Characterization of soil fertility and soil biodiversity with dsDNA as a covariate in a regression estimator for mean microbial biomass C

Abstract

The analytical determination of microbial biomass carbon is time‐consuming, which limits its use as a reference biochemical property for characterizing soil fertility and soil biodiversity of soil mapping units (SMUs). This paper explores whether the efficiency of sampling strategies for estimating the means of microbial biomass C (MBC) of SMUs can be increased with dsDNA as an ancillary variable in a regression estimator, leading to a model‐assisted sampling strategy. The map unit means of dsDNA are unknown; therefore, to implement the regression estimator a two‐phase sampling strategy is required. The two‐phase sampling design was tested in three soil units of the Brtonigla area (Istria, Croatia) that are widely involved in wine production. In the first phase, 20 locations per SMU were selected at which the ancillary variable dsDNA was determined in field‐moist and air‐dried soil. In the second phase, a subsample of ten locations per map unit were selected at which MBC was determined. The estimated sampling variances of the regression estimator were compared with the sampling variances of the design‐based estimator with a cost‐equivalent number of sampling points. The model‐assisted strategy was more accurate in two of the soil units in which the correlation between MBC and dsDNA was strong (adjusted R2 larger than 0.55). Measurements of dsDNA on field‐moist samples gave more precise estimated means of MBC than those on air‐dried samples.HighlightsUse of dsDNA to estimate the means of microbial biomass C (MBC) within soil mapping units (SMUs). Comparison of model‐assisted and design‐based sampling strategies to estimate mean MBC of SMUs. Model‐assisted strategy was more efficient in two SMUs; field‐moist dsDNA was better than air‐dried dsDNA. dsDNA increased the efficiency of sampling strategies to estimate mean MBC of SMUs with homogeneous land use.