Variations of the Oxidative Ratio across Ecosystem Components and Seasons in a Managed Temperate Beech Forest (Leinefelde, Germany)

Abstract

The oxidative ratio (OR) of organic material integrates the ratio of CO₂ sequestered in biomass vs. O₂ produced over longer timescales, but the temporal and spatial variability within a single ecosystem has received very limited attention. Between October 2017 and October 2019, we repeatedly sampled leaves, twigs, bark, outer stem wood, understorey vegetation and litter in a temperate beech forest close to Leinefelde (Germany) for OR measurements across a seasonal and spatial gradient. Plant component OR ranged from 1.004 ± 0.010 for fine roots to 1.089 ± 0.002 for leaves. Inter- and intra-annual differences for leaf and twig OR exist, but we found no correlation with sampling height within the canopy. Leaf OR had the highest temporal variability (minimum 1.069 ± 0.007, maximum 1.098 ± 0.002). This was expected, since leaf biomass of deciduous trees only represents the signal of the current growing season, while twig, stem and litter layer OR integrate multiple years. The sampling years 2018 and 2019 were unusually hot and dry, with low water availability in the summer, which could especially affect the August leaf OR. Total above-ground OR is dominated by the extremely stable stem OR and shows little variation (1.070 ± 0.02) throughout the two sampling years, even when facing extreme events.

1. Introduction

Land ecosystems play an important role in the global carbon cycle. Due to constant exchange with the biosphere via processes like respiration and assimilation, atmospheric carbon dioxide (CO₂) and oxygen (O₂) concentrations are subject to continuous temporal and spatial variation [1,2,3]. One key method to estimate the land carbon sink uses relative changes of O₂ and CO₂ concentrations in the atmosphere [3]. One of the parameters in this approach is the O₂:CO₂ ratio of the land biosphere exchange with the atmosphere [3]. In 2013, Worrall et al. [4] stated that calculating a precise terrestrial O₂:CO₂ ratio is a key factor to make more accurate estimates of the global land carbon sink.

O₂:CO₂ gas exchange ratios have been measured at some sites for branches, stems, and soils and show temporal variation and variation among individual component fluxes e.g., [5,6,7]. However, direct O₂:CO₂ gas exchange measurements are challenging on long time scales, due to high technical requirements for high precision O₂ measurements [5]. Thus, measurement of oxidative ratios (OR) of biomass can serve as a proxy of the long-term signal of the land biosphere O₂:CO₂ gas exchange. The oxidative ratio (OR) is the ratio of O₂ produced compared to the CO₂ sequestered in the terrestrial biosphere via photosynthesis, or equivalently, the ratio of O₂ consumed to CO₂ produced during respiration and decomposition [4,8,9]. OR and the oxidation state of organic carbon (C_ox) are fundamental properties of the terrestrial carbon cycle. They reflect the biogeochemistry at organism to ecosystem scale, like tissue composition and biochemical synthesis pathways and the environmental conditions at the time of the growth of the specific components.

Total ecosystem OR integrates the individual ORs of all ecosystem components, and, on a molecular level, of all biochemical components. The variability of ORs of different ecosystem components is large, typically between 0.25 to 1.50 [9]. For most plant biomass components like leaves and wood, however, variation is much smaller and values typically range from 1 to 1.1 resulting in an estimate of global vegetation OR of 1.03–1.06 ([1,2,3,4,5,6,7,8,9,10,11,12], Table 1). In terms of mass, plant biomass OR is dominated by lignin and cellulose, but other components like sugars, starch, soluble fats, waxes etc. also contribute to the overall OR, and can have very different individual ORs [9,13,14]. Variation of biochemical composition does occur within plants and between plant species, but can also occur within a plant over time. Marenco et al. [15] showed that the amount of the carbohydrates starch and sugar differs between sun and shade leaves. Starch and sugar accumulate during the growing season and during the whole life of a tree. As a consequence, old trees have a higher absolute amount of carbohydrates, but lower relative amount in regard to the whole tree mass than young trees [16]. Such variation in biochemistry can result in changes of OR. If a plant, e.g., produces less lignin (OR = 1.13) but more carbohydrates or cellulose (OR = 1.00), this would result in an overall reduction of biomass OR [9,17].

Table 1. Oxidative ratios (ORs) of common plant components, woody leaf, deciduous forest litter, woody stem and woody root and calculated OR for global terrestrial vegetation. * OR for carbohydrates is calculated from the empirical formula C_m(H₂O)_n.

