Quantifying Variations in Soil C:N Relationships for the Permafrost Region.

Soil property ratios are often used to characterize soil organic matter composition. However, mapping and spatial interpretation of soil property ratios is challenging, with no set standard, particularly for the heterogeneous profiles of permafrost-affected soils. Two different approaches - direct and indirect mapping - can be used. For direct mapping, the property ratios determined for each soil observation are used to predict soil property relationships within the landscape. With indirect mapping, each property is predicted independently and the ratio of the resulting two maps are used to calculate property ratios for each map pixel across the landscape. Observations of soil organic C and total N stocks for Alaska and their ratios were used to evaluate which mapping approach best captures soil property relationships for cold region soils. The specific objectives of this study were to (i) evaluate which approach best captures soil C:N relationships at high spatial resolution for a selected latitudinal transect in Alaska and (ii) identify which environmental covariates are important in capturing soil C:N relationships within this study area. We compared four prediction models: stepwise multiple linear regression, universal kriging, cubist, and random forest. Maps of predicted soil C:N relationships were generated at a spatial resolution of 34 m for three depth increments within the surface meter. Overall, random forest performed better than the other prediction models. Although the two mapping approaches performed similarly across the three depth increments, the predictive strength of the environmental covariates differed between approaches and varied with depth. However, both mapping approaches under-predicted the low observations and over-predicted the high observation values, because of the relatively low density and uneven distribution of soil observations available throughout the study area. Knowledge gained from these results will be applied to our ongoing efforts to map soil C:N relationships (as well as soil C and N stocks) for the state of Alaska at a spatial resolution of 34 m. Understanding the spatial and vertical relationships between soil C and N stocks will inform efforts to model and predict how cold region ecosystems will respond to a warming climate.