Aiming to determine the effects of different spatial sampling scales on the delineated characteristics of species’ realised niches we investigated 222 helocrenic springs in siliceous mountain ranges in central Germany and the north west of the Czech Republic. The probability of bryophyte and vascular plant species occurrence with respect to pH was modelled, using generalized additive models (GAM). To compare species optima and amplitudes, these were derived separately for two spatial scales, by pursuing an entire-spring and a subplot-based within-spring approach.

Species occurrence is driven by an acidity gradient, which is directly related to nutrient availability. The pH value best explains species composition, regardless of the scale of observation. But, the derivation of niche characteristics is not impartial to the sampling scale. Although the sampling scale did not systematically influence the species’ pH optima, we found that a smaller plot size (grain) narrowed their pH amplitudes significantly. Hence, the delineation of attributes of species’ realised niches is a matter of sampling scale, in particular for ecological amplitudes. This affects ecological applications, such as niche-based distribution models. The predictive capacity of these models is limited by the sampling scale (i.e. grain) that is chosen.