Component	OR	Reference
Lignin	1.13	[16]
Cellulose	1.00	[16]
Carbohydrates	1.00	*
Soluble phenolics	1.05	[17,18]
Woody leaf	1.054	[17,18]
Deciduous forest litter	1.045	[19]
Woody stem	1.041	[17,18]
Woody root	1.051	[17,18]
Global vegetation	1.03 ± 0.02	[4]
	1.07 ± 0.02	[11]
	1.040 ± 0.005	[12]
	1.06	[10]

Table 1. Oxidative ratios (ORs) of common plant components, woody leaf, deciduous forest litter, woody stem and woody root and calculated OR for global terrestrial vegetation. * OR for carbohydrates is calculated from the empirical formula C_m(H₂O)_n.

Component	OR	Reference
Lignin	1.13	[16]
Cellulose	1.00	[16]
Carbohydrates	1.00	*
Soluble phenolics	1.05	[17,18]
Woody leaf	1.054	[17,18]
Deciduous forest litter	1.045	[19]
Woody stem	1.041	[17,18]
Woody root	1.051	[17,18]
Global vegetation	1.03 ± 0.02	[4]
	1.07 ± 0.02	[11]
	1.040 ± 0.005	[12]
	1.06	[10]

Studies about oxidative ratios of whole ecosystem biomass only started only approximately 30 years ago [3,20]. While ORs are controlled by small molecular differences in the organic material, most research on natural ecosystem ORs so far has been on bigger scales, like a geographic transect [10] or the whole global terrestrial biosphere, or only parts of an ecosystem (e.g., only leaf litter and tree bole in [19]). For the global terrestrial OR, Severinghaus [20] first estimated 1.10 ± 0.05 in 1995. More recent values are lower, at 1.04–1.06 for the complete terrestrial biosphere (including soil) and 1.03–1.07 for terrestrial vegetation [4,10,11,12].

Clay and Worrall [11] and Clay et al. [10] are two of the very few publications that include all main organic components of ecosystems (wood, leaves, soil) with the aim to improve the global estimate and understanding the drivers of OR differences. Gallagher et al. [19,21] found that the highest differences in OR are seen when comparing different species, by comparing leaf litter and bole wood from different forests and different crops. Clay et al. [10] found an increase in ecosystem OR of 0.06 from southern Sweden to their northernmost sampling point in northern Norway. They explain the observation by the North–South climate gradient and land-use change from natural ecosystems to agriculture-dominated landscapes, building on the hypothesis by Randerson [17] that the terrestrial biosphere is becoming more oxidized because of accelerated land-use change. Out of all of those studies that deal with natural ecosystems, only one [19] extends over several years. However, this did not look at inter-annual changes and thus the inter- and intra-annual variations and their rates of change of a whole ecosystem are not well known. Studies using another tracer, i.e., δ¹³C, indicate that the biochemical composition of biomass components varies on temporal and spatial scales within a forest. These variations are caused by changes in water balance (e.g., water stress) or gradients in the canopy due to differences in the microclimate [22,23]. This leads to the question as to whether such variability also influences OR within a forest.

In this study, our objectives were to determine (a) the variation of OR of individual components of the biomass of a case-study ecosystem, i.e., a beech forest in Germany, as well as (b) the seasonal (spring, summer, autumn) variation of plant component ORs over two years. In particular, we investigate whether there is (c) an effect of height on the OR of canopy components (leaves and twigs) within a tree.

2. Materials and Methods

2.1. Study Site

This research was conducted at the Leinefelde tower site (DE-Lnf, https://doi.org/10.18140/FLX/1440150, accessed on 1 December 2021)), which is maintained by the University of Göttingen. The forest is an even-aged managed pure beech (Fagus sylvatica L.) forest with few tree individuals of other species like oak and maple, and is located in the Geney forest district, near Leinefelde, Germany (51°20′ N, 10°22′ E, 450 m a.s.l.). The sparse understorey vegetation includes Galium odoratum (L.), Melica nutans (L.), Milium effusum (L.), Oxalis acetosella (L.), Rubus sp., Urtica dioica (L.) and Stellaria holostea (L.). The forest has been managed as a shelterwood system since 1838. Regular thinning has occurred on a 10–20 year cycle, with the last three thinnings in 1982, 1999 and 2011. The stand where sampling took place has an approximate age of 130 years with a canopy height of about 35 m [24,25]. The maximum effective leaf area index of the stand surrounding the flux tower is approximately 5 m² m⁻² with a generally top-weighted leaf canopy structure. The total organic carbon pool of the Leinefelde forest is approximately 240 tC ha⁻¹, for the living tree biomass 160 tC ha⁻¹, and for soil 75 tC ha⁻¹. The soil is shallow, hitting the bedrock at approximately 65–100 cm [25]. At the specific stand, wood has the largest biomass share, with 68.90% of the total plant biomass being in the stems, 14.68% in the branches and twigs, 0.60% in the leaves, 14.13% in the coarse roots, 0.52% in the fine roots and 1.16% in the organic layer (litter) [24,25].

