Radiation and temperature drive diurnal variation of aerobic methane emissions from Scots pine canopy

Significance Reports of methane emissions from plant shoots led to a debate over the role of plants in the global methane cycle. Existing studies report strongly diverging results, likely due to differences in methodology and measurement artifacts, yet, few studies have quantified shoot methane fluxes under field conditions. Here, we measured the shoot methane exchange of boreal Scots pine trees from leaf-level laboratory to ecosystem-level field conditions. We find consistent and small methane emissions across scales; however, these emissions were orders of magnitude smaller than previously used emissions factors. We further show that shoot emissions exhibit clear diurnal trends that are driven by light and temperature.


Supporting Information Text
Supporting Methods -Detection Limit.We calculated the following detection limits primarily as an expression of the measurement precision primarily for method inter-comparison and as a determine if data points should not be interpreted individually because each datapoint by itself may have analytical noise.Note that datapoints below the detection limit should not be removed before statistical analysis (1).We further estimate the detection limit for the average of repeated measurements, which we provide primarily as a parameter describing the analytical sensitivity of our method.We do not rely on the so-calculated detection limits to determine the presence or absence of methane exchange between shoot and atmosphere.Rather, we use statistical tests (comparison between chambers with and without shoots) to evidence methane fluxes (see below) and use error propagation to estimate confidence intervals that include uncertainties derived from the uncertainties in both measurements and blanks.
Instrument Detection Limit.We calculated the instrument detection limit, that is, the minumum detectable flux based on analyzer precision alone, according to (2,3), as stated in eq. 1 where M DF stands for the minimum detectable flux in nmol g -1 dw h -1 , tc for the closure time in hours (7 minutes in forest and garden measurements, 20 minutes in greenhouse measurements), V for the chamber volume (5.2L in forest and garden measurements, 1.6 L in greenhouse measurements), m for the shoot foliar dry mass in g, p for the atmospheric pressure (assumed 101325 Pa), R for the ideal gas constant (831 L Pa K -1 mol -1 ), and T for the air temperature in the chamber (assumed 298 K).AA stands for the precision (3SD) of the analyser in ppb when integrated over time ps.We applied the analyser precision for specified by the manufacturer for either 5 minute (Picarro G2301, 1SD <0.22 ppb) or 100 sec integration (LGR UGGA, 1SD <0.25 ppb).The so-calculated analytical detection limits were 0.028 nmol g -1 dw h -1 in the forest and garden measurements and 0.013 nmol g -1 dw h -1 in the greenhouse measurements.In other words, the fluxes detected in our study were well above the minimal fluxes theoretically measurable with both analysers.
Method detection limit.The method detection limit includes measurement uncertainties resulting from analyser noise, but also uncertainties from other sources like chamber leakage (4).We calculated these method detection limits based on the best available blank measurement, that is measurements conducted with empty chambers placed in close proximity of the actual shoot chambers.The detection limits of these measurements were calculated as 3 standard deviations of the apparent flux observed in empty chambers.The minimum detectable fluxes based on the method detection limits were 0.387 nmol g -1 dw h -1 in the forest and garden measurements and 0.233 nmol g -1 dw h -1 in the greenhouse measurements.

Repeated measurements on individual shoots.
As measurement precision increases with number of measured replicates, the detection limit of decreases.In the case of normally distributed and independent measurement uncertainty ('white noise'), the detection limit decreases with the square root of the averaged measurements (5,6).These assumptions are typically proved with so-called Allen Variance plots, which are provided for manual and automated measurements in Fig. S8.These plots demonstrate that the measurement precision indeed improves as expected with increasing replicate numbers.We therefore follow calculate the detection limit for repeated measurements on one shoot according to (6).We aggregated, for example fluxes in three-hour time-of-day groups for statistical comparison.In the manual garden measurements, each of four shoots was measured on average 3.6 times in a three hour interval (combining the two measurement days, range 2-5).The theoretical detection limit for the average in a block was therefore 0.204 nmol g -1 dw h -1 .In the greenhouse experiment, an average 32.6 measurements per shoot and three-hour interval (once per day over 33 days, range 31-33) were available, and the theoretical detection limit for the average of a three-hour interval was 0.041 nmol g -1 dw h -1 .We refer to these detection limits because they were calculated based on statistical assumptions.We did not use these calculated detection limits to determine the presence or absence of methane fluxes between shoot and atmosphere, but rather made such inference based on statistical tests.

Repeated measurements on distinct shoots.
To account for repeated measurements on distinct shoots, we used mixed effects models which account for repeated measurements on individual shoots and random variation between distinct shoots (see results below).The average uncertainties (2 SE) of the estimated flux for a three-hour interval (fixed effects) were 0.102 nmol g -1 dw h -1 and 0.029 nmol g -1 dw h -1 in the manual and automatic measurements, respectively.These uncertainties can serve as an estimate of the minimal detectable average flux during a three-hour time-of-day interval after merging the data measured on distinct shoots.
We calculated the following detection limits primarily as an expression of the measurement precision primarily for method inter-comparison and as a determine if data points should not be interpreted individually because each datapoint by itself may have analytical noise.Note that datapoints below the detection limit should not be removed before statistical analysis (1).We further estimate the detection limit for the average of repeated measurements, which we provide primarily as a parameter describing the analytical sensitivity of our method.We do not rely on the so-calculated detection limits to determine the presence or absence of methane exchange between shoot and atmosphere.Rather, we use statistical tests (comparison between chambers with and without shoots) to evidence methane fluxes (see below) and use error propagation to estimate confidence intervals that include uncertainties derived from the uncertainties in both measurements and blanks.

Fig. S7 .Fig. S8 .
Fig. S7.Spectral irradiance received at a Scots pine shoot in the greenhouse experiment.The spectrum was measured under the cover of a shoot chamber in February 2020.

Table S1 . Literature review of shoot methane emissions measured in field and laboratory settings. Measurements on detached leaves were not included in the list.
a Positive value indicate methane emissions, negative values m,ethane uptake.Measurements below the detection limit (LOD) are stated as 0±95% confidence interval.b Mean ±95 % confidence interval.c Converted from original source based on specific leaf area (SLA) = 4.64 m 2 kg -1 (50).d Median.e Converted from original source based on SLA=5 m 2 kg -1 .e Medians of measurements on two shoots.g converted source based on SLA=4.38 m 2 kg -1 (36) h 95% confidence interval for two plants Lukas Kohl, Salla A. M. Tenhovirta , Markku Koskinen, Anuliina Putkinen, Henri M.P. Siljanen, Iikka Haikarainen, Tatu Polvinen, Marjo Patama, Luca Galeotti, Ivan Mammarella, Thomas Matthew Robson, Bartosz Adamczyk, Mari Pihlatie

Table S3 . Targeted metagenomics: Total and functional gene fasta reads. Total reads indicate the number of reads produced by sequencing, mcrA (methanogens) and nifH (nitrogen fixers, as positive control) reads indicate the number of reads assigned to these genese by HMMER- searches with inclusion threshold of Inc-E 0.0001, verified nifH reads indicate the number of reads that were confirmed by alignment with known nifH sequences in the reference gene database with using MAFFT (19).
Lukas Kohl, Salla A. M. Tenhovirta , Markku Koskinen, Anuliina Putkinen, Henri M.P. Siljanen, Iikka Haikarainen, Tatu Polvinen, Marjo Patama, Luca Galeotti, Ivan Mammarella, Thomas Matthew Robson, Bartosz Adamczyk, Mari Pihlatie