Measuring oxygen fluxes in a European beech forest - results from the OXYFLUX project

Ecosystem assimilation and respiration result in anti-correlated fluxes of oxygen (O₂) and carbon dioxide (CO₂). While the ecosystem O₂:CO₂ molar exchange ratio is usually assumed constant at ≈1.1 on longer timescales, variations for individual ecosystem compartments or shorter timescales have been reported in the past. We hypothesize that these exchange ratio variations can reveal information about underlying biotic and abiotic processes in plants or soil that cannot be inferred from traditional net ecosystem exchange measurements. To date, oxygen measurements have not been widely implemented in ecosystem research due to the technical challenge of detecting very small variations (ppm-level) against an atmospheric background of ≈21% (≈210,000 ppm).

We evaluate the performance and applicability of two commercial oxygen analyzers Integrated into custom-built gas handling and calibration systems, and report first results from measurements of O₂:CO₂ exchange ratios in a managed European beech forest in central Germany.

System 1, consisting of a relatively slow response differential fuel cell O₂ analyzer (Oxzilla FC-2, Sable Systems Inc., USA) together with a non-dispersive infrared CO₂ analyzer (LI-840, LI-COR Biosciences, USA), was used to simultaneously measure O₂ and CO₂ mole fractions in air sampled from soil, stem, and branch chambers. Chambers were operated in an open flow-through steady-state design aimed at equilibrium mole fractions within a few hundred ppm of atmospheric background. Using a multiplexer valve design, we measured chambers sequentially by directing chamber air at a controlled flow rate to the gas analyzing system.

Preliminary analysis of August to December 2018 data show that chamber-based flux estimates for O₂ and CO₂ were anti-correlated at all times, and that the O₂:CO₂ molar exchange ratios (defined as ‑Δ[O₂]/Δ[CO₂]) varied considerably over time and between the different ecosystem compartments (soil, stems, and branches) with a median (interquartile range) of 0.94 (0.75 to 1.09).

In system 2, CO₂, O₂ and water vapor (H₂O) measurements were performed with a fast response (5 Hz) gas analyzer using tunable infrared laser direct absorption spectroscopy (TILDAS, Aerodyne Research Inc., USA). We measured fluctuations in O₂:CO₂ exchange ratios in air sampled at 1.5 times the canopy height, i.e. a typical eddy covariance set-up.

Analysis of the high-frequency data revealed instrumental noise levels of ≈±12 ppm O₂. Fourier transformation of high-frequency data obtained during well-mixed boundary layer conditions indicate that turbulent fluctuations of the O₂ signal were insufficiently resolved when compared to the CO₂ power spectra. When averaging high-frequency data to 2-min aggregates, instrumental noise was reduced to ≈±1 ppm, similar to the precision of system 1. At this timescale, contemporaneous measurements of above-canopy air revealed agreement between the fuel cell and the laser systems, both in O₂ mole fraction (R² = 0.6 slope = 0.7, MAE = 1.6 ppm) and in estimated O₂:CO₂ exchange ratios of 1.01 and 0.97 for system 1 and 2, respectively.

Our presentation will expand on the applicability of both O₂ and CO₂ measurement systems with regard to micrometeorological flux techniques. Specifically, we elucidate on the potential of using O₂ flux measurements as a constraint for estimating ecosystem-scale gross primary production.