Making the most of scarce data: Mapping soil gradients in data‐poor areas using species occurrence records

Abstract

Abstract Maps of environmental characteristics are needed to improve our understanding of species distributions and ecosystem dynamics. Despite the growing demand for digital environmental maps, scarcity of environmental field samples to be used as input data often constrains the accuracy of such maps, especially for soils. We developed and tested a method that combines information on species–environment associations and the spatial distribution of indicator species (as retrieved from repositories such as GBIF) to improve mapping accuracy of environmental variables. Our approach includes: (a) Compile field data on the environmental variable of interest (direct environmental data) and documented occurrences of the species to be used as indicators; (b) define species optima for the environmental variable; (c) use georeferenced records of the indicator species to calculate species‐based environmental values (indirect environmental data); (d) generate maps using direct and indirect environmental data as input data for interpolation; (e) validate the maps. We applied the method to map the concentration of exchangeable base cations in Amazonian soils using fern and lycophyte species as indicators. Including soil values that had been indirectly estimated using indicator species represented a 12‐fold increase in the number of input data points used for mapping. At the same time, map accuracy improved considerably: the correlation between mapped soil cation concentration estimates and field‐measured values from an independent validation dataset increased from r = 0.48 to r = 0.71. Knowledge on species–environment relationships can be useful for modelling ecologically relevant environmental variables in areas where species occurrence data are more readily available than direct environmental measurements. The method works even with haphazard species occurrence points obtained from public repositories such as GBIF and can be applied to other environmental variables and other indicator groups, provided that the environmental variable of interest is relevant as a determinant of species occurrences in the indicator group. The Amazonian soil cation concentration maps produced (available at https://doi.pangaea.de/10.1594/PANGAEA.879542) can be used as digital layers in species distribution and habitat modelling, and to guide conservation actions in Amazonia.