Temporary reductions of stem CO2 efflux during rainfall events across tree species

Temporary reductions of stem CO2 efflux during rainfall events across tree species 1 2 Eva Darenova 3 Global Change Research Institute CAS, v.v.i., Belidla 4a, 603 00 Brno, Czech Republic 4 darenova[email protected] 5 6 7 8 Abstract 9 Stem respiration is an important component of forest carbon cycling, yet its accurate quantification 10 remains challenging. While stem respiration is commonly measured as CO2 efflux from the stem 11 surface (EA), several processes can decouple measured fluxes from actual respiratory activity. In 12 this study, we investigated the influence of rainfall events on EA across temperate tree species and 13 assessed the potential consequences for seasonal carbon budget estimates. Continuous automated 14 chamber measurements revealed distinct short-term decreases in EA coinciding with rain events. 15 Time-lapse camera records confirmed that stem surfaces were frequently covered by a water film 16 during rainfall, suggesting that wet bark temporarily restricted CO2 diffusion to the atmosphere. 17 Rain-related decreases in EA accounted for up to 21% of the available seasonal data (May–October) 18 and caused up to 12% underestimation of seasonal sums of measured EA. The largest impact was 19 observed in beech, followed by hornbeam and spruce, whereas oak and ash showed the smallest 20 underestimation. Our results demonstrate that rainfall can systematically bias EA measurements 21 and lead to underestimation of seasonal and annual carbon fluxes from tree stems. Recognizing 22 and accounting for these temporary but repeated events is therefore essential for improving the 23 accuracy of forest carbon budget assessments under both current and future precipitation regimes. 24 25 Keywords 26 Bark wetting, carbon flux, diffusion resistance, forest, rainfall, stem respiration 27 28 29 1 Introduction 30 Forest respiration is a key component of the global carbon cycle, as it represents a major flux of 31 carbon dioxide (CO2) from ecosystems to the atmosphere. Although stem respiration generally 32 accounts for only about 5–25% of total forest ecosystem respiration (Acosta et al., 2008; Khomik 33 et al., 2010; Kim et al., 2021; Rodríguez-Calcerrada et al., 2014; Song et al., 2023; Yang et al., 34 2016; Zha et al., 2007), it plays a non-negligible role. 35 Stem respiration is most commonly measured as CO2 efflux (EA) from the stem surface using 36 chamber systems. This method provides valuable insights into both short-term dynamics and long37 term seasonal trends. However, EA does not necessarily equal the actual respiration of stem tissues 38 at the measurement point. Several processes can alter the relationship, including transport of 39 dissolved CO2 by the xylem sap (Teskey et al. 2008; Tarvainen et al. 2018), partial CO2 refixation 40 by photosynthetically active cells in the bark (Dukat et al., 2024), and diffusional limitations within 41 woody tissues (Steppe et al., 2007). Nevertheless, EA remains the most widely applied and 42 technically feasible indicator of stem respiration. 43 Temporal EA dynamics is primarily driven by temperature (Darenova et al., 2018; Kim et al., 2021). 44 During the period of new wood formation, EA is enhanced due to extra energy demands (Darenova 45 et al., 2020; Rodríguez-calcerrada et al., 2019). From our experience and previous studies, the 46 diurnal courses of EA follows the temperature fluctuations. We, however, observed unexpected 47 temporal decreases in our EA datasets from automated continuous measurement systems. These 48 data were not flagged by the data quality indicator, therefore, they were not assessed as error 49 measurements. We did not find any reason stem respiration to drop suddenly. The decreases, 50 however, coincided with rain events. Time-laps camera observations confirmed that the stem 51 surface was covered by a water film (Fig. S1). We assumed that such surface wetting temporarily 52 restricted EA by limiting CO2 diffusion to the atmosphere. This phenomenon has so far received 53 little attention. Steppe et al. (2007) and Salomón et al. (2016) investigated the role of woody tissue 54 and bark water status in controlling CO2 diffusion resistance. They found that resistance to radial 55 CO2 diffusion was higher at night, when woody tissue and bark water reservoirs were refilled in 56 the absence of transpiration, than during the day, when transpiration lowered water content. This 57 pattern reflects the much slower diffusion of CO2 in water compared with air (Nobel, 2009). 