Uncertainty in coprophilous fungal spore concentration estimates

PERROTTI, ANGELINA G.; RAMIADANTSOA, TANJONA; OKEEFE, JENNIFER; OTAÑO, NOELIA NUÑEZ

Frontiers in Ecology and Evolution

Frontiers Media S.A.

Año: 2022 vol. 10 p. 1 - 10

The abundance of coprophilous (dung-inhabiting) fungal spores (CFS) in sedimentary records is an increasingly popular proxy for past megaherbivore abundance that is used to study megaherbivore-vegetation interactions, timing of megaherbivore population declines and extinctions, and the introduction of domesticated herbivores. This method often relies on counting CFS alongside pollen and tracers of known concentration such as exotic pollen or synthetic microspherules. Prior work has encouraged reporting CFS abundances as accumulation rates (spores/unit2/year) or concentration (spores/unit3) instead of percentages relative to the total pollen abundance, because CFS percentages can be sensitive to fluctuations in pollen influx. In this work, we quantify the uncertainty associated with estimating concentration values at different total counts and find that high uncertainty is associated with concentration estimates using low to moderate total counts (n = 20 to 200) of individual fungal spore types and tracers. We also demonstrate the effect of varying tracer proportions, and find that larger tracer proportions result in narrower confidence intervals. Finally, the probability of encountering a CFS spore from a specific taxon occurring in moderate concentrations (1,000 spores/unit2) dramatically decreases after a low tracer count (∼50). The uncertainties in concentration estimates caused by calculating tracer proportion are a likely cause of the high observed variance in many CFS time series, especially when CFS or tracer concentrations are low. Thus, we recommend future CFS studies increase counts and report the uncertainty surrounding concentration values. For some records, reporting spore data as presence/absence rather than concentrations or counts is preferable, such as when performing high counts is not feasible.