How far apart should I place my soil plots? Or: Spatial variation in soil properties in North American ecosystems

Background/Question/Methods Soil ecologists often have to decide how far apart to place soil plots to ensure that they are representative of the surrounding area of interest. These decisions are typically based on knowledge of the ecosystem and gut feeling, but this is approach is both subjective and difficult, especially if the experiment is being established in an unfamiliar ecosystem. Geostatistical techniques, e.g. semivariograms, can be used to determine the minimum distance required for plots to be considered spatially independent at a given scale, thus maximizing plot representativeness while minimizing plot number. However, these techniques are rarely used by researchers when planning an experiment due to a lack of time, money, and/or familiarity with geostatistics. We characterized spatial patterns of soil temperature and moisture at over 30 US sites to determine how far apart to place the long term soil monitoring plots at the National Ecological Observatory Network (NEON) sites. Temperature and moisture (0 - 15 cm deep) were measured during peak ecosystem greenness at over 100 points along transects at each site and at distances ranging from 10 cm to hundreds of meters. Temperature and moisture were chosen because they strongly influence soil properties.

Results/Conclusions Preliminary analyses indicated that the distance at which soils can be considered spatially independent (i.e. no longer autocorrelated) were typically greater for temperature (50-101 m) than moisture (8-35 m). The larger range distances based on temperature may be due to small topographic features (e.g. slight rises and dips in the landscape) which would likely influence drainage more than heating. Of the different ecosystems studied so far, the agricultural site had the smallest range distances (50 m for temperature and 8 m moisture), while range distance was largest at the forest site (101 m and 35 m), and intermediate at the grassland site (88 m and 35 m). The agricultural site was ploughed regularly, which likely increased homogeneity in soil properties and resulted in the relatively short range distances. Whereas, the larger vegetation size at the forest site probably created autocorrelation at a larger scale (e.g. due to canopy gaps or individual tree effects) and resulted in the larger range distances. Preliminary results suggest that soil plots should be located >100 m apart in forests ecosystems, whereas in homogeneous agricultural systems 50 m may be sufficient. As more sites are analyzed general rules of thumb will be developed for different ecosystems.