IntroductionElevation gradients are often used as a proxy for climate change as they allow comparisons of ecological responses over much larger temporal and spatial scales than is possible through experimental manipulations.MethodsHere, we tested how microarthropod communities (Collembola and Acari) are affected by climatic differences between sea level and 600 m a.s.l. on Navarino Island, in the Magellanic sub-Antarctic ecoregion of southern Chile (mean annual temperatures of 5.6 vs 3.1°C, respectively). We quantified microarthropod abundance, richness and community trait characteristics in dominant moss (Racomitrium lanuginosum and Polytrichum strictum) and lichen (Usnea trachycarpa, Pseudocyphellaria freycinetii and Stereocaulon alpinum) vegetation growing at both elevations. These moss and lichen genera are characterized by large morphological differences and allow testing of how habitat characteristics affect microarthropod community response across elevation gradients.ResultsCollembola and Acari community composition differed between the low and high elevation sites. Total abundance levels of Acari were maintained in each habitat across elevation, whereas Collembola richness strongly declined (50%) at high elevation in the moss habitats. Acari community differences across elevation were driven by relative abundance changes whereas the Collembola community lost species at higher elevation. An anticipated decline of smaller eudaphic Collembola at high elevation was only observed in the moss Racomitrium, reflecting potentially lower temperature buffering capacity and shelter options compared to Polytrichum. Lichens mostly supported larger epigeic species irrespective of elevation. There were no consistent patterns linking microarthropod communities with habitat water holding capacity or water loss rates across the studied habitats and elevation.DiscussionHabitat type and the genus of moss or lichen were associated with microarthropod community changes across elevation, including examples of declines, increases and no change. These findings highlight that community responses across gradients may not always relate to the generally hypothesized environmental variables (e.g. temperature variability) and that habitat characteristics should be taken into account when using elevation as a proxy for climate change.