Plot Level Spatial Variability of Soil Organic Carbon, Nitrogen, and Their Stable Isotopic Compositions in Temperate Managed Forest Soils of Atlantic Canada

Abstract

Spatial variability of soil organic matter is known to influence ecosystem processes at various scales. The aims of this study were to assess this variability in two paired managed forests differing in age and use the information to propose future sampling plans for these and similar forest soils. The two paired sites (Hattie Lake and Melrose) consisted of adjacent young (10–13 years) and mid-age (45–57 years) black spruce (Picea mariana Mill. BSP)–dominated forest plots. Soil samples were collected from the organic (forest floor (FF)) and mineral soil layers (0- to 15- and 15- to 30-cm depths) at 49 points on a square grid using 62 × 62–m area established at each of the four plots. Soil samples were analyzed for organic C, total N concentrations, and their stable isotopes. We used conventional statistics and geostatistics to evaluate the spatial variability of these soil properties. We also evaluated if mean values varied in relation to stand age. The young plot at Hattie Lake site contained areas of poor drainage compared with the other plots. Only this plot exhibited a very high coefficient of variation (CV, 81–92%) of C and N concentrations and a wider range of δN in the mineral soil layers compared with those of the other plots. Also moderate spatial dependence of C and N concentrations and mass was associated with the poorly drained plot while the other plots showed weak to no spatial dependence for these soil properties. Variability of C and N mass was very high (CV, 82–89%) in the FF, whereas intermediate variability (CV, 19–53%) was shown for mineral soil layers that did not consistently relate to stand age. Mean total soil C mass within the top 30 cm including the FF was significantly higher for the poorly drained plot (7.7 ± 0.9; mean ± 99% confidence interval) than the other plots (pooled mean, 5.3 ± 0.6). More intensive sampling is required for the FF than the mineral soil to estimate mean soil C mass. Although generalizations about spatial autocorrelation of forest soil properties studied could not be made in relation to stand age at the scale of our study, site drainage status was an important factor that needs to be considered in future sampling designs of these and similar managed forests.