Landscape Scale Drivers of Spatial Soil Carbon Variability on Three Northern California Vineyards

Vineyards make up over one million acres of US agricultural land and present a unique opportunity for enhanced soil carbon (C) sequestration through regenerative agriculture practices on typically lower quality soils. Vineyard landscapes are characterized by specific landscape features that promote grape growth, such as well draining soils and sloped terrain, differing from other agricultural products. To accurately quantify the effects of regenerative practices on vineyard soil C, it is necessary to understand how vineyard landscape features impact soil C concentrations. Stratification of samples by key landscape features allows a sample design to capture maximum C variation with minimal samples. Identifying which landscape features account for significant C variation allows for sampling designs that increase accuracy while minimizing sampling effort.

In this research, we measured the effect of seven unique landscape scale variables on soil C distribution on three vineyards in Northern California. Soil samples were collected at 0-15 cm and 15-30 cm depths for a total of 2,808 samples at a density of 4-5 samples an acre. The landscape variable values were matched to sample locations using the exactextractr package in R. Models for predicting soil C (at 0-15 cm, 15-30 cm, and 0-30 cm depths) for each vineyard using every possible combination of the landscape variables were ranked by their improvement to the model using the MuMIn package in R. Slope and NDVI were found to have the greatest impacts on soil C across vineyards and depths, while aspect and hillslope curvature, landscape features expected to affect C distributions, did not have as significant an impact. However, most landscape features have varying impacts depending on location and depth of samples, suggesting that these features do not predict soil C across vineyards, but rather are site specific. Information gained from more precise sampling can be used to determine the effects of regenerative viticulture practices at lower cost to land managers. This methodology can also be readily applied to other ecosystems to improve sampling design on future studies outside of viticulture.