58 Rain-related reductions in measured EA may systematically bias seasonal and annual forest carbon 59 budget estimates. Such discrepancies are likely influenced by interannual and site-specific variation 60 in precipitation regimes and may also differ among tree species. Therefore, the main objective of 61 this study was to quantify the effect of rainfall events on stem CO₂ efflux measurements in different 62 tree species. Specifically, we aimed i) to identify and quantify short-term decreases in EA during 63 and immediately after rain events, ii) to estimate the contribution of these decreases to total 64 seasonal stem CO₂ efflux, and iii) to compare the magnitude of the effect among tree species with 65 contrasting bark characteristics and crown architecture. The study included Norway spruce as a 66 representative conifer, along with broadleaved species—beech and hornbeam with smooth bark, 67 and oak and ash with rough bark. Our findings provide new insights into the processes that alter 68 measurements of stem respiration and support more accurate estimates of forest carbon fluxes. 69 70 2 Materials and Methods 71 72 2.1 Sites 73 Measurements were carried out at three sites: a beech forest (European beech, Fagus sylvatica L.), 74 a spruce forest (Norway spruce, Picea abies L.), and a broadleaved mixed forest dominated by 75 common hornbeam (Carpinus betulus L.), English oak (Quercus robur L.), and narrow-leaved ash 76 (Fraxinus angustifolia Vahl). Site and stand characteristics are summarised in Table S1. 77 78 2.2 CO2 efflux measurements 79 We analysed data from six chambers installed on beech and spruce trees, and four chambers on 80 hornbeam, oak, and ash trees. The chambers were a part of the 12-chmaber automated measurement 81 systems. The control unit consisted of an infrared gas analyser (Li-840, LI-COR, Lincoln, NE, 82 USA), and a personal computer equipped with control software, as well as additional analogue 83 input and digital output hardware. The chamber closing was enabled by pressure air supplied by a 84 compressor and regulated through a system of valves, and pneumatic pistons fixed on the chambers. 85 Each chamber (Fig. S2) comprised a frame and a chamber head. The frames were rectangular with 86 inner dimensions 7x12 cm on spruce and 9x15 cm on the other tree species, and they had a neoprene 87 seals against the tree trunk. A groove a few millimetres wide, filled with rubber, secured the seal 88 between the frame and chamber head. The chamber heads were made of stainless steel and had a 89 half-cylinder shape. Chambers were attached to the tree trunk at breast height on the northern side 90 using two belts. 91 For beech, hornbeam, and spruce (all with relatively smooth bark), the neoprene seal was sufficient. 92 In contrast, oak and ash had thick, rough outer bark that needed careful removal with a chisel to 93 avoid damaging living tissues. Bark was removed only along the frame circumference where the 94 neoprene seal was applied; the area inside the frame remained untouched. 95 Stem CO2 efflux (EA) was measured sequentially in all chambers, with data collected every 2 h at 96 each position. After chamber closure, air was circulated between the chamber and the analyzer, 97 followed by a 60-second equilibration period. The system then recorded 20 CO2 concentration 98 values at 10-second intervals for efflux calculation, using the following equation: 99 𝐸𝐴=𝑃∙𝑉∙(𝑐2−𝑐1) 𝑅∙𝑇∙𝑆∙𝑡 , 100 where P is air pressure (Pa), V is the volume of the system (m3), c1 and c2 are consecutive CO2 101 concentrations (mol mol–1), R is the molar gas constant, T is sample air temperature (K), S is the 102 stem surface area enclosed by the frame (m2) and t is measurement interval (10 s). 103 For each 200 s measurement, mean EA and standard deviation (SD) were calculated. SD served as 104 an indicator of data quality; if SD exceeded the efflux value, the data were excluded. Across 2020– 105 2024, 64–96% of seasonal data were available for analyses (Table 1) 106 107 2.3 Meteorological measurements 108 Stem temperature sensors (PT 100, Sensit, Rožnov pod Radhoštěm, Czech Republic) were 109 integrated into the stem CO2 efflux systems and measured temperature at the stem surface beneath 110 the chambers. Precipitation was continuously monitored with a rain gauge (MetOne 386, Met One 111 Instruments, Inc., Grants Pass, OR, USA) installed on towers above the forest stands. 112 113 2.4 Data processing 114 The effect of rain on stem CO₂ efflux was determined using the following approach (see also the 115 schematic in Fig. S3): 116 1. Detection of rain-related decreases in EA 117 We identified periods when measured stem EA showed unexpected declines that coincided 118 with rainfall events. These declines were recognized by discrepancies between the observed 119 EA and the values expected from the temperature–efflux relationship. 120 2. Modelling the expected EA 121 To obtain the reference efflux (Em), we fitted an exponential temperature–efflux model 122 using a 10-day moving window. For model fitting, we excluded data affected by the rain123 related decreases. This provided a smooth, temperature-driven prediction of EA under 124 unaffected conditions. 125 3. Quantifying the effect of decreases 126 For each rain-related decrease, we calculated the ratio of the measured EA to the modelled 127 Em at the same time. This ratio represents the relative reduction of EA due to the presence 128 of surface water films during rain events. 