Using digital soil maps to infer edaphic affinities of plant species in Amazonia: Problems and prospects.

摘要: Amazonia combines semi-continental size with difficult access, so both current ranges of species and their ability to cope environmental change have be inferred from sparse field data. Although efficient techniques for modeling distributions on the basis a small number occurrences exist, success depends availability relevant data layers. Soil are important in this context, because soil properties been found determine plant occurrence patterns Amazonian lowlands at all spatial scales. Here we evaluate potential purpose three digital maps that freely available online: SOTERLAC, HWSD, SoilGrids. We first tested how well they reflect local cation concentration as documented 1,500 widely distributed samples. measured differed by up two orders magnitude between sites mapped into same class. The best map-based predictor was obtained regression model combining classes HWSD exchange capacity (CEC) Next, evaluated what degree known edaphic affinities thirteen (as 1,200 sample sites) can maps. segregated clearly along gradient field, but only partially model-estimated gradient, hardly CEC gradient. main problems reducing predictive were insufficient resolution and/or georeferencing errors combined thematic inaccuracy absence most variables. Addressing these would provide better models environment ecological studies Amazonia.