Root distributions in a laboratory box evaluated using two different techniques (gravimetric and image processing) and their impact on root water uptake simulated with HYDRUS

Abstract Knowledge of the distribution of plant roots in a soil profile (i.e. root density) is needed when simulating root water uptake from soil. Therefore, this study focused on evaluating barley and wheat root densities in a sand-vermiculite substrate. Barley and wheat were planted in a flat laboratory box under greenhouse conditions. The box was always divided into two parts, where a single plant row and rows cross section (respectively) was simulated. Roots were excavated at the end of the experiment and root densities were assessed using root zone image processing and by weighing. For this purpose, the entire area (width of 40 and height of 50 cm) of each scenario was divided into 80 segments (area of 5×5 cm). Root density in each segment was expressed as a root percentage of the entire root cluster. Vertical root distributions (i.e. root density with respect to depth) were also calculated as a sum of root densities in each 5 cm layer. Resulting vertical root densities, measured evaporation from the water table (used as the potential root water uptake), and the Feddes stress response function model were used for simulating substrate water regime and actual root water uptake for all scenarios using HYDRUS-1D. All scenarios were also simulated using HYDRUS-2D. One scenario (areal root density of barley sown in a single row, obtained using image analysis) is presented in this paper (because most scenarios showed root water uptakes similar to results of 1D scenarios). The application of two root detecting techniques resulted in noticeably different root density distributions. Differences were mainly attributed to the fact that fine roots of high density (located mostly at the deeper part of the box) had lower weights in comparison to the weight of few large roots (at the box top). Thus, at the deeper part, higher root density (with respect to the entire root zone) was obtained using the image analysis in comparison to that from the gravimetric analysis. Conversely, lower root density was obtained using the image analysis at the upper part in comparison to that from the gravimetric analysis. On the other hand, fine roots overlapped each other and therefore were not visible in the image, which resulted in lower root density values from image analysis. Root water uptakes simulated with HYDRUS-1D using diverse root densities obtained for each cereal declined differently from the potential root water uptake values depending on water scarcity at depths of higher root density. Usually, an earlier downtrend associated with gradual root water up-take decreases and vice versa. Similar root water uptakes were simulated for the presented scenario using the HYDRUS-1D and HYDRUS-2D models. The impact of the horizontal root density distribution on root water uptake was, in this case, less important than the impact of the vertical root distribution resulting from different techniques and sowing scenarios.