129 4. Seasonal impact assessment 130 We replaced the periods of rain-related decreases with the corresponding modelled values 131 (Em). 132 Using these corrected time series, we calculated seasonal sums of CO₂ efflux both with and 133 without the rain-related EA decreases. The contribution of the decreases to seasonal totals 134 was determined by comparing these sums. Calculations of sums were based only on 135 available data, without gap-filling for missing values 136 137 To determine the EA-temperature relationship, Q10 was calculated as: 138 𝑄10 = 𝑒10∙𝛼, 139 where α is a function growth parameter derived from the exponential relationship between EA and 140 stem temperature. 141 A one-way repeated measures ANOVA was used to test differences between tree species in the 142 proportional impact of rain-related EA decreases on seasonal EA sums. Statistical analyses were 143 performed using SigmaPlot version 15.0 (Systat Software Inc., San Jose, CA, USA). 144 145 146 147 3 Results 148 149 3.1 Micrometeorology 150 Precipitation during the experimental seasons (May – October) was generally highest in spruce 151 forest and lowest in the broad-leaved mixed forest. Seasonal totals ranged from 345 to 624 mm in 152 the beech forest, from 602 to 1066 mm in the spruce forest, and from 304 to 517 mm in the broad153 leave mixed forest. A summary of precipitation and the number of rainy days for each season is 154 provided in Table 1. 155 156 3.2 Rain-related decreases in EA 157 Rain-related decreases in EA accounted for up to 21% of the available seasonal data (May–October) 158 (Table 1). The largest proportion was observed in beech (13 ± 5%), followed by hornbeam (9 ± 159 2%), spruce (8 ± 3%), ash (3 ± 1%), and oak (2 ± 2%). The biggest decreases occurred in beech, 160 where most reductions were concentrated between 40–60% (Fig. 1). In hornbeam, decreases 161 peaked at 20–30%, while in spruce they were more evenly distributed, ranging from 10 to 60%. 162 The smallest reductions were observed in oak and ash, generally between 10 and 30%. 163 The rain-related decreases in EA caused up to 12% underestimation of seasonal sums of measured 164 EA (Fig. 2). The biggest impact was observed in beech with mean underestimation of 5.3 ± 2.8%, 165 followed by hornbeam (3.3 ± 1.6%) and spruce (2.2 ± 2.4%). The smallest underestimation was in 166 oak (0.3 ± 0.5%) and ash (1.1 ± 1.4%). 167 168 3.3 After-rainfall period 169 To test whether the temporary inhibition of CO2 release from stems during rain events leads to an 170 increase in EA after the rainfall ends and the bark dries, we analyzed the difference between 171 measured and modeled EA in six hours following rain (more precisely three measurements after 172 measured EA returned modeled values). Within six hours after rainfall (corresponding to three EA 173 measurements), EA tended to increase in beech, spruce, and hornbeam, most frequently by 0–10% 174 (Fig. 1). In contrast, no clear post-rain increases were observed in oak or ash. 175 176 3.4 Effect of rain-related EA decreases on Q10 177 To illustrate the impact of rain-related EA decreases on EA interpretation, we selected one chamber 178 on beech (the species with the strongest rainfall effect on EA) in 2024. We compared Q10 values 179 calculated from the whole measured dataset and from a dataset in which the rain-related decreases 180 were removed. Using a 10-day moving window, Q10 in the corrected dataset fluctuated consistently 181 between 1.5 and 2.0 throughout the season (Fig. 3-A). In contrast, when based on all measured 182 data, Q10 exhibited several large deviations, reaching values as high as 15.4. At a monthly time 183 step, the discrepancy between the two datasets was reduced but remained substantial (Fig. 3-B). 184 185 4 Discussion 186 In our study, we observed a temporal decrease in EA that was so steep it could not be explained by 187 the typical temperature drop that often accompanies rainy weather. Moreover, there is no 188 physiological reason for such a pronounced decline in living cell respiration, which is the main 189 source of CO2 production. On the contrary, (Salomón et al., 2016) observed an increase in stem 190 CO2 concentrations after rain. The explanation must therefore be sought in factors affecting CO2 191 efflux from the stem to the atmosphere. 192 A thin water film covering tree stems during and shortly after rain events, or bark saturated with 193 water, can represent a temporary barrier for CO2 diffusion from the stem to the atmosphere. This 194 makes it more difficult to estimate the actual amount of CO2 respired by living stem cells and 195 released to the atmosphere. The effect of bark water content on radial CO2 diffusion resistance has 196 been confirmed by (Salomón et al., 2019). Generally, the rate of CO2 diffusion is approximately 197 10⁴ times slower in water than in air (Nobel, 2009). This fundamental difference in diffusion rates 198 highlights why even a thin water film may substantially alter CO2 efflux dynamics. 