Mean annual temperature is approximately 8 °C and mean annual precipitation is approximately 1000 mm. The site is a long-term eddy covariance flux site of the global FLUXNET network where measurements started in April 2002 [24].

2.2. Methods

Sampling of leaves, twigs, and bark from beeches, roots, understorey vegetation and litter (from the ground) took place in October 2017, May, August and October 2018 and May, August and October 2019. In October 2017, no roots were collected and five trees were sampled. On all other sampling dates, three trees were sampled. For each tree, we sampled a small branch with at least five leaves at three different heights of the tree canopy (24, 28 and 32 m). The small parts cut off from those branches are further referred to as twigs. When there were seeds on these branches, those were collected too. The understorey vegetation is a representative mixture of the vegetation within the tower’s footprint (see Section 2.1 Study Site). Root collection started in spring 2018, and roots were only taken from the uppermost approximately 10 cm of the soil. Stem wood was sampled in 2019 (see Table S1 in Supplementary Data) to compare with the wood from the twigs. However, it was only cut from the outermost part of the stem, as no stem cores could be taken from the same trees the twigs and leaves were collected from.

Sampling always took place within one or two days. In addition to the regular samples, in October 2017, leaves from three young beeches (ca. 2 m total height), in May 2018, dead wood, and in October 2019, single leaves from branch chambers which had been operating in 2019, were collected.

In summer 2020, stem cores were collected from different beech trees of similar size and had a minimum length of 18 cm and a maximum length of approximately 25–29 cm. They were cut into 1 cm sections, and the five outermost and two innermost sections were analyzed. Furthermore, leaves from the same three trees and heights as in 2019 were collected to be used as a comparison for 2018–19 summer leaves which were already analyzed at that point in time, to investigate a possible affect of the 2018–19 droughts.

Sample treatment included cleaning, drying and grinding. Roots were washed with water and all samples were dried at 70 °C for a minimum of 48 h, milled to a homogeneous fine powder using a disc mill (vibrating disc mill TS, Siebtechnik GmbH, Mühlheim an der Ruhr, Germany) and stored in flat-bottom soda glasses with snap-on lid (Laborversand A. Hafenstein, Würzburg, Germany). The five leaves from each of the branches were always milled together and further treated as one sample. Milled samples were dried again immediately before weighing aliquots into tin capsules for CN analysis and silver capsules for OH analysis.

Elemental and isotope (δ¹³C) analysis took place at the Center for Stable Isotope Research and Analysis, University Göttingen, using elemental analyzers at 1400 °C, for C and N by a NA1100 (CE-Instruments, Rodano, Milano, Italy) and for H and O by a TC/EA (Delta V plus, Thermo Fisher Scientific, Schwerte, Germany) via reduction on glassy fiber. Peach leaves (NIST SRM 1547 from Thermo Fisher Scientific, Schwerte, Germany) were used as a calibration standard, and an internal plant standard (“KOSI-B1-1”) as a control standard (

Table 2. Elemental concentrations of C, N, O and H of the three standards used in the elemental analysis (mean ± sd), with replicate number n in brackets.

Standard	C (m%)	N (m%)	O (m%)	H (m%)
Acetanilide	71.56 ± 1.72 (48)	10.47 ± 0.51 (47)	12.85 ± 0.47 (24)	6.34 ± 0.17 (25)
Peach leaves	47.04 ± 0.20 (42)	2.96 ± 0.03 (49)	37.10 ± 0.37 (49)	6.60 ± 0.13 (52)
KOSI-B1-1	41.69 ± 0.91 (25)	2.31 ± 0.07 (25)	41.03 ± 0.90 (25)	5.50 ± 0.19 (25)

). Initially, we used acetanilide instead of the peach leaves (originally as control standard), but switched later due to less reliable O and H measurements. For δ¹³C, Vienna PD Belemnite served as a standard.

In a first step, the oxidation state of organic carbon (C_ox) [13] was calculated as follows:

$$C_{ox}=\frac{2\left[O\right]-\left[H\right]+3\left[N\right]}{\left[C\right]}$$

(1)

where [O], [H], [N] and [C] are the molar concentrations of C, H, N, or O. This equation assumes that the contribution of S or P to C_ox is small. It has been shown that this assumption is valid in most cases and only causes an error of ± 0.0002 in OR [8,9]. In a second step, the OR was calculated as:

$$\mathrm{OR}=1-\frac{C_{ox}}{4}+\frac{3\left[N\right]}{4\left[C\right]}$$

(2)

The ecosystem, above-ground and below-ground ORs were calculated by weighting the ecosystem components based on their biomass share (see Section 2.1 Study Site).

