfall traps. Each sampling technique uniquely identified species of management concern further strengthening the importance of a multi-technique sampling approach. Main conclusions: Multiple sampling techniques were required to best characterize cave arthropod diversity. For techniques applied uniformly across all caves, each technique uniquely detected between ~40% and 67% of the total species observed. Also, sampling cave deep zones and nutrient resource sites was critical for both increasing the number of species detected and maximizing the likelihood of detecting management concern species.