199 Overall, EA was most affected by rain in beech, followed by hornbeam and spruce. The smallest 200 effect was observed in ash and oak. Tree species and stand structure strongly influence both the 201 amount of precipitation reaching the stem and the intensity of stemflow. In general, stemflow is 202 higher in broadleaves than in conifers due to canopy morphology (Barbier et al., 2009; Kantor and 203 Šach, 2009). This likely explains the smaller rain effect on EA in spruce compared with beech and 204 hornbeam, despite higher seasonal precipitation at the spruce site. Differences among deciduous 205 species can be explained by bark roughness (Novosadová et al., 2023). The smooth bark of beech 206 and hornbeam has a smaller soaking area and faster water flow down the trunk, whereas the rough 207 bark of oak and ash increases soaking capacity and slows water runoff. These bark-related 208 differences in water retention may therefore control both the intensity and duration of rain-related 209 EA reductions. A further methodological influence cannot be excluded: in oak and ash, a certain 210 layer of outer bark was removed to achieve a tight seal of the chamber, which may have introduced 211 an artificial barrier to stemflow around the frame. 212 In beech, spruce, and hornbeam, the strongest rain effects were observed in 2020, both in term of 213 the proportion of data classified as rain-related EA decreases and the magnitude of seasonal EA 214 underestimation (Table 1). Notably, 2020 experienced the highest seasonal precipitation sum, 215 accompanied by a larger number of rainy days compared with other years. This result highlights 216 the role of precipitation regime in affecting stem CO2 fluxes. 217 The data from rain-related decrease periods are of good technical quality, but they do not represent 218 the true temporal pattern of CO2 production in stems. This raises the question of whether such data 219 should be included in datasets used for estimating tree CO2 budgets. We may expect that CO2 not 220 released during wet periods could later be emitted in a different location or at a different time. 221 Higher stem CO2 concentrations relative to the atmosphere due to radial diffusion barriers are well 222 known (Steppe et al., 2007), and temporary strengthening of these barriers during rain can further 223 be released, potentially causing a transient acceleration of EA. Indeed, we observed increased EA 224 within six hours after rain-suppressed EA periods. This post-rain increases typically ranged from 0 225 to 20%, though it was not consistently observed. While the EA drop during rain was rapid, recovery 226 to expected values was often more gradual, probably depending on the speed of surface drying. 227 Moreover, the stem surface inside the chamber frame did not dry uniformly, which complicated 228 the determination of the precise end of rain-related wet conditions. However, the post-rain 229 increases were generally smaller and shorter than the decreases and did not appear to compensate 230 for the presumed CO2 accumulation within stems. Thus, the ultimate fate of the CO2 respired during 231 rain events remains uncertain and may involve its dissolution in xylem sap and subsequent upward 232 transport. CO2 can be then released to the atmosphere or re-assimilated by the photosynthetically 233 active cells later in upper stem parts, branches or leaves (Salomón et al., 2021; Teskey et al., 2008). 234 This would shift the spatial and temporal pattern of CO2 efflux beyond what stem chambers detect. 235 Although the rain-related decreases accounted for only a few percent of seasonal EA, including 236 them in datasets can strongly bias models based on the EA–temperature relationship. During rain, 237 air temperature typically drops. If such periods are included, the combination of lower EA and lower 238 temperature produces an artificially steep regression slope between EA and temperature. This can 239 substantially overestimate the temperature sensitivity parameter Q10. Using a 10-day moving 240 window for Q10 calculation, inclusion of rain-related decreases can yield values several times 241 higher than the true sensitivity (Fig. 3). Longer time windows reduced this overestimation but did 242 not eliminate it. However, longer periods are also influenced by additional factors, especially 243 differences in stem growth status (Darenova et al., 2018). Therefore, careful data filtering is 244 essential before applying EA–temperature models to avoid introducing systematic errors into 245 estimates of forest carbon dynamics. 246 247 5 Conclusions 248 Reduced CO2 efflux caused by wet bark generally led to only a slight underestimation – typically 249 a few percent – of seasonal CO2 emissions, which can be considered negligible. However, our 250 results demonstrate that in regions with high precipitation and in beech forests, or in stands 251 dominated by tree species with fine bark, this underestimation can become substantial. 252 Furthermore, including rain-related stem CO2 efflux data without correction can significantly bias 253 models that rely on the relationship between CO2 efflux and temperature. 254