Meteorological measurements at the site include air temperature (at 44 m above the ground; thermohygrometer HMP45A, Vaisala, Finland), precipitation (rain gauge, Thies Clima, Göttingen, Germany) and volumetric soil water content (at 0.08, 0.16, 0.32 and 0.64 m below the surface; ML-2x, Delta-T Devices Ltd., Burwell, UK). The calculation of relative extractable water (REW) follows Granier [26], using available data from the soil moisture detectors at the four different depths. Long-term meteorological data from the tower site are available on the Europe Flux website (http://www.europe-fluxdata.eu/, accessed on 1 December 2021). The specific part of the data which was used for this study is shown in the Supplementary Data (Table S3).

Statistical analyses were carried out in R version 4.0.0 and R Studio. An analysis of variance (ANOVA) was performed and the differences between groups of data were calculated by Tukey’s HSD test.

3. Results

Average weekly air temperature ranged from −4.3 to 19.9 °C in 2017, −8.6 to 24.7 °C in 2018 and −2.3 to 25.2 °C in 2019, with a mean annual temperature of 8.7, 10.2, and 10.6 °C in 2017, 2018, and 2019, respectively (Figure 1a). Annual precipitation was 690, 388, and 572 mm in 2017, 2018, and 2019, respectively, (Figure 1b) resulting in very low soil moisture conditions in 2018 and 2019 (Figure 1c). Overall, 2018 and 2019 were exceptionally dry and warm years compared to the long-term data record at this site (data not shown).

According to Granier et al. [26,27], the critical REW for different tree species is 0.3–0.4. At this point the transpiration of trees begins to decrease. In Leinefelde the REW was below 0.3 in summer to mid-autumn in both 2018 and 2019, in 2018 for 25 weeks and in 2019 for 17 weeks, while the minimum was 0.49 in 2017.

The oxidative ratio (OR) differed between components, especially between leaves and litter, leaves and stem wood, and between fine roots and all other components (Figure 2,

Table 3. C/N, C/O, C_ox, and OR of the ecosystem components independent of time (mean with standard error SE). Fresh leaf litter are leaves collected from the ground in October, still green in color, while litter is the deposited organic material (leaves, twigs, seeds) on the forest floor. Fine roots and coarse roots are not specifically from the beeches, but a mix of unidentified species. ¹ The biomass data are for branches and twigs [25], while the collected samples are twigs. The biomass share of only twigs is not known. Stem wood includes 6 samples from May and October 2019, as well as from the 2020 stem core sampling.

Sample Type	N Samples	Plant Biomass Share [%]	C/N (±SE)		C/O (±SE)		Cox (±SE)		OR (±SE)
Leaves	78	0.6	22.81	±0.52	1.22	±0.01	−0.228	±0.007	1.086	±0.002
Twigs	64	14.7 1	44.68	±1.69	1.21	±0.01	−0.258	±0.009	1.081	±0.002
Stem wood	36	68.9	119.24	±6.79	1.08	±0.01	−0.256	±0.007	1.069	±0.001
Bark	26	N/A	57.62	±2.85	1.24	±0.03	−0.264	±0.028	1.078	±0.007
Fresh leaf litter	8	N/A	42.67	±1.55	1.21	±0.02	−0.210	±0.020	1.068	±0.005
Litter (organic layer)	23	1.2	29.11	±1.10	1.18	±0.01	−0.072	±0.020	1.041	±0.004
Understorey vegetation	22		19.55	±0.65	1.08	±0.01	−0.140	±0.017	1.069	±0.004
Fine roots	12	0.5	34.04	±2.20	1.02	±0.04	0.063	±0.041	1.004	±0.010
Coarse roots	11	14.1	44.15	±5.17	1.07	±0.03	−0.138	±0.017	1.051	±0.003

). Leaves had the highest OR (1.086 ± 0.002, mean ± SE), with twigs (1.081 ± 0.002) and coarse roots (1.051 ± 0.003) being between leaves and stem wood (1.069 ± 0.001). Understorey vegetation (ground vegetation) and fresh leaf litter had similar ORs to stem wood, while for older, mixed litter (leaves, twigs, seed shells) the OR was lower, only higher than fine root OR (1.004 ± 0.010). However, both fine and coarse root and bark ORs may be affected by the variability and inhomogeneity of the samples more than the other components. For the roots, the sample sizes were smaller and proportions of dead and alive fine roots, as well as the species they belong to being unknown. Bark chemistry can vary depending on how far exactly we cut into the tree. Since it always was just 3–5 mm, the proportions of the outermost bark in relation to the connected outer wood varies.

