Who is behind the wheel? The drivers of soil N availability in high-elevation red spruce (Picea rubens Sarg.) forests along a gradient of atmospheric N deposition
Atmospheric deposition of reactive nitrogen (N) can cause considerable changes to nutrient cycling in forest ecosystems including accelerated N losses and changes in N retention and availability. Though this has been observed in a variety of forests, ecosystem responses can be highly variable due to confounding environmental factors. To test the extent to which rates of N deposition drive current soil N availability, we established seven 200-m sample plots in red spruce (Picea rubens Sarg.) forests along a modeled gradient of atmospheric N deposition. We used a variety of field and laboratory techniques to estimate soil N availability at each site along the spatial gradient. During June 2011 we measured C, N, and d15N content of soil and red spruce leaf samples. We also deployed PRSTM-probes at each sample plot to measure soil nutrient supply rates. In July 2012 we resampled soils to conduct a short-term lab incubation to measure specific rates of N transformations. To determine the influence of other environmental factors on N availability, we also measured relative abundance of hardwoods and soil pH at each stand. Additional variables including climate and stand history were also obtained.
We compared each index of N availability to several continuous predictor variables (e.g. atmospheric N deposition, community composition, soil pH, annual precipitation and temperature, and stand age) in a series of regression models. Each regression was compared using Akaike's Information Criterion (AIC) to determine the likelihood of that particular model. In a comparison of all indices of N availability, we determined that mineral soil C:N was most strongly correlated to specific rates of soil N transformations (nitrification, R2 = 0.67, p = 0.02; mineralization, R2 = 0.83, p = 0.004); therefore, soil C:N was used as a metric for N availability for all subsequent analyses. Our results show that soil C:N is most strongly related to relative abundance of hardwoods (R2 = 0.92, p = 0.0006) and soil pH (R2 = 0.64, p = 0.03); whereas, all other relationships (including N deposition) were not significant. These findings do not necessarily invalidate our modeled values of historical deposition; rather they suggest that current rates of deposition may be too minimal to detect a significant response. Instead differences in species composition between sites appear to be the strongest drivers of soil N availability along this spatial gradient.