Vegetation controls spatial patterns of soil water isotopes in a tropical dry forest and UAV’s can help to predict them


Analyzing water stable isotopes in soils and plants is a key method to identify the water sources for transpiration. However, the spatial representation of such studies is often limited and typically data from one or only a few soil water isotope profiles are used for analyzing plant water sources for much larger areas. Contrary, it is well known from soil sciences that soil physical and hydraulic properties are highly heterogeneous, even over small areas. Only few studies have investigated the spatial variability of soil water isotopes, despite its potential importance for water uptake depth analysis. Goldsmith et al., (2018) showed that vegetation can have a substantial influence on the spatial pattern of soil water isotopes in a tropical cloud forest. We extend the hypothesis that vegetation does not only have an influence on soil water isotopes in wet environments, but also under dry conditions: The isotopic enrichment of soil water isotopes under steady-state dry conditions is controlled by vegetation (canopy parameters). In order to test this hypothesis, we undertook a spatial sampling of ten soil water isotope depth profiles (at 6 depths up to 2m depth) and ~60 evergreen and deciduous trees at the peak of dry season in February 2019 in a tropical dry forest in the northwest of Costa Rica. We then correlated the spatial patterns of water content and isotopes of the soil with 12 vegetation indices and surface (leaf/soil) temperature derived from UAV (Unmanned aerial vehicles; drones) overflights (Jan-Apr 2019) in order to investigate if spatial patterns of soil water isotopes can be predicted using additional information. Finally, we interpolated (external drift kriging) the soil water isotope values using the highest correlated vegetation indices in order to provide a spatially distributed map of soil water isotope depth profiles. Our findings indicate that i.) soil water isotopes are (highly) spatially heterogeneous, even under steady-state conditions (no rain); ii.) this heterogeneity is particularly pronounced for the near-surface soil (first 50 cm) and diminishes with soil depth; iii.) there is a significant correlation between soil water isotopes and multiple vegetation indices. Surprisingly, the highest correlations (0.82 for water content, 0.75 for 𝛿2H and 0.62 for 𝛿18O, all for 10 cm soil depth) were found for indices based on color infrared (CIR) and the red-edge triangular vegetation index (RTVI), and not NDVI (Normalized Difference Vegetation Index). We proved the theoretical concept (more vegetation cover = lower soil temperatures = less fractionation) to hold true by correlating the soil surface temperatures at each sampling location to the water isotope values (R² = 0.75 for both 𝛿2H and 𝛿18O at the soil surface). This research demonstrates that classic approaches of assigning one or few soil water isotope profiles for characterization of water uptake depths of larger areas are highly error-prone. Vegetation and soil water isotopes affect each other and need to be incorporated into spatial analyses. The interpolated soil water isotope depth profiles we provide can act as a baseline for more robust spatial investigations of soil water uptake depths in highly heterogeneous environments.