Tukey’s HSD test showed highly significant differences (p = 0–0.0049) between the following components: leaves-litter, leaves-coarse roots, litter-twigs, litter-bark and bark-coarse roots. Fine roots were significantly different from all other components (p = 0). Furthermore, leaves-stem wood, litter-stem wood, litter-understorey vegetation as well as twigs-coarse roots were less significant (p = 0.005–0.01). On the other hand some components did not differ (p = 0.9–1), e.g., stem wood-understorey vegetation, stem wood-fresh leaf litter, twigs-bark, twigs-fresh leaf litter, leaves-bark, leaves-twigs, litter-bark, litter-coarse roots and bark-understorey vegetation (See

Table 4. Significance level matrix of components. *** = high significance (p ≤ 0.005), ** = medium significance (0.005 < p ≤ 0.01), * = low significance (0.01 < p ≤ 0.05), - = no significance (p > 0.05).

	Leaves	Stem Wood	Twigs	Litter	Fine Roots	Coarse Roots	Bark	Understorey Veg.
Stem wood	**
Twigs	-	*
Litter	***	**	***
Fine roots	***	***	***	***
Coarse roots	***	-	**	-	***
Bark	-	-	-	***	***	*
Understorey veg.	*	-	-	**	***	-	-
Fresh leaf litter	-	-	-	-	***	-	-	-

).

The carbon isotopic composition (δ¹³C) of leaves and twigs differed across heights within the canopy, decreasing from the top layer (−27.72 ± 0.36 ‰) to the middle (−28.91 ± 0.41‰) and lower layer (−29.91 ± 0.34‰) for leaves and −26.86 ± 0.18‰, −28.35 ± 0.26‰, −29.58 ± 0.28‰ for twigs respectively (Figure 3). There is seasonal variability, which is clearer at the middle and lower layer, with autumn typically having the most negative δ¹³C. However, summer and autumn 2018 behave differently from the other two years: the δ¹³C values did not decrease as much as in autumn 2017 and 2019, and in summer 2018 they were close to or even higher than in spring 2018.

Figure 3. Normalized carbon isotopic composition (δ¹³C) and oxidative ratios of leaves and twigs for three heights (24, 28 and 32 m) and 7 sampling times. For each panel the mean is taken from the 6 times from May 2018 to October 2019, and then every value is divided by the mean to show the deviation from the mean.In contrast to δ¹³C, we found no effect of sampling height on ORs in twigs and leaves (Figure 3 and Figure 4) This means, while the isotopic composition changes vertically with canopy position reflecting vertical changes in microclimate, the stoichiometry of the elements (C, N, O, H) is not affected by the location when looking at the same component.

Figure 3. Normalized carbon isotopic composition (δ¹³C) and oxidative ratios of leaves and twigs for three heights (24, 28 and 32 m) and 7 sampling times. For each panel the mean is taken from the 6 times from May 2018 to October 2019, and then every value is divided by the mean to show the deviation from the mean.In contrast to δ¹³C, we found no effect of sampling height on ORs in twigs and leaves (Figure 3 and Figure 4) This means, while the isotopic composition changes vertically with canopy position reflecting vertical changes in microclimate, the stoichiometry of the elements (C, N, O, H) is not affected by the location when looking at the same component.

Twigs differed in the OR between May 2018 and August 2019 (p = 0.001), and May 2018 and May 2019 (p = 0.024, where twig ORs in May 2018 were the lowest and highest in August 2019; second highest values were found in May 2019. Leaves from the canopy showed significant differences between many sampling dates, with the biggest difference between spring and autumn (Figure 5). The only other two components with significant differences were understorey vegetation and litter.

Stem cores, which represent the highest biomass share, showed no significant difference between the outer and inner segments. The highest difference occurred between the outermost (1.074 ± 0.00) and innermost (1.058 ± 0.01) part however (p = 0.183), while the outermost two sections (1-cm parts) are the same (p = 0.970), indicating a trend in wood OR towards the inner wood (see Supplementary Data, Table S2).

Leaf OR from September 2020 was higher than for August 2018 and 2019 and significantly different from August 2018 (August 2018 vs. August 2018: p = 0.024; August 2018 vs. September 2020: p < 0.001; August 2019 vs. September 2020: p = 0.247). The weather of 2020 was closer to the long-term mean than the previous two years. Therefore it can be used for comparison, when discussing how the drought affected oxidative ratios of the beech forest, but as a single point it is unclear how well it represents only the summer of 2020 or whether it is unaffected by the previous two years.

