Inference from eDNA-based field distributions vs laboratory analysis of isolated strains: physiological performance of non-marine Antarctic biota

Abstract

Environmental DNA (eDNA) is frequently used to infer distributions of microorganisms in Antarctica. Their distributions relative to environmental variables are, in turn, sometimes used to infer their physiological range (and a relationship between the two is generally assumed for conservation purposes). We sought to determine whether ecological inferences based on distributions accurately reflect tolerances of the organisms concerned, using 249 legacy non-marine samples from a latitudinal gradient between 72 and 86°S, Antarctica. A cyanobacterium, a heterotrophic bacterium, two eukaryotic algae, two fungi, and a moss were isolated into culture, and their field distributions inferred using eDNA analysis of the samples above. Tolerances of each organism with respect to environmental predictors were then inferred from the eDNA distribution and metadata using Generalised Additive Models. We then measured growth of the cultured isolates in response to a set of these predictors. Laboratory responses were then compared to inferences from the eDNA/metadata. Predictions from eDNA/metadata agreed with the results of physiological laboratory experiments for strains that were detected at high taxonomic resolution in the field samples. However, errors were never completely eliminated, and direct contradictions occurred when strains were represented at lower taxonomic resolution in the field data. We found that accurate ecological inference from eDNA studies would be best achieved via maximising both taxonomic resolution (through marker choice/read length) and ecological signal (through careful sampling design and rigorous metadata collection).

Bibliographic Information