Towards accurate monitoring of water content in woody tissue across tropical forests and other biomes

Abstract

Forest ecosystems face increasing drought exposure due to climate change, necessitating accurate measurements of vegetation water content to assess drought stress and tree mortality risks. Although Frequency Domain Reflectometry offers a viable method for monitoring stem water content by measuring dielectric permittivity, challenges arise from uncertainties in sensor calibration linked to wood properties and species variability, impeding its wider usage. We sampled tropical forest trees and palms in eastern Amazônia to evaluate how sensor output differences are controlled by wood density, temperature and taxonomic identity. Three individuals per species were felled and cut into segments within a diverse dataset comprising five dicotyledonous tree and three monocotyledonous palm species on a wide range of wood densities. Water content was estimated gravimetrically for each segment using a temporally explicit wet-up/dry-down approach and the relationship with the dielectric permittivity was examined. Woody tissue density had no significant impact on the calibration, but species identity and temperature significantly affected sensor readings. The temperature artefact was quantitatively important at large temperature differences, which may have led to significant bias of daily and seasonal water content dynamics in previous studies. We established the first tropical tree and palm calibration equation which performed well for estimating water content. Notably, we demonstrated that the sensitivity remained consistent across species, enabling the creation of a simplified one-slope calibration for accurate, species-independent measurements of relative water content. Our one-slope calibration serves as a general, species-independent standard calibration for assessing relative water content in woody tissue, offering a valuable tool for quantifying drought responses and stress in trees and forest ecosystems.

Keywords: dielectric permittivity; drought response; forest ecosystems; frequency domain reflectometry; hydraulic capacitance; palms; vegetation water content.

Figure 1

Factors affecting the calibration curve for frequency-domain reflectometry sensors. In contrast to previous findings, palms are grouped very closely to dicotyledonous tree species. Furthermore, while (A) wood density had no significant effect on the calibration curve, (B) random species effects accounted for more variation than WD. To improve readability, the graph has been simplified by aggregating data at the species level, omitting individual trees.

Figure 2

As species-specific differences were driven by intercept differences in the TTC, first-order differencing was applied to model the relative changes between StWC and . This led to the development of an OSC, which disregards species-specific intercept changes and focuses on the relative changes.

Figure 3

Linear model result of the temperature effect experiment conducted with uncalibrated Teros12 sensors in a water bucket. With increasing temperature, the dielectric permittivity decreases, affecting the FDR sensor reading.

Figure 4

Field data from Manilkara bidentata in Caxiuanã showing the unadjusted ΔStWC and the temperature-adjusted ΔStWC. The ΔT relative to the mean (25 °C) is given on the secondary y-axis. While larger daily fluctuations of T in the dry season (upper panel) are mainly driving artefactual diurnal WC dynamics, a true diurnal WC dynamic exists in the wet season starting end of December (middle panel), even after accounting for temperature effects. Large changes in StWC coincide well with precipitation events at the start of the wet season (lower panel). Due to low ΔT in the tropics, seasonal data are relatively unaffected by T (middle panel). However, care must be taken when analysing diurnal patterns (upper panel), especially in the dry season. Disregarding the temperature effect leads to false physiological and ecological conclusions. Care must be taken when interpreting figure parts (upper panel) and (middle panel), as the axes are differently scaled.

Figure 5

A single calibration (TTC) was established for the relationship between and StWC for tropical trees, including arborescent palms from this study. Interestingly, the slope (s = 0.223) established here was indistinguishable from the temperate tree model by He et al. (2021) and only the intercept was found to be different.