4. Discussion

4.1. Oxidative Ratio (OR) at Ecosystem Level

The above-ground ecosystem OR was stable throughout the whole sampling period, ranging from 1.069 to 1.070 across all 7 sampling times with a SE of 0.01–0.02, and was dominated by the OR of the woody components. Therefore, it is no surprise that average above-ground ecosystem OR is very close to average stem OR. Since any newly formed biomass in a certain growing season (leaves, new wood, etc.) is relatively small compared to the already existing stem biomass pool, major inter-annual changes of the ecosystem OR are highly unlikely and presumably would require significant changes in the ecosystem structure. A younger forest with a higher proportional contribution of leaf or twig biomass in relation to stem wood or a forest with different species will lead to a different ecosystem OR, but the OR of a forest where most of the biomass is represented by old beeches remains the same. The low variability in the stem core wood agrees with Gallagher et al. [19], where both wood from coniferous and deciduous trees showed differences over time smaller than measurement uncertainty. The deciduous forest in their study includes oak, hickory, and maple, with tree boles sampled from wild black cherry (Prunus serotina Ehrh.), white oak (Quercus alba L.), and white ash (Fraxinus americana L.), with a mean OR of 1.035 ± 0.007. The coniferous forest wood has an OR of 1.032 ± 0.003. Our stem wood OR is higher than both those (1.069 ± 0.001), however, European beech (Fagus sylvatica L.) was not included in Gallagher et al. [19] making direct comparisons difficult.

Gallagher et al. studied OR of different crops [21], and tree bole and leaf litter OR of a temperate deciduous and a temperate coniferous forest from 1995 to 2006 [19] and state that time and environmental factors like climate do not or only very slightly affect ecosystem OR, but the different plant species do. Here our question is how well this hypothesis works when looking at all parts of the above-ground biomass, including living leaves and twigs. Gallagher et al. [19,21] also note that nutrient availability, which is influenced by soil type or fertilization can influence ORs of plants. However, no fertilization took place at our site and we believe that nutrient availability did not change significantly during the sampling campaign, and since we do not have ORs from different beech forests to compare, these factors cannot be investigated.

4.2. OR and δ¹³C at Component Level

We found general differences in the OR of different components (leaves, twigs, bark, stem wood, roots and litter) of the beeches. Interestingly, we found a trend of decreasing OR the further away we sampled from the leaves of the canopy (leaves-twigs-stem wood-litter-fine roots). The OR of fresh leaf litter, collected in autumn (1.068 ± 0.005), was between that of fresh leaves from the canopy (1.089 ± 0.002), and that of old (leaf) litter from the forest floor (1.040 ± 0.006).

O₂:CO₂ exchange ratios in the air depend on seasons, time of the day and distance from ground and canopy [22,23,28]. However, our research has confirmed the hypothesis that the oxidative ratio of the plant biomass is a more long-term signal for most components. Gallagher et al. [19] found no differences in wood OR in 10 years and our stem cores showed no OR differences in wood that represents at least 15 years either.

As carbohydrates like cellulose have an OR of 1.00, the higher OR is influenced by the amount of compounds which are not cellulose, like lignin (1.13), lipids (mean OR 1.137) and proteins in general. Those have higher concentrations in leaves and seeds than in wood [9], and their amounts change within a growing season.

For some components we found a seasonal variation in OR and δ¹³C, but no effect of height on the oxidative ratio within the leaves or twigs. The differences in δ¹³C can largely be explained by differences in height, microclimate and season, both in leaves and wood, as explained by Göttlicher et al. [23] and Knohl et al. [22]. Göttlicher et al. [23] found that in a profile ranging from 5 to 20 m above the forest floor, δ¹³C increased with height in leaves, Knohl et al. [22] sampled understorey plants, forest canopy leaves (at 2, 10, 20 and 30 m) and litter, finding the same for leaves, but also differences in δ¹³C between August, September and October for the same components. Furthermore, isotopic ratios are most likely affected by the summer drought too, as higher δ¹³C values can indicate drought stress in plant leaves [28,29,30].

The low precipitation, higher average air temperature and critical REW in the summers of 2018 and 2019 show that both of our sampling years were exceptionally dry years, thus possibly resulting in atypical or stress plant metabolism. It is unknown how this might have affected the OR values, but they could differ from years with an average climate, especially for ecosystem components with relatively fast turnover like leaves, fine roots or small branches that are predominantly affected by the weather during a single vegetation period. Studying the effects of drought stress and recovery in trees, drought resistance and resilience is a growing field of research. For European beech, García-Plazaola and Becerril [31] found several changes in the photoprotective system, with generally decreasing pigment content and increasing antheraxanthin, zeaxanthin and tocopherol, Gallé and Feller [32] found that drought stomatal conductance and net photosynthesis decreased because of water stress. However, after four weeks under better conditions, net photosynthesis recovered again. Gebauer et al. [33] more recently also found a high decrease in stomatal conductance (80–85%) of beech seedlings upon drought stress, leading to a decline in net photosynthesis. Biomass of stems and roots was affected too.

Leaf chemistry and OR are not only affected by weather or climate events, but also by the sampling month, as there are chemical differences between leaves sampled in May, August and October just because of natural growth and senescence. Differences in understorey vegetation ORs between seasons can be explained by different herbs and shrubs growing in different seasons. However, as the species composition of the understorey was not quantified by weight, we cannot determine potential effects of a shifting composition on the understorey oxidative ratio. For leaf litter, we found the lowest OR in spring, supposedly reflecting advanced decomposition over the course of several months since leaf fall. In comparison, litter ORs in August and October represent a mixture of the OR of decomposing leaves from the previous year, and the fresh litter of the current year, resulting in higher OR values. Gunendehou et al. [34] studied decomposition and chemical composition of leaf litter in a West African forest. Although this was a tropical forest, their findings can be useful for interpretation. Concentrations of acid-hydolyzable, water-soluble and ethanol-soluble components decreased with progressing decomposition, but at different rates. Lignin also decreased, but in relation to the other three groups its concentration became higher, leaving behind a more lignin-rich litter. Since lignin has an OR of 1.13 [9,16], this cannot explain decreasing OR. Torres-Ruiz and Wehr [35] state that lipids do not decrease in leaf litter of all species. Several studies, e.g., d’Annunzio et al. [36] have studied decomposition in European beech litter in general, but focus lies on mass, carbon and nitrogen in relation to soil and decomposers. This means that explaining decreasing litter OR by only one or few factors is difficult.

Previous studies [18,19] also showed higher leaf OR than wood OR for several deciduous tree species, however leaf litter is not always the lowest value [10,18,19]. This may be caused by the different ways of litter collection and thus different stages of decay. If for example litter is collected in litter traps and dried very soon after the collection, leaf litter OR will be closer to that of fresh leaves than to that of litter that has been deposited on the ground several months ago.

Table 5. Oxidative ratios (mean ± SE) by component and sampling date and biomass share.

Sample Type	Plant BM Share [%]	Oct 2017	May 2018	Aug 2018	Oct 2018	May 2019	Aug 2019	Oct 2019	Sep 2020
		OR ± SE	OR ± SE	OR ± SE	OR ± SE	OR ± SE	OR ± SE	OR ± SE	OR ± SE
Leaves	0.6	1.069 ± 0.007	1.104 ± 0.005	1.077 ± 0.005	1.076 ± 0.003	1.098 ± 0.002	1.091 ± 0.003	1.087 ± 0.003	1.099 ± 0.002
Twigs	14.7 1	1.082 ± 0.004	1.064 ± 0.007	1.081 ± 0.005	1.083 ± 0.004	1.085 ± 0.003	1.094 ± 0.003	1.078 ± 0.004	-
Stem wood *	68.9	-	-	-	-	1.065 ± 0.004	-	-	1.071 ± 0.001
Bark	N/A	1.060 ± 0.007	1.114 ± 0.021	1.032 ± 0.010	1.115 ± 0.018	1.108 ± 0.001	1.076 ± 0.024	1.081 ± 0.022	-
Fresh leaf litter	N/A	1.013	-	-	-	-	-	1.068 ± 0.004	-
Litter	1.2	1.046 ± 0.003	1.029 ± 0.019	1.039 ± 0.001	1.036 ± 0.009	1.026 ± 0.025	1.050 ± 0.003	1.058 ± 0.010	-
(org. layer)
Understorey veg.	N/A	1.058 ± 0.008	1.088 ± 0.007	1.080 ± 0.009	1.089 ± 0.002	1.078 ± 0.008	1.069 ± 0.002	1.042 ± 0.001	-
AGB	85.4	1.070 ± 0.002	1.069 ± 0.002	1.069 ± 0.002	1.069 ± 0.002	1.070 ± 0.002	1.070 ± 0.002	1.070 ± 0.002	-
Fine roots	0.5	-	0.983	-	0.981 ± 0.024	1.016 ± 0.009	1.037 ± 0.021	0.981 ± 0.016	-
Coarse roots	14.1	-	1.058 ± 0.005	-	-	1.047 ± 0.004	1.059 ± 0.008	1.044 ± 0.006	-
BGB	14.6	-	1.055 ± 0.005	-	-	1.046 ± 0.004	1.058 ± 0.008	1.042 ± 0.006	-

¹ The biomass data are for branches and twigs, while the collected samples are twigs. The biomass share of twigs is not known. * An extensive stem wood sampling (5 tree cores of up to 29 cm length) was only done in 2020, and not from the same trees. As the stem wood represents an accumulation across many years, the same OR is used for the calculation of all seasons (mean value of all 36 stem wood samples). AGB = above-ground biomass, BGB = below-ground biomass, BGB = below-ground biomass. Plant BM share = Plant biomass share.

and Figure 5 show that leaf ORs in August 2018 were lower than in summer 2019 (p = 0.024), similar to autumn OR (August 2018–October 2018: p = 0.998; August 2018–October 2017: p = 0.757; August 2018–October 2019: p = 0.711). As the oxidative ratio is calculated from the concentrations of N, C, O and H, it is important to know how these elements vary with the seasons. While leaf N follows an expected pattern, with increased N in younger leaves and reduction due to leaf senescence in autumn [13,37], C, O and H differed between 2018 and 2019. As proteins that are removed from leaves during leaf senescence do not only have a higher N concentration than for example cellulose (no N), but also a higher H concentration [13,38,39], their variations can heavily influence OR via H differences. Due to how OR is calculated (see Equations (1) and (2)), H and O have the highest impact, and a lower concentration of H with higher concentrations of C and O at the same time lead to lower ORs. RuBisCO is the predominant protein in the leaves of C₃ plants, making up to 30% of the total leaf protein and up to 50% of soluble leaf proteins, and on average about 3% of the mass of a leaf is RuBisCO [13,40,41,42,43]. Leaf RuBisCO has an OR of 1.24, while lignin has 1.13 and cellulose 1.00. Knowing which exact molecules were present in the samples at the specific dates could help with explaining the OR differences, but it is likely that RuBisCO concentrations are lower in autumn due to senescence.

Feller et al. [40] and Ono et al. [41] both state that amino acids from RuBisCO that are degraded during leaf senescence are transported to sinks and recycled, and that this also can happen as a response to stress. Our results for leaves show that in August 2018 a state which in 2019 was reached two months later in the year only, did already exist, likely caused by the drought. In addition to that, there is a general trend of rising OR from October 2017 to October 2019 in leaf OR, meaning that there would be an effect of the drought on leaf level, but not ecosystem level.

Lastly, we want to mention that soil also has an important role, as it contains a large part of the organic carbon (in Leinefelde 31% [25]) and is actively taking part in the carbon cycle, but we were unable to determine reliable ORs for the soil. Because of a low organic matter content and high mineral content, the same methods used for plant material lead to O and H concentrations dominated by O and H from soil minerals. Thus the whole ecosystem OR is still unknown. The soil at the site is classified as Parabraunerde [25], which is an Alfisol in the USDA Soil Taxonomy [44], and has a mean oxidative ratio of 1.11, but within a range of 0.77 to 1.37 according to Clay and Worrall [11]. Because of this high range, those data cannot be used for the oxidative ratio of the soil of the forest from this study. A different method used to determine OR of soils, which was used by Hockaday et al. [8] is carbon ¹³NMR spectroscopy and it has to be considered that this method could be able to result in more reliable soil OR values than elemental analysis via combustion.

4.3. Implications for Future Research

Our research has shown that there are variations and changes of OR on several levels (between components, on ecosystem level and between seasons), but the causes of some of those could not be determined. Several questions remain, and could only be answered by different sampling designs or alternative methods. A first question is, how fast do changes in the oxidative ratio occur in response to certain events? We chose the temporal resolution of May, once leaves were fully developed, August, and October before leaf fall. But is water stress or any other environmental factor already visible in leaf OR within weeks or one or two months? How long does an OR of components that change faster stay at the same value if the conditions become more favorable again? This would require much more frequent sampling, but also supervision of the leaves to determine when the leaves that are analyzed started to grow.

A further question is, by which exact molecules are the oxidative ratios controlled? To answer this, the molecular composition of a certain component (e.g., lignin, cellulose and lipid concentration, etc. of a leaf of a certain species) has to be tracked over time. This would require molecular research, but could allow predictions in OR changes, like “a drought with certain changes in water availability will change leaf OR in a certain direction”. Cellulose [45,46] and lignin [47] can be extracted from leaves, however this includes extensive chemical procedures like acid hydrolysis, chlorination, alkaline extraction, and bleaching.

Furthermore, for many other unanswered questions about the oxidative ratio, the methods still have to be improved to account for the high natural variability and influences of the design of the methods, which depend on the available equipment at the research location, so far differing between the facilities which have conducted research on oxidative ratios of organic material. This includes how exactly samples are collected, dried (e.g., drying temperatures), milled, homogenized and measured. Different elemental analyzers use different combustion temperatures or different sample weights.

5. Conclusions

Ecosystem OR is dominated by stem wood OR and thus is stable within a period of several years represented in the stem cores. There are differences between seasons for other plant components, which are influenced by their typical growth cycle and climate. However, the seasonal differences are only significantly present in the leaves, leaf litter and leafy understorey vegetation, which have a very small biomass share in comparison to the OR-static stem wood. Overall, looking at the whole ecosystem OR, we see little variation between seasons or even after severe climate impacts (i.e., natural variation in OR within or between plant components is higher than intra- or inter-annual variation). This means that even in extremely dry years, the oxidative ratio remains stable within the above-ground biomass of the managed beech forest.